System for diagnosing operational faults of complex industrial chemical processes

Abstract

The application discloses a diagnosis system for operation failure of complex industrial chemical process, and relates to the technical field of intelligent fault diagnosis, comprising a rhythm acquisition module, a rhythm analysis module, a trend reconstruction module, a risk judgment module and a rhythm regulation module; the rhythm acquisition module acquires a liquid level operation time sequence, extracts a sampling time interval record from the liquid level operation time sequence, synchronously extracts a valve adjustment time interval record, and forms a sampling rhythm and adjustment rhythm comparison result according to the sampling time interval record and the valve adjustment time interval record. The application identifies the rhythm coincidence state and reconstructs the continuous rising trend of the hidden liquid level by analyzing the multiple corresponding relationship and phase difference change of the sampling time interval and the valve adjustment time interval, and forms a cumulative deviation track. Meanwhile, the rhythm coincidence is broken by the sampling misplacement deviation and the adjustment off-beat pause, so that the real change of the liquid level can be continuously presented, and the operation risk identification and dynamic control capability are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent fault diagnosis technology, and more specifically to a diagnostic system for operational faults in complex industrial chemical processes. Background Technology

[0002] A diagnostic system for operational faults in complex industrial chemical processes refers to a comprehensive technical system designed for continuous, large-scale, and multivariate coupled chemical production processes. This system performs real-time monitoring, anomaly identification, fault location, and causal analysis of equipment operation. Typically, this system revolves around multiple process stages such as reaction, separation, heat transfer, transportation, and storage, continuously collecting multidimensional operational data including temperature, pressure, flow rate, liquid level, component concentration, current, voltage, and vibration. It then combines process mechanism relationships with historical operating patterns to perform dynamic correlation analysis of the coupling changes between variables. When a variable exhibits deviation, abnormal fluctuation, or lag in response, the system identifies the anomaly propagation path and determines the fault type by tracing process topology relationships and material energy balance logic. Examples of fault types include equipment failure, sensor drift, process parameter mismatch, control loop instability, or raw material quality fluctuations. This enables early warning and accurate diagnosis of operational risks in complex chemical processes. Essentially, this system is an operational safety assurance technology system built within a high-dimensional, multi-temporal, and strongly coupled industrial environment, possessing continuous monitoring, anomaly detection, and causal analysis capabilities.

[0003] The existing technology has the following shortcomings: In existing technologies, fault diagnosis of complex industrial chemical processes typically involves acquiring liquid level data according to a fixed sampling period, while the control system adjusts the inlet and outlet valves according to a fixed control period. When the unit is operating with a continuous feed flow rate slightly higher than the discharge flow rate, the sampling period easily overlaps with the valve adjustment period, causing each sample to fall within the instantaneous equilibrium range after the control action. This results in diagnostic results exhibiting periodic fluctuations. In this situation, the true trend of a continuous, slow rise in liquid level is easily masked by these periodic fluctuations. The system struggles to identify cumulative deviations in a timely manner, easily misjudging the operation as stable, ultimately leading to a gradual rise in tank liquid level until overflow, posing a serious safety risk.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a diagnostic system for operational faults in complex industrial chemical processes, in order to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a diagnostic system for operational faults in complex industrial chemical processes, comprising a rhythm acquisition module, a rhythm analysis module, a trend reconstruction module, a risk assessment module, and a rhythm control module: The rhythm acquisition module collects the liquid level operation time series, extracts the sampling time interval record from the liquid level operation time series, and simultaneously extracts the valve adjustment time interval record. Based on the sampling time interval record and the valve adjustment time interval record, a comparison result of sampling rhythm and adjustment rhythm is formed. The rhythm analysis module analyzes the correlation between the sampling time interval and the valve adjustment time interval based on the comparison results of the sampling rhythm and the adjustment rhythm, calculates the phase difference between the sampling time interval and the valve adjustment time interval over time, and forms the rhythm coincidence judgment result based on the correlation between the multiples and the phase difference over time. The trend reconstruction module locates the position where the sampling time repeatedly falls into the short-term stable section after valve adjustment when the rhythm overlap judgment result indicates rhythm overlap. Based on the position where the sampling time repeatedly falls into the short-term stable section after valve adjustment, it reverses the continuous rising process of the liquid level being covered up, forming a cumulative offset change trajectory. The risk assessment module determines the start and end times of the continuous rise in liquid level based on the cumulative offset change trajectory, extracts the time segment corresponding to the overflow risk based on the start and end times of the continuous rise in liquid level, and forms the risk trigger result. The rhythm control module adjusts the sampling time interval based on the risk trigger result, introduces a misaligned rhythm on the basis of the original sampling time interval, and adds a misaligned pause in the valve adjustment time interval, thereby breaking the rhythm overlap between the sampling time interval and the valve adjustment time interval, restoring the real rising change process of the liquid level, and completing the dynamic control processing.

[0007] Preferably, the steps for generating the comparison results between sampling rhythm and adjustment rhythm are as follows: The liquid level operation time series is continuously organized and decomposed into time. The sampling time points corresponding to the liquid level values ​​are arranged in chronological order to form a sampling time point sequence, and the time difference between adjacent sampling time points is calculated to form a sampling time interval record sequence. The valve adjustment behavior records are processed by time extraction and interval decomposition. The valve adjustment time points are arranged in chronological order to form a valve adjustment time point sequence, and the time difference between adjacent valve adjustment time points is calculated to form a valve adjustment time interval record sequence. Under a unified time reference, the sampling time interval record sequence and the valve adjustment time interval record sequence are processed by segment-by-segment time mapping to establish a correspondence table between the sampling time interval record and the valve adjustment time interval record; A time mapping matrix is ​​formed based on the correspondence table, and the sampling time interval records and valve adjustment time interval records are integrated side by side in chronological order to form a comparison result of sampling rhythm and adjustment rhythm.

[0008] Preferably, the steps for forming the rhythm overlap determination result are as follows: The sampling time interval records are numbered consecutively on a unified time axis, and the valve adjustment time interval records within the corresponding time segment are extracted simultaneously. The time axis is divided into continuous analysis segments, and the number of sampling time interval records, the number of valve adjustment time interval records, and the total time length are statistically analyzed to determine the multiplier correspondence. Within the analysis segment corresponding to the multiple correspondence, the starting time of the sampling time interval and the starting time of the valve adjustment time interval are compared item by item, the time difference is extracted to form a phase difference sequence, and the phase difference changes over time is formed. The correlation between multiples and the change of phase difference over time are recorded side by side on a unified time axis, and candidate segments with overlapping rhythms are marked. The candidate segments of rhythm overlap are continuously integrated to form a rhythm overlap determination result that includes start time, end time, multiple correspondence, and phase difference distribution range.

[0009] Preferably, when recording the multiplier correspondence and the phase difference change over time in parallel, if the phase difference maintains a fixed time offset range within a continuous analysis segment and the multiplier correspondence remains consistent, the corresponding analysis segment is marked as a rhythm overlap candidate segment, and the rhythm overlap candidate segments that are continuous in time and have consistent multiplier correspondences are integrated into a rhythm overlap determination result.

[0010] Preferably, the cumulative offset change trajectory formation process is as follows: Extract the time points of valve adjustment actions within the time range of overlapping rhythms, divide the short-term stable section after valve adjustment, and form a segment sequence with start and end time identifiers; By comparing the sampling time points within the rhythm overlap time segment with the short-term stable segment after valve adjustment one by one, the position sequence of sampling times repeatedly falling into the short-term stable segment after valve adjustment is determined. By backtracking the time sequence of the sampling time repeatedly falling into the short-term stable section after valve adjustment, the corresponding liquid level operation time sequence is spliced ​​to form a continuous rising process of the liquid level that is masked. The liquid level values ​​during the continuous rise process where the liquid level is obscured are incrementally accumulated to construct a cumulative offset change trajectory.

[0011] Preferably, the continuous rise of the liquid level is masked by splicing the liquid level operation time sequence corresponding to the position sequence of the short-term stable section after the sampling time repeatedly falls into the valve adjustment in chronological order, and the start time and end time are determined by the continuous increasing section in the same direction. The cumulative offset change trajectory is formed by accumulating the difference between adjacent liquid level values ​​in the continuous increasing section in the same direction.

[0012] Preferably, the risk trigger result formation process is as follows: The cumulative offset change trajectory is read point by point along a unified time axis to identify continuously positive growth time points and form candidate segments for continuous liquid level rise; Boundary backtracking and backward extension positioning are performed on candidate segments of continuous liquid level rise to determine the start and end times of continuous liquid level rise, forming a set of time segments of continuous liquid level rise. The set of time segments in which the liquid level continues to rise is matched with the liquid level operation time series. The liquid level values ​​and cumulative offset values ​​are extracted and risk correlation analysis is performed to mark the overflow risk candidate segments. The candidate overflow risk segments are integrated into continuous segments to form the time segment corresponding to the overflow risk and generate the risk trigger result.

[0013] Preferably, the sampling time interval is adjusted based on the risk triggering result. A misaligned rhythm is introduced into the original sampling time interval, and a pause with a misaligned beat is added to the valve adjustment time interval to break the rhythm overlap between the sampling time interval and the valve adjustment time interval, thus restoring the true process of liquid level rise and change. The steps are as follows: On a unified time axis, read the start and end times of the time segment corresponding to the overflow risk corresponding to the risk triggering result, extract the sampling time points before the corresponding time segment to form the original sampling time interval benchmark sequence, and determine the sampling time start point. A fixed time offset is added to the original sampling time interval reference sequence to generate a new sampling time point sequence and determine the misalignment rhythm sequence. The timing of valve adjustment actions within the risk time interval is extracted to form the original valve adjustment time interval sequence, and pause time intervals are inserted into the original valve adjustment time interval sequence to form a new valve adjustment time interval sequence. The new sampling time point sequence is overlaid and compared with the new valve adjustment time interval sequence to form an alternating distribution of sampling time interval and valve adjustment time interval; The liquid level value corresponding to the new sampling time point is read along a unified time axis, and the time correspondence is performed by combining the cumulative offset change trajectory to complete the dynamic control processing.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention identifies specific time segments where the sampling and adjustment rhythms overlap by constructing a multiple correspondence between the sampling time interval and the valve adjustment time interval, as well as the phase difference over time. In these overlapping rhythm segments, it further pinpoints the location where the sampling time repeatedly falls within a short-term stable segment after valve adjustment. Furthermore, it reverse-calculates the continuous rise in liquid level that is being masked, forming a cumulative offset trajectory. This overcomes the limitation of relying solely on apparent fluctuation data for judgment, accurately reconstructing the continuous upward trend of liquid level against the backdrop of periodic fluctuations, and enhancing the ability to identify hidden risks.

[0015] After the risk triggering result is formed, this invention performs misalignment offset processing on the sampling time interval and inserts a misstep pause into the valve adjustment time interval, so that the sampling time interval and the valve adjustment time interval form an interleaved distribution relationship on the time axis, avoiding re-entry into the multiple correspondence and phase lock state, so that the real liquid level rise and change process can be continuously collected and recorded, realizing the active breaking of the rhythm overlap state and dynamic adjustment control, enhancing the safety assurance capability and risk prevention and control stability during the operation of complex industrial chemical processes. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a schematic diagram of the diagnostic system module for complex industrial chemical process operation faults according to the present invention.

[0018] Figure 2 This is a mind map of the diagnostic system for operational faults in complex industrial chemical processes according to the present invention. Detailed Implementation

[0019] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0020] This invention provides, for example Figure 1 The diagnostic system for operational faults in complex industrial chemical processes, as shown, includes a rhythm acquisition module, a rhythm analysis module, a trend reconstruction module, a risk assessment module, and a rhythm control module. The rhythm acquisition module collects the liquid level operation time series, extracts the sampling time interval record from the liquid level operation time series, and simultaneously extracts the valve adjustment time interval record. Based on the sampling time interval record and the valve adjustment time interval record, a comparison result of sampling rhythm and adjustment rhythm is formed. To establish a comparison between sampling and adjustment rhythms, the liquid level operation time series and valve adjustment behavior records were processed according to a unified time benchmark. The rhythm correlation was established through time decomposition, time mapping, and time reconstruction. The specific implementation steps are as follows: The liquid level operation time series undergoes continuous processing and time decomposition. The liquid level operation time series consists of liquid level values ​​arranged chronologically and their corresponding sampling time points. After obtaining the complete liquid level operation time series, the sampling time point corresponding to each liquid level value is extracted as an independent time marker, and these time points are arranged in chronological order to form a sampling time point sequence. After the sampling time point sequence is formed, the time difference between any two adjacent sampling time points is calculated, and each time difference is recorded as a sampling time interval record. Each sampling time interval record is assigned a start time and an end time identifier, where the start time corresponds to the previous sampling time point, and the end time corresponds to the next sampling time point. Subsequently, all sampling time interval records are arranged chronologically to form a sampling time interval record sequence, and each sampling time interval record is assigned a serial number for subsequent time segment matching. Through this process, the sampling time interval record sequence and the liquid level operation time series are within the same time coordinate system, ensuring the continuity of subsequent time comparisons.

[0021] The valve regulation behavior records undergo time extraction and interval decomposition. These records include valve opening and closing times. After obtaining complete records, all valve regulation times are arranged chronologically to form a valve regulation time point sequence. Following this sequence, the time difference between adjacent times is calculated, and each difference is recorded as a valve regulation time interval record. Each time interval record is assigned a start and end time identifier, with the start time corresponding to the preceding and end times corresponding to the following. All time interval records are then arranged chronologically to form a valve regulation time interval record sequence, and each record is assigned a serial number. After constructing the sequence, the time identifiers are mapped to the unified time axis of the aforementioned sampling time interval record sequence, placing them under the same starting point and unit of measurement, thus establishing a unified time reference.

[0022] After establishing a unified time reference, a segment-by-segment time mapping process is performed on the sampling time interval record sequence and the valve adjustment time interval record sequence. For each sampling time interval record, the corresponding time segment is determined based on its start and end times. Within the corresponding time segment, it is checked whether the valve adjustment time point is located within that time segment, and the specific positional relationship of the valve adjustment time point within the sampling time interval segment is recorded, including whether it is located at the beginning, middle, or end of the segment. Simultaneously, the same time mapping process is performed on each valve adjustment time interval record to determine the number and positional order of the sampling time points contained within the valve adjustment time interval segment. Through the above bidirectional time mapping process, a correspondence table between the sampling time interval records and the valve adjustment time interval records is established. This table clearly identifies the valve adjustment time interval record number corresponding to each sampling time interval record, as well as the time position difference between the two. After completing the correspondence table construction, all correspondences are arranged in chronological order to form a time mapping matrix, presenting the relative temporal positional relationship between the sampling time interval records and the valve adjustment time interval records in a continuous structural form.

[0023] After the time mapping matrix is ​​formed, based on the correspondence between each set of sampling time interval records and valve adjustment time interval records in the time mapping matrix, they are parallelized and integrated in chronological order to form a rhythm comparison data set containing the sampling time interval record number, valve adjustment time interval record number, sampling time interval start and end time, valve adjustment time interval start and end time, time segment number, and time position difference. Through this rhythm comparison data set, the sampling time interval records and valve adjustment time interval records are presented in a structured manner within a unified time framework, allowing the time distribution relationship between the sampling rhythm and the adjustment rhythm to be represented in units of clearly defined time segments. At this point, the comparison results between the sampling rhythm and the adjustment rhythm are formed, providing complete time-related basic data for subsequent analysis of multiple correspondences and calculation of phase difference changes over time.

[0024] The rhythm analysis module analyzes the correlation between the sampling time interval and the valve adjustment time interval based on the comparison results of the sampling rhythm and the adjustment rhythm, calculates the phase difference between the sampling time interval and the valve adjustment time interval over time, and forms the rhythm coincidence judgment result based on the correlation between the multiples and the phase difference over time. Based on the established comparison results of sampling rhythm and adjustment rhythm and the completion of unified time axis alignment, the time ratio and relative time position relationship between the sampling time interval and the valve adjustment time interval are continuously expanded. Rhythm overlap determination results are generated through time window decomposition, ratio structure identification, time offset tracking, and segment merging. The specific implementation steps are as follows: Sampling time interval records are sequentially numbered and arranged chronologically on a unified time axis. Valve adjustment time interval records within corresponding time segments are extracted simultaneously. The time axis is divided into several continuous analysis segments, each containing several consecutive sampling time interval records and valve adjustment time interval records within the same time range. Within each analysis segment, the number of sampling time interval records and the number of valve adjustment time interval records within the same time segment are counted. The total time length of all sampling time interval records and the total time length of all valve adjustment time interval records within the analysis segment are also calculated. Subsequently, the number of sampling time interval records and the number of valve adjustment time interval records are arranged in a corresponding order, and the total time length of sampling time intervals is compared with the total time length of valve adjustment time intervals. When the number of sampling time interval records and the number of valve adjustment time interval records exhibit a fixed integer ratio across multiple consecutive analysis segments, and the total time length of sampling time intervals and the total time length of valve adjustment time intervals exhibit a stable proportional structure, this proportional structure is defined as a multiple correspondence. A time segment number identifier is generated for this multiple correspondence, ensuring a continuous distribution of the multiple correspondence on the time axis.

[0025] Within the analysis segment where the multiplier correspondence is determined, the starting positions of the sampling time interval and the valve adjustment time interval are compared item by item. The starting time position of each sampling time interval record is marked on a unified time axis, and the starting time position of the corresponding valve adjustment time interval record is also marked. By measuring the time difference between the starting time of the sampling time interval and the starting time of the most recent valve adjustment time interval, the phase difference value corresponding to that sampling time interval is formed. Subsequently, as the time axis moves forward, the above time difference extraction process is repeated for each subsequent sampling time interval record, so that the phase difference values ​​form a continuous phase difference sequence on the time axis. After the phase difference sequence is formed, the phase difference sequence is numbered and organized in chronological order, so that each phase difference value is associated with the corresponding analysis segment number, thereby forming a continuous time trajectory of the phase difference changing over time. In this process, the time segment number of the multiplier correspondence is consistent with the time number of the phase difference sequence, so that the multiplier correspondence and the phase difference changing over time can be jointly analyzed in the same time coordinate system.

[0026] After obtaining the multiplier correspondence and the phase difference change over time, the two sets of time information are superimposed. On a unified time axis, the multiplier correspondence identifier and the phase difference value within each analysis segment are recorded side by side, forming a joint time series of the multiplier correspondence and phase difference change over time. In the joint time series, the trajectory of the phase difference value within multiple consecutive analysis segments is observed. When the phase difference value remains within a fixed time offset range within multiple consecutive analysis segments, and the multiplier correspondence remains unchanged within the corresponding time segment, the time segment is recorded as a rhythm locking candidate segment. At the same time, when the phase difference value exhibits a periodic repetition on the time axis, and the periodic repetition is consistent with the proportional structure of the multiplier correspondence, the time segment is further marked as a rhythm overlap candidate segment. In the above marking process, each candidate segment includes a clear start time, end time, multiplier ratio number, and phase difference value range.

[0027] After all candidate segments are marked, consecutive adjacent rhythm-overlapping candidate segments are merged. Segments with continuous time and consistent multiple correspondences are integrated into complete rhythm-overlapping time segments, and a rhythm-overlapping identifier is generated for each complete rhythm-overlapping time segment. After the rhythm-overlapping identifiers are formed, the start and end times of the rhythm-overlapping time segments are organized, and the corresponding multiple correspondences and the time distribution range of phase difference changes over time are recorded to form a structured rhythm-overlapping determination result data set. Through this rhythm-overlapping determination result data set, it can be clearly pointed out that in which time segment the sampling time interval and the valve adjustment time interval are in multiple correspondence and the phase difference remains stable, thereby determining the interval of rhythm-overlapping state occurrence in the time dimension. This provides a complete time determination basis for subsequent positioning of sampling times that repeatedly fall into short-term stable segments after valve adjustment and for estimating the continuous rise process of the liquid level being masked.

[0028] The trend reconstruction module locates the position where the sampling time repeatedly falls into the short-term stable section after valve adjustment when the rhythm overlap judgment result indicates rhythm overlap. Based on the position where the sampling time repeatedly falls into the short-term stable section after valve adjustment, it reverses the continuous rising process of the liquid level being covered up, forming a cumulative offset change trajectory. After the rhythm overlap determination result clearly defines the rhythm overlap time segment, the temporal relationship between the sampling time point and the valve adjustment behavior within the rhythm overlap time segment is processed layer by layer. The cumulative offset change trajectory is formed by precise definition of stable segment, spatial positioning of sampling time, reverse expansion of time segment, and continuous incremental accumulation. The specific implementation steps are as follows: Within the overlapping time range, all valve adjustment actions are extracted in chronological order, and each valve adjustment action is marked on a unified time axis. This time point serves as the starting point for dividing continuous time segments. During the division process, starting from the valve adjustment action, subsequent liquid level values ​​and their corresponding time points are read sequentially from the liquid level operation time sequence. The change between adjacent liquid level values ​​is calculated, and the direction and magnitude of the liquid level value change on the time axis are continuously observed. When the change between multiple consecutive liquid level values ​​remains within a preset fixed range and the direction of change does not reverse, the time segment from the valve adjustment action to the point where the trend changes is reversed is determined as the short-term stable segment after valve adjustment. The above division process is performed for each valve adjustment action, forming multiple short-term stable segments after valve adjustment with clear start and end time identifiers, which are then numbered and arranged in chronological order.

[0029] After numbering the short-term stable sections following valve adjustment, all sampling time points within the overlapping time intervals are compared one by one with each short-term stable section following valve adjustment. On a unified time axis, it is determined whether each sampling time point falls between the start and end times of a short-term stable section following valve adjustment. When a sampling time point is within the time range of a short-term stable section following valve adjustment, it is recorded as the sampling time of the stable section and associated with the corresponding short-term stable section number. Subsequently, the sampling times of the stable sections are arranged in chronological order. When multiple consecutive sampling time points correspond to consecutively numbered short-term stable sections following valve adjustment, the sequence of consecutive sampling time points is determined as the position sequence in which the sampling time repeatedly falls within the short-term stable section following valve adjustment. The consecutive start and end times are marked for this position sequence, so that the positions in which the sampling time repeatedly falls within the short-term stable section following valve adjustment form a complete interval identifier on the time axis.

[0030] After determining the sequence of locations where sampling times repeatedly fall within short-term stable segments following valve adjustments, the corresponding liquid level operation time series is processed in reverse. Taking the sampling time of each stable segment as the starting point, the time is traced back to the time interval between the corresponding valve adjustment action. All liquid level values ​​and their corresponding time points within this time interval are read and arranged in chronological order to form a single-segment liquid level change data set. Subsequently, the single-segment liquid level change data sets corresponding to multiple consecutive stable segment sampling times are spliced ​​together in chronological order to form a continuous liquid level change data set covering the entire overlapping time interval. After splicing, the continuous liquid level change data set is analyzed line by line to identify time intervals where the liquid level value increases continuously in the same direction on the time axis. This continuous upward time interval is defined as the masked continuous rise process of the liquid level, and the start and end times of this continuous rise process are recorded to give the masked continuous rise process of the liquid level a clear time boundary.

[0031] After determining the continuous rise of the concealed liquid level, the liquid level values ​​within the continuous rise process are incrementally accumulated item by item. The positive differences between adjacent liquid level values ​​are superimposed in chronological order to form a cumulative offset value sequence, and each cumulative offset value is associated with its corresponding time point. After the cumulative offset value sequence is formed, the cumulative offset values ​​are arranged in chronological order to construct a continuous data trajectory of cumulative offset changes over time. This trajectory can fully reflect the degree of continuous rise of the concealed liquid level within the overlapping time intervals. Through the above process, the location of the short-term stable section after the sampling time repeatedly falls into the valve adjustment is accurately located. The continuous rise of the concealed liquid level is completely reconstructed through time backtracking and segment splicing, ultimately forming a cumulative offset change trajectory with time start and end markers and incremental accumulation relationships. This provides a continuous and structured time data foundation for the subsequent extraction of the time interval corresponding to overflow risk.

[0032] The risk assessment module determines the start and end times of the continuous rise in liquid level based on the cumulative offset change trajectory, extracts the time segment corresponding to the overflow risk based on the start and end times of the continuous rise in liquid level, and forms the risk trigger result. After the cumulative offset change trajectory has been continuously arranged in chronological order, the cumulative offset change trajectory is segmented and analyzed and time-based. The risk triggering result is formed through continuous incremental identification, boundary backtracking positioning, risk segment extraction, and risk triggering information generation. The specific implementation steps are as follows: The cumulative offset change trajectory is read point by point along a unified time axis, and the cumulative offset value corresponding to each time point is recorded. The cumulative offset value of the current time point is compared with the cumulative offset value of the previous time point. When the cumulative offset value of the current time point is greater than the cumulative offset value of the previous time point, the time point is marked as a positive growth time point. When positive growth time points appear consecutively on the time axis, the time segment covered by the consecutive positive growth time points is determined as a candidate segment for continuous liquid level rise. During the candidate segment identification process, the cumulative offset change trajectory is fully traversed, all consecutive positive growth segments are recorded in sequence, and a candidate segment number is assigned to each consecutive positive growth segment. At the same time, the initial value of the start time and the initial value of the end time of the candidate segment are recorded, so that each candidate segment has a clear interval identification on the time axis.

[0033] Based on the formation of candidate segments for continuous liquid level rise, the initial value of the starting time of each candidate segment is processed backward. Starting from the time point corresponding to the initial value of the starting time, historical time points in the cumulative offset change trajectory are read point by point backward to find the turning point when the cumulative offset value changes from a state of no increase to a state of continuous increase. This turning point is determined as the starting time of continuous liquid level rise. The initial value of the ending time of the candidate segment is processed backward. Starting from the time point corresponding to the initial value of the ending time, subsequent time points in the cumulative offset change trajectory are read point by point backward to find the turning point when the cumulative offset value changes from a state of continuous increase to a state of no increase. This turning point is determined as the ending time of continuous liquid level rise. After completing the above backtracking and extension processing, a clear liquid level rise time segment is formed for each candidate segment, and the liquid level rise time segments are arranged in chronological order to form a set of liquid level rise time segments.

[0034] After the set of time segments with continuously rising liquid levels is formed, a risk correlation analysis is performed on each time segment. The changes in liquid level values ​​for the corresponding time segments with continuously rising liquid levels are read on a unified time axis, and the starting and ending values ​​of the liquid level values ​​within the time segment are recorded. At the same time, the cumulative offset value corresponding to the end time of the time segment is read. The ending value of the liquid level value is compared item by item with the preset upper limit operating standard. When the ending value of the liquid level value reaches the preset upper limit operating standard, or the cumulative offset value reaches the preset upper limit standard, the time segment with continuously rising liquid levels is marked as a candidate segment for overflow risk. During the marking process, the starting time of the continuous rise in liquid level, the ending time of the continuous rise in liquid level, the starting value of the cumulative offset value, and the ending value of the cumulative offset value are recorded for each candidate segment for overflow risk, so that the candidate segment for overflow risk has a complete numerical description on the time axis.

[0035] After all overflow risk candidate segments are marked, overflow risk candidate segments that are continuous in time and have the same risk status are merged. Candidate segments that are consecutive in time are integrated into complete overflow risk time segments, and a risk trigger identifier is generated for each complete overflow risk time segment. After the risk trigger identifier is generated, a risk trigger result dataset is constructed. The risk trigger result dataset includes the start time of the overflow risk time segment, the end time of the overflow risk time segment, the range of liquid level changes within the time segment, and the range of cumulative offset changes, so that the risk trigger results form a complete closed interval in the time dimension. Through the above processing, based on the cumulative offset change trajectory, not only are the start and end times of the continuous rise in liquid level accurately determined, but also the time segments related to overflow risk are extracted, forming risk trigger results with time boundaries and numerical basis, providing a clear time reference for subsequent sampling time interval adjustment and rhythm misalignment handling.

[0036] The rhythm control module adjusts the sampling time interval based on the risk trigger result, introduces a misaligned rhythm on the basis of the original sampling time interval, and adds a misaligned pause in the valve adjustment time interval, thereby breaking the rhythm overlap between the sampling time interval and the valve adjustment time interval, restoring the real rise and change process of the liquid level, and completing the dynamic control processing. After the risk triggering results have given the start and end times of the overflow risk corresponding to the time segment, the sampling time interval and the valve adjustment time interval are collaboratively reconstructed. Through time base extraction, sampling time interval misalignment construction, valve adjustment time interval misstep pause implantation, and rhythm stagger solidification process, the overlapping rhythm state is resolved and the actual liquid level rise process is restored. The specific implementation steps are as follows: On a unified time axis, read the start and end times of the overflow risk corresponding to the time segment of the risk trigger result, and extract the sampling time points within the three consecutive complete sampling periods before this time segment. Calculate the time difference between adjacent sampling time points to form the original sampling time interval benchmark sequence. After the original sampling time interval benchmark sequence is formed, count the value of each sampling time interval in the benchmark sequence and arrange them in chronological order to form a sampling time interval benchmark table. Use this sampling time interval benchmark table as the reference time basis for the subsequent misalignment offset rhythm construction. Re-determine the sampling time starting point at the start time of the risk time segment, and generate a new sampling time point sequence based on this sampling time starting point.

[0037] During the generation of the new sampling time point sequence, a fixed time offset is added to each time interval value of the original sampling time interval benchmark sequence, so that the new sampling time point is shifted forward by a fixed time offset relative to the original sampling time point on the time axis. After the fixed time offset is inserted, multiple consecutive sampling time points are time-arranged so that the new sampling time point sequence presents a different time distribution position on a unified time axis than the original sampling time point sequence. After the new sampling time point sequence is formed, the new sampling time point is compared with the valve adjustment action time point within the risk time segment one by one, and the time difference between the new sampling time point and the valve adjustment action time point is recorded. It is observed whether the time difference maintains a fixed proportional structure within the continuous sampling period. When the time difference no longer presents a fixed proportional structure, the current sampling time point sequence is determined as a misaligned offset rhythm sequence.

[0038] After determining the sampling time interval misalignment rhythm sequence, the valve adjustment time interval records within the risk time segment are processed by inserting a pause. The time point of each valve adjustment action within the risk time segment is extracted on a unified time axis, and the time difference between two adjacent valve adjustment action time points is calculated to form the original valve adjustment time interval sequence. A pause time segment is inserted into a fixed time segment within the original valve adjustment time interval sequence. The pause time segment length is a preset fixed time value. After inserting the pause time segment, the subsequent valve adjustment action time points are extended sequentially, creating a beat misalignment structure on the time axis for the new valve adjustment time interval sequence. After the new valve adjustment time interval sequence is formed, it is overlaid and compared with the new sampling time point sequence, creating an interleaved distribution relationship between the sampling time points and the valve adjustment action time points on the time axis.

[0039] After the sampling time interval misalignment and valve adjustment time interval misalignment occur simultaneously, the liquid level value corresponding to the new sampling time point is read along a unified time axis, and a new liquid level operation time series is formed in chronological order. The new liquid level operation time series is matched with the cumulative offset change trajectory formed before the risk trigger result, so that the new liquid level operation time series covers the time segment outside the short-term stable segment after the original valve adjustment. After the new sampling time point covers the liquid level change time segment that was not recorded before, the change trend of the liquid level value on the time axis can continuously reflect the continuous rise of the liquid level, so that the real rise and change of the liquid level can be fully presented in the time dimension.

[0040] After the actual rise and change of the liquid level is continuously recorded, the new sampling time interval sequence and the new valve adjustment time interval sequence are fixed as the time control benchmark for the subsequent operation stage. The overflow risk corresponding time segment in the risk trigger result is associated with the current time control benchmark and stored. When the cumulative offset change trajectory enters the continuous growth segment again in the subsequent operation, the sampling time interval misalignment offset and valve adjustment time interval misalignment pause processing is directly executed according to the current time control benchmark. This ensures that the sampling time interval and valve adjustment time interval are continuously kept in a non-overlapping distribution state on the time axis, thereby maintaining a complete record of the liquid level change process during dynamic operation, completing dynamic control processing and preventing the rhythm overlap state from forming again.

[0041] This invention identifies specific time segments where the sampling and adjustment rhythms overlap by constructing a multiple correspondence between the sampling time interval and the valve adjustment time interval, as well as the phase difference over time. In these overlapping rhythm segments, it further pinpoints the location where the sampling time repeatedly falls within a short-term stable segment after valve adjustment. Furthermore, it reverse-calculates the continuous rise in liquid level that is being masked, forming a cumulative offset trajectory. This overcomes the limitation of relying solely on apparent fluctuation data for judgment, accurately reconstructing the continuous upward trend of liquid level against the backdrop of periodic fluctuations, and enhancing the ability to identify hidden risks.

[0042] After the risk triggering result is formed, this invention performs misalignment offset processing on the sampling time interval and inserts a misstep pause into the valve adjustment time interval, so that the sampling time interval and the valve adjustment time interval form an interleaved distribution relationship on the time axis, avoiding re-entry into the multiple correspondence and phase lock state, so that the real liquid level rise and change process can be continuously collected and recorded, realizing the active breaking of the rhythm overlap state and dynamic adjustment control, enhancing the safety assurance capability and risk prevention and control stability during the operation of complex industrial chemical processes.

[0043] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A diagnostic system for operational faults in complex industrial chemical processes, characterized in that, It includes a rhythm acquisition module, a rhythm analysis module, a trend reconstruction module, a risk assessment module, and a rhythm control module: The rhythm acquisition module collects the liquid level operation time series, extracts the sampling time interval record from the liquid level operation time series, and simultaneously extracts the valve adjustment time interval record. Based on the sampling time interval record and the valve adjustment time interval record, a comparison result of sampling rhythm and adjustment rhythm is formed. The rhythm analysis module analyzes the correlation between the sampling time interval and the valve adjustment time interval based on the comparison results of the sampling rhythm and the adjustment rhythm, calculates the phase difference between the sampling time interval and the valve adjustment time interval over time, and forms the rhythm coincidence judgment result based on the correlation between the multiples and the phase difference over time. The trend reconstruction module locates the position where the sampling time repeatedly falls into the short-term stable section after valve adjustment when the rhythm overlap judgment result indicates rhythm overlap. Based on the position where the sampling time repeatedly falls into the short-term stable section after valve adjustment, it reverses the continuous rising process of the liquid level being covered up, forming a cumulative offset change trajectory. The risk assessment module determines the start and end times of the continuous rise in liquid level based on the cumulative offset change trajectory, extracts the time segment corresponding to the overflow risk based on the start and end times of the continuous rise in liquid level, and forms the risk trigger result. The rhythm control module adjusts the sampling time interval based on the risk trigger result, introduces a misaligned rhythm on the basis of the original sampling time interval, and adds a misaligned pause in the valve adjustment time interval, thereby breaking the rhythm overlap between the sampling time interval and the valve adjustment time interval and restoring the true rising and changing process of the liquid level.

2. The diagnostic system for operational faults in complex industrial chemical processes according to claim 1, characterized in that, The steps for generating the comparison results between sampling rhythm and adjustment rhythm are as follows: The liquid level operation time series is continuously organized and decomposed into time. The sampling time points corresponding to the liquid level values ​​are arranged in chronological order to form a sampling time point sequence, and the time difference between adjacent sampling time points is calculated to form a sampling time interval record sequence. The valve adjustment behavior records are processed by time extraction and interval decomposition. The valve adjustment time points are arranged in chronological order to form a valve adjustment time point sequence, and the time difference between adjacent valve adjustment time points is calculated to form a valve adjustment time interval record sequence. Under a unified time reference, the sampling time interval record sequence and the valve adjustment time interval record sequence are processed by segment-by-segment time mapping to establish a correspondence table between the sampling time interval record and the valve adjustment time interval record; A time mapping matrix is ​​formed based on the correspondence table, and the sampling time interval records and valve adjustment time interval records are integrated side by side in chronological order to form a comparison result of sampling rhythm and adjustment rhythm.

3. The diagnostic system for operational faults in complex industrial chemical processes according to claim 2, characterized in that, The steps for determining the rhythm overlap result are as follows: The sampling time interval records are numbered consecutively on a unified time axis, and the valve adjustment time interval records within the corresponding time segment are extracted simultaneously. The time axis is divided into continuous analysis segments, and the number of sampling time interval records, the number of valve adjustment time interval records, and the total time length are statistically analyzed to determine the multiplier correspondence. Within the analysis segment corresponding to the multiple correspondence, the starting time of the sampling time interval and the starting time of the valve adjustment time interval are compared item by item, the time difference is extracted to form a phase difference sequence, and the phase difference changes over time is formed. The correlation between multiples and the change of phase difference over time are recorded side by side on a unified time axis, and candidate segments with overlapping rhythms are marked. The candidate segments with overlapping rhythms are integrated into continuous segments to form the result of rhythm overlap determination.

4. The diagnostic system for operational faults in complex industrial chemical processes according to claim 3, characterized in that, When recording the correlation between multiples and the change of phase difference over time, if the phase difference maintains a fixed time offset range within a continuous analysis segment and the correlation between multiples remains consistent, the corresponding analysis segment is marked as a candidate segment for rhythm overlap. The candidate segments for rhythm overlap that are continuous in time and have consistent correlation between multiples are integrated into the rhythm overlap determination result.

5. The diagnostic system for operational faults in complex industrial chemical processes according to claim 3, characterized in that, The cumulative offset trajectory formation process is as follows: Extract the time points of valve adjustment actions within the time range of overlapping rhythms, divide the short-term stable section after valve adjustment, and form a segment sequence with start and end time identifiers; By comparing the sampling time points within the rhythm overlap time segment with the short-term stable segment after valve adjustment one by one, the position sequence of sampling times repeatedly falling into the short-term stable segment after valve adjustment is determined. By backtracking the time sequence of the sampling time repeatedly falling into the short-term stable section after valve adjustment, the corresponding liquid level operation time sequence is spliced ​​to form a continuous rising process of the liquid level that is masked. The liquid level values ​​during the continuous rise process where the liquid level is obscured are incrementally accumulated to construct a cumulative offset change trajectory.

6. The diagnostic system for operational faults in complex industrial chemical processes according to claim 5, characterized in that, The continuous rise of the liquid level, which is masked, is formed by splicing together the liquid level operation time sequence corresponding to the position sequence of the short-term stable section after the sampling time repeatedly falls into the valve adjustment in chronological order. The start time and end time are determined by the continuous increasing section in the same direction. The cumulative offset change trajectory is formed by accumulating the difference between adjacent liquid level values ​​in the continuous increasing section in the same direction.

7. The diagnostic system for operational faults in complex industrial chemical processes according to claim 5, characterized in that, The risk trigger result formation process is as follows: The cumulative offset change trajectory is read point by point along a unified time axis to identify continuously positive growth time points and form candidate segments for continuous liquid level rise; Boundary backtracking and backward extension positioning are performed on candidate segments of continuous liquid level rise to determine the start and end times of continuous liquid level rise, forming a set of time segments of continuous liquid level rise. The set of time segments in which the liquid level continues to rise is matched with the liquid level operation time series. The liquid level values ​​and cumulative offset values ​​are extracted and risk correlation analysis is performed to mark the overflow risk candidate segments. The candidate overflow risk segments are integrated into continuous segments to form the time segment corresponding to the overflow risk and generate the risk trigger result.

8. The diagnostic system for operational faults in complex industrial chemical processes according to claim 7, characterized in that, The sampling time interval is adjusted based on the risk triggering results. A staggered rhythm is introduced into the original sampling time interval, and a pause with a misaligned beat is added to the valve adjustment time interval. This breaks the rhythm overlap between the sampling time interval and the valve adjustment time interval, restoring the true process of liquid level rise and change. The steps are as follows: On a unified time axis, read the start and end times of the overflow risk corresponding to the time segment of the risk trigger result, extract the sampling time points before the corresponding time segment to form the original sampling time interval benchmark sequence, and determine the sampling time start point. A fixed time offset is added to the original sampling time interval reference sequence to generate a new sampling time point sequence and determine the misalignment rhythm sequence. The timing of valve adjustment actions within the risk time interval is extracted to form the original valve adjustment time interval sequence, and pause time intervals are inserted into the original valve adjustment time interval sequence to form a new valve adjustment time interval sequence. The new sampling time point sequence is overlaid and compared with the new valve adjustment time interval sequence to form an alternating distribution of sampling time interval and valve adjustment time interval; The liquid level value corresponding to the new sampling time point is read along a unified time axis, and the time correspondence is performed by combining the cumulative offset change trajectory to complete the dynamic control processing.

Images