A method, device, and medium for early warning of leak rate in vacuum equipment based on pressure monitoring.

Abstract

This invention discloses a method, device, and medium for leak rate early warning of vacuum equipment based on pressure monitoring, relating to the field of vacuum monitoring technology. The method includes: performing stable segment screening to determine a short-term pressure-holding isolation window from the original pressure monitoring sequence, and performing time-series alignment to generate a detection window sequence; implementing pressure-holding isolation control on the detection window sequence and performing high-frequency pressure sampling to generate a window pressure recovery curve; based on the window pressure recovery curve, segmenting and reconstructing the detection window sequence using a window pressure return fingerprint reconstruction algorithm, and extracting pressure return behavior features for fingerprint encoding to generate a window leakage fingerprint sequence; based on the window leakage fingerprint sequence, constructing a window progressive risk sequence by quantifying leakage fingerprint differences, and performing trend extrapolation to generate future out-of-tolerance risk information. This invention, by constructing a short-term pressure-holding isolation window and combining it with pressure return enhanced sampling, makes the minute leakage response masked by continuous pumping visible.

Description

Technical Field

[0001] This invention relates to the field of vacuum monitoring technology, and in particular to a method, device and medium for early warning of leak rate in vacuum equipment based on pressure monitoring. Background Technology

[0002] With the continuous development of photovoltaic manufacturing, heat treatment, and industrial processes relying on controlled vacuum environments, vacuum equipment faces increasingly higher requirements for the stability of chamber vacuum levels during production operations. Vacuum fluctuations often directly affect the consistency of the reaction environment, the stability of process execution, and the controllability of product batch quality. Current mainstream equipment typically features vacuum level over-limit detection capabilities, using pressure sensors to monitor pressure changes within the reaction chamber. When the detected value reaches predetermined alarm conditions, an alarm is triggered to the operator, allowing for appropriate handling of the machine's status. Using pressure signals as the core monitoring target, this method is relatively direct and highly compatible with equipment control systems, thus becoming a common means of monitoring vacuum equipment operation.

[0003] However, existing technologies still have significant shortcomings. On the one hand, current solutions generally focus on triggering alarms only when the vacuum level is abnormal or the leakage rate exceeds acceptable limits. While this can alert the equipment to malfunctions after problems appear, it often fails to prevent losses from process failures, batch defects, or rework that have already occurred. On the other hand, existing methods lack mechanisms for continuous preprocessing of cavity operation monitoring data, stable segment screening, short-term pressure holding isolation observation, extraction of window pressure recovery behavior, and progressive risk analysis between windows. As a result, while the equipment has alarm capabilities, it lacks proactive early warning capabilities, making it difficult to meet the requirements for early inspection and handling and prevention of fault escalation in continuous production scenarios. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for early warning of leakage rate in vacuum equipment based on pressure monitoring to solve the problems of delayed technical early warning and difficulty in proactively identifying out-of-tolerance risks.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a method for early warning of leakage rate in vacuum equipment based on pressure monitoring, comprising: collecting cavity operation monitoring data and preprocessing it to generate an original pressure monitoring sequence; performing stable segment screening on the original pressure monitoring sequence to determine a short-term pressure holding isolation window, and performing time sequence alignment to generate a detection window sequence; performing pressure holding isolation control on the detection window sequence and performing high-frequency pressure sampling to generate a window pressure recovery curve; based on the window pressure recovery curve, performing segmented reconstruction of the detection window sequence through a window pressure return fingerprint reconstruction algorithm, and extracting pressure return behavior features for fingerprint encoding to generate a window leakage fingerprint sequence; based on the window leakage fingerprint sequence, constructing a window progressive risk sequence by quantifying leakage fingerprint differences, and performing trend extrapolation to generate future out-of-tolerance risk information.

[0008] As a preferred embodiment of the vacuum equipment leak rate early warning method based on pressure monitoring according to the present invention, the cavity operation monitoring data includes cavity pressure data, pressure sampling timing information and equipment operation status information;

[0009] The preprocessing includes time alignment, outlier removal, and sorting.

[0010] As a preferred embodiment of the vacuum equipment leak rate early warning method based on pressure monitoring described in this invention, the steps of performing stable segment screening to determine short-term pressure holding isolation windows on the original pressure monitoring sequence and performing time sequence alignment to generate a detection window sequence are as follows:

[0011] The pressure slowly varying entropy valley capture algorithm is used to calculate the pressure slowly varying entropy value based on the distribution of adjacent pressure changes in the original pressure monitoring sequence, and to perform distribution analysis to generate a pressure entropy valley state map.

[0012] The pressure entropy valley state map is synchronously coupled with the original pressure monitoring sequence to extract local areas with continuous and consistent operating states and determine adjacent segments before and after, thereby generating candidate isolation windows.

[0013] The boundary sampling records of adjacent segments before and after the candidate isolation window are gradually stripped to retain segments with continuous and consistent operating status. Short-term pressure holding isolation windows are obtained, and the central local time period is extracted for location anchoring and sequential splicing to generate a detection window sequence.

[0014] As a preferred embodiment of the vacuum equipment leak rate early warning method based on pressure monitoring described in this invention, the step of performing pressure holding isolation control on the detection window sequence refers to performing time axis mapping on the detection window sequence in the original pressure monitoring sequence to form window anchoring information, and performing short-term pressure holding isolation control to obtain isolation response information.

[0015] As a preferred embodiment of the vacuum equipment leak rate early warning method based on pressure monitoring described in this invention, the specific steps for generating the window pressure recovery curve are as follows:

[0016] Based on the isolation response information, enhanced sampling is performed using a pressure-back enhanced sampling algorithm to obtain window pressure-back sampling data sets;

[0017] Window aggregation and time-series processing are performed on the window pressure return sampling data group to generate the window pressure recovery curve.

[0018] As a preferred embodiment of the vacuum equipment leak rate early warning method based on pressure monitoring described in this invention, the step of segmenting and reconstructing the detection window sequence based on the window pressure recovery curve using a window pressure return fingerprint reconstruction algorithm, and extracting pressure return behavior features for fingerprint encoding to generate a window leakage fingerprint sequence, is as follows:

[0019] The pressure recovery curve of the window is split into pressure return trajectories to obtain segmented pressure return trajectories, and the segmented pressure return trajectories are further divided into trajectory segments to obtain pressure return trajectory segment groups;

[0020] A pressure back leakage behavior mapping algorithm is used to analyze the pressure back behavior of the pressure back trajectory segment group, and the pressure back behavior feature value is calculated to obtain the pressure back behavior feature group;

[0021] The characteristic groups of pressure return behavior are combined, characterized, and sequentially encoded to generate a window leakage fingerprint sequence.

[0022] As a preferred embodiment of the vacuum equipment leak rate early warning method based on pressure monitoring described in this invention, the step of constructing a window-progressive risk sequence based on the window leak fingerprint sequence by quantifying the differences in leak fingerprints includes the following specific steps:

[0023] The leakage fingerprints of adjacent detection windows are compared one by one, and the leakage fingerprint difference index is calculated.

[0024] By using a progressive risk aggregation method, the differential indicators of each leakage fingerprint are sequentially accumulated and correlated to construct a windowed progressive risk sequence.

[0025] As a preferred embodiment of the vacuum equipment leakage rate early warning method based on pressure monitoring described in this invention, the future deviation risk information is generated by sequential analysis and trend extrapolation of the window progressive risk sequence to generate a risk extrapolation trajectory and risk extension identification.

[0026] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the vacuum equipment leak rate early warning method based on pressure monitoring as described in the first aspect of the present invention.

[0027] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the vacuum equipment leak rate early warning method based on pressure monitoring as described in the first aspect of the present invention.

[0028] The beneficial effects of this invention are as follows: by constructing a short-term pressure-holding isolation window and combining it with pressure-back sensitization sampling, the micro-leakage response masked by continuous evacuation can be made visible; by window pressure-back fingerprint reconstruction and progressive risk extrapolation, the discrete pressure recovery behavior is transformed into comparable and predictable leakage evolution information, thereby enabling early identification of future leakage rate deviation risks and improving the foresight and accuracy of early warning. Attached Figure Description

[0029] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. 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.

[0030] Figure 1 This is a flowchart of a method for early warning of leak rate in vacuum equipment based on pressure monitoring.

[0031] Figure 2 This is a flowchart for generating the detection window sequence.

[0032] Figure 3 A flowchart for generating a fingerprint sequence for window leakage.

[0033] Figure 4 A flowchart for generating future out-of-tolerance risk information. Detailed Implementation

[0034] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0035] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0036] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0037] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a method for early warning of leakage rate in vacuum equipment based on pressure monitoring, comprising the following steps:

[0038] S1: Collect cavity operation monitoring data and generate raw pressure monitoring sequences through preprocessing.

[0039] S1.1: Cavity operation monitoring data includes cavity pressure data, pressure sampling timing information, and equipment operation status information.

[0040] Specifically, during equipment operation, the pressure inside the reaction chamber is continuously sampled by a pressure sensor installed on the reaction chamber, and the pressure value obtained from each sample is recorded one by one in the order of sampling to form chamber pressure data.

[0041] The time position and sampling sequence of the sampling time corresponding to the cavity pressure data are recorded synchronously, so that each continuous pressure data of the reaction cavity corresponds to the pressure sampling time sequence information composed of the sampling time and sampling sequence.

[0042] According to the sampling time corresponding to each pressure sampling time sequence information, the equipment operation status record at the sampling time is read from the equipment control terminal, and the read equipment operation status record is matched with the corresponding cavity pressure data and pressure sampling time sequence information one by one to form equipment operation status information.

[0043] S1.2: Preprocessing includes time alignment, outlier removal, and sorting.

[0044] Specifically, based on each sampling time in the pressure sampling time sequence information, the cavity pressure data and equipment operating status information are matched one by one according to the same sampling time; when the recording position of the equipment operating status information is inconsistent with the corresponding sampling time, the equipment operating status information is adjusted to the position of the corresponding sampling time, so that the cavity pressure data, pressure sampling time sequence information and equipment operating status information correspond one-to-one at the same sampling time.

[0045] According to the order of pressure sampling time sequence information, compare the changes between the current pressure value and the adjacent pressure values. If the current pressure value has an isolated abrupt change (e.g., the previous pressure value is 100, the current pressure value suddenly changes to 160, and the next pressure value drops back to 102) and the equipment operating status information at the corresponding time does not change, the record corresponding to the current pressure value is judged as an abnormal record, and the corresponding cavity pressure data, pressure sampling time sequence information and equipment operating status information are deleted together.

[0046] According to the sampling sequence in the pressure sampling time sequence information, the retained cavity pressure data, pressure sampling time sequence information and equipment operating status information are rearranged, and the correspondence at the same sampling time is checked to see if it is complete, and the original pressure monitoring sequence is generated.

[0047] S2: Perform stable segment screening on the original pressure monitoring sequence to determine the short-term pressure holding isolation window, and perform time sequence alignment to generate the detection window sequence.

[0048] S2.1: The pressure slowly varying entropy valley capture algorithm is used to calculate the pressure slowly varying entropy value based on the distribution of adjacent pressure changes in the original pressure monitoring sequence, and to perform distribution analysis to generate a pressure entropy valley state map.

[0049] Specifically, according to the order of the pressure sampling time sequence information, consecutive adjacent pressure sampling records in the original pressure monitoring sequence are read sequentially, and consecutive adjacent pressure sampling records in which the equipment operating status information has not changed are continuously merged to form a continuous analysis segment.

[0050] For the cavity pressure data in each continuous analysis segment, the pressure slowly varying entropy value is calculated using the pressure slowly varying entropy valley capture algorithm, and each continuous analysis segment corresponds to a pressure slowly varying entropy value.

[0051] The formula for calculating the slowly varying entropy of pressure is:

[0052] ;

[0053] in, The entropy value is the slowly varying pressure value. For the first Number of pressure sampling records in each continuous analysis section This represents the number of pressure sampling records. For the first In the nth continuous analysis segment One pressure value, For the first In the nth continuous analysis segment One pressure value, For indexing continuous analysis segments, Record the corresponding pressure value for pressure sampling. The sequential position numbering of adjacent pressure sampling records, For the first In the nth continuous analysis segment The absolute value of each pressure change For the first In the nth continuous analysis segment The absolute value of each pressure change The position number is used to sum the absolute values ​​of the pressure changes in the order of their positions.

[0054] It should be noted that the absolute values ​​of adjacent pressure changes themselves have pressure dimensions, while the proportional term after normalization is dimensionless. Therefore, the entropy value of the pressure change is dimensionless.

[0055] By comparing the pressure entropy values ​​of all continuous analysis segments in the distribution of pressure entropy values, regions where the pressure entropy values ​​of multiple adjacent continuous analysis segments are lower than those of the adjacent consecutive analysis segments are identified as pressure entropy valley regions; regions where the pressure entropy values ​​of multiple consecutive continuous analysis segments are higher than those of the adjacent consecutive analysis segments are identified as non-pressure entropy valley regions, thus generating a pressure entropy valley state map.

[0056] It should be noted that the pressure gradual entropy valley capture algorithm can perform a holistic analysis of the continuous pressure change relationship in the original pressure monitoring sequence, identify the segments where the pressure change is continuous, gradual, and the fluctuations converge, and avoid the deviation caused by determining the detection window based solely on a single sampling point or instantaneous fluctuation. With the help of the pressure entropy valley state map, candidate areas suitable for short-term pressure holding isolation control can be screened more accurately, thereby improving the stability and specificity of the candidate isolation window determination, and providing a more reliable preliminary foundation for subsequent window pressure recovery curve extraction and leak rate early warning analysis.

[0057] S2.2: Synchronously couple the pressure entropy valley state map with the original pressure monitoring sequence, extract local areas with continuous and consistent operating states, determine adjacent segments before and after, and generate candidate isolation windows.

[0058] Specifically, according to the order of the pressure sampling time sequence information, each pressure entropy valley region in the pressure entropy valley state diagram is aligned with the corresponding segment in the original pressure monitoring sequence, so that each pressure entropy valley region corresponds to a continuous sampling record in the original pressure monitoring sequence.

[0059] Extract the equipment operating status information in the corresponding section, and compare the equipment operating status information one by one according to the arrangement order of the pressure sampling time sequence information. If the equipment operating status information corresponding to multiple consecutive sampling positions is the same, it is determined as a local area with continuous and consistent operating status; if the equipment operating status information corresponding to multiple consecutive sampling positions is different, it is not regarded as a local area with continuous and consistent operating status.

[0060] Using the start and end positions of a local region with a continuous and consistent operating state as boundaries, the system reads the continuous sampling records immediately adjacent to the start position and determines the preceding adjacent segment, and reads the continuous sampling records immediately adjacent to the end position and determines the following adjacent segment, so that each local region with a continuous and consistent operating state corresponds to a preceding adjacent segment and a following adjacent segment.

[0061] Local regions with consistent operating status, along with their corresponding preceding and following adjacent segments, are identified as candidate isolation windows.

[0062] S2.3: Stepwise strip the boundary sampling records of adjacent segments before and after the candidate isolation window, retain segments with continuous and consistent operating status, obtain short-term pressure holding isolation windows, and extract the central local time period for position anchoring and sequential splicing to generate a detection window sequence.

[0063] Specifically, according to the sequential position corresponding to the pressure sampling time sequence information, the continuous sampling records in the preceding and following adjacent segments of each candidate isolation window are read respectively, and the pressure change connection relationship between the preceding and following adjacent segments and the continuous sampling records inside the candidate isolation window is checked.

[0064] The continuous sampling records that connect to the candidate isolation window boundary are stripped one by one from the preceding adjacent sections, and the continuous sampling records that connect to the candidate isolation window boundary are stripped one by one from the following adjacent sections, so that the remaining continuous sampling records in the middle correspond only to local areas where the pressure changes are gentle and the equipment operating status information is continuous and consistent.

[0065] The continuous sampling records in the middle of the candidate isolation window are determined as the short-term pressure holding isolation window, and the central local time period located in the middle and with continuous connection between the sampling before and after is extracted from the short-term pressure holding isolation window. The central local time period is anchored according to the start and end positions corresponding to the pressure sampling time sequence information, so that each central local time period corresponds to a clear interval in the original pressure monitoring sequence.

[0066] Based on the chronological order of each local time period in the original pressure monitoring sequence, the local time periods of the completed location anchoring are sequentially spliced ​​to generate a detection window sequence.

[0067] S3: Perform pressure-holding isolation control on the detection window sequence and perform high-frequency pressure sampling to generate a window pressure recovery curve.

[0068] S3.1: Map the detection window sequence to the original pressure monitoring sequence on the time axis to form window anchoring information, and perform short-term pressure holding isolation control to obtain isolation response information.

[0069] Specifically, according to the order of the pressure sampling timing information, each detection window in the detection window sequence is mapped to a continuous sampling interval in the original pressure monitoring sequence, so that each detection window corresponds to the start and end sampling positions in the original pressure monitoring sequence.

[0070] Extract the cavity pressure data, pressure sampling timing information, and equipment operating status information from the corresponding interval of each detection window, and match each detection window with its corresponding interval to form window anchoring information (including the starting sampling position, ending sampling position, cavity pressure data, pressure sampling timing information, and equipment operating status information corresponding to the detection window). Use the starting sampling position corresponding to each detection window in the window anchoring information as the start position of short-term pressure holding isolation control, and use the ending sampling position corresponding to each detection window in the window anchoring information as the end position of short-term pressure holding isolation control. Switch the connection between the reaction cavity and the exhaust passage within the corresponding interval to keep the reaction cavity in a pressure holding isolation state during the corresponding time period of the detection window.

[0071] The cavity pressure data and equipment operating status information in the corresponding interval of the detection window are continuously read according to the arrangement order corresponding to the pressure sampling time sequence information. The continuous pressure change process after the start of short-term pressure holding isolation control and the corresponding sampling position are recorded one by one to obtain isolation response information.

[0072] S3.2: Based on the isolation response information, enhance sampling is performed through the pressure-back enhancement sampling algorithm to obtain the window pressure-back sampling data group.

[0073] Specifically, according to the start and end sampling positions corresponding to the detection window, the continuous pressure change records in the isolation response information are extracted window by window, so that each detection window corresponds to a continuously arranged isolation pressure change record.

[0074] According to the order of the pressure sampling time sequence information, adjacent pressure sampling records in the same detection window are read sequentially, and the pressure change between the current pressure sampling record and the previous pressure sampling record is compared one by one, and the pressure change between the next pressure sampling record and the current pressure sampling record is compared.

[0075] When the pressure change in a subsequent segment is consistently greater than that in the preceding segment, the corresponding position is designated as a continuous sampling position where the pressure change transitions from slow to fast. When the pressure change in a subsequent segment is consistently less than that in the preceding segment, the corresponding position is designated as a continuous sampling position where the pressure change transitions from fast to slow. When the pressure changes between multiple consecutive adjacent pressure sampling records remain close and the magnitude of the changes gradually decreases, the corresponding position is designated as a continuous sampling position where the pressure is trending towards stability.

[0076] When a continuous increase in pressure change is detected, the sampling frequency is increased; when a continuous decrease in pressure change is detected, the original sampling order is maintained. This allows the continuous pressure change process within the same detection window to form differentiated sampling records based on the change in pressure. The enhanced sampling records corresponding to each detection window are then assigned to the corresponding detection window according to the order of the detection windows, thus obtaining the window pressure return sampling data group.

[0077] It should be noted that the pressure return enhanced sampling algorithm extracts continuous pressure change records according to the detection window based on the isolation response information, and compares the pressure change between adjacent pressure sampling records one by one. The position where the pressure change increases continuously is taken as the enhanced sampling position, and the position where the pressure change decreases continuously is taken as the sequential sampling position. Through the pressure return enhanced sampling algorithm, more sampling records can be retained in the pressure return change, and basic temporal continuity can be maintained in the position where the pressure return change is relatively slow. This allows the window pressure recovery curve to maintain the integrity of the overall pressure return process, and provides a more reliable sampling basis for subsequent pressure return trajectory splitting, pressure return behavior feature extraction and window leakage fingerprint construction.

[0078] S3.3: Perform window aggregation and time-series processing on the window pressure return sampling data group to generate the window pressure recovery curve.

[0079] Specifically, based on the window sequence in the detection window sequence, the enhanced sampling records in the window pressure return sampling data group are merged window by window, so that each detection window corresponds to a set of independently arranged continuous pressure sampling records.

[0080] According to the sequential position corresponding to the pressure sampling time sequence information, the continuous pressure sampling records corresponding to each detection window are organized in time sequence. The continuous pressure sampling records in the same detection window are arranged continuously according to the starting sampling position corresponding to the detection window and the sequential position corresponding to the pressure sampling time sequence information, so that each continuous pressure sampling record corresponds to the continuous time position after the reaction chamber and the gas extraction passage switch to the pressure holding isolation state. The organized continuous pressure sampling records of each detection window are then connected continuously according to the sequential position corresponding to the pressure sampling time sequence information to form the window pressure recovery curve of the corresponding detection window.

[0081] S4: Based on the window pressure recovery curve, the detection window sequence is reconstructed in segments using the window pressure return fingerprint reconstruction algorithm, and the pressure return behavior features are extracted for fingerprint encoding to generate a window leakage fingerprint sequence.

[0082] S4.1: Decompose the pressure rebound curve of the window into segments, obtain segmented pressure rebound trajectories, and divide the segmented pressure rebound trajectories into segments to obtain pressure rebound trajectory segment groups;

[0083] Specifically, the continuous pressure sampling records in the window pressure recovery curve are read sequentially, and the pressure changes between the current pressure sampling record and the previous pressure sampling record, as well as the pressure changes between the next pressure sampling record and the current pressure sampling record, are compared one by one. When the pressure change relationship between consecutive adjacent pressure sampling records changes from continuous increase to continuous decrease, or from continuous decrease to continuous increase, the corresponding position is determined as the pressure return trajectory splitting position. The window pressure recovery curve is continuously segmented along each pressure return trajectory splitting position, so that the window pressure recovery curve forms multiple segmented pressure return trajectories that are connected one after the other.

[0084] Based on the sequential position of each segment of the pressure return trajectory in the window pressure recovery curve, all segmented pressure return trajectories in the same window pressure recovery curve are sequentially collected to form pressure return trajectory segment groups.

[0085] S4.2: The pressure back leakage behavior mapping algorithm is used to analyze the pressure back behavior of the pressure back trajectory segment group, and the pressure back behavior feature value is calculated to obtain the pressure back behavior feature group.

[0086] Specifically, the pressure backflow leakage behavior mapping algorithm is used to sequentially read the continuous pressure sampling records corresponding to each pressure backflow trajectory segment according to the arrangement order of the trajectory segments in the pressure backflow trajectory segment group. The position of the first continuous pressure sampling record corresponding to each pressure backflow trajectory segment in the window pressure recovery curve is determined as the starting position; the position of the last continuous pressure sampling record corresponding to each pressure backflow trajectory segment in the window pressure recovery curve is determined as the ending position; and the sampling positions continuously covered between the starting position and the ending position are taken as the continuous position range.

[0087] By comparing the pressure change at the sampling position before the start position of each pressure return trajectory segment, the pressure change at the sampling position after the end position of each pressure return trajectory segment, and the pressure change connection at the boundary position of two adjacent pressure return trajectory segments, the connection relationship between adjacent pressure return trajectory segments can be obtained.

[0088] Each segment of the compression trajectory is associated with a set of starting positions, ending positions, duration ranges, the connection relationships between adjacent compression trajectory segments, and pressure changes. The characteristic values ​​of the compression behavior are then calculated, expressed as follows:

[0089] ;

[0090] in, For the characteristic value of the pressure return behavior, This represents the pressure value corresponding to the starting position. For the starting position index, The pressure value corresponding to the end position. For the end position index, For the continuous position range of the first Pressure values ​​corresponding to each sampling location For the continuous position range of the first Pressure values ​​corresponding to each sampling location This represents the pressure value corresponding to the sampling position preceding the starting position. This represents the pressure value at the next sampling position after the end position. This represents the number of sampling locations included within the continuous location range.

[0091] It should be noted that the expressions for calculating the characteristic values ​​of pressure back behavior are all differences in pressure values, with consistent molecular weight dimensions. The denominator is the number of sampling locations and does not have physical dimensions, so the dimensions of the entire formula are consistent.

[0092] According to the chronological order of the trajectory segments in the compression trajectory segment group, all compression behavior feature values ​​are merged and sorted, and all compression behavior feature values ​​corresponding to the same compression trajectory segment group are determined as the compression behavior feature group.

[0093] It should be noted that the pressure backflow leakage behavior mapping algorithm can unify the scattered pressure change relationships in the pressure backflow trajectory segment group into comparable pressure backflow behavior feature values, avoiding the one-sided characterization caused by judging the leakage state based solely on a single pressure change. By uniformly mapping the behavioral differences of different pressure backflow trajectory segments, the leakage response characteristics in the window pressure recovery curve can be reflected more accurately, thereby improving the accuracy of window leakage fingerprint construction and providing a more reliable basis for subsequent leakage fingerprint difference quantification and future out-of-tolerance risk identification.

[0094] S4.3: Combine and sequentially encode the pressure return behavior feature group to generate a window leakage fingerprint sequence.

[0095] Specifically, according to the sequential position of the trajectory segments corresponding to each pressure return behavior feature value in the pressure return behavior feature group, the pressure return behavior feature values ​​corresponding to the same pressure return trajectory segment group are read sequentially, and all pressure return behavior feature values ​​are arranged continuously according to the order of appearance in the window pressure recovery curve; the pressure return behavior feature value that is arranged first is connected with the subsequent adjacent pressure return behavior feature values ​​one by one in chronological order, so that all pressure return behavior feature values ​​corresponding to the same pressure return trajectory segment group form a continuous feature combination record.

[0096] Each feature combination record is matched one-to-one with the corresponding detection window in the detection window sequence, so that each detection window corresponds to one feature combination record. According to the window order in the detection window sequence, the feature combination records corresponding to each detection window are concatenated with the corresponding feature combination records according to the detection window number, so that the position of each feature combination record in the detection window sequence is uniquely matched, forming the window leakage fingerprint of the corresponding detection window. The window leakage fingerprints corresponding to the detection window are arranged continuously according to their positions in the detection window sequence to generate a window leakage fingerprint sequence.

[0097] It should be noted that the window pressure return fingerprint reconstruction algorithm transforms the continuously changing window pressure recovery curve within the detection window into an identifiable and comparable window leakage fingerprint, giving the pressure return change process, which was originally difficult to compare directly, a unified representation form; it provides a basis for subsequent comparison of leakage fingerprint differences between adjacent detection windows, construction of window progressive risk sequences, and future identification of out-of-tolerance risks, thereby improving the accuracy and foresight of leakage rate early warning.

[0098] S5: Based on the window leakage fingerprint sequence, a window progressive risk sequence is constructed by quantifying the differences in leakage fingerprints, and trend extrapolation is performed to generate future deviation risk information.

[0099] S5.1: Compare the window leakage fingerprints corresponding to adjacent detection windows one by one and calculate the leakage fingerprint difference index.

[0100] Specifically, the window leakage fingerprints corresponding to adjacent detection windows in the window leakage fingerprint sequence are split into coded record groups arranged in sequence, and each coded record is marked with its position in the corresponding window leakage fingerprint.

[0101] Read each coded record one by one, and search for coded records with the same content in the coded record group corresponding to the immediately following window leak fingerprint. For example, if a coded record with the same content is found, record the coded record difference result as zero and obtain the coded record difference value; record the absolute value of the difference in the arrangement position of the two coded records in the coded record group as the position difference result. For example, if no coded record with the same content is found, record the coded record difference result as one and record the position difference result as zero, and obtain the position difference value; group two adjacent coded records together, compare the connection relationship between the two adjacent coded records in the first window leak fingerprint with the connection relationship between the corresponding coded records in the immediately following window leak fingerprint. For example, if the connection relationship is consistent, record the connection relationship difference result as zero; if the connection relationship is inconsistent, record the connection relationship difference result as one, obtain the connection relationship difference value, and calculate the leak fingerprint difference index, the expression of which is:

[0102] ;

[0103] in, For the leakage fingerprint difference index, The number of encoded records contained in the leaked fingerprint from the window. For the first The difference value between the coded records corresponding to each coded record. For the first The location difference value corresponding to the barcode record. For the first The difference in the connection relationship between adjacent coded records before and after a group.

[0104] It should be noted that the differences in the encoded records, the differences in the location, and the differences in the connection relationships are all dimensionless, and the denominator is the number of comparisons. Therefore, the overall dimensions of the leaked fingerprint difference index are uniform.

[0105] S5.2: By using a progressive risk aggregation method, the differential indicators of each leakage fingerprint are sequentially accumulated and correlated to construct a windowed progressive risk sequence.

[0106] Specifically, using the leakage fingerprint difference index as input, the leakage fingerprint difference index corresponding to each adjacent detection window is read sequentially according to the window order in the detection window sequence, and adjacent leakage fingerprint difference indices are matched one after the other. The case where the leakage fingerprint difference index at the later position is equal to the leakage fingerprint difference index at the previous position is determined to be a same-direction maintenance relationship; the case where the leakage fingerprint difference index at the later position is greater than the leakage fingerprint difference index at the previous position is determined to be an increasing relationship; and the case where the leakage fingerprint difference index at the later position is less than the leakage fingerprint difference index at the previous position is determined to be a decreasing relationship.

[0107] Based on the same-direction maintenance relationship, rising relationship, and falling relationship, the leakage fingerprint difference indicators that appear continuously along the detection window sequence are divided into segments. The leakage fingerprint difference indicators that are continuously in the same-direction maintenance relationship form the maintenance segment, the leakage fingerprint difference indicators that are continuously in the rising relationship form the rising segment, and the leakage fingerprint difference indicators that are continuously in the falling relationship form the falling segment.

[0108] The leakage fingerprint difference indicators in the holding section, rising section and falling section are organized in order according to the window position in the detection window sequence, so that the leakage fingerprint difference indicators in the same section form a corresponding continuous progressive arrangement, and a window progressive risk sequence is constructed.

[0109] It should be noted that the progressive risk aggregation method can continuously organize the leakage fingerprint difference indicators that are scattered between adjacent detection windows along the detection window sequence, transforming a single difference change into a continuous progressive risk change process, avoiding the biased judgment caused by judging future risks based solely on the difference results of a single detection window; by constructing a window progressive risk sequence, the development trend of leakage risk along the detection window sequence can be more clearly reflected, thereby improving the continuity and foresight of future out-of-tolerance risk identification.

[0110] S5.3: Perform sequential analysis and trend extrapolation on the window-progressive risk sequence to generate risk extrapolation trajectory, and identify risk extension to generate future deviation risk information.

[0111] Specifically, the progressive risk values ​​at the beginning and end of the progressive risk sequence are compared sequentially. If the progressive risk value at the later position is greater than that at the previous position, the direction of change is determined to be upward; if the progressive risk value at the later position is equal to that at the previous position, the direction of change is determined to be stable; if the progressive risk value at the later position is less than that at the previous position, the direction of change is determined to be downward. Then, according to the upward, stable, or downward direction, a continuous extension portion is formed after the end position of the progressive risk sequence, consistent with the corresponding direction. This continuous extension portion is then continuously connected to the progressive risk sequence to generate a risk extrapolation trajectory.

[0112] Using the risk extrapolation trajectory as the identification object, the continuous extension portion connected to the end position of the window progressive risk sequence in the risk extrapolation trajectory is extracted. The extension direction and extension length of the continuous extension portion relative to the end position of the window progressive risk sequence are compared. The case where the continuous extension portion continues to rise along the detection window sequence is identified as the future risk increase state. The case where the continuous extension portion remains unchanged along the detection window sequence is identified as the future risk maintenance state. The case where the continuous extension portion continues to fall along the detection window sequence is identified as the future risk decrease state. Then, the future risk increase state, future risk maintenance state, or future risk decrease state are sorted with the corresponding extension length to generate future deviation risk information.

[0113] This embodiment also provides a computer device applicable to the vacuum equipment leak rate early warning method based on pressure monitoring, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the vacuum equipment leak rate early warning method based on pressure monitoring as proposed in the above embodiment.

[0114] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0115] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for early warning of leak rate in vacuum equipment based on pressure monitoring as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0116] In summary, this invention achieves the early identification of future leakage rate deviation risks by constructing a short-term pressure holding isolation window and combining it with pressure return enhanced sampling, thereby making the micro-leakage response masked by continuous air extraction visible; and by reconstructing the window pressure return fingerprint and progressive risk extrapolation, transforming discrete pressure rise behavior into comparable and predictable leakage evolution information, thereby improving the foresight and accuracy of early warning.

[0117] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for early warning of leakage rate in vacuum equipment based on pressure monitoring, characterized in that, include: Collect cavity operation monitoring data and generate raw pressure monitoring sequences through preprocessing; Stable fragment screening was performed on the original pressure monitoring sequence to determine the short-term pressure holding isolation window, and time sequence alignment was performed to generate the detection window sequence; The detection window sequence is subjected to pressure-holding isolation control and high-frequency pressure sampling to generate a window pressure recovery curve; Based on the window pressure recovery curve, the detection window sequence is reconstructed in segments using the window pressure return fingerprint reconstruction algorithm, and the pressure return behavior features are extracted for fingerprint encoding to generate a window leakage fingerprint sequence. Based on the window leakage fingerprint sequence, a window progressive risk sequence is constructed by quantifying the differences in leakage fingerprints, and trend extrapolation is performed to generate future deviation risk information.

2. The vacuum equipment leak rate early warning method based on pressure monitoring as described in claim 1, characterized in that, The cavity operation monitoring data includes cavity pressure data, pressure sampling timing information, and equipment operating status information; The preprocessing includes time alignment, outlier removal, and sorting.

3. The vacuum equipment leak rate early warning method based on pressure monitoring as described in claim 2, characterized in that, The process of performing stable segment screening on the original pressure monitoring sequence to determine a short-term pressure holding isolation window, and performing time sequence alignment to generate a detection window sequence, is as follows: The pressure slowly varying entropy valley capture algorithm is used to calculate the pressure slowly varying entropy value based on the distribution of adjacent pressure changes in the original pressure monitoring sequence, and to perform distribution analysis to generate a pressure entropy valley state map. The pressure entropy valley state map is synchronously coupled with the original pressure monitoring sequence to extract local areas with continuous and consistent operating states and determine adjacent segments before and after, thereby generating candidate isolation windows. The boundary sampling records of adjacent segments before and after the candidate isolation window are gradually stripped to retain segments with continuous and consistent operating status. Short-term pressure holding isolation windows are obtained, and the central local time period is extracted for location anchoring and sequential splicing to generate a detection window sequence.

4. The vacuum equipment leak rate early warning method based on pressure monitoring as described in claim 1, characterized in that, The process of performing pressure holding and isolation control on the detection window sequence refers to mapping the detection window sequence to the original pressure monitoring sequence on a time axis to form window anchoring information, and then performing short-term pressure holding and isolation control to obtain isolation response information.

5. The vacuum equipment leak rate early warning method based on pressure monitoring as described in claim 1, characterized in that, The specific steps for generating the window pressure recovery curve are as follows: Based on the isolation response information, enhanced sampling is performed using a pressure-back enhanced sampling algorithm to obtain window pressure-back sampling data sets; Window aggregation and time-series processing are performed on the window pressure return sampling data group to generate the window pressure recovery curve.

6. The vacuum equipment leak rate early warning method based on pressure monitoring as described in claim 1 or 5, characterized in that, The method involves reconstructing the detection window sequence segment by segmenting it based on the window pressure recovery curve and extracting pressure return behavior features for fingerprint encoding to generate a window leakage fingerprint sequence. The specific steps are as follows: The pressure recovery curve of the window is split into pressure return trajectories to obtain segmented pressure return trajectories, and the segmented pressure return trajectories are further divided into trajectory segments to obtain pressure return trajectory segment groups; A pressure back leakage behavior mapping algorithm is used to analyze the pressure back behavior of the pressure back trajectory segment group, and the pressure back behavior feature value is calculated to obtain the pressure back behavior feature group; The characteristic groups of pressure return behavior are combined, characterized, and sequentially encoded to generate a window leakage fingerprint sequence.

7. The vacuum equipment leak rate early warning method based on pressure monitoring as described in claim 6, characterized in that, The method for constructing a progressive risk sequence based on the window leakage fingerprint sequence by quantifying the differences in leakage fingerprints involves the following steps: The leakage fingerprints of adjacent detection windows are compared one by one, and the leakage fingerprint difference index is calculated. By using a progressive risk aggregation method, the differential indicators of each leakage fingerprint are sequentially accumulated and correlated to construct a windowed progressive risk sequence.

8. The vacuum equipment leak rate early warning method based on pressure monitoring as described in claim 7, characterized in that, The aforementioned future deviation risk information refers to the generation of risk extrapolation trajectory by performing sequential analysis and trend extrapolation on the progressive risk sequence of the window, and then performing risk extension identification.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the vacuum equipment leak rate early warning method based on pressure monitoring as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the vacuum equipment leak rate early warning method based on pressure monitoring as described in any one of claims 1 to 8.

Images