Production safety monitoring and early warning method and system based on industrial internet

Abstract

The application discloses a production safety monitoring and early warning method and system based on an industrial internet, and relates to the technical field of monitoring and early warning.The continuity of time series and the stability of network paths are fully considered in the data processing process, so that the monitoring result can more truly reflect the on-site operation condition, and the false alarm and the missed alarm caused by single link fluctuation, local interference and other factors are reduced.Meanwhile, through the cross verification mechanism of multiple paths and multiple nodes, the confirmation process of abnormal information is more rigorous and reliable, the response speed can be ensured, and the accuracy of early warning can be improved.The application has strong adaptive capacity, can adapt to industrial network environments of different scales and different structures, and can maintain stable monitoring and analysis performance even in the case that the number of nodes is large, the link topology is complex, and the data fluctuation is frequent.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of monitoring and early warning, and particularly to a production safety monitoring and early warning method and system based on an industrial internet. BACKGROUND

[0002] With the rapid development of technologies such as industrial internet, internet of things, and big data, safety monitoring in production sites is gradually evolving towards real-time, networking, and intelligentization. Through distributed device nodes, multiple types of safety monitoring data such as temperature, pressure, vibration, and chemical concentration are collected, and combined with path identification information of network links, centralized monitoring and information sharing across regions and devices can be achieved. On this basis, technologies such as dynamic envelope analysis, anomaly detection, and multi-source data fusion are continuously applied in the field of production safety, in order to improve the timeliness and accuracy of accident early warning. However, in actual application, due to the diversity of industrial site data collection links, complexity of transmission paths, wide distribution of devices, and multiple running environment disturbances, the time sequence consistency of collected data, path stability, and accuracy of anomaly determination still face great challenges. Especially in the scenario of high concurrency, multi-node, and cross-path data fusion, the existing technologies are prone to delay, misjudgment, or false reporting in the process of abnormal signal discrimination and effectiveness confirmation, thereby affecting safety management decisions.

[0003] The existing production safety monitoring and early warning method mainly relies on single-path or single-node data for anomaly judgment, and lacks deep utilization of the correlation between multi-path and multi-time period data. In terms of path stability evaluation, most methods only rely on instantaneous link quality indicators, which cannot comprehensively reflect the stability trend of the path at different time periods, and are prone to misjudgment in dynamic changes. At the same time, there is a lack of consistency comparison of adjacent nodes or multi-path data trends in the anomaly determination process, which leads to false anomalies caused by local interference and data drift being incorrectly marked as risk events, reducing the reliability of early warning. SUMMARY

[0004] In view of the problems existing in the prior art production safety monitoring and early warning technology, the present application is proposed.

[0005] Therefore, the problem to be solved by the present application is how to improve the accuracy and robustness of production safety monitoring and reduce the risk of false positives and false negatives.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides a production safety monitoring and early warning method based on an industrial internet, which comprises: collecting safety monitoring data and link path identification information reported by a device node, establishing a joint data packet containing a collection timestamp, a path segment code and a data source identification; arranging the joint data packet in order of the collection timestamp, constructing a multi-time period tolerance window, and performing homologous path aggregation processing according to the path segment code, and extracting a path stability matrix and a source data sequence; constructing a dynamic envelope of running data based on the source data sequence, and if a current production safety monitoring data sequence exceeds the envelope boundary, combining the path segment stability matrix to output a to-be-confirmed abnormality mark; and performing joint comparison on the production safety monitoring data marked as to-be-confirmed abnormality according to the path segment code and adjacent device nodes, and if consistent fluctuation trends appear in multiple paths in the same time period, marking the data as effective early warning data.

[0008] As a preferred scheme of the production safety monitoring and early warning method based on the industrial internet, the construction of the joint data packet comprises: performing identification code configuration on a safety monitoring device deployed on site, binding physical installation position information of the safety monitoring device and a device unique identification to form a data source identification; collecting generated safety monitoring data uploaded by the safety monitoring device, extracting a relay node identification of a network link through which the data is uploaded, and sequentially splicing to generate a path segment code; when first running, setting a path segment code sequence generated in a first continuous time interval as an initial historical sequence reference; and attaching a uniform format collection timestamp to each generated safety monitoring data record, and encapsulating the collection timestamp, the path segment code and the data source identification into a joint data packet in a set order.

[0009] As a preferred scheme of the production safety monitoring and early warning method based on the industrial internet, the construction of the joint data packet further comprises: detecting whether the data source identification is repeatedly bound, whether the path segment code contains a continuous link breakpoint and whether the collection timestamp is in reverse order and misaligned, and if any abnormal condition is met, the corresponding data packet is excluded.

[0010] As a preferred scheme of the production safety monitoring and early warning method based on industrial internet, the construction of the multi-time interval tolerance window comprises: arranging the joint data packets in ascending order according to the collection timestamps, counting the collection interval time between adjacent joint data packets to form a collection interval sequence, and segmenting and aggregating the collection interval sequence according to a set sliding window length to generate a time interval set; calculating the variation amplitude and standard fluctuation trend characteristic of the collection time interval within each time interval, and if the variation amplitude continuously appears in a time interval deviating from the overall average collection time interval by more than a preset proportion threshold within the current time interval, the time interval is marked as a non-stable interval, and the remaining time intervals are marked as candidate stable intervals; merging the continuous time intervals in all candidate stable intervals into stable time intervals, and if the boundary interval time of two adjacent candidate stable intervals is lower than a dynamic interval threshold, the two adjacent candidate stable intervals are merged again; and the plurality of stable time intervals formed by the final merging are defined as a multi-time interval tolerance window set.

[0011] As a preferred scheme of the production safety monitoring and early warning method based on industrial internet, the homologous path aggregation processing according to the path segment code comprises: extracting the path segment code in each joint data packet, constructing a grouping table based on the path segment code, and initially grouping the joint data packets with the same path segment code into the same path group, while associating the corresponding time interval tolerance window number to form a path segment-window two-level index structure; calculating the appearance frequency of the corresponding path segment in the corresponding time interval tolerance window, the sequence position consistency of the path segment code in the joint data packet, and the difference fluctuation amplitude of the path segment collection time interval and the time interval tolerance window average interval in each path segment-window two-level index group, respectively generating the frequency index, the sequence consistency score and the interval deviation score; calculating the stability score by weighting the three score indexes, generating a path stability matrix with the path segment code as the index, and each element is the stability score of the path segment in the corresponding time interval tolerance window; after the path segment stability score matrix is aggregated according to the path segment code, the data of all joint data packets in the corresponding time interval tolerance window is extracted for each path segment, arranged in chronological order to form a source data sequence; and the sequence position consistency is whether the path segment code appearance order matches the initial historical order reference, and when the path segment code sequence change rate continuously exceeds the change rate threshold, the initial historical order reference is updated.

[0012] As a preferred scheme of the production safety monitoring and early warning method based on the industrial internet, the dynamic envelope of the running data comprises: in the source data sequence, taking the path segment code and the corresponding time period tolerance window number as the index, extracting all value points of the same path segment within the historical time period tolerance window before the current time period in chronological order, calculating the mean, fluctuation range and trend slope of the value points in each time period, and forming a feature triple set; the feature triple set is spliced in chronological order to form an initial envelope reference sequence; based on the initial envelope reference sequence, the upper and lower envelope boundaries are corrected according to the feature triple change amplitude of adjacent time periods; the correction proportion of the upper and lower boundaries is controlled by the stability score of the corresponding path segment in the path stability matrix, and the dynamic envelope of the running data is generated; the lower the stability score, the greater the envelope boundary adjustment amplitude.

[0013] As a preferred scheme of the production safety monitoring and early warning method based on the industrial internet, if the current production safety monitoring data sequence exceeds the envelope boundary, the combination of the path segment stability matrix outputs the to-be-confirmed abnormality mark, which comprises: in the dynamic envelope, taking the path segment code and the time period tolerance window number as the index, comparing the corresponding data points of the current production safety monitoring data sequence with the upper and lower boundaries of the dynamic envelope indexed one by one, marking all time periods and corresponding path segment codes that exceed the envelope boundary to form an out-of-limit data index table, and retaining the out-of-limit amplitude and direction information; for each record in the out-of-limit data index table, the stability score of the corresponding path segment code and time period tolerance window number in the path stability matrix is called, and the weight adjustment of the out-of-limit amplitude is performed according to the stability score; when the weight-adjusted out-of-limit amplitude reaches or exceeds the preset marking threshold, the corresponding record is added to the to-be-confirmed abnormality mark set; if the weight adjustment results of multiple path segment codes in the same time period all reach the preset marking threshold, an associated identifier is established for the corresponding time period in the to-be-confirmed abnormality mark set.

[0014] As a preferred scheme of the production safety monitoring and early warning method based on the industrial internet, the weight adjustment of the out-of-limit amplitude according to the stability score comprises: when the stability score is lower than the preset low score threshold, the out-of-limit amplitude is enlarged by m%; when the stability score is between the preset low score threshold and the preset high score threshold, the out-of-limit amplitude remains unchanged; when the stability score is higher than the preset high score threshold, the out-of-limit amplitude is reduced by n%; wherein m and n are constants.

[0015] As a preferred scheme of the production safety monitoring and early warning method based on the industrial internet, the joint comparison comprises: in the abnormal mark set to be confirmed, taking the path segment code as an index, retrieving the path topology-based adjacent equipment node set corresponding to the path segment code, and extracting the production safety monitoring data sequence of the adjacent equipment node in the same time period tolerance window number in the source data sequence to form an adjacent node comparison data set; meanwhile, the association between the collection time stamp of each adjacent node comparison data and the corresponding path segment code is retained, and the time and path consistency of the comparison is maintained; for each production safety monitoring data in the adjacent node comparison data set, the numerical deviation and trend direction consistency score of the corresponding record in the abnormal mark set to be confirmed are calculated, and when the numerical deviation absolute value and the trend direction consistency score simultaneously satisfy the preset joint comparison condition, the corresponding abnormal mark to be confirmed is updated to a consistent fluctuation mark; if not less than H% of the adjacent equipment node records are updated to the consistent fluctuation mark in the same time period, an effective early warning label is added to the corresponding time period; wherein H is a constant.

[0016] In a second aspect, the present application provides a vehicle remote monitoring system, which comprises:

[0017] A data collection module collects safety monitoring data and link path identification information reported by the equipment nodes, and establishes a joint data packet containing a collection time stamp, a path segment code, and data source identification;

[0018] A path aggregation module arranges the joint data packet in order of collection time stamp, constructs a multi-time period tolerance window, and performs homologous path aggregation processing according to the path segment code, extracts a path stability matrix and a source data sequence;

[0019] A dynamic envelope module constructs a dynamic envelope of running data based on the source data sequence, and outputs an abnormal mark to be confirmed if the current production safety monitoring data sequence exceeds the envelope boundary in combination with the path segment stability matrix;

[0020] An early warning judgment module performs joint comparison on the production safety monitoring data marked as abnormal to be confirmed according to the path segment code and the adjacent equipment nodes, and if consistent fluctuation trends appear in multiple paths at the same time period, the data is marked as effective early warning data.

[0021] In a third aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the computer program instructions are executed by the processor to realize the steps of the production safety monitoring and early warning method based on the industrial internet according to the first aspect of the present application.

[0022] In a fourth aspect, the present application provides a computer readable storage medium having stored thereon a computer program, wherein the computer program instructs a processor to implement the steps of the production safety monitoring and early warning method based on the industrial internet according to the first aspect of the present application.

[0023] The present application has the following advantages: the present application fully considers the continuity of time series and the stability of network paths in the data processing process, so that the monitoring result can more truly reflect the on-site operation condition, and the false positives and false negatives caused by single link fluctuation, local interference and other factors are reduced. At the same time, through the cross-validation mechanism of multiple paths and multiple nodes, the confirmation process of abnormal information is more rigorous and reliable, which can improve the accuracy of early warning while ensuring the response speed. The present application has strong adaptive ability and can adapt to industrial network environments of different scales and different structures. Even in the case of a large number of nodes, complex link topology and frequent data fluctuations, the present application can also maintain stable monitoring and analysis performance.

[0024] Overall, the present application not only improves the real-time performance and reliability of production safety management, but also provides more sufficient data support for subsequent accident prevention and emergency disposal, which helps to reduce safety risks and ensure production continuity and personnel and equipment safety. BRIEF DESCRIPTION OF DRAWINGS

[0025] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0026] Figure 1 The flowchart of the production safety monitoring and early warning method based on the industrial internet.

[0027] Figure 2 The structure diagram of the production safety monitoring and early warning system based on the industrial internet. DETAILED DESCRIPTION

[0028] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification.

[0029] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from the description, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited by the specific embodiments disclosed below.

[0030] Second, the "one embodiment" or "embodiments" referred to herein are meant to encompass a specific feature, structure, or characteristic in at least one implementation of the application. The appearances of the "in one embodiment" or "in an embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are they necessarily mutually exclusive.

[0031] As described in the above background, the existing production safety monitoring and early warning method mainly relies on single-path or single-node data for abnormality judgment, and lacks deep utilization of the correlation between multi-path and multi-time period data. In terms of path stability evaluation, most methods only rely on instantaneous link quality indicators, which cannot comprehensively reflect the stability trend of the path in different time periods, and are prone to misjudgment in dynamic changes. At the same time, there is a lack of consistency comparison of adjacent nodes or multi-path data trends in the abnormality judgment process, which leads to false abnormalities caused by local interference and data drift being incorrectly marked as risk events, reducing the reliability of the early warning.

[0032] Figure 1 The flowchart of the production safety monitoring and early warning method based on industrial internet according to the embodiments of the present application. As shown in Figure 1 The production safety monitoring and early warning method based on industrial internet includes the following steps:

[0033] S1: Collecting safety monitoring data and link path identification information reported by device nodes, and establishing a joint data packet containing collection timestamp, path segment code and data source identification.

[0034] S1.1: Identifying and coding the safety monitoring devices deployed on site, binding the physical installation location information of the safety monitoring devices and the unique identification of the devices to form the data source identification.

[0035] In operation, first, the device number is managed in sequence, and the number format is formatted into a fixed length string, so as to be uniformly used in data collection, transmission and subsequent comparison. Secondly, the installation location is digitally marked, and the coordinate information of the installation location is mapped and stored with the unique identification of the device, so as to accurately trace to the specific device and its spatial location in subsequent data collection.

[0036] After the identification and coding configuration is completed, the binding information of each device needs to be verified, including checking whether the device unique identification code is repeated, checking whether the location code conflicts with other devices, and immediately renumbering or adjusting the installation location of the device with problems. This operation ensures that each collected data has a clear source and spatial attribute when generating a joint data packet, and guarantees the uniqueness and traceability of the data source identification in the entire safety monitoring system.

[0037] S1.2: Collect the generated safety monitoring data uploaded by the safety monitoring device, extract the relay node identification of the uploaded network link, and splice the path segment code in sequence; when running for the first time, set the path segment code sequence generated in the first continuous time interval as the initial historical order reference. Breakpoint detection is performed on the initial historical order reference, and if there is a continuous link interruption, the reference reconstruction is triggered.

[0038] During operation, first, the route node information of the data passing through is obtained through the network layer protocol or communication link management, and is recorded in sequence according to the data transmission order. Subsequently, these route node identifications are spliced in sequence to form a complete path segment code sequence, which can accurately describe the entire transmission path of the data from the device collection point to the receiving end. The generation of the path segment code not only records the network structure information, but also reflects the transmission delay or link change that may occur in different time periods.

[0039] In the first system operation or path initialization phase, the path segment code sequence generated in the first continuous time interval needs to be set as the initial historical order reference. In operation, by collecting data at consecutive time points, the path segment code is extracted and arranged in chronological order to form a reference sequence stored in the system database. The reference sequence is used to compare the newly generated path segment code in subsequent operation to determine whether the path has changed or there is an abnormal link. Through this operation, path change detection, link stability analysis, and abnormal data filtering can be realized, providing reliable basic information for subsequent multi-path aggregation and dynamic envelope construction.

[0040] S1.3: Attach a uniform format acquisition timestamp to each generated safety monitoring data record, and package the acquisition timestamp, path segment code and data source identification in the order set as a joint data packet.

[0041] Specifically, first, the data source identification is placed at the head of the joint data packet according to the field order, followed by the path segment code as the middle field, and the acquisition timestamp as the tail field, so as to be indexed and retrieved in time, path and source order subsequently. The joint data packet structure needs to support standardized storage formats such as JSON, XML or binary serialization form, to ensure consistency of the system in data transmission, storage and parsing process.

[0042] S1.4: Detect whether the data source identification is repeatedly bound, whether the path segment code contains a continuous link breakpoint, and whether the acquisition timestamp is in reverse order, and if any of the abnormal conditions are met, the corresponding data packet is excluded.

[0043] Preferably, during operation, firstly, it is detected whether the data source identifier is repeatedly bound, i.e., it is checked whether the same unique device identifier corresponds to multiple installation positions or number conflicts; secondly, it is detected whether the path segment code contains a continuous link breakpoint, and by comparing the path segment code sequence with the historical order reference, it is confirmed whether the relay node is missing or misplaced; and finally, it is detected whether the collection timestamp is in reverse order, i.e., whether the same path segment appears in reverse order within a continuous collection time.

[0044] When any abnormal condition in the joint data packet is triggered, the data packet is automatically marked as abnormal and excluded from the data set. During operation, the exclusion not only includes physical deletion, but also recording the abnormal type, path segment code, timestamp and data source identifier in the log for subsequent tracing and analysis.

[0045] S2: arranging the joint data packet in the order of the collection timestamp, constructing a multi-time period tolerance window, and performing homologous path aggregation processing according to the path segment code, to extract the path stability matrix and the source data sequence.

[0046] S2.1: Constructing a multi-time period tolerance window.

[0047] S2.1.1: Arranging the joint data packet in ascending order according to the collection timestamp, counting the collection interval time between adjacent joint data packets to form a collection interval sequence, and segmenting and aggregating according to the set sliding window length to generate a time period set.

[0048] The arrangement operation not only involves time sorting, but also needs to ensure that the joint data packets of the same path segment and data source maintain a relative order in the sequence, so that the internal time interval characteristics of each time period can be accurately calculated when constructing the sliding window and the time period tolerance window subsequently.

[0049] Further, the sliding window length parameter is parameterized and managed, and in implementation, the fixed value or adaptive value sliding window length is allowed to be selected as , wherein, is the target coverage duration, is the data density, which is estimated by the sample count per unit time; when using the sliding window, the in-window statistics including the window mean, window standard deviation and window median need to be calculated in each window, and these statistics are written as the initial attributes of the time period into the time period metadata.

[0050] During the aggregation process, the start and end timestamps, sample index range and sample count of each window must be recorded to facilitate subsequent tracing and weight distribution. Among them, the fixed sliding window mode is more efficient when the data arrival rate is stable, and the adaptive sliding window mode can better maintain the time resolution and statistical effectiveness when the data bursts, is sparse or the load changes, and it is recommended to use both modes in parallel to cross-verify the consistency of the results.

[0051] S2.1.2: For the acquisition timestamp sequence in the set of time periods, calculate the variation amplitude and standard fluctuation trend characteristics of the acquisition time interval within each time period, and if the variation amplitude deviates from the overall average acquisition time interval by more than a preset proportion threshold in the current time period, mark the time period as a non-stable period, and mark the remaining time periods as candidate stable periods.

[0052] Specifically, the acquisition interval within each time period is compared with the average acquisition interval of the corresponding time period, and when a plurality of consecutive acquisition intervals deviate from the overall average value by more than a preset proportion threshold, the corresponding time period is marked as a non-stable period; the remaining time periods are marked as candidate stable periods.

[0053] The preset proportion threshold is set to be based on the statistical results of the deviation proportion of the acquisition interval in the historical stable acquisition time period relative to the average acquisition interval.

[0054] This determination rule ensures that when there is an abnormal acquisition fluctuation, the non-stable data segment can be isolated in time, without affecting the continuity of the overall candidate stable period, thereby providing a reliable basis for subsequent path stability scoring. Through this operation, abnormal fluctuations in data acquisition can be effectively identified while ensuring the continuity of the time period, thereby improving the accuracy and reliability of the multi-time period tolerance window.

[0055] S2.1.3: Merge the consecutive time periods in all candidate stable periods into stable time periods, and if the boundary interval time of two adjacent candidate stable periods is lower than a dynamic interval threshold, then merge again, and define the plurality of stable time periods formed by the final merging as a set of multi-time period tolerance windows.

[0056] In the embodiment of the application, the merging operation needs to be performed in two steps. The first step is boundary interval calculation: for adjacent candidate stable periods A and B, calculate the boundary interval , and calculate the average interval and the standard deviation of the adjacent stable periods.

[0057] The second step is dynamic threshold calculation: , wherein is the dynamic threshold, is the lower limit of the engineering to prevent the threshold from being too small to cause excessive segmentation, is the sensitivity coefficient. If , the merging operation is performed.

[0058] Through the continuous merging operation, the time period fragmentation caused by short time interval fluctuations can be reduced, the integrity and stability of the time period tolerance window can be improved, and accurate time period indexes can be provided for subsequent path aggregation and stability matrix calculation. The finally formed stable time period set is defined as a multi-time period tolerance window set, and each tolerance window number is associated with the start and end timestamps, providing a basis for the time index of the source data sequence.

[0059] S2.2: Homologous path aggregation processing according to path segment encoding includes:

[0060] Extract the path segment encoding in each joint data packet, construct a grouping table based on the path segment encoding, and preliminarily group the joint data packets with completely consistent path segment encoding into the same path group, while associating the corresponding time period tolerance window number, forming a path segment-window two-level index structure.

[0061] In each path segment-window two-level index group, the occurrence frequency of the corresponding path segment in the corresponding time period tolerance window, the sequence position consistency of the path segment encoding in the joint data packet, and the difference fluctuation amplitude of the path segment collection time interval and the time period tolerance window average interval are calculated, respectively generating the frequency index, the sequence consistency score and the interval deviation score.

[0062] Among them, the number of occurrences of the path segment encoding in the current time period is counted as the frequency index.

[0063] Preferably, the sequence position consistency score calculation includes:

[0064] All joint data packets of a certain path segment in a specified time period tolerance window are arranged in ascending order of collection time, and the initial historical order reference corresponding to the path segment is extracted. The reference is the order sequence when the path segment is first run or the last order update. The deviation between the current sequence position and the historical reference position, i.e. the absolute value of the difference between the two, is calculated. The average deviation is obtained by averaging all position deviations The average deviation reflects the overall deviation of the path segment sequence from the historical order in the entire time period; the sequence position consistency score is defined as , wherein, is the maximum possible deviation of the sequence length, which is used for normalization to ensure that the score range is between 0 and 1. The closer the score is to 1, the more consistent the path segment sequence is with the historical order; the lower the score, the greater the path segment order change.

[0065] The calculation of the difference fluctuation amplitude comprises: calculating a sequence of adjacent collection intervals for each path segment in a time period window based on the collection timestamps of consecutive joint data packets, denoted as an actual collection interval sequence; comparing the sequence with an average interval sequence of all collection intervals in the current time period tolerance window point by point to obtain a difference sequence for describing the amplitude of the deviation of the collection intervals from the overall collection rule; performing trend analysis and normalization processing on the difference sequence: if the difference sequence continuously remains positive or negative and gradually decreases, it is marked as an interval convergence mode; if the difference sequence alternates between positive and negative in the short term, but returns to around zero as a whole, it is marked as a short-term fluctuation mode; if the difference sequence continuously expands, it is marked as an interval divergence mode. Through this mode recognition, the difference fluctuation amplitude is mapped to the three scores set: the convergence mode gets a high score, the short-term fluctuation mode gets a medium score, and the divergence mode gets a low score.

[0066] The three score indicators are used to calculate a stability score by weighting, and a path stability matrix is generated with the path segment code as the index, and each element is the stability score of the path segment in the corresponding time period tolerance window.

[0067] After all the path segment stability score matrices are summarized according to the path segment code, the data of all joint data packets in the corresponding time period tolerance window are extracted for each path segment, arranged in chronological order, and a source data sequence is formed. Among them, the sequence position consistency is whether the path segment code sequence matches the initial historical order benchmark, and when the path segment code sequence change rate continuously exceeds the change rate threshold, the initial historical order benchmark is updated. When the initial historical order benchmark is updated, the sequence consistency scores of all path segments in the affected time period are recalculated.

[0068] S3: Based on the source data sequence, a dynamic envelope of running data is constructed, and if the current production safety monitoring data sequence exceeds the envelope boundary, an abnormality confirmation mark is output in combination with the path segment stability matrix.

[0069] S3.1: Constructing a dynamic envelope of running data.

[0070] In the source data sequence, all value points of the same path segment in the historical time period tolerance window before the current time period (excluding the current detection window) are extracted in chronological order with the path segment code and the corresponding time period tolerance window number as the index, and the mean, fluctuation range and trend slope of the value points in each time period are calculated to form a feature triple set.

[0071] In operation, the mean reflects the central level of the path segment in the time period, the fluctuation range is used to quantify the fluctuation amplitude of the numerical points, and the trend slope is used to describe the change direction and speed of the numerical value over time. In actual calculation, the mean is obtained by calculating the arithmetic mean of all numerical points in the path segment, the fluctuation range is represented by the difference between the maximum and minimum values, and the trend slope is obtained by fitting the slope parameter of the time series by linear regression or local least squares method.

[0072] Further, the feature triple set is spliced in time period number order to form an initial envelope reference sequence, and based on the initial envelope reference sequence, the envelope upper and lower boundaries are corrected according to the feature triple change amplitude of adjacent time periods. By time sequence splicing, it can ensure that the boundaries of the dynamic envelope have continuity and traceability, while avoiding inaccurate evaluation caused by mixing of data across time periods. This operation ensures the scalability of the dynamic envelope in the time dimension, enabling subsequent boundary adjustment to change continuously according to the time trend.

[0073] wherein the correction proportion of the upper and lower boundaries is weighted controlled by the stability score of the corresponding path segment in the path stability matrix, and the dynamic envelope of the generated running data is generated, and the weighted control includes that the lower the stability score, the greater the envelope boundary adjustment amplitude.

[0074] For example, in the upper and lower boundary correction process, first, the path stability matrix corresponding to the path segment stability score of the time period is retrieved according to the time period number, and the score is stored corresponding to the change amplitude of the feature triple of the current time period to form a paired record of stability score and change amplitude; then, the stability score is mapped to the control coefficient of the boundary correction proportion according to the preset classification mapping rule, and the classification mapping rule is formulated according to the matching results of different stability levels and envelope adaptability in the system configuration stage, and is solidified as a callable parameter table. When the stability score is at a high level, the control coefficient corresponds to a smaller adjustment amplitude of the envelope boundary to maintain the convergence of the boundary; when the stability score is at a low level, the control coefficient corresponds to a larger adjustment amplitude of the envelope boundary to cover the possible wide fluctuation interval; then, in calculating the envelope upper and lower boundary correction value of the current time period, first, the boundary value after correction of the previous time period is extracted as the initial reference, and then the result of the control coefficient multiplied by the change amplitude of the feature triple of the current time period is used as the adjustment amount, and the adjustment amount is added to the corresponding position of the upper and lower boundaries to obtain the corrected envelope boundary.

[0075] This processing ensures that the change trend of the boundary is continuous in the time dimension, and the adjustment amplitude of the boundary of different path segments has differentiated characteristics under the weighted control of the stability score. Finally, the corrected envelope boundary is written into the dynamic envelope sequence, and the result is used in the boundary correction operation of the next time period to realize the recursive update and stability adaptive adjustment of the dynamic envelope in the whole time sequence.

[0076] S3.2: Compare the current time period production safety monitoring data sequence with the envelope boundary, if the current production safety monitoring data sequence exceeds the envelope boundary, output the abnormality confirmation mark combined with the path segment stability matrix.

[0077] S3.2.1: In the dynamic envelope, index the corresponding data points of the current production safety monitoring data sequence with the path segment code and the time period tolerance window number, and compare them one by one with the upper and lower boundaries of the dynamic envelope under the index, mark all the time periods and corresponding path segment codes that exceed the envelope boundary to form the out-of-limit data index table, and keep the out-of-limit amplitude and direction information (higher than the upper boundary or lower than the lower boundary).

[0078] S3.2.2: For each record in the out-of-limit data index table, call the stability score of the corresponding path segment code and time period tolerance window number in the path stability matrix, and weight adjust the out-of-limit amplitude according to the stability score.

[0079] Wherein, the weight adjustment of the out-of-limit amplitude according to the stability score includes: when the stability score is lower than the preset low score threshold, the out-of-limit amplitude is enlarged by m% proportion; when the stability score is between the preset low score threshold and the preset high score threshold, the out-of-limit amplitude remains unchanged; when the stability score is higher than the preset high score threshold, the out-of-limit amplitude is reduced by n% proportion; wherein, m and n are constants. For example, when m = 20, it means that the out-of-limit amplitude lower than the low score threshold will be enlarged by 20% to increase the risk sensitivity; when n = 15, it means that the out-of-limit amplitude higher than the high score threshold will be reduced by 15% to reduce false positives.

[0080] Wherein, according to the stability score distribution of the path segment that exceeds the limit and confirms the abnormality in the historical operation data, the cumulative probability lower than the set score value (set in the experimental operation) is set as the low score threshold; the preset high score threshold is set according to the stability score distribution of the path segment that continuously remains without abnormality in the historical operation data, and the cumulative probability higher than the set score value (set in the experimental operation) is set as the high score threshold.

[0081] It should be noted that the adjustment of the upper and lower boundaries is to adjust the tightness of the boundaries according to the stability score when generating or updating the dynamic envelope, which belongs to preventive control, and the envelope is adapted to the path segments with different stability before judging the out-of-bound; the weight adjustment of the out-of-bound amplitude according to the stability score is to recalculate the out-of-bound amplitude based on the stability score after the out-of-bound is found, which is used to evaluate the credibility and severity of the anomaly, and belongs to post-modification and anomaly quantification, which is a processing step after out-of-bound detection. Although both use the stability score as a weight factor, the stages and purposes are different, the former controls the boundary change in the boundary generation stage, and the latter controls the abnormal amplitude calculation in the out-of-bound detection stage, so they are not repeated. The weight adjustment is independent of the envelope generation stage and is only used for anomaly marker credibility calibration.

[0082] S3.2.3: When the weight-adjusted out-of-bound amplitude reaches or exceeds the preset marking threshold, the corresponding record is added to the set of abnormal marker to be confirmed; if the weight adjustment results of multiple path segment encodings in the same time period all reach the preset marking threshold, an associated identifier is established for the corresponding time period in the set of abnormal marker to be confirmed.

[0083] S4: For the production safety monitoring data marked as abnormal to be confirmed, joint comparison is performed according to the path segment encoding and adjacent device nodes, and if consistent fluctuation trends occur in multiple paths in the same time period, the data is marked as valid early warning data.

[0084] In the set of abnormal markers to be confirmed, the path segment encoding is used as the index to retrieve the set of adjacent device nodes based on the path topology corresponding to the path segment encoding, and the production safety monitoring data sequence of the adjacent device nodes in the same time period tolerance window number is extracted from the source data sequence to form the adjacent node comparison data set; meanwhile, the association relationship between the collection time stamp of each adjacent node comparison data and the corresponding path segment encoding is preserved, and the time and path consistency of the comparison is maintained.

[0085] For each production safety monitoring data in the adjacent node comparison data set, the numerical deviation and trend direction consistency score of the corresponding record in the set of abnormal markers to be confirmed are calculated, and when the numerical deviation absolute value and the trend direction consistency score both satisfy the preset joint comparison condition, the corresponding abnormal marker to be confirmed is updated to a consistent fluctuation marker; if in the same time period, no less than H% of the adjacent device node records are updated to the consistent fluctuation marker, an effective early warning label is added for the corresponding time period; wherein H is a constant.

[0086] In operation, the numerical deviation is obtained by calculating the difference between the value to be confirmed as an abnormality and the data of the adjacent node corresponding to the time period, the absolute value reflects the deviation size, and the sign indicates the deviation direction. The trend direction consistency score is calculated by comparing whether the rising, falling or stable trend of the data point to be confirmed as an abnormality and the adjacent node sequence is consistent. Specifically, for the path segment sequence to be confirmed as an abnormality, one adjacent sampling point is taken before and after the time point, and the signs of the adjacent differences are mapped as: rising = 1, stable = 0, falling = -1; the two trend encoding sequences are compared element by element: 1 point for complete consistency, 0 point for opposite signs, and 0.5 point for any one being stable and the other being rising or falling; the scores of the two segments are averaged to obtain the final trend consistency score.

[0087] Further, the numerical deviation absolute value is normalized; for each adjacent node, the joint score is calculated as the mean of the trend consistency score and the normalized numerical deviation sum; when the joint score is greater than or equal to the median of the joint scores of all adjacent nodes in the current window, the adjacent node is marked as a consistent fluctuation marker, and is marked as valid early warning data.

[0088] Further, as shown in Figure 2 the embodiment also provides an industrial internet-based production safety monitoring and early warning system, which comprises,

[0089] a data acquisition module, which acquires safety monitoring data and link path identification information reported by a device node, and establishes a joint data packet containing an acquisition timestamp, a path segment code and a data source identification;

[0090] a path aggregation module, which arranges the joint data packet in order of acquisition timestamp, constructs a multi-time period tolerance window, and performs homologous path aggregation processing according to the path segment code, and extracts a path stability matrix and a source data sequence;

[0091] a dynamic envelope module, which constructs a dynamic envelope of running data based on the source data sequence, and outputs a to-be-confirmed abnormality marker if the current production safety monitoring data sequence exceeds the envelope boundary in combination with the path segment stability matrix;

[0092] an early warning judgment module, which jointly compares the production safety monitoring data marked as to-be-confirmed abnormality according to the path segment code and the adjacent device node, and marks it as valid early warning data if consistent fluctuation trends appear in multiple paths at the same time period.

[0093] The embodiment also provides a computer device suitable for the industrial internet-based production safety monitoring and early warning method, which comprises a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to realize the industrial internet-based production safety monitoring and early warning method proposed in the above embodiment.

[0094] The computer device can be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected by a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be achieved by WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0095] The embodiment also provides a storage medium having a computer program stored thereon, the program being executed by a processor to implement the method for monitoring and early warning of production safety based on an industrial internet.

[0096] In summary, the application fully considers the continuity of time series and the stability of network paths in the data processing process, so that the monitoring result can more truly reflect the on-site running condition, and the false positives and false negatives caused by single link fluctuations, local interference and other factors are reduced. At the same time, through the cross-validation mechanism of multiple paths and multiple nodes, the confirmation process of abnormal information is more rigorous and reliable, which can improve the accuracy of early warning while ensuring the response speed. The application has strong adaptive ability and can adapt to industrial network environments of different scales and different structures. Even in the case of a large number of nodes, complex link topology and frequent data fluctuations, the application can also maintain stable monitoring and analysis performance.

[0097] Overall, the application not only improves the real-time performance and reliability of production safety management, but also provides more sufficient data support for subsequent accident prevention and emergency disposal, which helps to reduce safety risks and ensure production continuity and personnel and equipment safety.

[0098] It should be noted that the above embodiments are only used to illustrate the technical solutions of the application rather than limit the application. Although the application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the application, and all of them should be covered in the scope of the claims of the application.

Claims

1. A production safety monitoring and early warning method based on an industrial internet, characterized in that: The method comprises the following steps: Collecting safety monitoring data and link path identification information reported by the acquisition device node, establishing a joint data packet containing acquisition time stamp, path segment code and data source identification; Arranging the joint data packet in order of acquisition time stamp, constructing a multi-time interval tolerance window, and performing homologous path aggregation processing according to the path segment code, and extracting a path stability matrix and a source data sequence; Based on the source data sequence, a dynamic envelope of running data is constructed, and if the current production safety monitoring data sequence exceeds the envelope boundary, the path segment stability matrix is combined to output a to-be-confirmed abnormal marker; For the production safety monitoring data marked as to-be-confirmed abnormal, joint comparison is performed according to the path segment code and adjacent device nodes, and if consistent fluctuation trends appear in multiple paths in the same time interval, the data is marked as effective early warning data; The construction of the multi-time interval tolerance window comprises the following steps: arranging the joint data packet in ascending order according to the acquisition time stamp, counting the acquisition interval time between adjacent joint data packets to form an acquisition interval sequence, and segmenting and aggregating the acquisition interval sequence according to a set sliding window length to generate a time interval set; calculating the variation amplitude and standard fluctuation trend characteristics of the acquisition time interval within each time interval in the time interval set, and if the variation amplitude continuously appears in a time interval that deviates from the overall average acquisition time interval by more than a preset proportion threshold within the current time interval, the time interval is marked as a non-stable interval, and the remaining time intervals are marked as candidate stable intervals; merging the continuous time intervals in all candidate stable intervals into stable time intervals, and if the boundary interval time of two adjacent candidate stable intervals is lower than a dynamic interval threshold, the two adjacent candidate stable intervals are merged again; and the multiple stable time intervals formed after the final merging are defined as a multi-time interval tolerance window set; The homologous path aggregation processing according to the path segment code comprises the following steps: extracting the path segment code in each joint data packet, constructing a grouping table based on the path segment code, initially grouping the joint data packets with the same path segment code into the same path group, and simultaneously associating the corresponding time interval tolerance window number to form a path segment-window two-level index structure; in each path segment-window two-level index group, the appearance frequency of the corresponding path segment in the corresponding time interval tolerance window, the sequence position consistency of the path segment code in the joint data packet, and the difference fluctuation amplitude of the path segment acquisition time interval and the time interval tolerance window average interval are calculated to generate a frequency index, a sequence consistency score and an interval deviation score respectively; the three score indexes are calculated by weighting to generate a stability score, and a path stability matrix is generated with the path segment code as the index, and each element is the stability score of the path segment in the corresponding time interval tolerance window; after all the path segment stability score matrices are summarized according to the path segment code, the source data sequence is formed by extracting all the joint data packet data in the corresponding time interval tolerance window of each path segment and arranging them in chronological order. 2.The industrial internet-based production safety monitoring and early warning method of claim 1, wherein: The construction of the joint data packet comprises: The security monitoring device is identified and coded, and the physical installation position information and the unique identification of the security monitoring device are bound to form a data source identification; When collecting the security monitoring data uploaded by the security monitoring device, the relay node identification of the network link is extracted, and the path segment code is spliced in sequence; when running for the first time, the path segment code sequence generated in the first continuous time interval is set as the initial historical sequence reference; A uniform format acquisition timestamp is attached to each generated security monitoring data record, and the acquisition timestamp, path segment code and data source identification are packaged into a joint data packet in a set order. 3.The production safety monitoring and early warning method based on industrial internet of claim 2, characterized in that: The construction of the joint data packet also includes: The joint data packet is detected respectively to determine whether the data source identification is repeatedly bound, whether the path segment code contains a continuous link breakpoint, and whether the acquisition timestamp is in reverse order, and if any abnormal condition is met, the corresponding data packet is excluded. 4.The industrial internet-based production safety monitoring and early warning method of claim 1, wherein: The dynamic envelope of the running data includes: In the source data sequence, the same path segment in the historical time interval tolerance window before the current time interval is extracted in time sequence, the mean value, fluctuation range and trend slope of the numerical points in each time interval are calculated, and a feature triplet set is formed; The feature triplet set is spliced in time interval number order to form an initial envelope reference sequence; Based on the initial envelope reference sequence, the upper and lower boundaries of the envelope are corrected according to the feature triplet change amplitude of adjacent time intervals; The correction proportion of the upper and lower boundaries is controlled by the stability score of the corresponding path segment in the path stability matrix, and the dynamic envelope of the running data is generated; The lower the stability score, the greater the envelope boundary adjustment amplitude. 5.The production safety monitoring and early warning method based on industrial internet according to claim 4, characterized in that: If the current production safety monitoring data sequence exceeds the envelope boundary, the output of the abnormal mark to be confirmed includes: In the dynamic envelope, the corresponding data points of the current production safety monitoring data sequence are compared with the upper and lower boundaries of the dynamic envelope under the index of path segment code and time interval tolerance window number, and all time intervals and corresponding path segment codes that exceed the envelope boundary are marked to form an out-of-bound data index table, and the out-of-bound amplitude and direction information are retained; For each record in the out-of-bound data index table, the stability score of the corresponding path segment code and time interval tolerance window number in the path stability matrix is called, and the out-of-bound amplitude is adjusted by weight according to the stability score; When the weight-adjusted out-of-bound amplitude reaches or exceeds the preset marking threshold, the corresponding record is added to the abnormal mark to be confirmed set; if the weight adjustment results of multiple path segment codes in the same time interval all reach the preset marking threshold, an associated identification is established for the corresponding time interval in the abnormal mark to be confirmed set. 6.The industrial internet-based production safety monitoring and early warning method of claim 5, wherein: The weight adjustment of the out-of-bound amplitude according to the stability score includes: When the stability score is lower than the preset low score threshold, the out-of-bound amplitude is enlarged by m%; When the stability score is between the preset low score threshold and the preset high score threshold, the out-of-bound amplitude remains unchanged; When the stability score is higher than a preset high score threshold, the out-of-range amplitude is reduced by n% proportion; Wherein, m and n are constants.

7. The production safety monitoring and early warning method based on the industrial internet according to claim 1, characterized in that: The joint comparison includes: In the abnormality confirmation mark set, the path segment code is used as the index to retrieve the path topology-based adjacent device node set corresponding to the path segment code, and the production safety monitoring data sequence of the adjacent device node under the same time period tolerance window number is extracted from the source data sequence to form an adjacent node comparison data set; meanwhile, the association between the collection time stamp of each adjacent node comparison data and the corresponding path segment code is retained to maintain the consistency of time and path of the comparison; For each production safety monitoring data in the adjacent node comparison data set, the numerical deviation and trend direction consistency score of the corresponding record in the abnormality confirmation mark set are calculated, and when the numerical deviation absolute value and the trend direction consistency score meet the preset joint comparison condition at the same time, the corresponding abnormality confirmation mark is updated to a consistent fluctuation mark; if not less than H% of the adjacent device node records are updated to the consistent fluctuation mark in the same time period, an effective early warning label is added to the corresponding time period; wherein, H is a constant.

8. An industrial internet-based production safety monitoring and early warning system based on any one of the industrial internet-based production safety monitoring and early warning methods of claims 1-7. Also includes: A data acquisition module acquires safety monitoring data and link path identification information reported by the device node, and establishes a joint data packet containing collection time stamp, path segment code and data source identification; A path aggregation module arranges the joint data packet in order according to the collection time stamp, constructs a multi-time period tolerance window, and performs homologous path aggregation processing according to the path segment code to extract a path stability matrix and a source data sequence; A dynamic envelope module constructs a dynamic envelope of the running data based on the source data sequence, and outputs an abnormality confirmation mark if the current production safety monitoring data sequence exceeds the envelope boundary in combination with the path segment stability matrix; An early warning judgment module performs joint comparison on the production safety monitoring data marked as abnormality confirmation according to the path segment code and the adjacent device node, and if consistent fluctuation trends appear in multiple paths at the same time, the data is marked as effective early warning data.

Images