A method for checking the timing of a vacuuming operation

Abstract

This invention relates to the field of data processing technology and discloses a method for verifying the timing of vacuuming operations. This method addresses the problems of inaccurate timing judgment, untimely anomaly identification, and insufficient traceability of operation records in traditional methods, which affect the standardization and safety of vacuuming operations. The method first collects vacuum level, temperature, flow rate, valve status, and operation declaration information, and aligns them to a unified time base. Based on the aligned multi-source data, it determines the operation session and extracts process nodes. Verification rules are established in conjunction with process specifications to correct the start and end nodes of each process and construct the operation timing chain. Then, it verifies the process duration, process sequence, and process interval, and executes interlocking control, alarm output, and record archiving based on the verification conclusions. This improves the accuracy of vacuuming operation timing judgment, the timeliness of anomaly handling, and process traceability.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a method for verifying the timing of vacuuming operations. Background Technology

[0002] During the maintenance, reassembly, and sulfur hexafluoride gas treatment of high-voltage switchgear, the independent switch chambers of high-voltage switchgear, circuit breakers, busbars, and other equipment typically require vacuuming first. After the vacuum level reaches the process requirements, vacuum holding or leak detection is performed to determine whether the residual gas, moisture, and sealing risks within the chamber are under control. This type of operation generally includes a series of steps such as task creation, chamber information confirmation, sensor connection, vacuuming initiation, vacuum level compliance confirmation, continued vacuuming, pump shutdown confirmation, pressure holding and waiting, vacuum level retesting, and operation report generation. During the operation, it is necessary to record not only the start time, initial vacuum level, time when the vacuum level reaches the threshold, pump shutdown time, pressure holding start and end time, and vacuum level retesting, but also to continuously monitor the vacuum level or pressure changes during the pressure holding period to determine whether the chamber sealing status is stable. In the prior art, the invention patent with authorization announcement number CN102360756B discloses a vacuum drying and oil injection method for the main transformer of an electric locomotive. It mainly completes the vacuum drying and oil injection process through steps such as vacuuming the oil storage tank, heating, vacuuming the oven, and oil injection. The invention patent with authorization announcement number CN111223643B discloses a transformer vacuum oil injection system and its oil injection method, which focuses on the improvement of the structure of the vacuum oil injection system and the oil injection process. Although the above-mentioned prior art involves vacuuming or vacuum oil injection operations, it mainly focuses on the control of transformer vacuum drying, oil injection system structure, or oil injection process. It does not establish a complete time sequence verification mechanism for vacuuming and vacuum leak detection operations of switch chambers such as high-voltage combined electrical appliances, circuit breakers, and busbars. For switch chamber operations, after the vacuum degree reaches the set threshold, it is usually necessary to continue vacuuming for a certain period of time. After the pump is stopped, it is necessary to enter the pressure holding stage. The operation is judged as qualified based on the changes in vacuum degree before and after pressure holding and the changes in pressure during the pressure holding process. Therefore, the occurrence sequence, duration, and connection relationship of each process node directly affect the authenticity and traceability of the operation results. However, in actual field operations, inconsistencies often arise between vacuum equipment, vacuum monitoring devices, mobile operation terminals, and back-end management software, including inconsistent start-up times, delayed manual confirmation, different data upload cycles, network anomalies, or local data retransmission. When the monitoring device starts later than the vacuum equipment, pump shutdown confirmation is not timely, the pressure holding timer start is inaccurate, or only the final vacuum level is recorded without saving the pressure changes during the holding period, it can easily lead to distortions in the time when the vacuum level reaches the target, the duration of continued vacuuming, the start time of pressure holding, the duration of pressure holding, and the leak detection results. These problems can cause inaccuracies in the same operation. The vacuuming and leak detection phases in the vacuuming operation of the gas shut-off chamber are difficult to objectively reconstruct, which affects the credibility of the operation report, the accuracy of the maintenance quality evaluation, and the completeness of subsequent quality traceability. Therefore, it is necessary to propose a vacuuming operation sequence verification method to uniformly calibrate the vacuum level data, operation application information, pump shutdown confirmation information, pressure changes during the pressure holding process, and data upload records. It should also automatically verify the operation nodes, duration, sequence relationship, process interval, and abnormal results to improve the standardization and traceability of vacuuming and vacuum pressure holding operations in the high-pressure switch gas chamber. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a method for verifying the timing of vacuuming operations, which solves the problems of inaccurate timing judgment, untimely anomaly identification, and insufficient traceability of operation records in traditional methods, thus affecting the standardization and safety of vacuuming operations.

[0004] To achieve the above objectives, the present invention provides the following technical solution: A method for verifying the timing of vacuuming operations includes: S1. Collect vacuum level, temperature, flow rate, valve status and work application information, and write a unified timestamp for each data based on the same clock reference and perform time alignment to form a multi-source time series; S2. Determine the work session based on the work application information and the status changes in the multi-source time series, and extract the time sequence nodes corresponding to each process; S3. Read the preset process specifications and establish verification rules corresponding to each process. S4. Determine the start and end nodes of each process based on the multi-source time series and verification rules, and form a work sequence chain according to the sequence of processes; S5. Verify the duration, sequence, and interval of each process according to the work sequence chain, and generate a verification conclusion. S6. Based on the verification results, execute interlock control, trigger alarms, or record and archive data.

[0005] Preferably, S1 includes: Vacuum level, temperature, flow rate, valve status, and work application information are all integrated into the verification execution terminal, and a unified timestamp is written using the same clock reference provided by the verification execution terminal. The same clock reference is achieved through any one or a combination of PTP, IRIG-B code, a high-precision crystal oscillator built into the verification execution terminal, and an interrupt response mechanism. For data accessed via the communication interface, the sampling time is calculated by combining the message length, interface transmission rate, and fixed processing delay of the device. For digital input data, record the event moment when the edge flips; Resampling is performed using a unified time grid. For data that has not been updated, the previous values ​​are used and missing states are marked according to the number of consecutive sampling periods, forming a multi-source time series that includes timestamps, data values, source identifiers, validity identifiers, update time identifiers, and reuse identifiers.

[0006] Preferably, S2 includes: The operation start declaration information is linked with the vacuum pump start-up, main vacuum valve operation, and vacuum change trend within the relevant time window to establish an operation session; Along a unified time axis, candidate timing nodes for each process are extracted based on valve status, vacuum changes, flow rate changes, and the continuous satisfaction of status conditions. The candidate timing nodes of each process are bound to the job session, and supplementary events, duplicate nodes, conflicting nodes and incomplete sessions are marked.

[0007] Preferably, S3 includes: The process documents are parsed into threshold rules, continuity rules, sequence rules, interval rules, and mutual exclusion rules; Establish rule templates and their applicable relationships according to equipment type, preceding process, data source, and exception conditions; Each rule is matched with the candidate time nodes or corresponding time intervals of the corresponding process, and the candidate nodes are classified into credibility levels, priority rankings and anomaly markings based on vacuum, valve and flow evidence.

[0008] Preferably, rule templates and their applicability relationships are established according to equipment type, preceding processes, data sources, and exception conditions, including: Based on the equipment number parsing results or work application information, determine the equipment type identifier, call the rule template corresponding to the equipment type, and load the corresponding threshold value, duration, sequence relationship and interval conditions; Write the process type identifier, preceding process identifier, data source requirements, and exception templates corresponding to maintenance sampling, no-load trial operation, and manual confirmation shutdown according to the correspondence of each process.

[0009] Preferably, S4 includes: Based on the rule conditions corresponding to the candidate time-series nodes, the node trust level, and the continuous changes in multi-source states, consistency correction is performed on the candidate time-series nodes. With the combined support of two types of evidence—vacuum changes, valve status changes, flow rate changes, and work report information—the earliest established time point is determined as the starting point, and the earliest time point that meets the termination conditions and does not subsequently reverse is determined as the ending point. Presumed nodes, pause segments, and conflict nodes are marked and connected according to their preceding process relationships to form a work sequence chain.

[0010] Preferably, consistency correction is performed on candidate time-series nodes based on the rule conditions corresponding to the candidate time-series nodes, the node trust level, and the continuous changes in multi-source states, including: Within the time interval corresponding to the candidate timing node, the consistency between the valve status, vacuum changes, and flow rate changes and the process rules is verified by combining the node's credibility level. For the time point when the conditions for node establishment are first met, backtrack to the nearest unified time grid boundary; If a short-term reverse state recovery occurs, or if the conditions for subsequent processes to enter are not yet stable, continue to verify the stable entry point.

[0011] Preferably, S5 includes: The duration of each process in the work sequence chain is compared after deducting pauses; Perform sequential verification according to the completion conditions of the preceding processes; The maximum interval condition and minimum interval condition are checked based on the time difference between the start and end points of adjacent processes. It also generates itemized verification information and task-level verification conclusions by combining evidence integrity level, anomaly priority, and control recommendation level.

[0012] Preferably, sequential verification is performed according to the completion conditions of the preceding process, including: Read the identifier of the preceding process in the current process of the job sequence chain, and locate the end node, completion rule and node trust level of the corresponding preceding process; Determine if the end node of the preceding process exists, and verify whether the preceding process simultaneously meets the corresponding threshold rules, continuity rules, and trust level requirements; The marking order is abnormal when the start node of the current process is earlier than the valid end node of the previous process, or when the previous process has not met the completion rules but the current process has already started. Write the current process, preceding process, triggering evidence, and exception level corresponding to the sequence anomaly into the sub-item verification information, and provide the sequence verification results to the job-level verification conclusion.

[0013] Preferably, S6 includes: Interlock control includes one or more of the following: prohibiting the start of the next process, cutting off the execution authority of the valve or actuator corresponding to the abnormal process, and switching to manual control. According to the control recommendation level corresponding to the work-level verification conclusion, switch the interlock status, alarm output status and manual takeover status. For observed anomalies, the next process should be frozen and authorization should be granted while data collection continues. For anomalies that still exist in the waiting window, the process should be prohibited from continuing or require manual confirmation. For critical anomalies, the corresponding execution permissions should be revoked. Archive the verification conclusions and abnormal evidence fragments. If archiving fails, perform local temporary storage and sequential retransmission. Release the interlock control when manual authorization, exception template matching and current state resampling are satisfied at the same time.

[0014] Compared with the prior art, the present invention provides a method for verifying the timing of vacuuming operations, which has the following beneficial effects: 1. This invention unifies the time base alignment of vacuum level, temperature, flow rate, valve status, operation report information, and pump stop confirmation information during vacuuming operations of switch chambers in high-voltage switchgear, circuit breakers, busbars, etc., and merges key nodes such as vacuuming start, vacuum level reaching a set threshold, continuation of vacuuming, pump stop confirmation, pressure holding start, pressure holding end, and result judgment into the same operation session. This enables previously scattered, inconsistent, and difficult-to-corresponding operation data to be correlated on the same time axis. Furthermore, by establishing threshold rules, continuity rules, sequence rules, and interval rules in conjunction with process specifications, it addresses the starting point of each process. By correcting the stop point and constructing the operation sequence chain, the sequential relationship and duration between the vacuuming stage, the continued vacuuming stage, and the pressure holding and leak detection stage can be clearly defined. By uniformly verifying the duration of continued vacuuming after the vacuum degree reaches the set threshold, the start time of pressure holding after the pump stops, the difference in vacuum degree before and after pressure holding, and the pressure change process during pressure holding, the distortion of operation results caused by delayed monitoring start-up, untimely manual confirmation, misjudgment of single-point vacuum values, unclear start and end of pressure holding, and inaccurate calculation of duration can be reduced. This improves the accuracy of timing judgment, reliability of vacuum leak detection results, and traceability of data throughout the entire process for vacuuming operations in high-pressure switch chambers.

[0015] 2. This invention, by transforming the vacuuming and vacuum holding procedures into rule templates corresponding to equipment type, chamber information, preceding procedures, data sources, and exceptional conditions, and by setting up processing mechanisms such as candidate node reliability classification, anomaly classification, waiting observation, manual takeover, controlled release, data archiving, and anomaly retransmission during the verification process, enables the vacuuming operation of the switch chamber to select the corresponding handling method based on the rule matching results and anomaly level when situations such as short-term pressure fluctuations, vacuum recovery during the holding stage, missing sensor data, network upload anomalies, early termination of operation, or manual retransmission occur. This reduces misjudgments caused by differences in manual on-site judgment and single interlocking logic, avoids directly identifying short-term disturbances or low-reliability anomalies as serious anomalies, and promptly triggers alarms, takeover, or operation stoppage when critical anomalies occur. This improves the standardization of vacuuming monitoring result judgment, the classification of anomaly handling, and the safety and traceability of switch chamber vacuuming operations in complex field environments.

[0016] 3. By saving information such as gas chamber information, start and end times of operation, initial vacuum level, time and vacuum level to reach the threshold, pump stop time and vacuum level, pressure holding start and end times, vacuum level difference before and after pressure holding, pressure change curve, verification conclusions, and executed actions during the record archiving process, and by temporarily storing and sequentially re-uploading the operation data locally when remote upload fails, the vacuuming and vacuum leak detection operations can form a complete operation record that is searchable, exportable, and retransmittable. On this basis, even if the on-site network is unstable or the back-end management software is temporarily unavailable, the loss of critical operation data can be avoided, and the temporal sequence relationship between the vacuuming stage, pump stop stage, and pressure holding and leak detection stage can be preserved. This provides continuous, complete, and traceable data for subsequent work report generation, anomaly investigation, responsibility review, and maintenance quality assessment. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of a vacuuming operation timing verification method according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1: Figure 1 A method for timing verification of vacuuming operations is presented, including: S1. Collect vacuum level, temperature, flow rate, valve status and work application information, and write a unified timestamp for each data based on the same clock reference and perform time alignment to form a multi-source time series; S2. Determine the work session based on the work application information and the status changes in the multi-source time series, and extract the time sequence nodes corresponding to each process; S3. Read the preset process specifications and establish verification rules corresponding to each process. S4. Determine the start and end nodes of each process based on the multi-source time series and verification rules, and form a work sequence chain according to the sequence of processes; S5. Verify the duration, sequence, and interval of each process according to the work sequence chain, and generate a verification conclusion. S6. Based on the verification results, execute interlock control, trigger alarms, or record and archive data.

[0020] This method is applicable to industrial site vacuuming operation sequence verification scenarios. It can be applied to the vacuum drying and vacuum oil injection processes of oil-immersed power transformers, high-voltage switchgear gas chambers and switch chambers, busbar switch chambers, and also to vacuuming operation scenarios with multiple processes and time sequence constraints, such as vacuum heat treatment furnaces, semiconductor cavity maintenance, and refrigeration system evacuation and dehumidification. The participants include at least on-site personnel, vacuuming equipment, verification execution terminals, and multi-source acquisition terminals related to the vacuuming operation. Input information includes at least vacuum level data, temperature data, flow rate data, valve status data, and operation declaration information. Vacuum level data characterizes the degree and stage changes of vacuuming; temperature data characterizes changes in the working environment and the state of the processed object; flow rate data characterizes the medium introduction or discharge behavior; valve status data characterizes the action switching time; and operation declaration information characterizes the intention of manual start-up, termination, and process switching. The processing target... The goal is to convert on-site data from different sources, with asynchronous timing and inconsistent granularity, into a unified time series that can be used for process timing verification. Based on the process specifications, it automatically determines the start and end points, duration, sequence, and interval relationships of processes, outputting verification conclusions and corresponding interlocking controls, alarm prompts, and record archiving results. The operation of this method is based on a unified time base and full-process timing verification. First, multi-source data is aligned to the same clock base to solve the problem of inconsistent time bases from different data sources. Then, through job session identification and process node extraction, relevant data from different time segments are merged into the same job process. Next, the process specifications are converted into executable verification rules to support rule-based judgment. Subsequently, through start and end node correction and job timing chain construction, the true boundaries and sequential relationships of each process are determined. Finally, control, prompts, and archiving processes are executed based on the verification conclusions, thus forming a complete closed loop.

[0021] S1. Collect vacuum level, temperature, flow rate, valve status, and work report information, and write a unified timestamp to each data point based on the same clock reference and perform time alignment to form a multi-source time series. The specific implementation is as follows: Before the vacuuming operation begins, vacuum sensors, temperature sensors, flow meters, valve position feedback contacts, and operation reporting terminals are all connected to the verification execution terminal. Vacuum level data is collected by composite vacuum gauges installed at the top and bottom of the vacuuming container, with a range selectable from 100,000 Pa to 0.1 Pa. This range covers the actual working conditions of the vacuuming operation, from the roughing stage (approximately atmospheric pressure) to the high-pressure stage and the low-pressure stage (vacuum maintenance). Temperature data is collected by temperature sensors located at the environmental and object locations to distinguish between environmental changes and process changes. Flow rate data is collected by flow meters installed in the oil injection line or medium introduction line to identify the start, continuation, pause, and end of medium introduction. Valve status data is collected by valve position feedback contacts or actuator feedback terminals to record the valve opening, closing, and reversal times. Operation reporting information is collected by the human-machine interface terminal and includes at least the operation number, equipment number, operation start report, process switch report, and operation end report. The above data items correspond to the actual process steps of the vacuuming operation and can meet the input requirements for subsequent operation session recognition and process sequence verification. To eliminate the impact of local clock drift, communication delay, and inconsistent recording times on timing judgment from different devices, a unified timestamp is written to the access data using the same clock reference provided by the verification execution terminal. This unified clock reference refers to a clock reference in which the verification execution terminal uniformly writes timestamps to each data acquisition source, ensuring that vacuum changes, valve actions, flow changes, and work report events can be sorted on the same timeline. This unified clock reference is not limited to a specific protocol and can be implemented through one or a combination of PTP, IRIG-B code, a high-precision crystal oscillator built into the verification execution terminal, and an interrupt response mechanism. The verification execution terminal can employ a time synchronization mechanism supporting PTP hardware timestamps to control the timestamp alignment error of multi-source data within ±100μs. The hardware synchronization capability of the terminal is compatible, and it can meet the accuracy requirements for unified sequencing of valve switching, vacuum change edge, and work reporting actions. For data accessed through the communication interface, the reception time of the data entering the verification execution terminal is recorded first, and then the corresponding sampling time is calculated by combining the message length, interface transmission rate, and fixed processing delay of the equipment. The fixed processing delay of the equipment can use the factory calibration value of the equipment. When the field communication link, acquisition interface, or equipment firmware version is inconsistent with the factory conditions, field calibration can be performed through standard signal injection, and the average time difference obtained from multiple tests is used as the fixed processing delay of the corresponding equipment. For switch data such as valve status, an event timestamp is written at the moment of edge flipping to preserve the true sequence of valve actions. Regarding sampling frequency settings, vacuum level data can be selected at 100 times per second. This value corresponds to the main sampling configuration because vacuum level directly participates in process switching and compliance judgment, and needs to cover critical edge periods of vacuum changes and valve switching. In scenarios with limited computing power or power consumption, vacuum level data can be reduced to 25 times per second to balance sampling accuracy and on-site resource constraints. Temperature data can be selected at 50 times per second to meet the synchronous recording of ambient temperature and the temperature of the object being processed. When temperature is only used as an auxiliary judgment quantity and system resources are limited, it can be reduced to 10 times per second. Flow rate data can be selected at 10 times per second to identify continuous changes in media introduction, pause, and termination. Valve status data is recorded using an event-triggered method without setting a fixed polling cycle. This is because valve status is characterized by discrete flips, and event triggering can directly retain the moment of action. Furthermore, the state flip response time of the valve position feedback contact is no more than 1ms, which can meet the timing recording requirements in vacuum process switching scenarios. After writing the unified timestamp, time alignment processing is performed on various types of data. The unified time grid can be selected as 10ms, which corresponds to the resampling processing method under the 100Hz common time reference. This can maintain consistency with the 100Hz vacuum sampling while taking into account the time resolution capability of valve state reversal and the initial stage of vacuum change. In scenarios with limited edge computing power or storage resources, the unified time grid can be widened to 40ms, which corresponds to one sampling period when the vacuum is sampled at 25Hz. This is beneficial for maintaining time alignment consistency under alternative operating conditions. For multiple sampled values ​​falling into the same time grid, the latest valid value is retained as the corresponding value of the current time grid. For data sources where no new value appears in the current time grid, the valid value of the previous time grid is used and a reuse flag is written to distinguish whether the value belongs to a new sampled value or a retained value. To facilitate subsequent session identification, a missing data status check is also implemented. The missing data check is determined based on the number of consecutive sampling periods, without setting a fixed time threshold independent of the sampling frequency. Vacuum level data can be considered missing after five consecutive sampling periods of no update, corresponding to 50ms under 100Hz sampling conditions and 200ms under 25Hz sampling conditions. This setting is because vacuum level is a key parameter for vacuuming operations, requiring rapid response to data interruptions, and five consecutive sampling periods of no update avoids misjudging single communication jitter as missing data. Temperature data can also be considered missing after five consecutive sampling periods of no update, under 50Hz sampling conditions. The corresponding time threshold is 100ms at 10Hz and 500ms at 10Hz sampling. This setting is because the rate of temperature change is lower than the rate of vacuum change, allowing for a wider update tolerance time than vacuum. Flow data can be considered missing if it has not been updated for three consecutive sampling cycles, corresponding to 300ms at 10Hz sampling. This setting is because flow is mainly used to assist in identifying media introduction, pause, and termination. 300ms can cover short-term transmission jitter in the field and is significantly shorter than the duration of normal media introduction state changes, thus balancing judgment sensitivity and engineering stability. When the sampling frequency is switched, the corresponding absolute time threshold is adjusted synchronously with the number of consecutive sampling cycles. After unified access, unified timestamp writing, unified time grid alignment, and missing status marking, a multi-source time series is formed. Each time series record in the multi-source time series includes at least the following fields: timestamp, data source identifier, data value, validity identifier, update time identifier, and continuation identifier. This multi-source time series serves as the input for subsequent processing, used to jointly analyze work application behavior and equipment status changes on the same time axis, thereby identifying work session boundaries and extracting time series nodes of each process.

[0022] S2. Determine the work session based on the work application information and the status changes in the multi-source time series, and extract the time sequence nodes corresponding to each process. The specific implementation is as follows: Using the multi-source time series formed earlier as input, we perform correlation analysis between work application information and equipment status changes to establish work sessions and extract candidate time sequence nodes of the corresponding work sessions. Vacuuming operations are usually affected by factors such as manual application, vacuum pump operation, valve switching, vacuum changes, and media introduction simultaneously on site. Therefore, we do not use a single vacuum parameter change to define a work session, but instead use a dual constraint method of work application information and equipment status changes to identify work sessions. Equipment status changes include at least any change in vacuum pump startup, main vacuum valve opening, and vacuum level entering a continuous downward trend. The identification of the start of a work session adopts a dual-trigger condition: the first trigger condition is receiving a work start declaration information, and the second trigger condition is detecting a change in equipment status. A correlation time window is set between the two trigger conditions, which can be selected from 30 seconds to 120 seconds, with 60 seconds being the preferred value. This setting is because in on-site operations, there are usually tens of seconds of manual operation delay and communication delay from the start of the declaration to the control issuance, actuator action, and status feedback. When the correlation time window is less than 30 seconds, it is easy to misjudge normal preparation processes as invalid declarations; when the correlation time window is greater than 120 seconds, it is easy to merge irrelevant actions into the same work session. The correlation time window can be dynamically configured based on historical work data or equipment response characteristics. When historical records are available, the delay distribution between work declaration, vacuum pump start-up, main vacuum valve action, and the appearance of vacuum change trends in multiple operations of similar equipment can be statistically analyzed, and the coverage... The statistical value of the normal operation delay range is used as the configuration basis; after receiving the start declaration information, if a change in equipment status is detected within the associated time window, an operation session is established; if the start declaration information has been formed, but no change in equipment status is detected within the longest observation period, the pending confirmation status is canceled, and no operation session is established; the longest observation period can be taken as the upper limit of the associated time window; if the equipment status has changed but the start declaration information has not yet been received, the status change is recorded as an automatic start candidate event and written with a supplementary entry flag, allowing operators to supplement the start declaration information within the aforementioned associated time window, and then the supplementary entry time is matched with the automatic start candidate event; if the supplementary entry is not completed within the time window, the event is retained as an abnormal start record and is not confirmed as a normal operation session; through the above processing, it can be compatible with the timing deviation caused by reporting before action, action before supplementary entry, and communication delay; After the work session is established, candidate time nodes for each process are extracted along a unified timeline. Candidate time nodes are not identified solely based on a single parameter reaching a certain value, but rather by combining process intent information, process status changes, and stability confirmation results. Candidate nodes for the roughing process can be determined based on vacuum pump startup, roughing valve opening, and a continuously decreasing vacuum level. Candidate nodes for the high-pressure process can be determined based on the opening of the high-pressure valve, the closing of the roughing valve or switching to the high-pressure channel, a phased change in the vacuum descent rate, and the vacuum level entering a candidate vacuum range matching the high-pressure process. This candidate vacuum range characterizes the vacuum state as it continues to decrease from the roughing stage and enters a lower operating range, and is subsequently formally verified in conjunction with the process specifications. Candidate nodes for the vacuum maintenance process can be determined based on… The candidate nodes for the injection or extraction valves are determined by the following criteria: the valves remain stationary, the vacuum fluctuation amplitude decreases, and the duration reaches a stable observation window. The stable observation window can be selected from 20 to 60 seconds, with 30 seconds being the preferred value. This setting is because there is usually a short-term pressure equalization process after valve switching, and 30 seconds can better distinguish between excessive fluctuations and stable holding states. The candidate nodes for the oil injection process are determined by the opening of the oil injection valve, the flow rate changing from zero to a continuous positive value, and the vacuum change changing from a simple decrease to a complex change accompanied by fluctuations. The candidate nodes for the post-injection pressure holding process are determined by the oil injection flow rate returning to zero, the oil injection valve closing, the relevant pressure holding valve positions stabilizing, and the vacuum range re-entering a stable holding state. The candidate nodes for the vacuum breaking process are determined by the opening of the vacuum breaking valve, the vacuum degree recovering, and the vacuum degree maintaining a recovery direction. To avoid identifying transient jitter, short-term false triggering signals, or edge noise as candidate timing nodes for the process, a minimum duration is set for different types of state changes. The vacuum degree decrease trend can be maintained continuously for 5 to 15 seconds, with 10 seconds being preferred. This setting is because vacuum pump startup and valve switching are usually accompanied by pressure fluctuations on the order of several seconds, and 10 seconds of continuous observation can better distinguish the actual evacuation trend from transient disturbances. A continuous positive flow rate can be maintained continuously for 3 to 10 seconds, with 5 seconds being preferred. This setting is because flow rate changes usually directly reflect the medium introduction action, and 5 seconds can cover short-term pulse fluctuations without significantly delaying the identification of the oil injection node. The state maintenance time after valve flipping can be maintained for 1 to 3 seconds, with 2 seconds being preferred. This time is used to confirm that the actuator has completed its action and entered a stable state; it does not correspond to the millisecond-level response time of the valve position feedback contact, but rather to the stabilization confirmation time after valve switching. Only when the corresponding state reaches the corresponding minimum duration is the corresponding time point registered as a candidate timing node for the process. After the process candidate time sequence nodes are extracted, they are bound to the work session to form a set of process candidate time sequence nodes corresponding to the work session. For the same type of node that appears repeatedly in the same work session, such as multiple coarse sampling process candidate nodes appearing due to vacuuming after a work interruption, they are distinguished as first-time candidate nodes and re-entry candidate nodes, and the corresponding reason identifiers are recorded. If a re-entry candidate node appears before the previous process has formed a valid completion state, the time period is marked as an abnormal time sequence branch, and the original node information is retained for subsequent rule matching and node correction processing. If the equipment has been started, but no at least two consecutive process candidate time sequence nodes have been formed within the longest observation period, the work session is marked as an incomplete session and transferred to manual review. If there is a conflict between multiple source data, such as a valve showing as closed while the flow rate is continuously positive, the conflicting nodes are recorded and the original node set is retained, and they are not directly deleted in the current stage. After the above processing, the output job session and its corresponding set of candidate time nodes for the process are generated. The node record includes at least the node type, node generation time, trigger evidence, continuous status and anomaly identifier, which are used to establish verification rules and correct the start and end nodes of the process. Through job session identification and extraction of candidate time nodes for the process, data from different sources, different stages and different granularities can be merged into the same vacuuming operation, thereby improving the consistency and reliability of subsequent time sequence determination.

[0023] S3. Read the preset process specifications and establish verification rules corresponding to each process. The specific implementation is as follows: Using the previously formed work sessions and the set of candidate time nodes for processes as input, the process requirements in the process documents are converted into structured verification rules, and a correspondence is established between the verification rules and the candidate time nodes for processes. The process requirements for vacuuming operations usually come from work instructions, equipment operating procedures, or process cards. Their content generally includes requirements such as maintaining the vacuum level for no less than a specified time after reaching the target range, the next process can only proceed after a certain process is completed, the transition time requirements between adjacent processes should be met, and certain combinations of states cannot be met simultaneously. Since such requirements are difficult to directly participate in automatic judgment if they are kept in their original text form, they are first organized into computable rule objects. Verification rules can be categorized by their function into process threshold rules, process continuity rules, process sequence rules, process interval rules, and mutual exclusion rules. Process threshold rules define the parameter range that a process must satisfy when it enters or completes. Process continuity rules define the minimum duration a process should maintain. Process sequence rules define the sequential relationship between preceding and following processes. Process interval rules define the allowable transition time between adjacent processes. Mutual exclusion rules define the state combinations that cannot be simultaneously established within the same time interval. Through the above classification, requirements scattered across different process documents can be uniformly organized into a set of rules with clearly defined objects and judgment conditions. To facilitate project implementation, the verification rules can be set as single-value rules or range rules according to equipment type, operating conditions, and process level. For the vacuum oil filling scenario of oil-immersed transformers, the completion condition for the high-vacuum pumping process can be selected as a vacuum degree not exceeding 133 Pa. This value corresponds to the target vacuum requirement in the high-vacuum oil filling process and is used to characterize that the equipment has entered the target vacuum state after high-vacuum pumping. The duration of the vacuum holding process can be selected as not less than 24 hours, which corresponds to the conventional operation time for degassing and dehumidification of the transformer body. The duration of the post-filling pressure holding process can be selected as not less than 4 hours, which corresponds to the conventional process requirement of continuing to maintain vacuum and remove residual air bubbles after oil filling. For adjacent... Transition control between processes can be achieved by setting process interval rules based on equipment switching and valve operation requirements. For example, in the scenario of vacuum oil injection for oil-immersed transformers, the transition time from the end of coarse pumping to the start of high pumping can be set to no more than 10 minutes. This time range is used to accommodate valve switching, evacuation path conversion, and state stabilization processes. Exceeding this time usually indicates an operation interruption or switching anomaly. For other scenarios such as vacuum heat treatment furnaces and evacuation of refrigeration systems, the corresponding rule template can be called according to the equipment type identifier, and the threshold value, duration, sequence relationship, and interval conditions corresponding to the scenario can be loaded. The equipment type identifier can be obtained by parsing the equipment number or can be directly given from the operation declaration information. When establishing verification rules, in addition to the rule content, the applicable conditions of the rules are also written. The applicable conditions of the rules include at least the equipment type identifier, process type identifier, preceding process identifier, data source requirements, and abnormal exception conditions. Taking the preceding process identifier as an example, the oil injection process rule can establish a preceding relationship with the vacuum maintenance process. Only when the vacuum maintenance process meets the completion conditions will the oil injection process enter the allowed state. Taking the aforementioned data source requirements as an example, some vacuum threshold rules can require both the top and bottom vacuum data to meet the conditions simultaneously to be considered as compliant. This is because a single point vacuum reaching the target value does not fully represent that the overall space has entered a stable high vacuum state. Abnormal exception conditions are used to handle situations that do not fully follow the normal process chain, such as maintenance and test pumping, no-load test operation, and manual confirmation of shutdown. In such cases, the corresponding exception template can be called to avoid directly judging the exception condition as a violation condition. After the rules are established, node matching relationships are established for various processes. The roughing entry rule corresponds to the candidate nodes of the roughing process, the high-pressure completion rule corresponds to the candidate nodes of the high-pressure process, the vacuum holding duration rule corresponds to the time interval between the vacuum holding entry node and the vacuum holding end node, the oil injection rule corresponds to the candidate nodes of the oil injection process and the flow rate change interval, the post-injection pressure holding rule corresponds to the stable holding interval after the oil injection is completed, and the vacuum breaking rule corresponds to the candidate nodes of the vacuum breaking process. For the candidate vacuum interval in the high-pressure process, it can first be given by the rule template corresponding to the equipment type to represent that the vacuum state has continued to decrease from the roughing stage and entered the candidate range corresponding to the high-pressure process. Then, the formal matching is completed by combining the high-pressure completion conditions. Through the above correspondence, each rule can have a clear matching object, instead of comparing indiscriminately among all candidate nodes. When multiple candidate nodes exist in the same process, evidence integrity classification and priority ranking are performed first, and other matching results are not directly deleted at the current stage. Evidence integrity can be classified and evaluated by combining vacuum continuity before and after the node, valve status consistency, and flow-assisted evidence integrity. When at least two types of evidence among vacuum change, valve status change, and flow change can jointly support the establishment of the node, the node can be marked as a high-confidence candidate node. When only a single piece of evidence supports it, it can be marked as a low-confidence candidate node and retained for further confirmation by subsequent node correction. This is because node identification in vacuuming operations usually relies on the joint support of multi-source data. If it is triggered by only a single signal, it is easily affected by instantaneous noise, sensor drift, or communication anomalies. In case of anomalies, if no candidate node matches the corresponding rule for a certain process, the process is marked as a missing node process; if multiple matching results exist for a certain process, all matching results are retained and given priority, awaiting subsequent start and end node correction processing; if the triggering basis of a candidate node clearly conflicts with the process requirements, such as a vacuum candidate node appearing before the high-pressure pumping completion condition is met, the candidate node is transferred to the abnormal node set and is not used as normal process chain input; through the above processing, information can be avoided from being deleted too early in the rule establishment stage, thus preserving sufficient basis for subsequent node correction and timing chain construction; After the above processing, the output is a set of verification rules corresponding to the job session and a set of candidate time nodes for the process that have completed the preliminary process semantic screening. The set of verification rules includes at least threshold conditions, continuity conditions, sequence conditions, interval conditions and mutual exclusion conditions. The set of candidate time nodes for the process includes at least node type, matching rules, priority and anomaly identification. As a result, subsequent processing no longer deals with the full amount of original time series data, but directly deals with the set of candidate nodes with established rule constraints, thus providing a foundation for subsequent correction of process start and end nodes and construction of job time series chains.

[0024] S4. Determine the start and end nodes of each process based on the multi-source time series and verification rules, and form a work sequence chain according to the sequence of processes. The specific implementation is as follows: Using the verification rule set and the set of candidate time nodes for the process as input, consistency correction is performed on the candidate time nodes. Although the candidate time nodes already have preliminary process semantics, they may still be affected by factors such as early manual reporting, delayed action of the actuator, inconsistent arrival of communication messages, sampling jitter, or short-term abnormal signals, and cannot be directly used as start and end nodes. Therefore, the candidate nodes are corrected by combining the rule conditions corresponding to the candidate nodes, the node credibility level, and the continuous state changes in the multi-source time series, and the corrected start and end nodes are output. The vacuuming operation consists of successive processes such as roughing, high-pressure pumping, vacuum maintenance, vacuum oil injection, post-injection pressure maintenance, and vacuum breaking. There are clear time constraints between each process. Node correction is used to adjust the original candidate nodes to time points that can truly reflect the process boundaries. When calibrating nodes, priority is given to using time points supported by at least two types of evidence as process nodes. Under the premise of meeting the process rules, the earliest established time point is selected as the starting node, and the earliest time point that meets the termination conditions and does not subsequently reverse the recovery is selected as the ending node. The evidence here can be selected from any two of the following: vacuum changes, valve status changes, flow rate changes, and work declaration information. This is because process switching in vacuuming operations is usually accompanied by changes in multiple sources of status. If the node is determined based on only a single signal, it is easy to misjudge short-term noise, false triggering, or recording lag as the formal switching moment. For the calibration of the starting node, check whether the relevant valve status, vacuum change trend and flow change continuously meet the process rules within the corresponding candidate time interval. Taking the high-pressure pumping process as an example, when the high-pressure pumping valve is in the open state, the roughing valve is in the closed state or has been switched to the high-pressure pumping loop, and the vacuum change maintains the candidate change characteristics corresponding to the high-pressure pumping process within the continuous observation window and enters the candidate vacuum interval corresponding to the high-pressure pumping process, the time point when the above conditions are first met simultaneously is backtracked to the nearest unified time grid boundary as the starting node of the high-pressure pumping process. The continuous observation window can be selected from 5 seconds to 15 seconds, with 10 seconds being the preferred value. This setting is because the instantaneous pressure disturbance after the evacuation of the pipeline and valve switching usually decays within a few seconds, and continuing to observe for 10 seconds can take into account both response speed and stability confirmation requirements. Backtracking to the time grid boundary is to unify the timing benchmark of bus messages, switch quantity toggling and analog quantity sampling, which is convenient for subsequent calculation of duration and process interval. For the correction of the end point, a single valve action or a single manual declaration is not used as the sole basis. Instead, it is determined by combining the entry conditions of subsequent processes and the exit conditions of the current process. Taking the vacuum maintenance process as an example, the end point of the vacuum maintenance process is not directly based on the manual oil injection declaration time. Instead, the end point of the vacuum maintenance process is determined when the oil injection valve is opened, the flow rate changes from zero to a continuous positive value, or the vacuum change no longer meets the stable maintenance conditions corresponding to the vacuum maintenance process. The stable maintenance condition can be selected as maintaining the vacuum fluctuation amplitude within the allowable range corresponding to the vacuum maintenance rules within a continuous 30-second observation window. When this condition is broken, it indicates that the current process has exited the stable maintenance state. The continuous observation window can be dynamically configured based on the historical vacuum fluctuation recovery time, valve action feedback time, and pressure equalization time. When the on-site statistics show that the overall normal disturbance recovery time is extended or shortened, the observation window can be adjusted accordingly. This approach is because there may be situations where the declaration is made before the action is taken in the field. If the previous process is directly cut off based on the declaration time, it is easy to shorten the actual duration of the vacuum maintenance process and affect the subsequent verification of the requirement of maintaining the vacuum for no less than 24 hours. When a reverse state recovery occurs shortly after a candidate time point, the candidate point is not directly determined as the start or end point. Instead, the search continues to look for the time point when the first stable state is established. Examples of reverse state recovery include the valve immediately returning to its original state after switching, the flow rate quickly returning to zero after starting, or the vacuum change direction failing to be maintained. The correction window used to determine the reverse state recovery can be selected from 5 to 10 seconds, with 10 seconds being the preferred value. This setting is because the transitional disturbances caused by valve switching, pump start-up and shutdown, and pipeline impacts are usually completed within a few seconds. 10 seconds can cover most transitional disturbances without significantly delaying node confirmation. This time is also consistent with the aforementioned continuous observation window, which is beneficial for using a unified time scale throughout the time series identification process. After correcting the start and end points of each process, a work sequence chain is constructed based on the process sequence. This work sequence chain is not a simple time sorting of all nodes, but rather a connection based on the relationships between preceding processes, process types, node reliability levels, and rule matching results. The starting point of the sequence chain can be the start point of the work session, followed by sequentially connecting the time intervals of processes such as roughing, high-pressure pumping, vacuum maintenance, oil injection, post-injection pressure maintenance, and vacuum breaking, forming a work sequence chain corresponding to the actual vacuuming process. If a process lacks a formal node, but its preceding and following process nodes are clear, and there are continuous and interpretable state changes in between, a presumed node can be generated and written with a presumed identifier. A presumed node refers to a process node supplemented based on preceding and following process nodes and continuous state changes; for example, the high-pressure pumping end node... If the vacuum maintenance start point is clearly defined, and the vacuum state has met the conditions for high-pressure pumping completion and remained stable in the preceding time period, then the nearest high-vacuum stabilization time point before the vacuum maintenance start point can be used as the presumed end point of high-pressure pumping. The presumed point is only used for non-critical safety action procedures and is not used to replace critical action points involving safety interlocks. When critical actions such as vacuum oil injection start and vacuum breaking start are involved, the formal point must be determined based on clear evidence first, and a pure presumed method should not be used as a substitute. This is because critical action points are directly related to subsequent interlock control and anomaly judgment. If they are presumed based solely on the relationship between the preceding and following points, it is easy to amplify safety risks. For non-critical procedures, the presumed point can maintain the integrity of the time sequence in the case of missing data, which is convenient for subsequent calculation of duration and process interval. When constructing the job sequence chain, it also handles job interruption and restart situations. When the interval between adjacent processes exceeds the upper limit of the corresponding process interval rule and continues to reach the pause judgment time, a pause segment is inserted into the sequence chain. The cancellation judgment time can be selected as 20 minutes. This time is higher than the normal transition upper limit of no more than 10 minutes from the end of coarse pumping to the start of high pumping in the vacuum oil injection scenario of oil-immersed transformers. It is used to distinguish between normal switching delay and actual job interruption. When the interval continues to reach 20 minutes, it can usually be regarded as equipment waiting, fault handling or manual interruption. The pause segment is not included in the process duration, but is retained as the basis for subsequent process interval anomaly analysis. If the same process entry node reappears after the pause, the pause segment is retained in the original sequence chain and continues to connect subsequent nodes to reflect the job restart process. In case of anomalies, if the time intervals of two different processes overlap, such as the oil injection time interval and the vacuum holding time interval being valid at the same time, the process node that is consistent with the valve status change and flow rate change is retained first, and the other node is transferred to the conflict node set; then the conflict node set is judged as an anomaly according to the mutual exclusion rule. This is because overlap usually indicates that there is a conflict in the process switching identification. If all conflict nodes are deleted directly at the current stage, it is easy to lose the evidence required for subsequent anomaly analysis. Therefore, retaining the conflict information first is more conducive to subsequent verification. After the above processing, the output includes the corrected start and end nodes and the operation sequence chain formed by sequentially connecting the time intervals of each process. The operation sequence chain includes at least the process type, start node time, end node time, node nature, node reliability level, pause flag, and conflict flag. Thus, the start time, end time, sequential connection relationship, and whether there is a pause or restart of each process in the same vacuuming operation can be clearly expressed on a unified time axis, thereby providing direct input for subsequent verification of process duration, process sequence, and process interval.

[0025] S5. Verify the duration, sequence, and interval of each process based on the work sequence chain, and generate a verification conclusion. The specific implementation is as follows: Using the job sequence chain as input, the duration, sequence, and interval of each process are checked, and a check conclusion is generated for subsequent control and archiving. The job sequence chain has already given the start node, end node, pause segment, conflict identifier, and node confidence level of each process. Therefore, the check is performed directly on the time interval corresponding to the same vacuuming operation. When verifying the duration, first calculate the time difference between the end node and the start node of each process, then combine the pause segment correction to obtain the effective duration, and compare it with the corresponding process duration rule; if there is a pause segment within the process time interval, the pause segment duration is removed from the duration, and the number of pauses and the total pause duration are recorded; if the equipment does not continuously execute the corresponding process during the pause, it is not included in the effective process holding time; the number of pauses and the total pause duration are used to reflect the equipment operation stability and operation execution status; for the vacuum oil filling scenario of oil-immersed transformers, the effective duration of the vacuum holding process can be selected as not less than 24 hours, which corresponds to the conventional operation time of degassing and dehumidification of the transformer body; the effective duration of the pressure holding process after filling can be selected as not less than 4 hours, which corresponds to the conventional process requirement of continuing to maintain vacuum and remove residual air bubbles after oil filling; during verification, the effective duration is obtained by subtracting the total pause duration from the process time interval, and then it is determined whether the effective duration meets the corresponding process requirements; During sequence verification, the adjacent processes in the work sequence chain are checked one by one to see if they meet the rules of the preceding processes corresponding to the pre-set process specifications. Sequence verification not only compares the order of processes on the time axis, but also checks whether the preceding process meets the completion conditions. If the following process is after the preceding process in time, but the preceding process does not meet the corresponding continuity rule or threshold rule requirements, the sequence verification is still judged as not meeting the requirements. For example, in the vacuum oil filling scenario of an oil-immersed transformer, the oil filling process should be after the vacuum maintenance process, and the vacuum maintenance process should have met the corresponding continuity rule. If the vacuum maintenance process is ahead, but the effective duration is insufficient, the sequence condition of the oil filling process is still not met. The process sequence reflects both the time sequence and the succession relationship of the process completion status. During interval verification, the time difference between the end point of the previous process and the start point of the next process is compared with the corresponding process interval rules. Process interval rules may include maximum interval conditions and minimum interval conditions. The maximum interval condition is used to limit the excessive gap between processes to avoid work interruption, vacuum leakage, or abnormal waiting. The minimum interval condition is used to limit the necessary preparation time between adjacent processes to ensure that valve action is completed, status confirmation is completed, or medium switching is completed. For the vacuum oil injection scenario of oil-immersed transformers, the maximum switching interval from the end of coarse pumping to the start of high pumping can be selected to be no more than 10 minutes. This time limit corresponds to the normal range of valve switching, air extraction path conversion, and status stabilization process. If this time is exceeded, it is judged as an interval abnormality. For processes that need to set a minimum interval, the minimum interval can be taken as the sum of the valve action completion time and the status confirmation time, or the minimum transition time given by the corresponding equipment rule template, so that the minimum interval is consistent with the on-site equipment action process, and does not set a fixed value without basis outside the process. To improve the reliability of the verification conclusions, a sub-item verification result is output for each type of verification item, rather than simply outputting "qualified" or "unqualified." The sub-item verification result should include at least the verification category, target process, rule number, actual value, rule value, deviation direction, evidence integrity level, and anomaly indicator. Evidence integrity levels are divided into three levels: high, medium, and low. A high level indicates that at least two of the three main types of evidence (vacuum, valve, flow) jointly support the conclusion, and there is no obvious conflict. A medium level indicates that at least one main type of evidence supports the conclusion, and the remaining evidence does not form an obvious conflict. A low level indicates that only a single piece of evidence supports the conclusion, or there is a conflict among the main pieces of evidence, or key data is missing. When the lack of key data prevents the formation of more than two types of valid evidence, the level is preferentially downgraded to low. For abnormal situations, if the start or end point of a process is missing, the duration check is marked as undeterminable and not directly considered as unqualified; if the sequence of processes is reversed, such as the oil injection process being completed before the vacuuming process, the sequence check is marked as a serious abnormality; if the process interval exceeds the maximum interval condition and there is a pause in the middle, the result is preferentially classified as an operation interruption type abnormality, rather than directly classified as a process violation type abnormality; operation interruption type abnormalities and process violation type abnormalities usually correspond to different handling methods in subsequent control actions; When multiple abnormal conclusions exist simultaneously in the same process, a hierarchical conclusion strategy is adopted to determine the main conclusion. Within the same process, the priority of abnormal conclusions can be selected as follows: sequential abnormality is higher than duration abnormality, and duration abnormality is higher than interval abnormality. Incorrect process sequence usually indicates that the process logic has been disrupted, and its risk is higher than that of simply insufficient duration or excessive interval. Insufficient duration directly affects the degassing, dehumidification, or defoaming effect, and its risk is usually higher than that of general interval abnormality. This priority is only used for the selection of the main conclusion when multiple abnormalities coexist in the same process, and does not replace the retention of other abnormal information. Other abnormal information is still recorded as supplementary conclusions. After all checks are completed, a job-level check conclusion is generated. The job-level check conclusion includes at least the overall compliance status, the set of abnormal processes, the set of abnormal categories, the control recommendation level, and the archiving identifier. The control recommendation level can be divided into five categories: allow to continue, pause and wait, alarm prompt, prohibit to continue, and require manual confirmation. It can be further mapped to the action token or interlock action type of the control execution end. The control recommendation level is determined comprehensively based on the severity of the abnormality, the level of evidence integrity, and the scope of the abnormality's impact: when the conclusion is undeterminable or the level of evidence integrity is low, it will first enter the stage of requiring manual confirmation; when there are only minor interval exceedances or general non-critical abnormalities, it can enter the stage of alarm prompt or pause and wait; when there are critical sequence abnormalities, critical duration insufficient, or high-confidence level serious abnormalities, it will first enter the stage of prohibiting to continue. The generated job-level check conclusion serves as the direct input for subsequent control and archiving processes.

[0026] S6. Based on the verification results, execute interlock control, trigger alarms, or record and archive data. The specific implementation is as follows: Using the aforementioned work-level verification conclusion as input, interlock control, alarm output, and record archiving are executed according to the conclusion level. When necessary, waiting observation, manual takeover, re-archiving, and controlled deactivation processes are performed. In this application, interlock control refers to the control processing by which the verification execution end restricts or switches the authorization for subsequent process starts, the execution authority of valves or actuators, and the status of manual takeover based on the verification conclusion. Since the work-level verification conclusion already provides the overall compliance status, the set of abnormal processes, the set of abnormal categories, the control recommendation level, and the archiving identifier, this stage does not re-evaluate the original time-series data but directly drives the execution action based on the verification conclusion. When the verification conclusion is that the operation can continue, the current interlock status remains unchanged, and the current verification result is written into the operation record, waiting for subsequent processes to continue input; when the verification conclusion is an alarm prompt, an alarm message is output to the field interactive terminal, while the current authorized status of the valve and vacuum pump is not changed; the alarm message includes at least the operation number, abnormal process name, abnormality type, actual value, rule value, and suggested inspection items; for cases with only minor over-limits, short-term fluctuations, or non-critical abnormalities, alarm prompts guide operators to continue observation or inspection without changing the existing control status; When the verification conclusion is to pause and wait, an interlock waiting command is output, freezing the start authorization of the next process, while continuing to collect relevant data for the current process and maintaining the monitoring of the current process; the pause and waiting time is not set separately from the equipment action process, but is determined by the rule template based on the valve action completion time, pressure equalization time, status confirmation time, and short-term fluctuation recovery time; for the vacuum oil injection scenario of oil-immersed transformers, the pause and waiting time can be selected as 120 seconds; this time can cover the common observation cycle of valve stabilization, pressure equalization, and status recovery, and is significantly lower than the operation interruption judgment time, which is suitable for continuing to observe short-term anomalies; if the anomaly disappears within the waiting window, the waiting state is lifted and the verification conclusion is updated; if the anomaly still exists, the process is switched to prohibiting continuation or manual confirmation based on the updated conclusion; When the verification conclusion is that continuation is prohibited, an interlock control action is output to the control execution terminal to revoke the start authorization of the next process or cut off the execution permission corresponding to the abnormal process, and maintain the current alarm output status. For the vacuum oil filling scenario of oil-immersed transformers, when anomalies occur such as attempting to enter the oil filling process before vacuum maintenance is completed, the vacuum breaking action and oil filling flow rate are established simultaneously, or a process switch occurs when the critical vacuum range is not met, the start authorization of the oil filling-related valves can be closed to prohibit entry into the next process. Such anomalies have directly affected the process continuity or safety boundary and should not be eliminated by simply continuing to observe. When the verification conclusion requires manual confirmation, the current operation is switched to manual takeover mode. No mandatory actions are automatically issued. The current output status remains unchanged, subsequent status changes are continuously recorded, and a confirmation request is sent to the operation leader. Situations requiring manual confirmation include at least low evidence integrity level, missing key data, conflicting verification items, or the current scenario being an off-standard maintenance state. The manual takeover method is adopted to avoid executing mandatory interlocks based solely on automatic conclusions under low credibility or special working conditions. In addition to executing control actions, records are also archived. The archived content not only saves the final verification conclusion but also key evidence fragments supporting that conclusion. The archived information includes at least the job number, equipment number, job start time, job end time, verification rule version, start and end times of each process, duration results, sequential verification results, interval verification results, conclusion level, executed actions, and anomaly identification. For critical anomalies, the original multi-source time series fragments before and after the anomaly are also saved. The saving window for the anomaly fragments is determined by the archiving template corresponding to the anomaly type. For the vacuum oil injection scenario of oil-immersed transformers, 120 seconds before and after the anomaly point can be selected. This time can cover most of the anomaly formation, persistence, and elimination processes, facilitating subsequent review and cause analysis. If remote archiving fails due to network interruption, data is written to local non-volatile storage first, with an "upload pending" tag attached. Once the network is restored, data is re-uploaded in the order of job time to ensure no records are lost. If local storage space is insufficient, full data for abnormal and manually taken-over jobs is retained first, while summary records are retained for fully compliant jobs. Archived conclusions can optionally be fed back to the rule update end for subsequent rule parameter adjustments. When the same type of equipment experiences the same type of boundary anomaly multiple times under the same environmental conditions, and subsequent verification shows that the anomaly did not cause actual process failure, the corresponding historical records can be transferred to rule correction analysis to assist in adjusting relevant process time limits, observation windows, or exception templates. If a prohibition on continuing action has been issued, but subsequent manual confirmation indicates that the current scenario falls under an allowed maintenance exception, then controlled release processing can be performed. Controlled release must simultaneously meet at least three conditions: manual authorization, exception template matching, and current state resampling. Interlock control can only be released when all three conditions are met. After the above processing, the work-level verification conclusion is converted into on-site executable interlock actions, alarm outputs, and archived results, and a closed-loop handling process is formed through mechanisms such as waiting for observation, manual takeover, re-transmission of archives, and controlled release.

[0027] Example 2: Based on Example 1, the specific application process of a vacuuming operation timing verification method is further explained: Taking a 220kV / 180MVA oil-immersed transformer as an example, the monitored object is under normal steady-state conditions, with an ambient temperature of 25℃ and an altitude of 50m. Two vacuum gauges are configured on-site, connected to the top of the transformer body and the bottom of the oil tank respectively. Four temperature acquisition points are configured and distributed in the upper and lower oil channels of the transformer body. A flow acquisition device is connected in series at the outlet of the vacuum oil filter. The verification gateway is set to the PTP master clock domain, and the synchronization deviation of each slave node is controlled within ±50μs. In scenarios where PTP deployment is not possible, IRIG-B code or the high-precision crystal oscillator and interrupt response mechanism built into the verification gateway can be used to provide the same clock reference. The corresponding process requirements include at least the following conditions: the pressure drops to ≤2000Pa within 30 minutes during the rough pumping stage, the pressure is maintained at ≤133Pa for no less than 24 hours after high pumping, the vacuum continues for no less than 4 hours when the pressure is 200mm from the top of the tank, and the vacuum change rate is maintained within the allowable range during the vacuum maintenance period and the pressure maintenance period after injection. At the start of the operation, the verification gateway receives the operation start intention at 08:00:00.000 and simultaneously receives human-machine interaction information from the vacuum gauge, temperature acquisition point, flow acquisition device, valve status contact and host computer. The verification gateway first writes a unified timestamp to the multi-source data based on the same clock reference and establishes the vacuuming operation session on the unified time axis. Then, it associates the vacuum pump start status, pumping path switching status, vacuum change trend and manual reporting information to form a candidate node sequence for processes such as rough pumping, high pumping, vacuum maintenance and oil injection. Since the arrival time of data messages from different sources is not completely consistent, the arrival time of a single message is not directly used as the process switching time. Instead, the continuous state changes under the unified time grid are used as the basis to correct the process candidate nodes, thereby determining the formal start and end boundaries of each process. During the roughing stage, the vacuum change trend and its holding time are continuously monitored. When the body pressure drops to 1980 Pa and is maintained for 10 seconds, this moment is identified as the roughing standard anchor point, and the roughing stage is determined to meet the process requirements. Subsequently, the status of valves, the status of the pumping path, and the vacuum change trend related to the high-pressure pumping are continuously tracked. At 10:48:09.770, the vacuum is detected to have stabilized at 121 Pa, which corresponds to the high-pressure pumping completion rule. Thus, the state after this moment is identified as entering the vacuum maintenance stage. At this time, not only is it checked whether the vacuum value has entered the target range, but the valve status and continuous stability are also combined to confirm the process switch, so as to avoid prematurely determining the end of high-pressure pumping based on a single vacuum value. During the vacuum holding phase, the effective holding time was continuously calculated along the operation sequence chain, and any pauses or interruptions were checked. At 11:02:30.117 the next day, the system calculated the effective vacuum holding duration to be 24 hours, 14 minutes, and 20 seconds, which is greater than the corresponding process requirement of 24 hours. Therefore, the vacuum holding process was deemed compliant and completed. Subsequently, the oil injection process was checked. By using the cumulative flow information from the flow acquisition device and the correspondence between the flow rate and volume, the critical moment when the oil was injected to 200 mm from the top of the tank was located, and this moment was used as the switching boundary for the subsequent post-injection pressure holding process. Afterward, vacuuming continued for 4 hours, 12 minutes, and 31 seconds, which met the process requirement of no less than 4 hours. Therefore, the post-injection pressure holding process was also deemed compliant. Thus, no abnormal conclusions were found in the three dimensions of duration, process sequence, and process interval in this operation. In this operational scenario, since the roughing operation met the requirements, the high-pressure pumping was completed, the vacuum holding time was sufficient, and the post-injection pressure holding time all met the process requirements, and no sequence reversal, abnormal pauses, or conflicting nodes occurred during the operation, the generated operation-level verification conclusion was compliant. Correspondingly, the control output remained in the allowed-to-continue state, without issuing interlocking blocking commands or triggering mandatory alarms, only maintaining normal operation authorization, and writing the full-process verification results into the operation record. This operation process shows that this method does not only make instantaneous judgments on a single threshold, but also uniformly summarizes the operation session identification, rule matching, node correction, time sequence chain construction, and sub-item verification results before forming an operation-level conclusion, thereby objectively reflecting whether the entire vacuuming operation meets the process requirements. During the archiving phase, the system signs and stores the complete data, verification conclusions, and execution results of this operation, and generates corresponding archive records. The archive records include at least the operation start time, the time when the coarse pumping meets the standard, the time when the high-pressure pumping is completed, the time when the vacuum holding is completed, the time when the key switching of oil injection is completed, the time when the post-injection pressure holding is completed, and the corresponding verification conclusions. If subsequent verification is required, the entire time chain of the operation can be reproduced based on the above archive records. If a network interruption or remote archiving failure occurs during subsequent operation, the system will first write the archive results to local non-volatile storage and re-transmit them in chronological order after the network is restored. Thus, this method can not only complete the online verification during the vacuuming operation, but also retain complete process evidence after the operation is completed for verification, traceability, and subsequent rule correction. In other operating scenarios, if the system detects inconsistencies between the top and bottom vacuums during the vacuum maintenance phase, inconsistencies between valve status and flow rate changes during the oil injection switching phase, or waiting intervals exceeding the rule limits between adjacent processes, the aforementioned compliance conclusions will no longer be applied. Instead, the corresponding results will be transferred to alarm prompts, pause waiting, prohibition of continuation, or manual confirmation processing paths according to the logic of node correction, anomaly classification, and control recommendations, to ensure that control actions are consistent with the actual process status. This processing method is consistent with the technical boundaries of this approach, which is oriented towards multi-source asynchronous data in industrial settings, emphasizes a unified time base, and provides full-process traceability verification.

[0028] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0029] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented in software, the above embodiments can be implemented in whole or in part by a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions of the embodiments of this application are implemented in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted wirelessly or wiredly from one website, computer, server, or data center to another website, computer, server, or data center. Wired methods include optical fiber, twisted pair, coaxial cable, etc. Wireless methods include infrared, microwave, etc. Available media include any available media that can be accessed by a computer or data storage devices such as servers and data centers that contain one or more sets of available media. Available media can be magnetic media (floppy disks, hard disks, magnetic tapes), optical media (DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0030] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0031] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for verifying the timing of vacuuming operations, characterized in that, include: S1. Collect vacuum level, temperature, flow rate, valve status and work application information, and write a unified timestamp for each data based on the same clock reference and perform time alignment to form a multi-source time series; S2. Determine the work session based on the work application information and the status changes in the multi-source time series, and extract the time sequence nodes corresponding to each process; S3. Read the preset process specifications and establish verification rules corresponding to each process. S4. Determine the start and end nodes of each process based on the multi-source time series and verification rules, and form a work sequence chain according to the sequence of processes; S5. Verify the duration, sequence, and interval of each process according to the work sequence chain, and generate a verification conclusion. S6. Based on the verification results, execute interlock control, trigger alarms, or record and archive data.

2. The method for verifying the timing of vacuuming operations according to claim 1, characterized in that, S1 includes: Vacuum level, temperature, flow rate, valve status, and work application information are all integrated into the verification execution terminal, and a unified timestamp is written using the same clock reference provided by the verification execution terminal. The same clock reference is achieved through any one or a combination of PTP, IRIG-B code, a high-precision crystal oscillator built into the verification execution terminal, and an interrupt response mechanism. For data accessed via the communication interface, the sampling time is calculated by combining the message length, interface transmission rate, and fixed processing delay of the device. For digital input data, record the event moment when the edge flips; Resampling is performed using a unified time grid. For data that has not been updated, the previous values ​​are used and missing states are marked according to the number of consecutive sampling periods, forming a multi-source time series that includes timestamps, data values, source identifiers, validity identifiers, update time identifiers, and reuse identifiers.

3. The method for verifying the timing of vacuuming operations according to claim 1, characterized in that, S2 includes: The operation start declaration information is linked with the vacuum pump start-up, main vacuum valve operation, and vacuum change trend within the relevant time window to establish an operation session; Along a unified time axis, candidate timing nodes for each process are extracted based on valve status, vacuum changes, flow rate changes, and the continuous satisfaction of status conditions. The candidate timing nodes of each process are bound to the job session, and supplementary events, duplicate nodes, conflicting nodes and incomplete sessions are marked.

4. The method for verifying the timing of vacuuming operations according to claim 1, characterized in that, S3 includes: The process documents are parsed into threshold rules, continuity rules, sequence rules, interval rules, and mutual exclusion rules; Establish rule templates and their applicable relationships according to equipment type, preceding process, data source, and exception conditions; Each rule is matched with the candidate time nodes or corresponding time intervals of the corresponding process, and the candidate nodes are classified into credibility levels, priority rankings and anomaly markings based on vacuum, valve and flow evidence.

5. The method for verifying the timing of vacuuming operations according to claim 4, characterized in that, Establish rule templates and their applicable relationships according to equipment type, preceding processes, data sources, and exception conditions, including: Based on the equipment number parsing results or work application information, determine the equipment type identifier, call the rule template corresponding to the equipment type, and load the corresponding threshold value, duration, sequence relationship and interval conditions; Write the process type identifier, preceding process identifier, data source requirements, and exception templates corresponding to maintenance sampling, no-load trial operation, and manual confirmation shutdown according to the correspondence of each process.

6. The method for verifying the timing of vacuuming operations according to claim 1, characterized in that, S4 includes: Based on the rule conditions corresponding to the candidate time-series nodes, the node trust level, and the continuous changes in multi-source states, consistency correction is performed on the candidate time-series nodes; With the combined support of two types of evidence—vacuum changes, valve status changes, flow rate changes, and work report information—the earliest established time point is determined as the starting point, and the earliest time point that meets the termination conditions and does not subsequently reverse is determined as the ending point. Presumed nodes, pause segments, and conflict nodes are marked and connected according to their preceding process relationships to form a work sequence chain.

7. The method for verifying the timing of vacuuming operations according to claim 6, characterized in that, Based on the rule conditions corresponding to the candidate time-series nodes, the node trust level, and the continuous changes in multi-source states, consistency correction is performed on the candidate time-series nodes, including: Within the time interval corresponding to the candidate timing node, the consistency between the valve status, vacuum changes, and flow rate changes and the process rules is verified by combining the node's credibility level. For the time point when the conditions for node establishment are first met, backtrack to the nearest unified time grid boundary; If a short-term reverse state recovery occurs, or if the conditions for subsequent processes to enter are not stably met, continue to verify the stable state at the next stable point.

8. The method for verifying the timing of vacuuming operations according to claim 1, characterized in that, S5 includes: The duration of each process in the work sequence chain is compared after deducting pauses; Perform sequential verification according to the completion conditions of the preceding processes; The maximum interval condition and minimum interval condition are checked based on the time difference between the start and end points of adjacent processes. It also generates itemized verification information and task-level verification conclusions by combining evidence integrity level, anomaly priority, and control recommendation level.

9. The method for verifying the timing of vacuuming operations according to claim 8, characterized in that, Perform sequential verification according to the completion conditions of the preceding processes, including: Read the identifier of the preceding process in the current process of the job sequence chain, and locate the end node, completion rule and node trust level of the corresponding preceding process; Determine if the end node of the preceding process exists, and verify whether the preceding process simultaneously meets the corresponding threshold rules, continuity rules, and trust level requirements; The marking order is abnormal when the start node of the current process is earlier than the valid end node of the previous process, or when the previous process has not met the completion rules but the current process has already started. Write the current process, preceding process, triggering evidence, and exception level corresponding to the sequence anomaly into the sub-item verification information, and provide the sequence verification results to the job-level verification conclusion.

10. The method for verifying the timing of vacuuming operations according to claim 1, characterized in that, S6 includes: Interlock control includes one or more of the following: prohibiting the start of the next process, cutting off the execution authority of the valve or actuator corresponding to the abnormal process, and switching to manual control. According to the control recommendation level corresponding to the work-level verification conclusion, switch the interlock status, alarm output status and manual takeover status. For observed anomalies, the next process should be frozen and authorization should be granted while data collection continues. For anomalies that still exist in the waiting window, the process should be prohibited from continuing or require manual confirmation. For critical anomalies, the corresponding execution permissions should be revoked. Archive the verification conclusions and abnormal evidence fragments. If archiving fails, perform local temporary storage and sequential retransmission. Release the interlock control when manual authorization, exception template matching and current state resampling are satisfied at the same time.

Images