An information processing method and system for engineering quality inspection

Abstract

This invention discloses an information processing method and system for engineering quality inspection, relating to the field of engineering quality inspection technology. The method includes: continuously receiving raw records from various sources and writing them into a raw record buffer; extracting real-time features from a feature extraction entry point to form window features; updating the current reference state through a stable window; generating event triggering results by combining window features, initial reference state, current reference state, planned process sequence, record integrity within the window, and identity consistency; determining the relevant source set based on the contribution of the window features corresponding to each source to the event triggering results; and generating window judgment records. For the raw records in the raw record buffer during the inspection process, based on the window state labels represented by the window judgment records, continuous segmentation is performed to form regular evidence blocks and event evidence blocks, while simultaneously generating event switching records. This invention achieves dynamic identification of the inspection process status and directional positioning of relevant sources.

Description

Technical Field

[0001] This invention relates to the field of engineering quality inspection technology, and in particular to an information processing method and system for engineering quality inspection. Background Technology

[0002] Engineering quality inspection involves the collaborative processing of heterogeneous information from multiple sources, including testing equipment data, image data, on-site records, operation trajectories, and environmental parameters. Existing information-based methods for engineering quality inspection typically revolve around task filing, sampling data access, feature extraction, status determination, result storage, and report generation. By centrally managing and analyzing data from the testing process, the standardization, digitalization, and traceability of testing activities can be improved.

[0003] In conventional methods, the real-time identification link and the original evidence retention link are often organized separately, which can easily lead to coordination difficulties between the timeliness requirements of anomaly identification and the continuous preservation requirements of all original evidence. In addition, data in different states are usually organized in a similar way, resulting in insufficient connection and expression of key records before and after the event, which affects the fine tracing of the anomaly evolution process. Summary of the Invention

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

[0005] Therefore, this invention provides an information processing method for engineering quality inspection to solve the problems of difficulty in coordinating real-time discrimination and continuous preservation of original evidence, as well as insufficient connection between key records before and after an event, in the existing technology.

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

[0007] In a first aspect, the present invention provides an information processing method for engineering quality inspection, comprising: collecting basic task information of engineering quality inspection activities and generating a unique inspection task identifier; generating an initialization evidence block by establishing a source mapping relationship and a unified time benchmark; generating an initial reference state and a planned process sequence based on the initialization evidence block; establishing an original record cache and a feature extraction entry point; continuously receiving original records from various sources and writing them into the original record cache; extracting real-time features from the feature extraction entry point to form window features; updating the current reference state through a stable window; generating an event triggering result by combining window features, initial reference state, current reference state, planned process sequence, record integrity within the window, and identity consistency; determining the relevant source set based on the contribution of the window features corresponding to each source to the event triggering result; generating a window judgment record; continuously segmenting the original records in the original record cache during the inspection process according to the window state label represented by the window judgment record to form a regular evidence block and an event evidence block, while generating an event switching record; generating a solidified summary for the initialization evidence block, window judgment record, regular evidence block, event evidence block, and event switching record respectively; and connecting the solidified summaries in a unified time order to form an evidence chain and generate a conclusion record and an inspection report record.

[0008] As a preferred embodiment of the information processing method for engineering quality inspection described in this invention, the step of collecting basic task information of engineering quality inspection activities and generating a unique inspection task identifier, and generating an initialization evidence block by establishing a source mapping relationship and a unified time benchmark, includes: collecting basic task information, and for each source, collecting the source number, source category, installation location, the object being inspected bound to the source, and the source's recording rules; performing integrity and consistency checks, and combining and encoding the engineering project number, the object being inspected number, the inspection batch number, the inspection project number, and the inspection start date to obtain a unique inspection task identifier; establishing a source mapping relationship and a unified time benchmark, and collecting standard reference information, self-inspection records before the start of inspection, and the first stable record after the start of inspection that meets the initial stable inclusion conditions to generate an initialization evidence block.

[0009] As a preferred embodiment of the information processing method for engineering quality inspection described in this invention, the step of generating an initial reference state and a planned process sequence based on an initialization evidence block, and establishing an original record cache and a feature extraction entry point includes: extracting baseline feature components from the initialization evidence block, generating an initial reference vector and an initial discrete matrix, and forming an initial reference state; generating a planned process sequence according to the action requirements and source mapping relationship in the detection basis; establishing an original record cache using a unique detection task identifier, source number, and unified timestamp as indexes, and establishing a feature extraction entry point that reads original records only from the original record cache.

[0010] As a preferred embodiment of the information processing method for engineering quality inspection described in this invention, the step of continuously receiving raw records from various sources and writing them into the raw record buffer, and extracting real-time features from the feature extraction entry point to form window features includes: continuously receiving raw records generated from various sources and writing them into the raw record buffer; reading the raw records from the feature extraction entry point; extracting real-time features by calling feature extraction rules according to the source number and source category; normalizing all extracted real-time features; setting a sliding time window; and aligning and aggregating the real-time features within the current sliding time window according to a unified timestamp to form window features.

[0011] As a preferred embodiment of the information processing method for engineering quality inspection described in this invention, the step of updating the current reference state through a stable window and generating an event-triggered result by combining window features, initial reference state, current reference state, planned process sequence, record integrity within the window, and identity consistency includes: calculating the record integrity within the window, identity consistency within the window, and the matching degree of the planned process sequence based on the original records and action records within the current sliding time window; calculating the double-reference deviation of the current sliding time window by combining window features, initial reference state, and current reference state; determining the current sliding time window as a stable window when the double-reference deviation, record integrity within the window, identity consistency within the window, and the matching degree of the planned process sequence each meet the corresponding stable threshold, and updating the current reference state based on the window features; and generating an event-triggered result by using an event-triggered coupling algorithm by combining window features, initial reference state, current reference state, planned process sequence, record integrity within the window, and identity consistency.

[0012] As a preferred embodiment of the information processing method for engineering quality inspection described in this invention, the step of determining the relevant source set and generating a window judgment record based on the contribution of window features corresponding to each source to the event triggering result includes: extracting window feature sub-vectors corresponding to each source and reference sub-vectors corresponding to each source in the current reference state; characterizing the contribution of window features corresponding to each source to the event triggering result based on the deviation between the window feature sub-vectors corresponding to each source and the reference sub-vectors, and using the proportion of the deviation of each source to the total deviation of all sources; determining the sources with a contribution level not less than the source contribution threshold as the relevant source set; when no source meets the source contribution threshold, selecting the source with the largest contribution level, and including the operation trajectory source and the on-site record source in the relevant source set; and generating a window judgment record based on the event triggering result and the relevant source set.

[0013] As a preferred embodiment of the information processing method for engineering quality inspection described in this invention, the step of continuously segmenting the original records in the original record buffer during the inspection process according to the window state label represented by the window determination record to form regular evidence blocks and event evidence blocks includes: when the window determination record represents a regular state, using the source number as the smallest segmentation object, forming a regular evidence block according to the regular segmentation rules controlled by the regular segmentation duration; when the window determination record first represents an event state, generating an event number and determining the initial relevant source set, using the start time of the current sliding time window as the event start boundary; for the original records corresponding to the relevant source set before the event start boundary, with a length equal to the previous window backfill duration, forming an event evidence block according to the event segmentation rules; and continuing to form regular evidence blocks for the original records corresponding to non-relevant sources according to the regular segmentation rules.

[0014] As a preferred embodiment of the information processing method for engineering quality inspection described in this invention, the simultaneous generation of event switching records includes: when the current event has not ended and subsequent window judgment records continue to characterize the event state, reading the relevant source set corresponding to the subsequent window judgment records; applying the same front window backfilling rules and event segmentation rules as when the event is first triggered to newly added sources; when the window judgment record recovers from the event state to the normal state, using the recovery confirmation counting rules to perform an end judgment on the current event, determining the event end boundary, and generating an event switching record; after the current event ends, resuming the execution of the normal segmentation rules for each source in the relevant source set of the current event.

[0015] As a preferred embodiment of the information processing method for engineering quality inspection described in this invention, the step of generating solidified summaries for initialization evidence blocks, window judgment records, regular evidence blocks, event evidence blocks, and event switching records, and then connecting them in a unified chronological order to form an evidence chain and generate conclusion records and inspection report records includes: performing standardized organization on initialization evidence blocks, window judgment records, regular evidence blocks, event evidence blocks, and event switching records, and generating solidified summaries for each using a standard hash algorithm; sorting various objects in a unified chronological order, and connecting them in a unified chronological order to form an evidence chain; and extracting corresponding objects from the evidence chain based on the unique inspection task identifier, the inspected object number, and the inspection item number to generate conclusion records and inspection report records.

[0016] Secondly, this invention provides an information processing system for engineering quality inspection, comprising: a task filing module for collecting basic task information, generating a unique inspection task identifier, establishing source mapping relationships and a unified time benchmark, and generating initialization evidence blocks, initial reference states, planned process sequences, original record cache areas, and feature extraction entry points; a state determination module for receiving original records from various sources, extracting real-time features and forming window features, and generating event triggering results, related source sets, and window determination records; an evidence segmentation module for performing regular segmentation and event segmentation on the original records based on the window state labels represented by the window determination records, forming regular evidence blocks and event evidence blocks, and performing front window backfilling and related source updates in the event state; and an evidence output module for generating solidified summaries for the initialization evidence blocks, window determination records, regular evidence blocks, event evidence blocks, and event switching records, connecting them to form an evidence chain, and generating conclusion records and inspection report records.

[0017] The beneficial effects of this invention are as follows: by forming window features and generating event triggering results, dynamic identification of the detection process status and directional location of related sources are realized; by continuously performing segmentation based on the window status label represented by the window determination record, regular evidence blocks and event evidence blocks are formed, and event switching records are generated at the same time, supporting the continuous association of solidified summary, evidence chain, conclusion record and detection report record, taking into account process tracking, event reconstruction and result traceability. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of an information processing method used for engineering quality inspection.

[0020] Figure 2 This is a schematic diagram of an information processing system used for engineering quality inspection.

[0021] Figure 3 A flowchart generated for window determination records.

[0022] Figure 4 A flowchart for generating the chain of evidence. Detailed Implementation

[0023] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0024] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0025] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0026] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides an information processing method for engineering quality inspection, including the following steps:

[0027] S1. Collect basic information on engineering quality inspection activities and generate a unique inspection task identifier. By establishing source mapping relationships and a unified time benchmark, generate an initialization evidence block. Based on the initialization evidence block, generate an initial reference state and a planned process sequence, and establish an original record cache and feature extraction entry point.

[0028] Collect basic information on engineering quality testing activities and generate a unique testing task identifier. Establish source mapping relationships and a unified time benchmark. Gather standard reference information, self-inspection records before the start of testing, and the first stable record after the start of testing that meets the initial stability inclusion conditions, and generate an initialization evidence block.

[0029] Furthermore, collect basic task information corresponding to the current engineering quality inspection activities; for each source, collect the source number, source category, installation location, the object being inspected bound to the source, and the source's recording rules; after the basic task information is collected, perform integrity and consistency checks on the basic task information.

[0030] It should be noted that the basic information of the task includes, but is not limited to, the project number, the number of the object being tested, the batch number, the item number, the testing location information, the testing start date, the identity information of the operators, the identity information of the witnesses, and a list of data sources involved in this project quality testing activity.

[0031] The data source list includes one or more of the following: detection equipment sources, image acquisition sources, on-site record sources, operation trajectory sources, and environmental record sources.

[0032] It should be noted that integrity checks are used to determine the existence of missing fields; consistency checks are used to determine the existence of conflicts between the object number being checked, the batch number being checked, the location information being checked, and the source binding relationship.

[0033] Furthermore, after both integrity and consistency checks pass, the project number, the object being tested number, the batch number, the test item number, and the test start date are combined and encoded according to the field order to obtain a unique test task identifier.

[0034] The unique detection task identifier is used for the unified attribution of all subsequent original records, window judgment records, event switching records, evidence blocks, conclusion records, and detection report records.

[0035] It should be noted that the specific encoding method for the unique inspection task identifier is to use a fixed field sequence concatenation, fixed length truncation, and additional check bits. For example, the first 6 digits of the project number, the last 8 digits of the object being inspected, the first 4 digits of the inspection batch number, the first 3 digits of the inspection item number, and the inspection start date (year, month, day) are all taken as 8 digits. Then, 2 check bits are added, thus forming a unique inspection task identifier with a total length of 31 digits. The check bits are used to check for encoding errors during the field truncation and concatenation process, and are generated using modulo check or XOR check.

[0036] Furthermore, after generating a unique detection task identifier, a source mapping relationship for this project's quality detection activity is established, and a source mapping record is created for each source.

[0037] It should be noted that each source mapping record includes a source number, source category, corresponding detected object number, recording frequency rules, identity verification rules, allowed time deviation, and allowed location deviation. For example, for a detection equipment source, the source mapping record includes the detection equipment number, the detected object number corresponding to the detection equipment, and the sampling rules of the detection equipment. For an image acquisition source, the source mapping record includes the image acquisition device number, image acquisition location, and image acquisition frequency. For a field recording source, the source mapping record includes the generation rules for barcode scanning records, photo recordings, and manual entry records. For an operation trajectory source, the source mapping record includes the recording rules for actions such as starting detection, pausing detection, retest confirmation, and signature submission. For an environmental recording source, the source mapping record includes the environmental parameter type, acquisition cycle, and associated detection location.

[0038] It should be noted that by establishing a source mapping relationship, any subsequent original record can be directly traced back to the source category, object binding relationship, and record rules based on the source number.

[0039] It should be noted that the allowable time deviation is set jointly by the source type, source recording frequency, and time synchronization accuracy. For periodically output detection equipment sources, the allowable time deviation is set to twice the recording period of the detection equipment source, with a lower limit of 0.02 seconds and an upper limit of 0.5 seconds; for image acquisition sources, the allowable time deviation is set to 1 second; for on-site recording sources and operation trajectory sources, the allowable time deviation is set to 2 seconds; and for environmental recording sources, the allowable time deviation is set to 5 seconds. Therefore, the range of allowable time deviation is usually from 0.02 seconds to 5 seconds.

[0040] For periodically outputting detection equipment sources, since the recording period is known, time synchronization accuracy is high, and the detection equipment source usually directly corresponds to the measured value changes of the detected object, the allowable time deviation should be as small as possible. By setting the allowable time deviation to twice the recording period of the detection equipment source, short-term communication jitter and write latency are absorbed, while preventing confusion between adjacent period records. Setting the lower limit to 0.02 seconds is to avoid meaningless misjudgments due to excessively small allowable time deviations for high-frequency sources, and setting the upper limit to 0.5 seconds is to avoid weakening time alignment accuracy due to excessively large allowable time deviations for low-frequency sources. For image acquisition sources, the image acquisition process usually includes triggering, imaging, compression, and writing processes, thus having a larger system-level latency compared to detection equipment sources. Therefore, the allowable time deviation is set to... The allowable time deviation is set to 1 second, which can cover common image acquisition and writing delays while still ensuring the time correspondence between image recording and the same detection action. For on-site recording sources and operation trajectory sources, since these two types of sources usually involve manual triggering, manual confirmation, or mobile terminal upload processes, the time jitter is greater than that of detection equipment sources and image acquisition sources. Therefore, the allowable time deviation is set to 2 seconds, which can accommodate the natural delay caused by manual operation and mobile terminal transmission without distorting the sequence of actions. For environmental recording sources, since environmental parameters usually change slowly and are more used to reflect the environmental background during the detection period rather than the instantaneous correspondence of a single point, the allowable time deviation is appropriately relaxed to 5 seconds. This can cover the fluctuation of the environmental acquisition cycle without affecting the correspondence between the environmental background and the detection period.

[0041] It should be noted that the allowable positional deviation is set jointly by the source type, positioning method, and detection scenario. For fixed-installation detection equipment sources, the allowable positional deviation is set to 0.5 meters; for image acquisition sources and on-site recording sources, the allowable positional deviation is set to 3 meters; and for environmental recording sources, the allowable positional deviation is set to 5 meters. Therefore, the range of allowable positional deviation is usually from 0.5 meters to 5 meters.

[0042] For sources of fixed-installation detection equipment, since these devices are typically fixed at specific detection points and their outputs must strictly correspond to the specific object being detected or the specific detection location, the allowable positional deviation should be minimized. Setting the allowable positional deviation to 0.5 meters can absorb installation coordinate calibration errors and minor positioning errors while still ensuring accurate record attribution. For sources of image acquisition and on-site recording, these are typically generated by handheld terminals or mobile devices, which are susceptible to operator movement, terminal positioning drift, and indoor positioning errors. Therefore, setting the allowable positional deviation to 3 meters balances common handheld acquisition positioning errors and the detection area coverage while still ensuring a reasonable correspondence with the target detection area. For sources of environmental recording, since environmental parameters themselves have spatially gradual variations and typically reflect the environmental background within a certain area, strict correspondence to a single point is not required. Therefore, the allowable positional deviation is further relaxed to 5 meters, which can accommodate slight drift in environmental acquisition locations without affecting the representativeness of the environmental background recording.

[0043] Furthermore, the start time of the engineering quality inspection activity is selected as the unified start time of the task, and the unified start time of the task is distributed to each source; after each source returns the local recorded time, the time offset of each source relative to the unified start time of the task is recorded.

[0044] It should be noted that when each source generates an original record subsequently, the local record time is combined with the time offset of the corresponding source to convert it into a unified timestamp, and the unified timestamp is written into the original record.

[0045] It should be noted that the unified time base is used in conjunction with the allowable time deviation parameter in the source mapping relationship. Specifically, for periodic sources, the sequence of times to be recorded is calculated based on the recording frequency rules in the source mapping record. When there is no actual original record within the allowable time window corresponding to the time to be recorded, the time to be recorded is marked as a candidate position for integrity anomaly. The upper and lower boundaries of the allowable time window are determined by the allowable time deviation of the current source. For actual original records that have already arrived, when the deviation of the unified timestamp of the actual original record relative to the time to be recorded exceeds the allowable time deviation of the current source, the current actual original record is marked as a candidate record for time consistency anomaly. For actual original records that have already arrived, when the position corresponding to the current actual original record deviates from the allowable position deviation in the corresponding source mapping record, the current actual original record is marked as a candidate record for position consistency anomaly.

[0046] Furthermore, after completing the source mapping relationship and unified time reference, an initialization evidence block is generated, and when the original records within the current initial observation period meet the initial stability inclusion conditions, the original records within the current initial observation period are included in the initialization evidence block.

[0047] It should be noted that the initial evidence block is the base data set for the subsequent generation of the initial reference state.

[0048] It should be noted that the initialization evidence block includes standard reference information, self-test records before the start of the test, and the first stable record after the start of the test.

[0049] It should be noted that standard reference information refers to the reference range, reference value, allowable fluctuation range, and sequence of actions to be performed, given by the testing basis, testing procedures, or project technical requirements.

[0050] It should be noted that the self-inspection record before the start of the test refers to the original record formed when the testing equipment source, image acquisition source, and on-site recording source perform self-inspection actions before the test officially begins. For example, the test equipment no-load record, the image acquisition device clarity test record, and the barcode connection record of the on-site recording source.

[0051] It should be noted that the first stable record after the start of the test refers to a set of original records formed within the initial observation period after the test officially begins.

[0052] It should be noted that the initial observation duration is jointly set by the source recording frequency and the detection initial stability requirement. Specifically, the larger of 5 times the maximum value of the periodic source recording period and 10 seconds is taken as the initial observation duration, and the initial observation duration does not exceed 60 seconds; when there is no periodic source, the initial observation duration is 30 seconds.

[0053] It should be noted that the larger of 5 times the maximum value of the recorded cycles of all periodic sources and 10 seconds is used as the initial observation duration. Specifically, the initial observation duration needs to cover the output cycles of several complete periodic sources simultaneously, and also needs to cover the identity verification, initial action confirmation, and short-term stability observation process after the detection begins. When the maximum value of the recorded cycles of all periodic sources is too small, the initial observation duration may be too short to fully cover the key actions in the initial stage of detection, so 10 seconds is used as the lower limit. When the maximum value of the recorded cycles of all periodic sources is too large, taking 5 times the maximum value of the recorded cycles of all periodic sources can ensure that the continuous multi-cycle output of the slowest periodic source is observed, and the initial stability can be judged more reliably. The reason for limiting the initial observation duration to no more than 60 seconds is that if the initial observation duration is too long, it will cause records that have entered an unstable state or entered the actual abnormal evolution stage after the detection begins to be mistakenly included in the first stable record.

[0054] It should be noted that the reason for setting the initial observation duration to 30 seconds when there is no periodic source is that, in the absence of a periodic source, it is necessary to ensure that the on-site recording source and operation trajectory source can fully cover at least one round of identity verification, detection start, recording trigger and initial confirmation process. At the same time, the observation length of 30 seconds will not be too long and will include subsequent unstable records in the first stable record.

[0055] It should be noted that when the original records within the current initial observation period meet the initial stability inclusion conditions, the original records within the current initial observation period will be included in the initial evidence block to ensure that the first stable record can truly reflect the initial stable working conditions of this engineering quality inspection activity.

[0056] It should be noted that the initial stabilization inclusion criteria include the first, second, third, fourth, and fifth initial stabilization criteria. When all the initial stabilization inclusion criteria are met, the initial stabilization inclusion criteria are deemed to be met.

[0057] It should be noted that the first condition for initial stability is that the source binding relationship and the number of the object being tested are consistent.

[0058] It should be noted that the second condition for initial stabilization refers to successful identity verification.

[0059] It should be noted that the third condition for initial stability refers to the requirement that the continuity of the original records is met. Specifically, for periodic sources, the actual number of original records arriving within the initial observation period is not less than 95% of the number of original records that should have arrived. For non-periodic sources, all the key action records that must occur within the initial observation period have occurred, and the order of actions is consistent with the initial segment of the planned process sequence.

[0060] For periodic sources, the actual number of original records arriving within the initial observation period should not be less than 95% of the expected number of original records. The reason for setting the number of original records to 95% is that it allows for a small number of missing records caused by network jitter, write delay, or momentary asynchrony during startup, while still ensuring that the overall continuity and periodic structure of the records are not disrupted. If the value is set too low, obviously discontinuous records will be misjudged as stable records; if the value is set too high, a very small number of occasional missing records will lead to excessive exclusion of usable starting segments. Therefore, 95% of the number of original records is used as the continuity criterion to balance anti-interference and stability.

[0061] It should be noted that the fourth condition for initial stability refers to the absence of any of the following markers: manual pause, manual cancellation, and manual retest.

[0062] It should be noted that the fifth condition for initial stability refers to the absence of records that exceed the allowable time deviation and the allowable position deviation.

[0063] Based on the initial evidence block, the initial reference component and the initial discrete component are calculated by constructing a consistency coefficient, forming an initial reference state and generating a planned process sequence.

[0064] Furthermore, baseline feature components are extracted from the initial evidence block; for each baseline feature component, a consistency coefficient is constructed using the baseline consistency coefficient algorithm.

[0065] The number of extracted baseline feature components is denoted as . .

[0066] It should be noted that the consistency coefficient is used to prevent excessive traction on the initial reference state when the self-test record feature value and the first stable record feature value conflict with each other.

[0067] It should be noted that the baseline consistency coefficient algorithm is expressed as follows:

[0068] ;

[0069] in, Indicates the first Consistency coefficient of each baseline feature component Indicates the first The self-test record feature values ​​corresponding to each baseline feature component Indicates the first The first stable record feature value corresponding to each baseline feature component Indicates the first Standard reference values ​​corresponding to each baseline feature component This represents a small positive number used to avoid a denominator of zero.

[0070] It should be noted that, Based on numerical stability requirements, the value range is typically set as follows: to , Example taken as , Take as The specific reasons are as follows: Take as Effectively avoids zero denominator, and at the same time will not affect , as well as The relative differences cause substantial disturbances.

[0071] It should be noted that, Used to represent the difference between the feature value of the self-test record and the feature value of the first stable record; Used to represent the overall deviation scale of both relative to the standard reference value; when and The closer they get to each other, and and Compared to When the deviation trend is more consistent, The larger; when and The greater the difference between them, The smaller; when and The closer they get, The closer it is to 1.

[0072] Furthermore, after obtaining the consistency coefficient, the initial reference component is calculated using the reference component constraint correction algorithm.

[0073] It should be noted that the reference component constraint correction algorithm is expressed as follows:

[0074] ;

[0075] in, Indicates the first The initial reference component of each baseline feature component.

[0076] It should be noted that, in On this basis, if and only if and right Only when the offset trend is consistent can the standard reference value be effectively corrected; when and When there is a conflict, the correction magnitude is automatically reduced.

[0077] It should be noted that, for the reference component constraint correction algorithm, when and Compared to When the offset directions are consistent, the numerator is positive, indicating that both types of field information support correction of the standard reference value in the same direction; when... and Compared to When the offset direction is opposite, the correction term is weakened to prevent conflicting data from directly pulling the initial reference state off-center; through Controlling the correction intensity ensures more thorough corrections when there is consistency between on-site information and more conservative corrections when there is inconsistency.

[0078] Furthermore, after obtaining the initial reference components, the initial discrete components are calculated using a discrete component dynamic generation algorithm to determine the fluctuation boundary of the current baseline characteristic components.

[0079] It should be noted that the algorithm for dynamically generating discrete components is expressed as follows:

[0080] ;

[0081] in, Indicates the first The initial discrete components of each baseline feature component.

[0082] It should be noted that when the consistency coefficient is low, the initial discrete components refer more to the deviation of the self-test record feature value and the first stable record feature value from the standard reference value; when the consistency coefficient is high, the initial discrete components refer more to the subtle differences between the self-test record feature value and the first stable record feature value.

[0083] Furthermore, for all baseline feature components, an initial reference vector and an initial discrete matrix are constructed to form an initial reference state. Based on the action requirements and source mapping relationships in the detection criteria, a planned process sequence is generated. The key actions required for this engineering quality inspection activity are extracted from the detection criteria, such as identity verification, detection start, self-inspection confirmation, data acquisition, image acquisition, pause recording, retest recording, completion confirmation, and result submission. The source responsible for recording each key action is determined according to the source mapping relationship, and they are sorted according to their sequential dependencies to form a planned process sequence.

[0084] It should be noted that the initial reference vector is represented as:

[0085] ;

[0086] in, This represents the initial reference vector for the quality inspection activities of this project. This represents the initial reference component of the first baseline characteristic component. This represents the initial reference component of the second baseline characteristic component. Indicates the first The initial reference component of each baseline feature component, This indicates transpose.

[0087] It should be noted that the initial discrete matrix is ​​expressed as:

[0088] ;

[0089] in, This represents the initial discrete matrix for the quality inspection activities of this project. This indicates that the matrix is ​​constructed in diagonal form. This represents the initial discrete component of the first baseline characteristic component. This represents the initial discrete component of the second baseline characteristic component. Indicates the first The initial discrete components of each baseline feature component.

[0090] It should be noted that the initial reference state is composed of the initial reference vector and the initial discrete matrix.

[0091] It should be noted that the planned process sequence is represented as follows:

[0092] ;

[0093] in, Indicates the planned process sequence. This indicates the first planned action. This indicates the second planned action. Indicates the first One planned action, This indicates the total number of planned actions.

[0094] Establish a raw record cache and a feature extraction entry point.

[0095] Furthermore, after generating the initial evidence block, initial reference state, and planned process sequence, a raw record cache is established using three dimensions: unique detection task identifier, source number, and unified timestamp as indexes; at the same time as establishing the raw record cache, a feature extraction entry point is established.

[0096] The original record cache is used to continuously receive and save the original records generated by each source during the detection process. It uses a unique detection task identifier, source number and unified timestamp as the index structure to ensure that the original records can be quickly located by source and time in the same detection task.

[0097] It should be noted that the feature extraction entry point reads the original records only from the original record cache area, and does not read data directly from external sources. This is to ensure that subsequent real-time features, window features, and window judgment records can all be traced back to the underlying original records in the original record cache area, and to avoid the real-time judgment results and the original record saving process belonging to two separate data chains.

[0098] It should be noted that the feature extraction entry point calls the feature extraction rules corresponding to the detection equipment source, image acquisition source, on-site record source, operation trajectory source, and environmental record source according to the source number and source category.

[0099] S2. Continuously receive raw records from various sources and write them to the raw record buffer. Extract real-time features from the feature extraction entry point to form window features. Update the current reference state through a stable window. Combine the window features, initial reference state, current reference state, planned process sequence, record integrity within the window, and identity consistency to generate event triggering results. Determine the relevant source set based on the contribution of the window features corresponding to each source to the event triggering results and generate window judgment records.

[0100] The system continuously receives raw records from various sources and writes them into the raw record buffer. Real-time features are extracted from the feature extraction entry point. All extracted real-time features are normalized and aligned and aggregated according to a unified timestamp within the current sliding time window to form window features.

[0101] Furthermore, after acquiring the unique detection task identifier, source mapping relationship, unified time base, initial evidence block, initial reference state, planned process sequence, raw record cache, and feature extraction entry point, the system continuously receives raw records generated by each source and writes them into the raw record cache. The feature extraction entry point reads raw records from the raw record cache and calls the corresponding feature extraction rules based on the source number and source category to extract real-time features. All extracted real-time features are normalized. A sliding time window is set, and the real-time features within the current sliding time window are aligned and aggregated according to a unified timestamp to form window features.

[0102] It should be noted that each original record includes a unique detection task identifier, source number, unified timestamp, original payload, identity association information, location association information, and context information.

[0103] Specifically, for sources from detection equipment, the original payload includes the original measurement sequence, original waveform, and original equipment messages; for sources from image acquisition, the original payload includes the original image frames and original video clips; for sources from on-site recording, the original payload includes barcode scanning records, photo records, location records, and manual input records; for sources from operation trajectories, the original payload includes action records of starting detection, pausing detection, retest confirmation, and signature submission; and for sources from environmental records, the original payload includes environmental parameter acquisition records.

[0104] It should be noted that, for sources from detection equipment, the extracted real-time features include, but are not limited to, amplitude, mean, variance, dominant frequency position, local fluctuation intensity, and degree of abrupt change; for sources from image acquisition, the extracted real-time features include, but are not limited to, edge density, texture change rate, candidate anomaly region position, and sharpness score; for sources from on-site recording, the extracted real-time features include barcode matching status, positional deviation status, and manual input consistency status; for sources from operation trajectory, the extracted real-time features include action type, action time interval, and action sequence status; and for sources from environmental recording, the extracted real-time features include the mean of environmental parameters, the rate of change of environmental parameters, and the intensity of fluctuations in environmental parameters.

[0105] It should be noted that the sliding time window is set by the window duration and the sliding step size. Specifically, the maximum value between 3 times the maximum value of the periodic record period of all periodic sources and 5 seconds is taken as the window duration, and the window duration shall not exceed 30 seconds; half of the window duration is taken as the sliding step size, and the sliding step size shall not be less than 1 second.

[0106] It should be noted that the window duration is set jointly by the source recording frequency and the requirements for observing abnormal evolution. Taking three times the maximum value of the recording period of all periodic sources can ensure that the slowest periodic source contributes several recording points within a single window, which is convenient for judging the continuous offset trend. Using 5 seconds as the lower limit of the window duration can ensure that a statistically significant observation interval is still covered in high-frequency source scenarios. The reason for limiting the window duration to no more than 30 seconds is that an excessively long window duration will reduce the temporal resolution of event triggering.

[0107] It should be noted that the sliding step size is set by both real-time requirements and repetitive coverage requirements. Taking half the window duration can form overlapping coverage between adjacent windows, making it easier to continuously capture the start and end of anomalies. The reason for limiting the sliding step size to no less than 1 second is that a sliding step size that is too small will lead to excessive repetitive calculations and reduce online processing efficiency.

[0108] It should be noted that the window features are formed as follows: for each baseline feature dimension, all valid real-time feature values ​​belonging to the current baseline feature dimension are extracted within the current sliding time window. After performing outlier removal processing on all valid real-time feature values, the median value is taken as the window feature component of the current baseline feature dimension within the current sliding time window. All window feature components are arranged sequentially to form the window features.

[0109] The outlier removal process is implemented using either the interquartile range removal rule or the absolute median difference removal rule.

[0110] It should be noted that when there are no valid real-time feature values ​​for the baseline feature dimension within the current sliding time window, the current baseline feature dimension is marked as missing and deducted in the record integrity calculation in subsequent windows.

[0111] The stable window is determined based on the steady-state update conditions. Then, the event-triggered coupling algorithm is used to generate the event-triggered result by combining the window characteristics, initial reference state, current reference state, planned process sequence, record integrity within the window, and identity consistency within the window.

[0112] Furthermore, based on the original records and action records within the current sliding time window, the integrity of records within the window, the consistency of identities within the window, and the degree of matching of the planned process sequence are calculated.

[0113] It should be noted that the record integrity within the window is obtained by calculating the ratio of the actual number of records arriving in the current sliding time window to the number of records that should arrive in the sliding time window based on the source mapping relationship and the record frequency rule; when there are no records that should arrive in the sliding time window, the record integrity within the window is directly set to 1.

[0114] It should be noted that when the record integrity within the window is closer to 1, it indicates a higher record integrity within the current sliding time window; when the record integrity within the window is lower, it indicates a greater number of missing records within the current sliding time window.

[0115] It should be noted that identity consistency within the window is obtained by calculating the ratio of the number of identity items that have been matched within the current sliding time window to the total number of identity items that have participated in the verification within the sliding time window.

[0116] It should be noted that when there is no identity to be verified in the sliding time window, the identity consistency within the window is directly set to 1.

[0117] It should be noted that when the identity consistency within the window is closer to 1, it indicates a higher level of identity consistency within the current sliding time window; when the identity consistency within the window is lower, it indicates a more significant identity conflict within the current sliding time window.

[0118] Furthermore, the matching degree of the planned process sequence is calculated based on the degree of difference between the actual action sequence and the planned process sequence within the current sliding time window.

[0119] It should be noted that the degree of matching of the planned process sequence is expressed as:

[0120] ;

[0121] in, Indicates the degree of matching between the planned process sequence. This represents the actual sequence of actions within the current sliding time window. This represents the edit distance between the actual sequence of actions and the planned sequence of actions. This indicates the total number of planned actions.

[0122] It should be noted that the planned process sequence includes at least one planned action; therefore, Not less than 1.

[0123] It should be noted that when The closer a value is to 1, the more consistent the actual action sequence within the current sliding time window is with the planned flow sequence; when The lower the value, the more obvious the missing actions, inserted actions, and deviations in the action sequence exist within the current sliding time window.

[0124] Furthermore, by combining window features, initial reference state, and current reference state, the dual reference deviation of the current sliding time window is calculated by calculating the initial deviation and current deviation and performing geometric coupling. According to the steady-state update condition, the current reference state is updated through a stable window. When all conditions in the steady-state update condition are satisfied simultaneously, the steady-state update condition is determined to be satisfied, and the current sliding time window is determined to be a stable window.

[0125] The deviation of the window features from the initial reference state is denoted as... The deviation of the window features from the current reference state is denoted as... The current reference state includes the current reference vector and the current discrete matrix. The current reference vector is denoted as... Let the current discrete matrix be denoted as

[0126] It should be noted that the initial reference vector and the initial discrete matrix are used simultaneously as the initial values ​​for the current reference state. Before the start of the first sliding time window, the values ​​are... and .

[0127] It should be noted that the double-reference deviation of the current sliding time window is calculated as follows:

[0128] ;

[0129] ;

[0130] ;

[0131] in, Indicate window characteristics, This indicates the double reference deviation of the current sliding time window.

[0132] It should be noted that the reason for using dual reference deviation is that when only the initial reference state is used as the reference, it reflects the degree of deviation of the current sliding time window relative to the initial stable working condition of the detection, but cannot reflect the slow drift during the detection process; when only the current reference state is used as the reference, it can adapt to the slow changes during the detection process, but is not sensitive to the deviation of the initial working condition.

[0133] It should be noted that the steady-state update conditions include the first steady-state update condition, the second steady-state update condition, the third steady-state update condition, and the fourth steady-state update condition.

[0134] It should be noted that the first condition for steady-state updates is that the deviation between the two references does not exceed the first stability threshold.

[0135] It should be noted that the second condition for steady-state updates is that the integrity of records within the window is not lower than the second stable threshold.

[0136] It should be noted that the third condition for steady-state updates is that the consistency of identities within the window is not lower than the third stable threshold.

[0137] It should be noted that the fourth condition for steady-state updates is specifically the degree of matching between the planned process sequence and the actual process sequence. Not lower than the fourth stability threshold.

[0138] It should be noted that the first stability threshold is set by the allowable deviation range of the initial reference state and the stability window screening sensitivity, and the value range is usually from 0.5 to 2.5.

[0139] When the first stability threshold is small, only windows that are close to both the initial reference state and the current reference state will be considered stable windows, which helps to improve the purity of the current reference state update. When the first stability threshold is too large, it can accommodate more obvious slow drift during the detection process, but it increases the risk of mistakenly including abnormal edge windows into stable windows. Therefore, the first stability threshold is preferably set to 1.2.

[0140] It should be noted that the second stability threshold is set by the allowable percentage of missing records and the stability window integrity requirement, and its value ranges from 0.9 to 1.

[0141] When the second stability threshold is too low, it can accommodate more short-term packet loss and write latency, but it will reduce the integrity requirements of the stability window; when the second stability threshold is too high, it can ensure that the window used to update the current reference state has higher record integrity. Therefore, the second stability threshold is preferably set to 0.95.

[0142] It should be noted that the third stability threshold is set by the strictness of identity verification and the tolerance for identity loss, and the value range is usually from 0.95 to 1.

[0143] Specifically, a higher third stability threshold ensures stronger identity consistency for windows used to update the current reference state; a lower third stability threshold maintains the ability to update stable windows even when a small number of identity items have not yet been returned. Therefore, the preferred value for the third stability threshold is 0.98. The reason for this value is that identity consistency is one of the most crucial factors constraining authenticity in subsequent event triggering results. If the third stability threshold is too low, windows with significant identity mismatch risks will be incorrectly included in the stable window. If the third stability threshold is too high, windows that could be used to update the current reference state will be excessively excluded due to delays in the return of a very small number of identity items. Therefore, a preferred value of 0.98 strikes a balance between strict identity verification and tolerance for minor return delays.

[0144] It should be noted that the fourth stability threshold is set by the rigidity of the planned process sequence execution and the tolerance for action deviation, and the value range is usually from 0.8 to 1.

[0145] Specifically, when the fourth stability threshold is set too high, the actual action sequence must be highly consistent with the planned process sequence, which helps to exclude abnormal windows in the process. When the fourth stability threshold is set too low, it can accommodate a small number of missing non-critical actions and minor deviations in the action sequence. Therefore, the fourth stability threshold is preferably set to 0.9. The reason for setting it to 0.9 is that in engineering quality inspection activities, there may be occasional delays, merged reporting, or short-term order disturbances in the recording of non-critical actions. If the fourth stability threshold is set too high, windows caused by a small number of non-critical action deviations will be excluded from the stable window. If the fourth stability threshold is set too low, windows with significant process deviations will be allowed to participate in the current reference state update. Therefore, the fourth stability threshold is preferably set to 0.9 to accommodate a small number of non-critical action deviations while maintaining constraints on the critical process sequence.

[0146] Furthermore, when the current sliding time window is a stable window, the current reference state is updated according to the update rules of the current reference vector and the current discrete matrix, based on the window characteristics of the current sliding time window.

[0147] It should be noted that the update rules for the current reference vector and the current discrete matrix are expressed as follows:

[0148] ;

[0149] ;

[0150] in, Represents the identity matrix. Indicates the first The current reference vector corresponding to each sliding time window Indicates the first The current discrete matrix corresponding to each sliding time window This represents a small constant used for matrix regularization during the window determination phase. This represents the update coefficient.

[0151] It should be noted that, The value is set by combining the current reference state tracking speed and the degree of historical stability window preservation; the value typically ranges from 0.05 to 0.30. When the value is small, the current reference state updates more smoothly, which is beneficial for preserving historical stable information; when... When the value is larger, the current reference state responds faster to the most recent stable window, therefore, The preferred value is 0.1. The reason for choosing 0.1 is that... If the value is too small, the current reference state will follow the real and stable drift during the detection process too slowly, resulting in the subsequent dual reference deviation remaining too high for a long time. When the value of is too large, it will cause the current reference state to be overly influenced by the most recent single stable window, reducing the smoothing effect of historical stability information. Therefore, A value of 0.1 is preferred to achieve a balance between response speed and stability.

[0152] It should be noted that, The value range is typically set based on the numerical stability requirements of the current reference state update phase. to , Example taken as , Take as The reason is that, Take as It can ensure that the main diagonal elements of the current discrete matrix are strictly positive, avoid the occurrence of singular and near-singular problems in the subsequent inversion process of the current discrete matrix, and at the same time, it will not substantially amplify the true discreteness of the window features.

[0153] It should be noted that when the first If the first sliding time window does not meet the stable window condition, the current reference state remains unchanged, and the second sliding time window is retained. The current reference vector and the current discrete matrix corresponding to each sliding time window.

[0154] Furthermore, by combining window features, initial reference state, current reference state, planned process sequence, integrity of records within the window, and identity consistency, an event-triggered coupling algorithm is used to calculate the event triggering result.

[0155] It should be noted that the event-triggered coupling algorithm is expressed as follows:

[0156] ;

[0157] in, Indicates the result of the event triggering. This indicates the completeness of the records within the window. This indicates that the identities within the window are consistent.

[0158] It should be noted that in event-triggered coupling algorithms, Used to suppress excessive amplification of results when the deviation between the two references is too large, and to prevent excessive jumps in event-triggered results caused by instantaneous spikes in a single feature; Used to characterize the degree of deviation from the planned process sequence. Used to characterize the degree of record missing. Used to characterize the degree of identity conflict.

[0159] The deviation degree between the window feature sub-vectors corresponding to each source and the reference sub-vector is extracted. The proportion of the deviation degree corresponding to each source in the total deviation of all sources represents the contribution degree of the window features corresponding to each source to the event triggering result. The relevant source set is determined based on the contribution degree of the source, and a window judgment record is generated by comparing the event triggering result with the event triggering threshold.

[0160] Furthermore, based on the contribution of the window features corresponding to each source to the event triggering result, the window feature sub-vectors corresponding to each source within the current sliding time window and the reference sub-vectors corresponding to each source in the current reference state are extracted. The total deviation of all sources is calculated, and the proportion of the deviation between the window feature sub-vectors corresponding to each source and the reference sub-vectors in the total deviation of all sources is used to characterize the contribution of the window features corresponding to each source to the event triggering result, thus determining the relevant source set.

[0161] It should be noted that the total deviation from all sources is calculated and expressed as:

[0162] ;

[0163] in, This represents the total deviation from all sources. This represents the source index in the current source set. This represents the set of sources participating in the calculation within the current sliding time window. express The corresponding window feature vector, express The corresponding reference subvector, Represents the vector norm.

[0164] It should be noted that when the total deviation At that time, the contribution level of the current source is calculated using the source contribution normalization algorithm.

[0165] It should be noted that the source contribution normalization algorithm is expressed as:

[0166] ;

[0167] in, This indicates the level of contribution corresponding to the current source. This represents the window feature subvector corresponding to the current source within the current sliding time window. This represents the reference subvector corresponding to the current source in the current reference state.

[0168] It should be noted that the contribution of the window features corresponding to each source to the event triggering result is not a numerical breakdown of the event triggering result itself, but rather is represented by the proportion of the deviation between the current source's window feature sub-vector and the reference sub-vector in the total deviation of all sources. This reflects the strength of the association between the current source and the current window entering the event state. When determining the relevant source set, the contribution is represented by the source deviation proportion, which is used to identify sources that are more strongly associated with the current event state.

[0169] It should be noted that when the total deviation At that time, the contribution level of all sources is directly set to 0.

[0170] It should be noted that for the current sliding time window, the relevant source set is determined based on the contribution of the source. When the contribution of the current source to the event triggering result is not less than the source contribution threshold, the current source is included in the candidate relevant source. All candidate relevant sources are counted to obtain the relevant source set.

[0171] It should be noted that the source contribution threshold is jointly set by the source focus degree and the multi-source coverage requirement, and the value range is usually from 0.05 to 0.5. The preferred source contribution threshold is 0.15. The reason for choosing 0.15 is that if the source contribution threshold is too low, a large number of sources with only slight deviations will be included in the relevant source set, resulting in an excessively large segmentation range for subsequent events. If the source contribution threshold is too high, only a very small number of sources will be retained, resulting in the omission of sources that play a supporting role in the formation of the current event. Therefore, the preferred source contribution threshold of 0.15 can achieve a balance between source focus and multi-source coverage.

[0172] It should be noted that when no source meets the source contribution threshold, the source with the largest contribution is selected, and the source of the operation trajectory and the source of the on-site record are included in the relevant source set to prevent the relevant source set from being empty.

[0173] Furthermore, an event trigger threshold is set. By comparing the event trigger result with the event trigger threshold, a window status label is set, and a window determination record is generated based on the event trigger result and the relevant source set.

[0174] It should be noted that the window determination record includes the window status label, event trigger result, relevant source set, window characteristics of the current sliding time window, current reference state snapshot, record integrity within the window, identity consistency within the window, and the degree of matching of the planned process sequence.

[0175] It should be noted that the event trigger threshold is set by the abnormal capture sensitivity and the false trigger suppression requirement, and the value range is usually from 0.6 to 2.5.

[0176] Specifically, when the event trigger threshold is set too low, it is easier to determine the edge abnormal window as an event state, which is beneficial to improving the sensitivity of anomaly detection. When the event trigger threshold is set too high, it tends to enter the event state only when multiple factors deviate significantly, which is beneficial to suppressing false triggers. Therefore, the event trigger threshold is preferably set to 1.1. The reason for setting it to 1.1 is that in the event trigger coupling algorithm, when the double reference deviation is small and the matching degree of the planned process sequence, the integrity of the records in the window, and the consistency of the identity in the window are all kept high, the event trigger result is usually kept at a low level. Setting the event trigger threshold to 1.1 requires that there is a certain degree of double reference deviation in the current sliding time window at the same time, and that the event state is only entered after deterioration occurs in the process, integrity, and identity. This improves the sensitivity of anomaly detection while suppressing false triggers caused by a single slight disturbance.

[0177] Specifically, when the event trigger result is lower than the event trigger threshold, the window status label is set to the normal state; when the event trigger result is not lower than the event trigger threshold, the window status label is set to the event state.

[0178] S3. For the original records in the original record buffer during the detection process, based on the window state label represented by the window determination record, continuously perform segmentation to form regular evidence blocks and event evidence blocks, and at the same time generate event switching records.

[0179] During the detection process, the original records in the original record buffer are continuously segmented. When the window determines that the record represents a normal state, a normal evidence block is formed according to the normal segmentation rules controlled by the normal segmentation duration.

[0180] Furthermore, after generating the window determination record, the original records in the original record cache are continuously read, and the corresponding segmentation rules are executed according to the window state label in the window determination record; when the window determination record indicates that the window state label corresponding to the current sliding time window is in a normal state, a normal evidence block is formed according to the normal segmentation rules.

[0181] It should be noted that the regular segmentation rule is controlled by the regular segmentation duration. The source number is the smallest segmentation object. The regular segmentation progress is maintained for different sources. For any source, when the unified timestamp span of the original record corresponding to the current source in the original record cache reaches the regular segmentation duration, all original records belonging to the current source within the current time period are extracted from the original record cache to form a regular evidence block.

[0182] It should be noted that the standard segmentation time is set based on the standard traceability granularity requirements and the storage efficiency requirements under standard conditions. The standard segmentation time typically ranges from 10 seconds to 120 seconds, with 30 seconds being the preferred value. When the standard segmentation time is set too low, it can improve the traceability resolution of the original records under standard conditions, but it will increase the number of evidence blocks and index maintenance overhead. When the standard segmentation time is set too high, it can reduce the number of evidence blocks, but it will reduce the backtracking accuracy under standard conditions. Therefore, the standard segmentation time is preferably set at 30 seconds to achieve a balance between standard traceability accuracy and storage efficiency.

[0183] It should be noted that a standard evidence block includes a unique detection task identifier, source number, block start time, block end time, sequence of original records within the block, number of original records within the block, and source mapping information corresponding to the original records within the block.

[0184] It should be noted that regular evidence blocks are formed independently with source number as the smallest organizational granularity, in order to avoid mixing original records from different sources in the same regular evidence block.

[0185] When determining the record representing the event state in the window, the original record before the window corresponding to the event state is replenished according to the relevant source set, the event number and the initial relevant source set are generated, and only the sources corresponding to the relevant source set are used to form event evidence blocks according to the event segmentation rules, while the original records corresponding to non-relevant sources are used to form regular evidence blocks according to the regular segmentation rules.

[0186] Furthermore, when the window determination record indicates that the window state label corresponding to the current sliding time window is an event state, the event segmentation process is initiated; upon first entering the event state, the event number corresponding to the current event is generated, and the relevant source set corresponding to the current window determination record is determined as the initial relevant source set of the current event.

[0187] It should be noted that the event number is generated using a sequential numbering method associated with a unique inspection task identifier. For example, in the same engineering quality inspection activity, the first event number is generated when the event state is entered for the first time, and the second event number is generated when the event state is entered for the second time.

[0188] It should be noted that the event number is used for the unified attribution of all subsequent event evidence blocks and event switching records.

[0189] Furthermore, upon first entering an event state, the start time of the current sliding time window is used as the event start boundary, and the original records before the corresponding window of the event state are replenished based on the relevant source set.

[0190] It should be noted that the front window re-filling is controlled by the duration of the front window re-filling.

[0191] It should be noted that the front window backfill duration is jointly set by the requirements for pre-event traceability integrity and backfill data redundancy control. The value range is usually from 5 seconds to 60 seconds, with 15 seconds being the preferred value. If the front window backfill duration is too small, it cannot fully cover the gradual abnormal process before the event occurs; if the front window backfill duration is too large, it will introduce too many regular original records that are not related to the event. Therefore, the front window backfill duration is preferably set at 15 seconds to achieve a balance between pre-event traceability integrity and data redundancy control.

[0192] It should be noted that for each source belonging to the current event-related source set, the original records of the corresponding source within the time interval before the event start boundary and with a length equal to the front window backfill duration are extracted from the original record cache and used as the front window backfill original records.

[0193] It should be noted that the front window backfilling of original records is the extraction and reorganization of existing original records in the original record cache area. It does not delete the previously formed regular evidence blocks. The event evidence blocks formed by the front window backfilling are allowed to partially overlap with the previously formed regular evidence blocks in terms of time range. The event evidence blocks formed by the front window backfilling are used for event tracing, while the previously formed regular evidence blocks are used for full-process regular tracing.

[0194] Furthermore, under the event state, only the sources corresponding to the relevant source set are used to form event evidence blocks according to the event segmentation rules, while the original records corresponding to unrelated sources continue to be used to form regular evidence blocks according to the regular segmentation rules. For any source belonging to the relevant source set of the current event, the starting point of the front window backfill interval before the event start boundary is taken as the event segmentation starting point, and the original records are continuously segmented according to the event segmentation duration starting from the event segmentation starting point, continuously forming event evidence blocks during the event duration. When the current event end boundary is determined, the event end boundary is taken as the event segmentation endpoint for the corresponding source in the current event, and event segmentation stops for the original records beyond the event end boundary. When the last event segmentation interval is less than a complete event segmentation duration, the insufficient part is still retained as an independent event evidence block. By maintaining the continuous connection between the front window backfill part and the event duration part in time, they together constitute the event evidence chain.

[0195] It should be noted that the event segmentation rule is controlled by the event segmentation duration. The event segmentation duration is jointly set by the event tracing resolution requirements and the retrieval efficiency requirements under the event state. The value range of the event segmentation duration is usually from 1 second to 10 seconds, with 3 seconds being preferred. When the event segmentation duration is too small, it can improve the temporal resolution of the original records during the event, but it will significantly increase the number of event evidence blocks. When the event segmentation duration is too large, it will reduce the positioning accuracy of the abnormal evolution process within the event. Therefore, the event segmentation duration is preferably set at 3 seconds to achieve a balance between event tracing resolution and evidence block size control.

[0196] It should be noted that the duration of regular segmentation is longer than that of event segmentation, ensuring that the segmentation granularity under event conditions is significantly finer than that under regular conditions.

[0197] It should be noted that for sources that do not belong to the current event-related source set, the event segmentation rule is not switched, but the regular segmentation rule is continued to form regular evidence blocks. During the same event, the relevant source set forms event evidence blocks, and the unrelated source set forms regular evidence blocks. This ensures the accuracy of event tracing while controlling the expansion of evidence blocks from irrelevant sources.

[0198] It should be noted that the event evidence block includes a unique detection task identifier, event number, source number, block start time, block end time, sequence of original records within the block, number of original records within the block, and source mapping information corresponding to the original records within the block.

[0199] It should be noted that the differences between event evidence blocks and regular evidence blocks are reflected in three aspects: block affiliation, segmentation granularity, and tracing purpose. Event evidence blocks belong to a certain event number, adopt a finer event segmentation duration, and mainly serve event tracing; regular evidence blocks do not belong to a specific event number, adopt a coarser regular segmentation duration, and mainly serve full-process regular tracing.

[0200] During the event, the relevant source set is updated, and front window replenishment and event segmentation are performed on newly added relevant sources; when the window determination record is restored from the event state to the normal state, an event switching record is generated according to the restoration confirmation count rule, and normal segmentation is restored after the event ends.

[0201] Furthermore, when the current event has not yet ended and subsequent window determination records continue to represent the event state, the relevant source set corresponding to the subsequent window determination records is read, and the relevant source set of the current event is updated.

[0202] It should be noted that when a new source that has not yet been included in the current event's relevant source set exists in the relevant source set corresponding to the subsequent window determination record, the new source will be merged into the current event's relevant source set.

[0203] It should be noted that for newly added sources included in the current event-related source set, the same front window backfilling rule as when the event was first triggered is adopted. Based on the start time of the corresponding window when the new source first enters the current event-related source set, the original records with a length equal to the front window backfilling time are backfilled forward, and event evidence blocks are formed for the corresponding new sources according to the event segmentation rule.

[0204] Furthermore, when the window determination record returns from the event state to the normal state, the recovery confirmation counting rule is used to perform an end determination on the current event and generate an event switching record.

[0205] It should be noted that the recovery confirmation count rule is as follows: if a window status label of "normal" appears consecutively after the start of the current event, the current event is determined to end; if the number of consecutive window status label "normal" appears is less than the recovery confirmation count, the current event is determined to continue.

[0206] It should be noted that the recovery confirmation count is set jointly by the event end jitter suppression requirement and the end response speed requirement, and the value range is usually from 2 to 4, with 2 being preferred. When the recovery confirmation count is smaller, the event end response is faster, but it is easily affected by short-term window state jitter and the event ends too early. When the recovery confirmation count is larger, it can enhance the robustness of the event end determination, but it will delay the event end time. Therefore, the recovery confirmation count is preferably set to 2 to achieve a balance between jitter resistance and end response speed.

[0207] Furthermore, when the recovery confirmation count rule is met, the end time of the window whose state label is the last window in the current event is determined as the event end boundary.

[0208] It should be noted that the event start boundary is the start time of the first window in the current event whose state label is event state; the event end boundary is the end time of the last window in the current event whose state label is event state; and the front window backfill interval is the time interval before the event start boundary, with a length equal to the front window backfill duration.

[0209] Furthermore, after determining the event number, event start boundary, event end boundary, front window backfill interval, and the set of sources related to the current event, an event switching record is generated.

[0210] It should be noted that the event switching record includes at least the event number, the event start boundary, the event end boundary, the front window backfill interval, the relevant source set, and switching description information.

[0211] The switching description information is used to record the window number that triggered the start and end of the current event, the corresponding changes in the window status label, and the update status of the relevant source set.

[0212] It should be noted that the event switching record represents the start and end boundaries of the event, and at the same time represents the evolution process of the current event as determined by the window.

[0213] Furthermore, once the current event ends, event segmentation is stopped for each source in the set of sources related to the current event, and the regular segmentation rules are resumed.

[0214] It should be noted that when resuming the execution of the regular segmentation rules, the event end boundary is used as the new starting alignment time for subsequent regular segmentations to avoid overlapping between the event segmentation interval and the regular segmentation interval after the event.

[0215] S4. Generate solidified summaries for initialization evidence blocks, window judgment records, regular evidence blocks, event evidence blocks, and event switching records respectively. Based on the solidified summaries, connect them in a unified time sequence to form an evidence chain and generate conclusion records and test report records.

[0216] For the initialization evidence block, window judgment record, regular evidence block, event evidence block, and event switching record, a solidified digest is generated using a standard hash algorithm.

[0217] Furthermore, the initialization evidence block, window judgment record, regular evidence block, event evidence block, and event switching record are standardized and organized; a standard hash algorithm, such as the SM3 hash algorithm, is used to generate a fixed summary for each type of object after standardization and organization.

[0218] It should be noted that the initialization evidence block, window judgment record, regular evidence block, event evidence block, and event switching record are standardized. Specifically, for the initialization evidence block, the fields are organized in a fixed order: unique detection task identifier, initialization evidence block type, block start time, block end time, standard reference information, self-test record index, and first stable record index. For the regular evidence block, the fields are organized in a fixed order: unique detection task identifier, source number, block start time, block end time, original record sequence within the block, number of original records within the block, and source mapping information. For the event evidence block, the fields are organized in a fixed order: unique detection task identifier, event number, source number, and... For block start time, block end time, original record sequence within the block, number of original records within the block, and source mapping information, the fields are organized in a fixed order. For window determination records, the fields are organized in a fixed order: unique detection task identifier, window sequence number, window start time, window end time, window status label, event trigger result, related source set, integrity of records within the window, identity consistency within the window, degree of matching of planned process sequence, and current reference status snapshot index. For event switching records, the fields are organized in a fixed order: unique detection task identifier, event number, event start boundary, event end boundary, previous window backfill interval, related source set, and switching description information.

[0219] It should be noted that during the standardization process, a unified timestamp format is used for time fields, a preset sorting rule is used for source sets and index sets, null value fields are filled with null identifiers, and text fields are encoded using a character encoding standard to ensure that the same object obtains the same standardized byte sequence when processed repeatedly.

[0220] It should be noted that the initialization evidence block, window judgment record, regular evidence block, event evidence block, and event switching record all generate their own corresponding solidified summaries, and each solidified summary is bound and saved to its corresponding object one by one.

[0221] It should be noted that after the solidified summary is generated, the original object remains. The solidified summary serves only as a fixed representation of the original object and does not replace the original object itself.

[0222] Based on the fixed summary, evidence is linked together in a unified chronological order to form a chain of evidence; a conclusion record is generated based on the chain of evidence, and a test report record is generated based on the conclusion record.

[0223] Furthermore, after generating solidified summaries for all types of objects, the objects are sorted according to a uniform chronological order, and a chain of evidence is formed based on the solidified summaries.

[0224] It should be noted that the unified time order is determined by the effective time of the objects; specifically, the effective time of the initialization evidence block is taken as the block start time of the initialization evidence block; the effective time of the regular evidence block and the event evidence block is taken as the block start time of the corresponding evidence block; the effective time of the window judgment record is taken as the window start time of the corresponding sliding time window; the effective time of the event switching record is taken as the event start boundary; when there are several objects with the same effective time, they are sorted according to the priority of object type.

[0225] It should be noted that in this embodiment, the priority order of object types is as follows: initialization evidence block, window judgment record, regular evidence block, event evidence block, and event switching record. When the object types are still the same, they are further sorted according to the source number, event number, and generation order to ensure that the evidence chain connection order is uniquely determined.

[0226] Furthermore, the solidified summary corresponding to each object is read sequentially according to the sorting result, and evidence chain entries are formed.

[0227] It should be noted that each evidence chain entry includes a chain sequence number, object type, object identifier, object effective time, current solidified summary, previous chain entry identifier, and previous chain entry solidified summary.

[0228] In this case, the identifier of the preceding chain entry and the solidified summary of the preceding chain entry of the first chain of evidence entry are both set to null values; subsequent chain of evidence entries retain the association information with the previous chain of evidence entry.

[0229] It should be noted that the evidence chain is linked item by item through the current solidified summary and the associated information of the preceding chain items, ensuring that any object can be verified independently and can be traced continuously along the preceding and following chain items; and ensuring that when the test report record or conclusion record needs to be traced, it is possible to trace back to the corresponding object from any item, or to continue to trace back to the related objects before the current object, forming a complete time chain tracing path.

[0230] Furthermore, after the chain of evidence is formed, a head identifier and a tail identifier are generated for the chain of evidence, and the total number of objects contained in the chain, the distribution of object types, and the corresponding unique detection task identifier are recorded.

[0231] The chain head identifier is used to indicate the starting position of the evidence chain corresponding to the current engineering quality inspection activity, and the chain tail identifier is used to indicate the ending position of the evidence chain corresponding to the current engineering quality inspection activity.

[0232] Furthermore, after the evidence chain is formed, based on the unique detection task identifier, the number of the object being detected, and the detection item number, the initialization evidence block, window judgment record, regular evidence block, event evidence block, and event switching record corresponding to the current engineering quality detection activity are extracted from the evidence chain to generate a conclusion record.

[0233] It should be noted that the conclusion record is not a simple copy of the original record, but a summary record of the process status, event situation and traceable evidence scope of this engineering quality inspection activity. The conclusion record includes a unique inspection task identifier, the inspected object number, the inspection item number, the conclusion type, the conclusion description information, the corresponding evidence block summary set, the corresponding window judgment summary set, the corresponding event switch summary set, the evidence chain head identifier and the evidence chain tail identifier.

[0234] It should be noted that the conclusion type is determined jointly by the conclusion classification rules in the testing basis and the status information in the chain of evidence. The conclusion types include process stability conclusion, process abnormality conclusion, and conclusion requiring review.

[0235] It should be noted that the process stability conclusion is used to characterize the absence of an event state in the evidence chain, or the occurrence of a short-term event state, but which is determined to not affect the overall stability of the detection process according to the conclusion classification rules in the detection basis and in combination with the state information in the evidence chain; the process abnormal conclusion is used to characterize the existence of a clear event state in the evidence chain and the corresponding event switching record has been formed; the conclusion requiring review is used to characterize the existence of data missing, identity conflict and process deviation in the evidence chain, which makes it impossible to directly give a single conclusion of stability or abnormality.

[0236] It should be noted that the purpose of the conclusion type is to provide a standardized expression of process conclusions for subsequent test reports, rather than replacing the professional test value judgment of the specific test item itself.

[0237] Furthermore, a test report is generated based on the conclusion record, evidence chain location information, and object statistical information.

[0238] It should be noted that the object statistics include the total number of objects contained in the chain of evidence and the distribution of object types.

[0239] It should be noted that the test report records include the report number, unique test task identifier, project number, tested object number, test item number, test time range, source list, number of regular evidence blocks, number of event evidence blocks, event number list, window status statistics, conclusion type, conclusion description information, and evidence chain location information.

[0240] The window status statistics include the number of regular status windows and the number of event status windows; the evidence chain location information includes the chain head identifier, the chain tail identifier, and the total number of evidence chain objects.

[0241] It should be noted that the test report record simultaneously includes index information for the evidence block summary set, window judgment summary set, and event switching summary set corresponding to the conclusion record, so as to subsequently locate the corresponding entries in the evidence chain from the test report record; the test report record is a result record for users to view, and also an index entry for traceability.

[0242] Furthermore, when subsequent verification of the test report records is required, the corresponding evidence chain is first located based on the evidence chain location information in the test report records. Then, the corresponding object body is located based on the object identifier in the evidence chain and the current solidified summary. The object body is then re-normalized and subjected to standard hash processing. When the regenerated summary is consistent with the solidified summary stored in the evidence chain, it is determined that the current object has not been tampered with. When the regenerated summary is inconsistent with the solidified summary stored in the evidence chain, it is determined that the current object has a consistency anomaly risk, and the corresponding object is marked as an object to be reviewed.

[0243] This embodiment also provides an information processing system for engineering quality inspection, including:

[0244] The task documentation module is used to collect basic task information, generate a unique detection task identifier, establish source mapping relationships and a unified time benchmark, and generate initialization evidence blocks, initial reference states, planned process sequences, as well as original record cache areas and feature extraction entry points.

[0245] The status determination module is used to receive raw records from various sources, extract real-time features and form window features, and generate event triggering results, related source sets and window determination records;

[0246] The evidence segmentation module is used to perform regular segmentation and event segmentation on the original record based on the window state label representing the record, forming regular evidence blocks and event evidence blocks, and performing front window replenishment and related source updates in the event state;

[0247] The evidence output module is used to generate solidified summaries from initialization evidence blocks, window judgment records, regular evidence blocks, event evidence blocks, and event switching records, connect them to form an evidence chain, and generate conclusion records and test report records.

[0248] In summary, this invention achieves dynamic identification of the detection process status and directional location of related sources by forming window features and generating event triggering results; by continuously performing segmentation based on the window status label represented by the window determination record, regular evidence blocks and event evidence blocks are formed, and event switching records are generated at the same time, supporting the continuous association of solidified summaries, evidence chains, conclusion records and detection report records, taking into account process tracking, event reconstruction and result traceability.

[0249] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications and substitutions are covered within the scope of the claims of the present invention.

Claims

1. An information processing method for engineering quality inspection, characterized in that, include: Collect basic information on engineering quality inspection activities and generate a unique inspection task identifier. By establishing source mapping relationships and a unified time benchmark, generate an initialization evidence block. Based on the initialization evidence block, generate an initial reference state and a planned process sequence, and establish an original record cache and feature extraction entry point. The system continuously receives raw records from various sources and writes them to the raw record buffer. Real-time features are extracted from the feature extraction entry point to form window features. The current reference state is updated through a stable window. The event triggering result is generated by combining the window features, the initial reference state, the current reference state, the planned process sequence, the integrity of records within the window, and the identity consistency. The relevant source set is determined based on the contribution of the window features corresponding to each source to the event triggering result, and a window judgment record is generated. For the original records in the original record buffer during the detection process, the window state label representing the record is determined by the window and the segmentation is continuously performed to form regular evidence blocks and event evidence blocks, while generating event switching records. A solidified summary is generated for the initialization evidence block, window judgment record, regular evidence block, event evidence block, and event switching record. Based on the solidified summary, the evidence chain is formed by connecting them in a unified time sequence and a conclusion record and a test report record are generated.

2. The information processing method for engineering quality inspection as described in claim 1, characterized in that, The process of collecting basic information on engineering quality inspection activities and generating unique inspection task identifiers, along with establishing source mapping relationships and a unified time benchmark, to generate initialization evidence blocks includes: Collect basic task information, and for each source, collect source number, source category, installation location, the object being detected bound to the source, and the source's recording rules; Perform integrity and consistency checks, and combine and encode the project number, the object being tested number, the batch number, the test item number, and the test start date to obtain a unique test task identifier. Establish source mapping relationships and a unified time benchmark, collect standard reference information, self-inspection records before the start of testing, and the first stable record after the start of testing that meets the initial stable inclusion conditions, and generate an initial evidence block.

3. The information processing method for engineering quality inspection as described in claim 2, characterized in that, The process of generating an initial reference state and planned flow sequence based on the initialization evidence block, and establishing an original record cache and feature extraction entry point includes: Baseline feature components are extracted from the initial evidence block to generate an initial reference vector and an initial discrete matrix, which together form the initial reference state. Generate a planned process sequence based on the action requirements and source mapping relationships in the testing criteria; A raw record cache is established using a unique detection task identifier, source number, and unified timestamp as indexes, and a feature extraction entry point is established that reads raw records only from the raw record cache.

4. The information processing method for engineering quality inspection as described in claim 3, characterized in that, The continuous reception of raw records from various sources and writing them into the raw record buffer, followed by the extraction of real-time features from the feature extraction entry point to form window features, includes: The system continuously receives raw records from various sources and writes them to the raw record cache. The feature extraction entry point reads the raw records and extracts real-time features based on the source number and source category by calling the feature extraction rules. All extracted real-time features are normalized, a sliding time window is set, and the real-time features within the current sliding time window are aligned and aggregated according to a unified timestamp to form window features.

5. The information processing method for engineering quality inspection as described in claim 4, characterized in that, The step of updating the current reference state through a stable window, combined with window features, the initial reference state, the current reference state, the planned process sequence, the integrity of records within the window, and identity consistency, generates the event triggering result, including: Based on the original records and action records within the current sliding time window, calculate the completeness of records within the window, the consistency of identities within the window, and the degree of matching of the planned process sequence. By combining window features, initial reference state, and current reference state, the dual reference deviation of the current sliding time window is calculated; When the deviation of the dual references, the integrity of the records within the window, the consistency of the identity within the window, and the degree of matching of the planned process sequence meet the corresponding stable thresholds, the current sliding time window is determined to be a stable window, and the current reference state is updated according to the window characteristics. By combining window features, initial reference state, current reference state, planned process sequence, integrity of records within the window, and identity consistency, an event-triggered coupling algorithm is used to generate event-triggered results.

6. The information processing method for engineering quality inspection as described in claim 5, characterized in that, The step of determining the relevant source set and generating window determination records based on the contribution of window features corresponding to each source to the event triggering result includes: Extract the window feature sub-vectors corresponding to each source and the reference sub-vectors corresponding to each source in the current reference state. Based on the degree of deviation between the window feature sub-vectors corresponding to each source and the reference sub-vectors, the contribution of the window features corresponding to each source to the event triggering result is characterized by the proportion of the deviation degree corresponding to each source in the total deviation of all sources. Sources whose contribution level is not less than the source contribution threshold are identified as the relevant source set. When no source meets the source contribution threshold, the source with the largest contribution level is selected, and the operation trajectory source and the on-site record source are included in the relevant source set. A window determination record is generated based on the event triggering result and the relevant source set.

7. The information processing method for engineering quality inspection as described in claim 1, characterized in that, The original records in the original record buffer during the detection process are continuously segmented according to the window state label representing the record, forming regular evidence blocks and event evidence blocks, including: When the window determines the record to represent the normal state, the source number is used as the smallest segmentation object, and a normal evidence block is formed according to the normal segmentation rules controlled by the normal segmentation duration; When the window determines the first time the event state is characterized, an event number is generated and the initial set of related sources is determined, with the start time of the current sliding time window as the event start boundary; For the original records in the relevant source set that are before the start boundary of the source backfill event and have a length equal to the backfill duration of the front window, form event evidence blocks according to the event segmentation rules; For original records from unrelated sources, continue to form regular evidence blocks according to the usual segmentation rules.

8. The information processing method for engineering quality inspection as described in claim 7, characterized in that, The simultaneous generation of event switching records includes: While the current event has not ended and subsequent window determination records continue to represent the event state, read the relevant source set corresponding to the subsequent window determination records; For newly added sources, the same front window backfilling rules and event segmentation rules as when the event is first triggered are used; when the window judgment record is restored from the event state to the normal state, the restoration confirmation count rule is used to perform the end judgment of the current event, determine the event end boundary, and generate the event switching record; After the current event ends, the regular segmentation rules are resumed for each source in the set of sources related to the current event.

9. The information processing method for engineering quality inspection as described in claim 1 or 8, characterized in that, The process of generating solidified summaries for initialization evidence blocks, window determination records, regular evidence blocks, event evidence blocks, and event switching records, and then concatenating these solidified summaries in a unified chronological order to form an evidence chain and generate conclusion records and test report records includes: The initialization evidence block, window judgment record, regular evidence block, event evidence block, and event switching record are standardized and organized, and solidified digests are generated using a standard hash algorithm. All types of objects are sorted in a uniform chronological order and linked together to form an evidence chain based on the fixed summary. The corresponding objects are extracted from the evidence chain based on the unique detection task identifier, the number of the object being detected, and the number of the detection project, and conclusion records and detection report records are generated.

10. An information processing system for engineering quality inspection, based on the information processing method for engineering quality inspection according to any one of claims 1 to 9, characterized in that, include: The task documentation module is used to collect basic task information, generate a unique detection task identifier, establish source mapping relationships and a unified time benchmark, and generate initialization evidence blocks, initial reference states, planned process sequences, as well as original record cache areas and feature extraction entry points. The status determination module is used to receive raw records from various sources, extract real-time features and form window features, and generate event triggering results, related source sets and window determination records; The evidence segmentation module is used to perform regular segmentation and event segmentation on the original record based on the window state label representing the record, forming regular evidence blocks and event evidence blocks, and performing front window replenishment and related source updates in the event state; The evidence output module is used to generate solidified summaries from initialization evidence blocks, window judgment records, regular evidence blocks, event evidence blocks, and event switching records, connect them to form an evidence chain, and generate conclusion records and test report records.

Images