A multi-dimensional dynamic monitoring method and system for petroleum resin production

Abstract

The present application relates to the field of data processing, and particularly relates to a multi-dimensional dynamic monitoring method and system for petroleum resin production.The method comprises the following steps: obtaining a fault event of a petroleum resin production line; obtaining device elements, obtaining a co-occurrence degree of target co-occurrence device elements; determining a global order sequence and a global score order sequence of the device elements; obtaining a transaction, and obtaining continuous prefixes and discontinuous prefixes in the transaction; determining a score of each device element in the transaction of the fault event, calculating a first final score of the continuous prefixes and a second final score of the discontinuous prefixes; calculating a priority score of the transaction of each fault event; determining a high-priority branch and a low-priority branch; constructing an optimized FP tree according to the high-priority branch and the low-priority branch, and determining a to-be-detected device element from the optimized FP tree; and detecting the to-be-detected device element.The present application can improve the fault detection efficiency of the petroleum resin production line.

Description

Technical Field

[0001] This invention relates to the field of data processing, and specifically to a multi-dimensional dynamic monitoring method and system for petroleum resin production. Background Technology

[0002] The production process of petroleum resins mainly includes four key steps: raw material pretreatment, polymerization reaction, hydrogenation refining, and product post-treatment. Each step involves various treatments of the materials and utilizes a variety of reaction vessels and catalysts.

[0003] Current methods, when detecting anomalies in petroleum resin production, such as reactor overheating, typically employ a blanket emergency shutdown as a protective measure, followed by a reactive troubleshooting process. This process requires maintenance personnel to check multiple related components, including the cooling system, mixing equipment, and feeding unit, relying on manual experience for on-site inspection and data verification. This sequential "shutdown-troubleshooting-location-response" operation mode results in low efficiency in fault location, significantly extends equipment recovery time, and causes substantial unplanned production losses. Summary of the Invention

[0004] This invention provides a multi-dimensional dynamic monitoring method and system for petroleum resin production to solve existing problems.

[0005] The present invention provides a multi-dimensional dynamic monitoring method for petroleum resin production, which adopts the following technical solution: One embodiment of the present invention provides a multi-dimensional dynamic monitoring method for petroleum resin production, the method comprising the following steps: Acquire fault events of the petroleum resin production line, wherein each fault event includes the faulty equipment and the corresponding abnormal data of the faulty equipment; The faulty device and its corresponding abnormal data are combined in the fault event to obtain the device element of each faulty device. The target device element and the associated device element are determined as the target co-occurrence device element. The co-occurrence degree of the target co-occurrence device element of each device element is obtained. The target device element is any device element in the device elements, and the associated device element is any device element in the device elements except the target device element. Based on the co-occurrence of the target co-occurring device elements, determine the global sorting sequence of the device elements, and obtain the score of each device element in the global sorting sequence to obtain the global score sorting sequence; Retrieve the transactions for each failure event, and obtain the consecutive and non-consecutive prefixes within the transactions for each failure event; Based on the global score sorting sequence, determine the score of each device element in the transaction for each fault event, and calculate the first final score for each consecutive prefix and the second final score for each non-consecutive prefix based on the score of each device element in the transaction for each fault event. Calculate the transaction priority score for each failure event based on the first final score for each consecutive prefix and the second final score for each non-consecutive prefix; Transactions with failure events whose priority scores are greater than a preset priority threshold are identified as high-priority branches, and transactions with failure events whose priority scores are less than or equal to the preset priority threshold are identified as low-priority branches. Based on the high-priority and low-priority branches, an optimized FP tree is constructed, and the device elements to be detected are determined from the optimized FP tree and then detected.

[0006] Optionally, the co-occurrence degree of each target co-occurring device element for each device element is obtained, specifically including: Obtain the target fault event and co-occurring fault events from the fault events, wherein the target fault event includes at least one of the target device element and the associated device element, and the co-occurring fault events include the target device element and the associated device element; The ratio of the number of co-occurring fault events to the number of target fault events is determined as the co-occurrence degree of the target co-occurring device element. In the device elements, each device element except the target device element is identified as an associated device element, and the co-occurrence degree of the target co-occurring device elements for each target co-occurring device element is obtained; Each device element is identified as a target device element, and the co-occurrence degree of each target co-occurring device element for each device element is obtained.

[0007] Optionally, a global sorting sequence of device elements is determined based on the co-occurrence degree of the target co-occurring device elements, specifically including: Obtain the co-occurrence degree of each target co-occurring device element of the a-th device element and sum them to obtain the co-occurrence degree sum of the a-th device element; Get the co-occurrence sum for each device element; The device element with the highest co-occurrence is determined as the first element, and the first element is removed from the device elements to obtain the device elements after the first update; Get the co-occurrence of the target co-occurrence device element between the b-th device element and the first element in the device elements after the first update, and get the co-occurrence of the target co-occurrence device element between each device element (excluding the b-th device element) and the b-th device element in the device elements after the first update and sum them to get the co-occurrence sum of the b-th device element. The product of the co-occurrence degree of the b-th device element and the target co-occurrence device element of the first element, and the sum of the co-occurrence degrees of the b-th device element, is determined as the ranking value of the b-th device element. Get the sort value of each device element in the device elements after the first update; The device element with the largest sort value is designated as the second element; Remove the second element from the device element after the first update to obtain the device element after the second update; In the device elements after the second update, obtain the co-occurrence degree of the target co-occurrence device element of the c-th device element and the second element, and obtain the co-occurrence degree of each device element except the c-th device element with the target co-occurrence device element of the c-th device element and sum them to obtain the co-occurrence degree sum of the c-th device element. The product of the co-occurrence degree of the target co-occurrence device element of the c-th device element and the co-occurrence degree of the c-th device element is determined as the ranking value of the c-th device element. Get the sort value of each device element in the device elements after the second update; The device element with the largest sort value is designated as the third element; Get the Nth element, and determine the global sorting sequence of the device elements from the first element to the Nth element.

[0008] Optionally, the score of each device element in the global sort sequence is obtained to obtain the global score sort sequence, specifically including: The score of the last device element in the global sort sequence is set to 1, and the score of the second-to-last device element in the global sort sequence is set to 2. The score of the Mth device element from the end of the global sorting sequence is determined as M, thus obtaining the global score sorting sequence, where M is the number of elements in the global sorting sequence.

[0009] Optionally, the transaction for each failure event is obtained, and the consecutive and non-consecutive prefixes in the transaction for each failure event are obtained, specifically including: Retrieve the device elements for each fault event; Obtain the sequence number of the e-th device element in the global sort sequence in the d-th fault event, and get the transaction sequence number of the e-th device element in the d-th fault event; Obtain the transaction sequence number of each device element in the d-th fault event; Sort the transaction sequence number of each device element in the d-th fault event in ascending order to obtain the transaction of the d-th fault event; Acquire transactions for each failure event; The transactions of each failure event are divided according to the global sorting sequence to obtain continuous prefixes and non-contiguous prefixes in each failure event's transactions. The continuous prefix is ​​a subset of continuous device elements in the global sorting sequence, and the number of device elements in the continuous prefix is ​​greater than 1. The non-contiguous prefix is ​​a subset of device elements in the global sorting sequence, and the number of device elements in the non-contiguous prefix is ​​equal to 1. When at least two non-contiguous prefixes are adjacent in the transactions of the failure event, the at least two non-contiguous prefixes are merged into one non-contiguous prefix.

[0010] Optionally, based on the global score sorting sequence, the score of each device element in the transaction for each failure event is determined. Based on the score of each device element in the transaction for each failure event, a first final score for each consecutive prefix and a second final score for each non-consecutive prefix are calculated, specifically including: The score of the f-th device element in the global score sorting sequence is determined as the score of the h-th device element in the transaction of the g-th failure event, wherein the f-th device element in the global score sorting sequence and the h-th device element in the transaction of the g-th failure event are the same device element. Get the score of each device element in the transaction for the g-th failure event; Get the score of each device element in the transaction for each failure event; Based on the score of each device element in the k-th consecutive prefix in the transaction of the j-th failure event, calculate the first final score of the k-th consecutive prefix in the transaction of the j-th failure event; Based on the score of each device element in the l-th discontinuous prefix in the transaction of the j-th failure event, calculate the second final score of the l-th discontinuous prefix in the transaction of the j-th failure event.

[0011] Optionally, the transaction priority score for each failure event is calculated based on the first final score for each consecutive prefix and the second final score for each non-consecutive prefix, specifically including: Based on continuous and non-contiguous prefixes, the transactions of the failure event are divided into transaction groups, where any transaction group in the transaction group has either a continuous prefix or a non-contiguous prefix; The frequency of each transaction group within the transactions that acquire failure events; The product of the frequency and final score of the m-th transaction group in the transaction of the j-th failure event is determined as the third final score of the m-th transaction group in the transaction of the j-th failure event. Specifically, when the m-th transaction group in the transaction of the j-th failure event is the k-th consecutive prefix in the transaction of the j-th failure event, the first final score of the k-th consecutive prefix in the transaction of the j-th failure event is determined as the final score of the m-th transaction group in the transaction of the j-th failure event; when the m-th transaction group in the transaction of the j-th failure event is the l-th non-consecutive prefix in the transaction of the j-th failure event, the second final score of the l-th non-consecutive prefix in the transaction of the j-th failure event is determined as the final score of the m-th transaction group in the transaction of the j-th failure event. Obtain the third final score of each transaction group in the transaction of the j-th failure event and sum them to get the priority score of the transaction of the j-th failure event; Get the priority score of the transaction for each failure event.

[0012] Optionally, an optimized FP-tree is constructed based on the high-priority and low-priority branches, and the device element to be detected is determined from the optimized FP-tree. The detection of the device element to be detected includes: Each high-priority branch and each low-priority branch are identified as the original data, and each device element in each failure event transaction is identified as a node. An FP tree is constructed to obtain the optimized FP tree. In the construction process, each high-priority branch participates in the construction of FP with the root node of the FP tree as the root node of the high-priority branch. Obtain the faulty device in each fault event to obtain the faulty device sequence; The target faulty device and the associated faulty device are combined into a faulty device pair. The frequency of the faulty device pair in the faulty device sequence is obtained to obtain the co-occurrence of the target faulty device and the associated faulty device. The target faulty device is any faulty device in the faulty device, and the associated faulty device is any faulty device in the faulty device except the target faulty device. The associated faulty devices are sorted in descending order of co-occurrence to obtain the co-occurrence sequence of the target faulty device; Each faulty device is identified as a target faulty device, and the co-occurrence sequence of each faulty device is obtained. Obtain the target abnormal device, obtain the associated faulty devices of the target abnormal device from the co-occurrence device sequence of the target abnormal device, determine the device element to be detected based on the target device and associated faulty devices, and perform detection on the device element to be detected from the optimized FP tree.

[0013] Optionally, the target anomalous device is obtained, and its associated faulty devices are obtained from the co-occurrence device sequence. Based on the target device and its associated faulty devices, and from the optimized FP tree, the device element to be detected is determined, and the device element to be detected is detected, specifically including: Calculate the co-occurrence of each device element in the high-priority branch and the co-occurrence of each device element in the low-priority branch, respectively. Retrieve the target abnormal device element of the target abnormal device; In the sequence of co-occurrence devices of the target abnormal device, identify the first associated faulty device of the target abnormal device. From the optimized FP tree, determine the set of fault paths that include the target anomalous device element and the first associated faulty device; Determine the first fault path from the set of fault paths; From the first fault path, determine the device element corresponding to the first associated faulty device to obtain the first faulty device element; Obtain the abnormal trend of the first faulty device element; When the abnormal trend of the first faulty device element is greater than the preset abnormal threshold, the first faulty device element is identified as the device element to be detected. When the abnormal trend of the first faulty device element is less than or equal to the preset abnormal threshold, the co-occurrence of the high-priority branch of the target abnormal device element and the co-occurrence of the low-priority branch of the first faulty device element are obtained respectively. When the co-occurrence of the high-priority branch of the target abnormal device element is greater than the co-occurrence of the low-priority branch of the first faulty device element, a second fault path is obtained from the first target branch. When the co-occurrence of the high-priority branch of the target abnormal device element is less than or equal to the co-occurrence of the low-priority branch of the first faulty device element, a second fault path is obtained from the second target branch. The first target branch is the branch obtained from the optimized FP tree with the high-priority branch as the original data, and the second target branch is the branch obtained with the low-priority branch as the original data. From the second fault path, determine the device element corresponding to the first associated faulty device to obtain the second faulty device element; Obtain the abnormal trend of the second faulty device element; When the abnormal trend of the second faulty device element is greater than the preset abnormal threshold, the second faulty device element is identified as the device element to be detected. In the sequence of co-occurrence devices of the target abnormal device, the fault path set of each associated fault device is obtained in turn. Then, from the fault path set of each associated fault device, the fault device element corresponding to each associated fault device is determined in turn until the abnormal trend of the fault device element corresponding to the associated fault device is greater than the preset abnormal threshold. The fault device element corresponding to the associated fault device is then determined as the device element to be detected.

[0014] This invention proposes a multi-dimensional dynamic monitoring system for petroleum resin production, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of the multi-dimensional dynamic monitoring method for petroleum resin production as described above.

[0015] The beneficial effects of the technical solution of the present invention are: In this embodiment of the invention, an intelligent traceability mechanism that links multiple systems' data is constructed, automatically initiating diagnosis the moment an anomaly is detected. The system synchronously calls upon real-time and historical data from multiple dimensions, such as vibration, flow rate, pressure, and temperature, and uses a fusion analysis model to quickly pinpoint the root cause of the fault. This method achieves a leap from "single-point alarm" to "precise traceability," directly indicating specific causes such as "insufficient flow due to cooling valve jamming." This upgrades the handling approach from blind shutdown to targeted maintenance, significantly shortening diagnosis and recovery time and effectively improving production continuity and operational efficiency. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart of a multi-dimensional dynamic monitoring method for petroleum resin production provided in one embodiment of the present invention; Figure 2 This is an example diagram illustrating the traditional method of constructing an FP-tree. Figure 3 This is an example diagram illustrating the optimized FP-tree construction method in this embodiment; Figure 4 This is a structural diagram of a multi-dimensional dynamic monitoring system for petroleum resin production provided in one embodiment of the present invention. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a multi-dimensional dynamic monitoring method for petroleum resin production proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0020] The following description, in conjunction with the accompanying drawings, details a specific scheme for a multi-dimensional dynamic monitoring method for petroleum resin production provided by the present invention.

[0021] This invention provides a multi-dimensional dynamic monitoring method and system for petroleum resin production. Please refer to [link / reference]. Figure 1 The diagram illustrates a flowchart of a multi-dimensional dynamic monitoring method for petroleum resin production according to an embodiment of the present invention. The method includes the following steps: S101. Obtain the fault events of the petroleum resin production line, wherein each fault event includes the faulty equipment and the abnormal data corresponding to the faulty equipment.

[0022] For example, in a petroleum resin production line, multiple pieces of equipment operate in cooperation and various data flows are monitored to ensure the stable operation of the petroleum resin production line.

[0023] For example, in the production process of petroleum resin, various sensors are needed to collect data from specific equipment, such as reactor temperature and agitator speed.

[0024] In this embodiment, an event in the petroleum resin production line represents a complete process of producing petroleum resin once. An event in a petroleum resin production line can include various devices and various data generated by each device.

[0025] For example, in a reactor: data on internal temperature, stirrer speed, and operating pressure are collected to monitor the core reaction status.

[0026] Feed pump: Collects data on outlet flow rate, motor current, and bearing vibration to assess material conveying stability and equipment health.

[0027] Cooler: Collects data on cooling water inlet and outlet temperatures, refrigerant pressure, and valve opening to accurately control the heat exchange efficiency of the process system.

[0028] Fault events in a petroleum resin production line are defined as the processes that generate abnormal data during the production of petroleum resin. In each fault event, the equipment that causes the fault is identified as the faulty equipment, and the corresponding fault data is identified as the abnormal data for that equipment. Examples include abnormal reactor temperature and reactor pressure data.

[0029] Furthermore, in this embodiment, the device that causes the faulty device to malfunction is also identified as the faulty device. For example, if the temperature of the reactor is normal, but the refrigerant pressure of the cooler is abnormal, the cooler is identified as the faulty device, and the reactor that causes the malfunction is also identified as the faulty device.

[0030] S102. Combine each faulty device and the corresponding abnormal data in the fault event to obtain the device element of each faulty device. Determine the target device element and the associated device element as the target co-occurrence device element, and obtain the co-occurrence degree of the target co-occurrence device element of each target co-occurrence device element of each device element. The target device element is any device element in the device elements, and the associated device element is any device element in the device elements except the target device element.

[0031] In this embodiment, obtaining the co-occurrence degree of each target co-occurring device element for each device element specifically includes: Obtain the target fault event and co-occurring fault events from the fault events, wherein the target fault event includes at least one of the target device element and the associated device element, and the co-occurring fault events include the target device element and the associated device element; The ratio of the number of co-occurring fault events to the number of target fault events is determined as the co-occurrence degree of the target co-occurring device element. In the device elements, each device element except the target device element is identified as an associated device element, and the co-occurrence degree of the target co-occurring device elements for each target co-occurring device element is obtained; Each device element is identified as a target device element, and the co-occurrence degree of each target co-occurring device element for each device element is obtained.

[0032] For example, for each fault event, abnormal data for each faulty device in that fault event is obtained. For instance, in a fault event, the reactor is the faulty device, and the reactor's temperature data is abnormal data. (Faulty device, temperature data) is denoted as a device element. Co-occurrence is the probability that two device elements, or data points, or devices occur simultaneously.

[0033] The process for calculating the co-occurrence of any two device elements for all historical failure events is as follows: Obtain the total number of times N1 occurs in the fault event for these two device elements. Here, the total number of times N1 indicates that the fault event contains either of these two device elements. Record the number of times the two device elements appear together in the fault event as N2. Record the ratio of N2 to N1 as the co-occurrence degree of the two target co-occurring device elements.

[0034] For example, calculate the co-occurrence degree of two target co-occurring device elements: (reactor, temperature) and (stirrer, speed). Fault events include: Fault event 1, (reactor, temperature), (stirrer, speed), (separator, speed); Fault event 2, (reactor, temperature), (stirrer, speed), (separator, speed); Fault event 3, (reactor, temperature), (stirrer, temperature), (separator, speed).

[0035] In the example data, there are three failure events involving two-element devices. The device elements (reactor, temperature) and (stirrer, speed) appear twice in these three events, therefore the co-occurrence rate of the target co-occurring device element is 2 / 3. Among these, failure events 1, 2, and 3 are the target failure events, and failure events 1 and 2 are co-occurring failure events.

[0036] For each device element, the co-occurrence degree of that device element with each other can be obtained.

[0037] S103. Based on the co-occurrence degree of the target co-occurring device elements, determine the global sorting sequence of the device elements, and obtain the score of each device element in the global sorting sequence to obtain the global score sorting sequence.

[0038] In this embodiment, the global sorting sequence of device elements is determined based on the co-occurrence degree of the target co-occurring device elements, specifically including: Obtain the co-occurrence degree of each target co-occurring device element of the a-th device element and sum them to obtain the co-occurrence degree sum of the a-th device element; Get the co-occurrence sum for each device element; The device element with the highest co-occurrence is determined as the first element, and the first element is removed from the device elements to obtain the device elements after the first update; Get the co-occurrence of the target co-occurrence device element between the b-th device element and the first element in the device elements after the first update, and get the co-occurrence of the target co-occurrence device element between each device element (excluding the b-th device element) and the b-th device element in the device elements after the first update and sum them to get the co-occurrence sum of the b-th device element. The product of the co-occurrence degree of the b-th device element and the target co-occurrence device element of the first element, and the sum of the co-occurrence degrees of the b-th device element, is determined as the ranking value of the b-th device element. Get the sort value of each device element in the device elements after the first update; The device element with the largest sort value is designated as the second element; Remove the second element from the device element after the first update to obtain the device element after the second update; In the device elements after the second update, obtain the co-occurrence degree of the target co-occurrence device element of the c-th device element and the second element, and obtain the co-occurrence degree of each device element except the c-th device element with the target co-occurrence device element of the c-th device element and sum them to obtain the co-occurrence degree sum of the c-th device element. The product of the co-occurrence degree of the target co-occurrence device element of the c-th device element and the co-occurrence degree of the c-th device element is determined as the ranking value of the c-th device element. Get the sort value of each device element in the device elements after the second update; The device element with the largest sort value is designated as the third element; Get the Nth element, and determine the global sorting sequence of the device elements from the first element to the Nth element.

[0039] Obtain the score of each device element in the global sort sequence to obtain the global score sort sequence, specifically including: The score of the last device element in the global sort sequence is set to 1, and the score of the second-to-last device element in the global sort sequence is set to 2. The score of the Mth device element from the end of the global sorting sequence is determined as M, thus obtaining the global score sorting sequence, where M is the number of elements in the global sorting sequence.

[0040] For example, after obtaining the co-occurrence degree of the a-th device element with each other device element, the sum of the co-occurrence degrees of all co-occurring device elements can be calculated, and the sum of the co-occurrence degrees of each device element can be obtained.

[0041] The device element with the highest co-occurrence sum is determined as the first element. For the remaining device elements excluding the first element, the co-occurrence sum g1 of the target co-occurring device element with the first element is obtained, and the co-occurrence sum h1 of each device element with other device elements is obtained; the product of g1 and h1 is calculated, and the device element with the largest product is determined as the second element. For the remaining device elements excluding the first and second elements, the co-occurrence sum g2 of the target co-occurring device element with the second element is obtained, and the co-occurrence sum h2 of each device element with other device elements is obtained; the product of g2 and h2 is calculated, and the device element with the largest product is determined as the third element, and so on, until each device element has been sorted, resulting in a globally sorted sequence.

[0042] For example, for a globally sorted sequence with 9 elements, the score of the first element is recorded as 9, the score of the second element as 8, and so on, to obtain the score of each element.

[0043] S104. Obtain the transaction for each failure event, and obtain the continuous prefix and non-continuous prefix in the transaction for each failure event.

[0044] In this embodiment, the transaction for each fault event is obtained, and the continuous and non-contiguous prefixes in the transaction for each fault event are obtained, specifically including: Retrieve the device elements for each fault event; Obtain the sequence number of the e-th device element in the global sort sequence in the d-th fault event, and get the transaction sequence number of the e-th device element in the d-th fault event; Obtain the transaction sequence number of each device element in the d-th fault event; Sort the transaction sequence number of each device element in the d-th fault event in ascending order to obtain the transaction of the d-th fault event; Acquire transactions for each failure event; The transactions of each failure event are divided according to the global sorting sequence to obtain continuous prefixes and non-contiguous prefixes in each failure event's transactions. The continuous prefix is ​​a subset of continuous device elements in the global sorting sequence, and the number of device elements in the continuous prefix is ​​greater than 1. The non-contiguous prefix is ​​a subset of device elements in the global sorting sequence, and the number of device elements in the non-contiguous prefix is ​​equal to 1. When at least two non-contiguous prefixes are adjacent in the transactions of the failure event, the at least two non-contiguous prefixes are merged into one non-contiguous prefix.

[0045] For example, for a certain failure event, all device elements in the failure event are arranged in the order of elements in the global sorting sequence to determine the transaction of this failure event.

[0046] For example, the global sort sequence can be [G, J, K, L, T], and the fault event can be [K, T, G]. Then, in the fault event, the transaction number of device element K is 3, the transaction number of device element T is 5, the transaction number of device element G is 1, and the transaction of the fault event is [G, T, K].

[0047] The transactions for each failure event are divided according to the global sorting sequence to obtain the continuous and non-contiguous prefixes in the transactions of each failure event. Specifically, this can be done as follows: For each element in a transaction, calculate the local match with the global permutation sequence, record the matching parts as continuous prefixes, obtain all continuous prefixes, and then divide the elements in each transaction into continuous prefixes and non-continuous prefixes.

[0048] For example, given the global permutation sequence [A, B, C, D, E, F, G, H], and a sorted transaction [A, B, C, E, G, H], ABC and GH are consecutive prefixes, and E is a non-consecutive prefix. As another example, given the global permutation sequence [A, B, C, D, E, F, G, H, J], and a sorted transaction [A, B, C, E, G, J], ABC is a consecutive prefix, and EG and J are two distinct non-consecutive prefixes with lengths of 2 and 1 respectively.

[0049] S105. Based on the global score sorting sequence, determine the score of each device element in the transaction of each fault event, and calculate the first final score of each consecutive prefix and the second final score of each non-consecutive prefix based on the score of each device element in the transaction of each fault event.

[0050] In this embodiment, based on the global score sorting sequence, the score of each device element in the transaction of each fault event is determined. Based on the score of each device element in the transaction of each fault event, the first final score of each consecutive prefix and the second final score of each non-consecutive prefix are calculated, specifically including: The score of the f-th device element in the global score sorting sequence is determined as the score of the h-th device element in the transaction of the g-th failure event, wherein the f-th device element in the global score sorting sequence and the h-th device element in the transaction of the g-th failure event are the same device element. Get the score of each device element in the transaction for the g-th failure event; Get the score of each device element in the transaction for each failure event; Based on the score of each device element in the k-th consecutive prefix in the transaction of the j-th failure event, calculate the first final score of the k-th consecutive prefix in the transaction of the j-th failure event; Based on the score of each device element in the l-th discontinuous prefix in the transaction of the j-th failure event, calculate the second final score of the l-th discontinuous prefix in the transaction of the j-th failure event.

[0051] Based on the score of each device element in the k-th consecutive prefix within the transaction of the j-th failure event, calculate the first final score of the k-th consecutive prefix within the transaction of the j-th failure event, specifically including: The sum of the scores of each device element in the k-th consecutive prefix in the transaction of the j-th fault event is obtained by summing the scores of the k-th consecutive prefix in the transaction of the j-th fault event. The ratio of the sum of the scores of the k-th consecutive prefixes in the transaction of the j-th fault event to the sum of the scores of each device element in the transaction of the j-th fault event is determined as the percentage of the score of the k-th consecutive prefix in the transaction of the j-th fault event. The product of the percentage of the score of the k-th consecutive prefix in the transaction of the j-th failure event and the sum of the scores of the k-th consecutive prefix in the transaction of the j-th failure event is determined as the first final score of the k-th consecutive prefix in the transaction of the j-th failure event.

[0052] Based on the score of each device element in the l-th discontinuous prefix within the transaction of the j-th failure event, calculate the second final score of the l-th discontinuous prefix within the transaction of the j-th failure event, specifically including: Sum the scores of each device element in the l-th discontinuous prefix within the transaction of the j-th fault event to obtain the sum of the scores of the l-th discontinuous prefix within the transaction of the j-th fault event. The product of the number of device elements in the l-th discontinuous prefix in the transaction of the j-th fault event and the sum of the scores of the l-th discontinuous prefix in the transaction of the j-th fault event is determined as the second final score of the l-th discontinuous prefix in the transaction of the j-th fault event.

[0053] For example, the score of each device element in each transaction of each fault event is determined according to the global score sorting sequence. A specific implementation can be as follows: the global sorting sequence is [A, B, C, D, E, F, G, H, J], and a sorted transaction is [A, B, C, E, G, J]. In this case, ABC in the transaction is a continuous prefix, and EG and J are two different non-contiguous prefixes. The score of each device element in this transaction is [9, 8, 7, 5, 3, 1].

[0054] For each transaction, the product of the score and percentage in each consecutive prefix and the sum of the scores of each device element in that prefix is ​​calculated and denoted as the first final score s1. Here, the score and percentage refer to the ratio of the sum of the scores of device elements in the consecutive prefix to the sum of the scores of all device elements in the transaction. Using this value as a multiplier ensures that consecutive elements in the earlier parts of the global permutation sequence are constructed earlier on the FP-tree, becoming shared paths for more other transactions, thus allowing all transactions to be represented by a tree structure with fewer nodes. For consecutive prefixes, calculating the product of the score and percentage and the sum of the scores distinguishes between earlier and later consecutive prefixes. This calculation widens the score difference between earlier and later consecutive prefixes in the global permutation sequence, highlighting the earlier consecutive prefixes. For example, in the example above, ABC is an earlier consecutive prefix, and GH is a later consecutive prefix. For each transaction, the product of the number of each non-consecutive prefix and the sum of the scores of its device elements is calculated and denoted as s2. This calculation increases the score difference between consecutive and non-consecutive prefixes. Consecutive prefixes can be considered shared nodes, thus assigning higher priority scores to transactions with consecutive prefixes. Therefore, the first final score s1ABC of the consecutive prefix ABC can be calculated as:

[0055] The calculation method for consecutive prefixes GH is the same as that for ABC.

[0056] The second final score s2EG of the discontinuous prefix EG can be calculated as:

[0057] The calculation method for discontinuous prefix J is the same as that for EG.

[0058] S106. Calculate the transaction priority score for each failure event based on the first final score of each consecutive prefix and the second final score of each non-consecutive prefix.

[0059] In this embodiment, the priority score of each failure event is calculated based on the first final score of each consecutive prefix and the second final score of each non-consecutive prefix, specifically including: Based on continuous and non-contiguous prefixes, the transactions of the failure event are divided into transaction groups, where any transaction group in the transaction group has either a continuous prefix or a non-contiguous prefix; The frequency of each transaction group within the transactions that acquire failure events; The product of the frequency and final score of the m-th transaction group in the transaction of the j-th failure event is determined as the third final score of the m-th transaction group in the transaction of the j-th failure event. Specifically, when the m-th transaction group in the transaction of the j-th failure event is the k-th consecutive prefix in the transaction of the j-th failure event, the first final score of the k-th consecutive prefix in the transaction of the j-th failure event is determined as the final score of the m-th transaction group in the transaction of the j-th failure event; when the m-th transaction group in the transaction of the j-th failure event is the l-th non-consecutive prefix in the transaction of the j-th failure event, the second final score of the l-th non-consecutive prefix in the transaction of the j-th failure event is determined as the final score of the m-th transaction group in the transaction of the j-th failure event. Obtain the third final score of each transaction group in the transaction of the j-th failure event and sum them to get the priority score of the transaction of the j-th failure event; Get the priority score of the transaction for each failure event.

[0060] For example, each transaction is divided into multiple transaction groups based on consecutive prefixes and non-consecutive prefixes.

[0061] For example, if the global permutation sequence is [A, B, C, D, E, F, G, H], and a sorted transaction is [A, B, C, E, G, H], where ABC and GH are consecutive prefixes and E is a non-consecutive prefix, then three transaction groups are obtained.

[0062] This allows us to obtain all transaction groups within all transactions, then obtain the frequency of each transaction group in all transactions, calculate the ratio of this frequency to the number of transactions, and finally obtain the frequency of each transaction group.

[0063] For example: The global permutation sequence is: [A, B, C, D, E, F, G, H]. Transaction 1 has three groups: ABC, E, and GH. Transaction 2 has two transaction groups: AB and DE. Transaction 3 has two transaction groups: ABC and DE. Then there are a total of 5 non-repeating transaction groups in transactions 1, 2, and 3: ABC, E, GH, AB, and DE. ABC appears twice, with a frequency of 2 / 3, and E appears once, with a frequency of 1 / 3.

[0064] For each transaction group within each transaction, the frequency of that transaction group is used as the weight of its final score (continuous prefixes are the first final score, and non-continuous prefixes are the second final score). The priority score of each transaction is obtained by weighted summation.

[0065] S107. Transactions with fault events whose priority scores are greater than the preset priority threshold are identified as high-priority branches, and transactions with fault events whose priority scores are less than or equal to the preset priority threshold are identified as low-priority branches.

[0066] Optionally, in this embodiment, the preset priority threshold can be set according to the actual situation. Here, the specific value of the preset priority threshold is not limited. In a preferred embodiment, it can be the lowest priority score among the top 5% of priority scores.

[0067] S108. Based on the high-priority branches and low-priority branches, construct an optimized FP tree, determine the device element to be detected from the optimized FP tree, and perform detection on the device element to be detected.

[0068] In this embodiment, an optimized FP-tree is constructed based on high-priority branches and low-priority branches, and the device element to be detected is determined from the optimized FP-tree. The detection of the device element to be detected specifically includes: Each high-priority branch and each low-priority branch are identified as the original data, and each device element in each failure event transaction is identified as a node. An FP tree is constructed to obtain the optimized FP tree. In the construction process, each high-priority branch participates in the construction of FP with the root node of the FP tree as the root node of the high-priority branch. Obtain the faulty device in each fault event to obtain the faulty device sequence; The target faulty device and the associated faulty device are combined into a faulty device pair. The frequency of the faulty device pair in the faulty device sequence is obtained to obtain the co-occurrence of the target faulty device and the associated faulty device. The target faulty device is any faulty device in the faulty device, and the associated faulty device is any faulty device in the faulty device except the target faulty device. The associated faulty devices are sorted in descending order of co-occurrence to obtain the co-occurrence sequence of the target faulty device; Each faulty device is identified as a target faulty device, and the co-occurrence sequence of each faulty device is obtained. Obtain the target abnormal device, obtain the associated faulty devices of the target abnormal device from the co-occurrence device sequence of the target abnormal device, determine the device element to be detected based on the target device and associated faulty devices, and perform detection on the device element to be detected from the optimized FP tree.

[0069] The process involves: acquiring the target anomalous device; obtaining associated faulty devices from the co-occurrence device sequence; determining the device elements to be detected based on the target device and associated faulty devices, and from the optimized FP tree; and performing detection on these device elements, specifically including: Calculate the co-occurrence of each device element in the high-priority branch and the co-occurrence of each device element in the low-priority branch, respectively. Retrieve the target abnormal device element of the target abnormal device; In the sequence of co-occurrence devices of the target abnormal device, identify the first associated faulty device of the target abnormal device. From the optimized FP tree, determine the set of fault paths that include the target anomalous device element and the first associated faulty device; Determine the first fault path from the set of fault paths; From the first fault path, determine the device element corresponding to the first associated faulty device to obtain the first faulty device element; Obtain the abnormal trend of the first faulty device element; When the abnormal trend of the first faulty device element is greater than the preset abnormal threshold, the first faulty device element is identified as the device element to be detected. When the abnormal trend of the first faulty device element is less than or equal to the preset abnormal threshold, the co-occurrence of the high-priority branch of the target abnormal device element and the co-occurrence of the low-priority branch of the first faulty device element are obtained respectively. When the co-occurrence of the high-priority branch of the target abnormal device element is greater than the co-occurrence of the low-priority branch of the first faulty device element, a second fault path is obtained from the first target branch. When the co-occurrence of the high-priority branch of the target abnormal device element is less than or equal to the co-occurrence of the low-priority branch of the first faulty device element, a second fault path is obtained from the second target branch. The first target branch is the branch obtained from the optimized FP tree with the high-priority branch as the original data, and the second target branch is the branch obtained with the low-priority branch as the original data. From the second fault path, determine the device element corresponding to the first associated faulty device to obtain the second faulty device element; Obtain the abnormal trend of the second faulty device element; When the abnormal trend of the second faulty device element is greater than the preset abnormal threshold, the second faulty device element is identified as the device element to be detected. In the sequence of co-occurrence devices of the target abnormal device, the fault path set of each associated fault device is obtained in turn. Then, from the fault path set of each associated fault device, the fault device element corresponding to each associated fault device is determined in turn until the abnormal trend of the fault device element corresponding to the associated fault device is greater than the preset abnormal threshold. The fault device element corresponding to the associated fault device is then determined as the device element to be detected.

[0070] For example, when a traditional FP-tree is constructed using the insertion method, it follows a random insertion method. In this embodiment, due to the large number of combinations of device types and data types, the number of device elements is large, resulting in a large amount of data in the constructed tree structure. Therefore, in order to facilitate subsequent queries, the data devices that appear more frequently and conform more to the global element sorting sequence are given higher insertion priority. Then, for high-priority data, instead of matching according to the traditional method and sharing nodes, a branch is opened for the data to facilitate quick location during queries.

[0071] In a traditional FP-tree, each node (such as the node labeled "CCC") is linked to all nodes with the same name through a linked list corresponding to the header table. When tracing a specific transaction pattern (such as a path containing {AAA, CCC}), the algorithm must traverse all nodes under the "CCC" header table and check their complete paths one by one. This process requires a large number of judgments and filtering in many related nodes, and most of the computation is consumed on redundant branches unrelated to the target pattern, resulting in low efficiency.

[0072] For each failure event, all device elements in that failure are retrieved and used to participate in the construction of the FP-tree. Then, a priority score is obtained based on the score of each device element within the transaction of the failure event. A higher priority score increases the probability of creating a separate branch without sharing nodes with other transactions, preserving the singleness of the transaction's structure in the FP-tree and facilitating subsequent queries. Device elements appearing earlier in the global permutation sequence within a transaction participate in the FP-tree construction earlier. This allows for the formation of shared nodes for subsequent device elements earlier, resulting in a more streamlined tree structure and stronger data compression capabilities.

[0073] For fault events ranking in the top 5% by priority score, even if there are shareable nodes during construction, the elements in the transaction are not inserted through shared nodes, but instead a new branch is created directly from the root node. This insertion method constructs the device characteristic FP-tree. The branches ranking in the top 5% by priority score in the FP-tree are designated as high-priority branches, and those to the right are designated as low-priority branches. In this tree, any path from the root node to a leaf node represents a high-probability fault propagation chain. Elements closer to the root node are more common fault initiators. Through data-driven sorting rules, the FP-tree construction process automatically organizes frequently co-occurring fault elements into directed paths. Because the sorting prioritizes "influencers" over "affected," the direction of fault propagation is naturally reflected in the tree structure, ultimately making nodes near the root node the "originating point" of faults, while the leaf node area becomes the "converging point" of faults.

[0074] like Figure 2 As shown, Figure 2 This is an example diagram illustrating the traditional method of constructing an FP-tree. Figure 3 This is an example diagram illustrating the optimized FP-tree construction method in this embodiment. In the traditional FP-tree construction method, the sorted original data are shown as T1, T2, T3, T4, and T5. These are inserted sequentially in the order of T1-T5, resulting in an FP-tree as shown below. Figure 2 As shown. For example: when tracing the origin of point C, through... Figure 2 The FP-tree in the example can yield three complete paths: DCBA-null, ECBA-null, and CBA-null.

[0075] In one specific embodiment, according to Figure 2 From the original data, T1 and T2 have the highest priority scores, so they can be inserted first. T1 has the highest priority, so it's inserted first. When inserting T2, there's no tree branch sharing a prefix with T1's branch; instead, a new branch is created. This method divides transactions into high-priority and low-priority transactions. Therefore, during queries, high-priority and low-priority transactions don't share nodes. This significantly reduces the number of nodes that need to be traced when tracing back to the source of a query. Figure 3 The FP-tree in the traditional method can yield three complete paths. However, the improved FP-tree divides the FP-tree into a high-priority region on the left and a low-priority region on the right. Therefore, it first analyzes the two paths on the left, reducing the computational cost of calculating the number of paths during path tracing compared to the traditional method. When the total amount of data is large, it can significantly reduce query time.

[0076] Furthermore, for each fault event, all faulty devices related to that fault can be obtained, resulting in a faulty device sequence. All faulty devices in the sequence are then paired to obtain faulty device pairs. For each faulty device pair, the frequency of its occurrence in the faulty device sequence is calculated, yielding the co-occurrence rate. A higher co-occurrence rate indicates a greater likelihood that both devices in the pair failed together, meaning there is a high probability that the failure of one device caused the failure of the other.

[0077] For a given faulty device, its co-occurrence with every other faulty device can be obtained. These co-occurrences can then be sorted in descending order to obtain the co-occurrence sequence of that faulty device. This allows us to obtain the co-occurrence sequence of each faulty device.

[0078] Calculate the co-occurrence of each device element in the high-priority branches and the co-occurrence of each device element in the low-priority branches, specifically including: Get the high-priority device elements in the high-priority branch, determine the target high-priority device element from the high-priority device elements, and form a high-priority device element pair with the target high-priority device element and other high-priority device elements. Get the co-occurrence of the high-priority device element pair with the target high-priority device element. The target high-priority device element is any high-priority device element in the high-priority branch, and other high-priority device elements are any high-priority device elements in the high-priority branch other than the target high-priority device element. Each high-priority device element in the high-priority branch is identified as the target high-priority device element, and the co-occurrence of high-priority device element pairs for each high-priority device element is obtained; The co-occurrence of high-priority device element pairs for each high-priority device element is determined as the high-priority branch co-occurrence of each device element. Get the low-priority device elements in the low-priority branch, determine the target low-priority device element from the low-priority device elements, and form a low-priority device element pair with the target low-priority device element and other low-priority device elements. Get the co-occurrence of the low-priority device element pair of the target low-priority device element. The target low-priority device element is any low-priority device element in the low-priority branch, and other low-priority device elements are any low-priority device elements in the low-priority branch other than the target low-priority device element. Each low-priority device element in the low-priority branch is identified as the target low-priority device element, and the co-occurrence of low-priority device element pairs for each low-priority device element is obtained. The co-occurrence of low-priority device element pairs for each low-priority device element is determined as the low-priority branch co-occurrence of each device element.

[0079] The first fault path is determined from the set of fault paths, specifically including: Determine the length of the target faulty device element and the leaf node in each path from the set of fault paths; The path with the shortest length is identified as the first faulty path.

[0080] Obtain the abnormal trend of the first faulty device element, specifically including: Obtain historical data and historical normal data for the first faulty device element; The Z-score method is used to compare the historical data of the first faulty device element with the historical normal data to obtain the anomaly of the first faulty device element; When the abnormal time point of the first faulty device element is before the abnormal time point of the target faulty device element, obtain the abnormality sequence of the first faulty device element and the abnormality sequence of the target faulty device element. Calculate the cosine similarity between the anomaly sequence of the first faulty device element and the anomaly sequence of the target faulty device element to obtain trend consistency; Based on the anomaly and trend consistency of the first faulty equipment element, calculate the abnormal trend of the first faulty equipment element; If the abnormal time point of the first faulty device element is not before the abnormal time point of the target faulty device element, the abnormal trend of the first faulty device element is determined to be 0.

[0081] Based on the anomaly and trend consistency of the first faulty equipment element, the abnormal trend of the first faulty equipment element is calculated, specifically including: Incrementing the trend consistency by one yields the adjusted trend consistency; The product of the anomaly of the first faulty equipment element and the consistency of its adjustment trend is determined as the anomaly trend of the first faulty equipment element.

[0082] Optionally, the preset abnormal threshold can be set according to actual needs. There is no specific limitation on the value of the preset abnormal threshold here. In a preferred embodiment, it can be 0.6.

[0083] In the high-priority branches, the co-occurrence of high-priority branches for each device element is denoted as k1, and in the low-priority branches, the co-occurrence of low-priority branches for each device element is denoted as k2.

[0084] Real-time monitoring of relevant data for each device; when an anomaly is detected in the data of device AA (ii): If a shutdown is required, first inspect equipment AA. If the problem lies with equipment AA, repair it. If the problem does not lie with equipment AA, determine which related causes led to the problem. Find the first device BB in the co-occurrence sequence of device AA. Then, in the optimized FP-tree, first find all paths in the high-priority branches that simultaneously contain (AA, ii) and (BB, #), where "#" represents all data types corresponding to device element BB. Among all paths, select the path with the (AA, ii) node closest to the leaf node. This path represents the most typical and direct failure scenario for (AA, ii) as the final manifestation.

[0085] Find the device element of device BB along this path, denoted as (BB, YY), where YY is the specific data type most likely to cause the problem. Obtain historical data for (BB, YY) for a period of time before the failure occurred. Calculate the anomaly using the Z-score method with the historical data and historical normal data. For example, the normal range of the standard value is [20%, 80%] (i.e., L=20, U=80). Current value: 92%, anomaly = 92-80=12 (or (92-80) / (100-80)=0.6).

[0086] Then, if the anomaly start time (which can be obtained manually) is earlier than (AA, ii), the trend consistency of the anomaly changes of (BB, YY) and (AA, ii) is calculated, and the real-time anomaly values ​​of the two types of data are obtained as sequences. Then, the cosine similarity is used as the trend consistency.

[0087] The product of anomaly and (1 + trend consistency) is used as the probability that the chain reaction caused by the abnormal trend of data YY of device BB will lead to the abnormality of data ii of AA. If the probability value is greater than 0.6, it is considered that B caused the abnormality of A. Then, targeted repairs are carried out on the problems that may cause the abnormality of data YY of device BB.

[0088] If the conditions are not met, then: obtain k1 and k2 of the two device elements (AA, ii) and (BB, YY). If k1 > k2, then search for the matching conditions in the high priority area first; otherwise, search in the low priority area. Use the same method to search for the matching conditions until the matching device element to be detected is found, and then perform targeted repair.

[0089] In the high-priority area, the first priority is to find the faulty device elements in other fault paths in the fault path set of device BB, and calculate the abnormal trend of the faulty device elements. When device BB has no abnormality, the second device is selected from the co-occurrence device sequence, and the calculation process of device BB is referred to until the device element to be detected is found.

[0090] In this embodiment, an intelligent traceability mechanism that links multiple systems' data is constructed, automatically initiating diagnosis the moment an anomaly is detected. The system synchronously calls upon real-time and historical data from multiple dimensions, such as vibration, flow rate, pressure, and temperature, and uses a fusion analysis model to quickly pinpoint the root cause of the fault. This method achieves a leap from "single-point alarm" to "precise traceability," directly indicating specific causes such as "insufficient flow due to cooling valve jamming." This upgrades the handling approach from blind shutdown to targeted maintenance, significantly shortening diagnosis and recovery time and effectively improving production continuity and operational efficiency.

[0091] This invention also proposes a multi-dimensional dynamic monitoring system for petroleum resin production; please refer to [link / reference]. Figure 4 The diagram shows a structure of a multi-dimensional dynamic monitoring system for petroleum resin production provided by an embodiment of the present invention. The system includes: a data acquisition module 101, a data processing module 102, and an anomaly detection module 103.

[0092] The data acquisition module 101 is used to acquire fault events of the petroleum resin production line, wherein each fault event includes the faulty equipment and the abnormal data corresponding to the faulty equipment. Data processing module 102 is used to combine each faulty device and its corresponding abnormal data in a fault event to obtain the device element of each faulty device. It identifies the target device element and associated device elements as target co-occurring device elements and obtains the co-occurrence degree of each target co-occurring device element for each device element. The target device element is any device element among the device elements, and the associated device element is any device element other than the target device element. Based on the co-occurrence degree of the target co-occurring device elements, it determines the global sorting sequence of device elements and obtains the score of each device element in the global sorting sequence to obtain a global score sorting sequence. It also obtains the transaction information for each fault event. The process involves obtaining the continuous and non-contiguous prefixes in the transactions of each fault event; determining the score of each device element in the transactions of each fault event based on the global score sorting sequence; calculating the first final score of each continuous prefix and the second final score of each non-contiguous prefix based on the score of each device element in the transactions of each fault event; calculating the priority score of the transactions of each fault event based on the first final score of each continuous prefix and the second final score of each non-contiguous prefix; identifying the transactions of fault events with priority scores greater than a preset priority threshold as high-priority branches, and identifying the transactions of fault events with priority scores less than or equal to the preset priority threshold as low-priority branches. The anomaly detection module 103 is used to construct an optimized FP tree based on high-priority branches and low-priority branches, and to determine the device element to be detected from the optimized FP tree and to detect the device element to be detected.

[0093] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the multi-dimensional dynamic monitoring system for petroleum resin production and the multi-dimensional dynamic monitoring method for petroleum resin production provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0094] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0095] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0096] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-dimensional dynamic monitoring method for petroleum resin production, characterized in that, include: Acquire fault events of the petroleum resin production line, wherein each fault event includes the faulty equipment and the corresponding abnormal data of the faulty equipment; The faulty device and its corresponding abnormal data are combined in the fault event to obtain the device element of each faulty device. The target device element and the associated device element are determined as the target co-occurrence device element. The co-occurrence degree of the target co-occurrence device element of each device element is obtained. The target device element is any device element in the device elements, and the associated device element is any device element in the device elements except the target device element. This includes obtaining the co-occurrence degree of each target co-occurring device element for each device element, including: The number of times a fault event that simultaneously contains both the target device element and the associated device element occurs is recorded as the first occurrence. The number of occurrences of a fault event that includes at least one of the target device element and the associated device element is denoted as the second count; The ratio of the first count to the second count is determined as the co-occurrence degree of the target device element and its associated device elements. Based on the co-occurrence of the target co-occurring device elements, determine the global sorting sequence of the device elements, and obtain the score of each device element in the global sorting sequence to obtain the global score sorting sequence; Based on the co-occurrence of the target co-occurring device elements, determine the global sorting sequence of the device elements, including: The sum of the co-occurrence rates of the target co-occurring device elements corresponding to each device element is determined as the sum of the co-occurrence rates of each device element, and the device element with the largest sum of co-occurrence rates is determined as the first element; From all device elements except the first element, the product of the co-occurrence degree of the target co-occurrence device elements between any device element and the previous element whose sorting has been determined, and the sum of the co-occurrence degrees of the target co-occurrence device elements between any device element and the remaining unsorted device elements, is determined as the sorting value of any device element. The device element with the largest sorting value is determined as the next element after the previous element. Repeat the step of determining the next element until all device elements are sorted to obtain a global sorted sequence; Retrieve the transactions for each failure event, and obtain the consecutive and non-consecutive prefixes within each transaction for that failure event, including: Sort the device elements in any fault event according to their order in the global sorting sequence to obtain the transaction for that fault event; For each device element in any transaction, perform local matching between each device element and the global sort sequence, determine the consecutive device elements that are successfully matched as consecutive prefixes, and determine the device elements in the transaction other than consecutive prefixes as non-consecutive prefixes. Based on the global score sorting sequence, determine the score of each device element in the transaction for each failure event; Based on the score of each device element in the transaction for each failure event, calculate the first final score for each consecutive prefix and the second final score for each non-consecutive prefix, specifically including: For each consecutive prefix in the transaction of each failure event, the first final score of the consecutive prefix is ​​determined by multiplying the score and percentage of the consecutive prefix by the sum of the scores of each device element in the consecutive prefix, where the score and percentage of the consecutive prefix is ​​the ratio of the sum of the scores of all device elements in the consecutive prefix to the sum of the scores of all device elements in the transaction. For each discontinuous prefix in the transaction of each failure event, the second final score of the discontinuous prefix is ​​determined by multiplying the number of device elements in the discontinuous prefix by the sum of the scores of each device element in the discontinuous prefix. Based on the first final score for each consecutive prefix and the second final score for each non-consecutive prefix, the transaction priority score for each failure event is calculated, specifically including: Based on the continuous and non-contiguous prefixes in the transactions of each failure event, the transactions of each failure event are divided into transaction groups, where each transaction group consists of either continuous or non-contiguous prefixes. Obtain the frequency of each transaction group in all transactions across all failure events; For each transaction group in the transaction of each failure event, the third final score of the transaction group is determined by multiplying the frequency of the transaction group by the final score of the transaction group, wherein the final score of the transaction group is either the first final score of the continuous prefix or the second final score of the non-continuous prefix. The sum of the third final scores of each transaction group in the transaction of each failure event is determined as the priority score of the transaction in that failure event. Transactions with failure events whose priority scores are greater than a preset priority threshold are identified as high-priority branches, and transactions with failure events whose priority scores are less than or equal to the preset priority threshold are identified as low-priority branches. Based on the high-priority and low-priority branches, an optimized FP-tree is constructed. The device elements to be detected are then determined from the optimized FP-tree, and detection is performed on these elements, specifically including: Each high-priority branch and each low-priority branch are identified as the original data, and each device element in each failure event transaction is identified as a node. An FP tree is constructed to obtain the optimized FP tree. In the construction process, each high-priority branch participates in the construction of FP with the root node of the FP tree as the root node of the high-priority branch. Obtain the faulty device in each fault event to obtain the faulty device sequence; The target faulty device and the associated faulty device are combined into a faulty device pair. The frequency of the faulty device pair in the faulty device sequence is obtained to obtain the co-occurrence of the target faulty device and the associated faulty device. The target faulty device is any faulty device in the faulty device, and the associated faulty device is any faulty device in the faulty device except the target faulty device. The associated faulty devices are sorted in descending order of co-occurrence to obtain the co-occurrence sequence of the target faulty device; Each faulty device is identified as a target faulty device, and the co-occurrence sequence of each faulty device is obtained. Calculate the co-occurrence of each device element in the high-priority branch and the co-occurrence of each device element in the low-priority branch, respectively. Obtain the target abnormal device element; obtain the first associated faulty device of the target abnormal device in the co-occurrence device sequence of the target abnormal device; From the optimized FP tree, determine the set of fault paths that include the target anomalous device element and the first associated faulty device; Determine the first fault path from the set of fault paths; determine the device element corresponding to the first associated faulty device from the first fault path to obtain the first faulty device element; obtain the abnormal trend of the first faulty device element; When the abnormal trend of the first faulty device element is greater than the preset abnormal threshold, the first faulty device element is identified as the device element to be detected. When the abnormal trend of the first faulty device element is less than or equal to a preset abnormal threshold, the co-occurrence of the high-priority branch of the target abnormal device element and the co-occurrence of the low-priority branch of the first faulty device element are obtained respectively. When the co-occurrence of the high-priority branch of the target abnormal device element is greater than the co-occurrence of the low-priority branch of the first faulty device element, a second fault path is obtained from the first target branch. When the co-occurrence of the high-priority branch of the target abnormal device element is less than or equal to the co-occurrence of the low-priority branch of the first faulty device element, a second fault path is obtained from the second target branch. Here, the first target branch is a branch obtained from the optimized FP tree using the high-priority branch as the original data, and the second target branch is a branch obtained from the low-priority branch as the original data. The device element corresponding to the first associated faulty device is determined from the second fault path to obtain the second faulty device element. The abnormal trend of the second faulty device element is obtained. When the abnormal trend of the second faulty device element is greater than the preset abnormal threshold, the second faulty device element is identified as the device element to be detected. In the sequence of co-occurrence devices of the target abnormal device, the fault path set of each associated fault device is obtained in turn. Then, from the fault path set of each associated fault device, the fault device element corresponding to each associated fault device is determined in turn until the abnormal trend of the fault device element corresponding to the associated fault device is greater than the preset abnormal threshold. The fault device element corresponding to the associated fault device is then determined as the device element to be detected.

2. The multi-dimensional dynamic monitoring method for petroleum resin production according to claim 1, characterized in that, The acquisition of the co-occurrence degree of each target co-occurring device element for each device element specifically includes: Obtain the target fault event and co-occurring fault events from the fault events, wherein the target fault event includes at least one of the target device element and the associated device element, and the co-occurring fault events include the target device element and the associated device element; The ratio of the number of co-occurring fault events to the number of target fault events is determined as the co-occurrence degree of the target co-occurring device element. In the device elements, each device element except the target device element is identified as an associated device element, and the co-occurrence degree of the target co-occurring device elements for each target co-occurring device element is obtained; Each device element is identified as a target device element, and the co-occurrence degree of each target co-occurring device element for each device element is obtained.

3. The multi-dimensional dynamic monitoring method for petroleum resin production according to claim 1, characterized in that, The step of determining the global sorting sequence of device elements based on the co-occurrence degree of target co-occurring device elements specifically includes: Obtain the co-occurrence degree of each target co-occurring device element of the a-th device element and sum them to obtain the co-occurrence degree sum of the a-th device element; obtain the co-occurrence degree sum of each device element; The device element with the highest co-occurrence is determined as the first element, and the first element is removed from the device elements to obtain the device elements after the first update; Get the co-occurrence of the target co-occurrence device element between the b-th device element and the first element in the device elements after the first update, and get the co-occurrence of the target co-occurrence device element between each device element (excluding the b-th device element) and the b-th device element in the device elements after the first update and sum them to get the co-occurrence sum of the b-th device element. The product of the co-occurrence degree of the b-th device element and the target co-occurrence device element of the first element, and the sum of the co-occurrence degrees of the b-th device element, is determined as the ranking value of the b-th device element. Get the sort value of each device element in the device elements after the first update; determine the device element with the largest sort value as the second element; remove the second element from the device elements after the first update to obtain the device elements after the second update; In the device elements after the second update, obtain the co-occurrence degree of the target co-occurrence device element of the c-th device element and the second element, and obtain the co-occurrence degree of each device element except the c-th device element with the target co-occurrence device element of the c-th device element and sum them to obtain the co-occurrence degree sum of the c-th device element. The product of the co-occurrence degree of the target co-occurrence device element of the c-th device element and the co-occurrence degree of the c-th device element is determined as the ranking value of the c-th device element. Get the sort value of each device element in the device elements after the second update; determine the device element with the largest sort value as the third element; get the Nth element, and determine the first element to the Nth element as the global sort sequence of device elements.

4. The multi-dimensional dynamic monitoring method for petroleum resin production according to claim 1, characterized in that, The step of obtaining the score of each device element in the global sorting sequence to obtain the global score sorting sequence specifically includes: The score of the last device element in the global sort sequence is set to 1, and the score of the second-to-last device element in the global sort sequence is set to 2. The score of the Mth device element from the end of the global sorting sequence is determined as M, thus obtaining the global score sorting sequence, where M is the number of elements in the global sorting sequence.

5. The multi-dimensional dynamic monitoring method for petroleum resin production according to claim 1, characterized in that, The step of obtaining the transaction for each failure event, and obtaining the continuous and non-contiguous prefixes in the transaction for each failure event, specifically includes: Retrieve the device elements for each fault event; Obtain the sequence number of the e-th device element in the global sort sequence in the d-th fault event, and get the transaction sequence number of the e-th device element in the d-th fault event; Obtain the transaction sequence number of each device element in the d-th fault event; Sort the transaction sequence number of each device element in the d-th fault event in ascending order to obtain the transaction of the d-th fault event; Acquire transactions for each failure event; The transactions of each failure event are divided according to the global sorting sequence to obtain continuous prefixes and non-contiguous prefixes in each failure event's transactions. The continuous prefix is ​​a subset of continuous device elements in the global sorting sequence, and the number of device elements in the continuous prefix is ​​greater than 1. The non-contiguous prefix is ​​a subset of device elements in the global sorting sequence, and the number of device elements in the non-contiguous prefix is ​​equal to 1. When at least two non-contiguous prefixes are adjacent in the transactions of the failure event, the at least two non-contiguous prefixes are merged into one non-contiguous prefix.

6. The multi-dimensional dynamic monitoring method for petroleum resin production according to claim 1, characterized in that, The process of determining the score of each device element in each transaction for each fault event based on the global score sorting sequence, and calculating the first final score for each consecutive prefix and the second final score for each non-consecutive prefix based on the score of each device element in each transaction for each fault event, specifically includes: The score of the f-th device element in the global score sorting sequence is determined as the score of the h-th device element in the transaction of the g-th failure event, wherein the f-th device element in the global score sorting sequence and the h-th device element in the transaction of the g-th failure event are the same device element. Get the score of each device element in the transaction for the g-th failure event; Get the score of each device element in the transaction for each failure event; Based on the score of each device element in the k-th consecutive prefix in the transaction of the j-th failure event, calculate the first final score of the k-th consecutive prefix in the transaction of the j-th failure event; Based on the score of each device element in the l-th discontinuous prefix in the transaction of the j-th failure event, calculate the second final score of the l-th discontinuous prefix in the transaction of the j-th failure event.

7. The multi-dimensional dynamic monitoring method for petroleum resin production according to claim 1, characterized in that, The calculation of the transaction priority score for each failure event based on the first final score of each consecutive prefix and the second final score of each non-consecutive prefix specifically includes: Based on continuous and non-contiguous prefixes, the transactions of the failure event are divided into transaction groups, where any transaction group in the transaction group has either a continuous prefix or a non-contiguous prefix; The frequency of each transaction group within the transactions that acquire failure events; The product of the frequency and final score of the m-th transaction group in the transaction of the j-th failure event is determined as the third final score of the m-th transaction group in the transaction of the j-th failure event. Specifically, when the m-th transaction group in the transaction of the j-th failure event is the k-th consecutive prefix in the transaction of the j-th failure event, the first final score of the k-th consecutive prefix in the transaction of the j-th failure event is determined as the final score of the m-th transaction group in the transaction of the j-th failure event; when the m-th transaction group in the transaction of the j-th failure event is the l-th non-consecutive prefix in the transaction of the j-th failure event, the second final score of the l-th non-consecutive prefix in the transaction of the j-th failure event is determined as the final score of the m-th transaction group in the transaction of the j-th failure event. Obtain the third final score of each transaction group in the transaction of the j-th failure event and sum them to get the priority score of the transaction of the j-th failure event; Get the priority score of the transaction for each failure event.

8. A multi-dimensional dynamic monitoring system for petroleum resin production, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is executed by the processor, it implements the steps of a multi-dimensional dynamic monitoring method for petroleum resin production as described in any one of claims 1-7.

Images