An industrial big data driven speed field measurement data analysis method and system

Abstract

The application relates to the technical field of industrial big data analysis, and discloses an industrial big data driven speed field measurement data analysis method and system, which comprises the following steps: receiving data packets from distributed speed measurement nodes, determining a time confidence interval for each data packet; when a speed field model at a target moment is constructed, whether the data packets from different distributed speed measurement nodes belong to the same moment physical event is judged based on the time confidence interval of the data packets; if it is judged that the data packets belong to the same moment physical event, the length of the time confidence interval of the data packets is used to assign a weight of the data packets in the speed field model. The application effectively solves the limitation of the traditional method in time alignment, and improves the construction accuracy and reliability of the speed field model.

Description

Technical Field

[0001] This application relates to the field of industrial big data analysis technology, and more specifically, to a method and system for analyzing velocity field measurement data driven by industrial big data. Background Technology

[0002] In modern smart factories, equipment health management typically involves deploying a monitoring network composed of numerous distributed speed measurement nodes to continuously collect minute vibration data from various parts of the equipment and calculate local speed information. The initial design goal of the system was to accurately time-align these speed measurement data from different locations and, based on this, build a holistic speed field model covering the entire factory's equipment cluster to detect early equipment failures in a timely manner. The synchronization processing module used to align the timestamps of data from different sources is optimized and constructed based on the stable and low-latency network connection characteristics between all distributed speed measurement nodes and the central clock source. It assumes that the path transmission time of data packets from any speed measurement node to the central processing unit is relatively constant and predictable, thereby achieving high-precision time synchronization through a simple compensation mechanism.

[0003] However, in actual industrial production environments, this ideal stable state is often disrupted by factory layout adjustments and network infrastructure upgrades. When a large influx of data occurs, the pressure on the internal packet forwarding queue increases significantly, leading to small but cumulative fluctuations in network transmission latency. These fluctuations do not exhibit a fixed pattern but rather display randomness that is difficult to predict precisely, making it difficult for traditional network monitoring methods to effectively capture and quantify these subtle, non-linear latency changes.

[0004] Existing synchronization algorithms typically rely on statistical estimates or periodic measurements of network latency, assuming these latency levels are relatively stable over short periods. When network latency fluctuations become rapid and unpredictable, these algorithms cannot update their internal latency models in a timely and accurate manner, leading to discrepancies between the calculated clock offset and the actual offset. This discrepancy causes a discrepancy between the recorded time of data collected by different speed measurement nodes and the actual time of physical events after the data is received by the system, making it impossible to accurately reconstruct the true order of data collection.

[0005] Due to the unpredictable discrepancies in timestamps between different data sources, the big data processing architecture cannot find a unified and reliable time reference when attempting to integrate data streams from all speed measurement nodes. This means that when building a speed field model, the system cannot ensure accurate correlation between speed measurement data from different spatial locations at the same physical moment. This inaccurate data time alignment caused by network latency fluctuations makes the spatial location interpolation calculation method relied upon for building the speed field model unable to accurately capture the true speed distribution and subtle changes of the equipment when processing input data with slight temporal misalignments. Ultimately, this makes it difficult for the system to build a high-precision, globally consistent speed field model in real time, severely impacting its ability to identify and warn of early equipment failure modes.

[0006] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this application provides an industrial big data-driven speed field measurement data analysis method and system to solve the problem that in industrial production environments, network latency fluctuations lead to inaccurate time alignment of distributed speed measurement node data, which in turn affects the accuracy of speed field model construction and the ability to identify early equipment faults.

[0008] In a first aspect, this application discloses an industrial big data-driven method for velocity field measurement data analysis, including: Receive data packets from distributed speed measurement nodes and determine a time confidence interval for each data packet; When constructing the velocity field model at the target time, the time confidence interval of the data packets is used to determine whether data packets from different distributed velocity measurement nodes belong to the same physical event at the same time. If the data packets are determined to be physical events occurring at the same time, then their weights in the velocity field model are assigned based on the length of the data packet's time confidence interval.

[0009] This technical solution effectively addresses complex network latency fluctuations in industrial environments. By introducing a time confidence interval to quantify the time uncertainty of data packets, and based on this, determining the physical event attribution and weighting of data packets, it overcomes the limitations of traditional methods in time alignment and improves the accuracy and reliability of velocity field model construction.

[0010] Furthermore, the step of determining the time confidence interval for each data packet includes: The following factors are considered to determine the time confidence interval: historical transmission delay characteristics of the data packet source node, real-time network load information of the data packet transmission path, and consistency of time information within the data packet.

[0011] This technical solution allows for a comprehensive consideration of various factors affecting the time reliability of data packets, making the determination of the time confidence interval more accurate and robust, thus providing a more reliable basis for subsequent physical event judgment and weight allocation.

[0012] Based on this, the steps for evaluating the historical transmission delay characteristics of the data packet source node include: Maintain a transmission delay sliding window for each distributed speed measurement node, and record the transmission delay of historical data packets within the transmission delay sliding window; The average transmission delay is calculated based on a sliding window of transmission delay. If the average transmission delay exceeds a preset delay deviation value, the confidence level of the data stamp carried in the data packet is reduced and the time confidence interval of the data packet is expanded.

[0013] In some preferred embodiments, the step of evaluating the real-time network load status of the packet transmission path includes: Monitor the operating parameters of key network switching devices along the data packet transmission path; The system determines whether network congestion exists based on operating parameters and adjusts the length of the time confidence interval based on the degree of network congestion.

[0014] Furthermore, the steps for evaluating the consistency of time information within data packets include: Parse the data packet to retrieve at least two levels of timestamps; Calculate the difference between timestamps of adjacent levels. If the difference exceeds the preset normal range, it is determined that there is inconsistency in the time information within the data packet, and the time confidence interval of the data packet is expanded.

[0015] As a technological improvement, based on the time confidence interval of data packets, the steps to determine whether data packets from different distributed speed measurement nodes belong to the same physical event at the same time include: Set the build time window; Data packets whose time confidence intervals overlap with the preset construction time window are filtered out to form a candidate data packet set; For any two data packets from different distributed speed measurement nodes in the candidate data packet set, if the time confidence intervals of any two data packets meet the overlap condition and the overlap degree reaches the preset overlap degree threshold, they are judged to belong to the same physical event at the same time.

[0016] Preferably, the overlap condition is that the time confidence intervals of any two data packets intersect; The overlap degree is obtained by calculating the length of the overlapping portion of the time confidence intervals of any two data packets, and then calculating the ratio of the length of the overlapping portion to the length of the longest time confidence interval among the two data packets.

[0017] To improve the scheme, if the data packets are determined to belong to the same physical event at the same time, the steps for assigning weights to the data packets in the velocity field model based on the length of the time confidence interval of the data packets include: Based on the length of the time confidence interval of the data packet, the weights of data packets belonging to the same physical event at the same time are assigned using an inverse proportional relationship or a Gaussian function.

[0018] As a further improvement, when the length of the time confidence interval of the data packet exceeds a preset length threshold, or when the overlap between the time confidence intervals of the data packet and the data packets in the candidate data packet set does not reach a preset overlap threshold, the method further includes: The data packets are marked as having questionable time reliability, and their display priority in the velocity field model is reduced. In the velocity field interpolation calculation in the velocity field model, data packets are assigned weight values ​​lower than the preset normal weights; Trigger an alarm to prompt operations and maintenance personnel to check the distributed speed test node corresponding to the data packet.

[0019] Secondly, this application also discloses an industrial big data-driven velocity field measurement data analysis system, which includes: The data receiving and processing module is used to receive data packets from distributed speed measurement nodes and determine a time confidence interval for each data packet; The physical event judgment module is used to determine whether data packets from different distributed velocity measurement nodes belong to the same physical event when constructing the velocity field model at the target time, based on the time confidence interval of the data packets. The weight allocation module is used to assign weights to data packets in the velocity field model based on the length of the time confidence interval of the data packets if they are determined to be physical events at the same time.

[0020] This application provides a system for implementing the above method through this technical solution. Through modular design, it can efficiently receive and process data packets, determine the attribution of physical events and assign weights, providing reliable hardware and software support for speed field measurement data analysis driven by industrial big data.

[0021] In summary, this application provides an industrial big data-driven speed field measurement data analysis method and system. The method receives data packets from distributed speed measurement nodes and determines a time confidence interval for each packet, effectively quantifying the time uncertainty caused by network latency fluctuations during transmission. When constructing a speed field model for a target time, this method determines whether data packets from different distributed speed measurement nodes belong to the same physical event based on the time confidence interval of the data packets. This solves the problem of traditional methods failing to accurately align timestamps from different sources when facing nonlinear and cumulative latency fluctuations. If the data packets are determined to belong to the same physical event, their weights in the speed field model are assigned based on the length of their time confidence intervals, allowing data packets with higher time reliability to play a greater role in model construction. Through the above technical solution, this application overcomes the shortcomings of existing technologies where inaccurate data time alignment due to network latency fluctuations leads to ambiguous and inaccurate speed field models in the time dimension. It can construct high-precision, globally consistent speed field models in real time, significantly improving the ability to identify and warn of early equipment failure modes. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating an industrial big data-driven velocity field measurement data analysis method provided in an embodiment of this application.

[0023] Figure 2 This is a schematic diagram of the structure of an industrial big data-driven velocity field measurement data analysis system provided in an embodiment of this application.

[0024] Labeling Explanation: 210, Data Reception and Processing Module; 220, Physical Event Judgment Module; 230, Weight Allocation Module. Detailed Implementation

[0025] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0026] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0027] Traditional industrial equipment health management systems, when integrating data collected by distributed speed measurement nodes, suffer from fluctuating data transmission delays due to the complexity and dynamic changes of the network environment. This makes it difficult to accurately align the timestamps between different speed measurement data. This temporal misalignment severely impacts the accuracy of the velocity field model, thereby weakening the ability to identify and warn of early equipment failures. If this problem is not addressed, when the system attempts to depict the overall motion state of the equipment at a given instant, it is essentially mixing local information from different time points, resulting in a vague and inaccurate velocity field model in the temporal dimension. This data temporal alignment inaccuracy caused by network latency fluctuations makes the spatial interpolation calculation methods relied upon for constructing the velocity field model unable to accurately capture the true velocity distribution and subtle changes of the equipment when processing input data with slight temporal misalignments. Ultimately, this makes it difficult for the system to construct a high-precision, globally consistent velocity field model in real time, severely impacting its ability to identify and warn of early equipment failures.

[0028] In this regard, firstly, referring to Figure 1 This application proposes a method for analyzing velocity field measurement data driven by industrial big data, including: S1. Receive data packets from distributed speed measurement nodes and determine a time confidence interval for each data packet; S2. When constructing the velocity field model at the target time, based on the time confidence interval of the data packets, determine whether the data packets from different distributed velocity measurement nodes belong to the same physical event at the same time. S3. If it is determined that the data packets belong to the same physical event at the same time, then assign the weights of the data packets in the velocity field model based on the length of the time confidence interval of the data packets.

[0029] Distributed velocity measurement nodes refer to sensors or equipment units deployed in industrial sites to collect equipment vibration data and calculate local velocity information. These nodes are typically distributed across different locations within the factory, transmitting the collected data packets to a central processing system via a network. A data packet is a data unit carrying velocity measurement data and related time information. A time confidence interval is a time range representing the interval within which the actual occurrence time of the physical event recorded by the data packet may fall; it reflects the reliability and uncertainty of the data packet's timestamp. A velocity field model is a mathematical model used to describe the overall velocity distribution of a group of equipment at a specific target time, typically constructed through interpolation calculations.

[0030] The industrial big data-driven velocity field measurement data analysis method of this application effectively solves the problem of inaccurate data time alignment in complex industrial network environments by introducing the concept of time confidence interval and judging whether data packets belong to the same physical event at the same time based on this, and allocating weights according to the length of the time confidence interval.

[0031] Specifically, this application first receives data packets from distributed speed measurement nodes and determines a time confidence interval for each data packet. This time confidence interval does not simply rely on the timestamp inherent in the data packet, but comprehensively considers various factors that may cause time deviations, thereby more accurately reflecting the actual occurrence time range of the physical events carried by the data packet. For example, by evaluating the historical transmission delay characteristics of the data packet's source node, the real-time network load of the transmission path, and the consistency of the time information within the data packet, the length and position of the time confidence interval can be dynamically adjusted to adapt to the nonlinear, cumulative delay fluctuations in industrial networks.

[0032] Subsequently, when constructing the velocity field model at the target time, this application determines whether data packets from different distributed velocity measurement nodes truly belong to the same physical event at that time based on the time confidence intervals of these data packets. Traditional methods often rely solely on the comparison of a single timestamp, which is easily affected by network latency fluctuations and can lead to misjudgments. This application, by comparing time confidence intervals, can more robustly identify physical events that truly overlap in time, avoiding the confusion of data from different times. For example, by setting a construction time window and filtering out data packets whose time confidence intervals overlap with that window, a candidate data packet set is formed. For any two data packets from different distributed velocity measurement nodes in the candidate data packet set, they are only determined to belong to the same physical event at that time if their time confidence intervals meet the overlap condition and the overlap degree reaches a preset threshold. This mechanism significantly improves the accuracy of data alignment.

[0033] Finally, if the data packets are determined to belong to the same physical event at the same time, this application assigns them corresponding weights in the velocity field model based on the length of their time confidence intervals. A shorter time confidence interval indicates higher temporal reliability, and thus a greater contribution to the velocity field model construction, resulting in a higher weight. Conversely, a longer time confidence interval indicates lower temporal reliability, leading to a correspondingly lower weight. This mechanism of dynamically assigning weights based on time reliability allows the velocity field model to prioritize data with higher temporal accuracy during construction, effectively reducing errors introduced by inaccurate time alignment and improving the overall accuracy and reliability of the velocity field model.

[0034] Specifically, this application proposes a method for comprehensively evaluating multiple factors when determining the time confidence interval for each data packet.

[0035] The steps for determining the time confidence interval for each data packet include: evaluating the following factors to determine the time confidence interval: historical transmission delay characteristics of the data packet source node, real-time network load information of the data packet transmission path, and consistency of time information within the data packet.

[0036] The historical transmission delay characteristics of the data packet source node refer to the network latency experienced by the distributed speed test node when sending data packets over a past period. This information reflects the inherent transmission performance and stability of a specific node, and may include statistical data such as average transmission delay and latency jitter range. Real-time network load information of the data packet transmission path refers to the current busy level of network links and devices (e.g., routers, switches) traversed by the data packet during its transmission from the distributed speed test node to the receiving end. Higher network load may result in greater data packet transmission delay and jitter; this may include metrics such as network bandwidth utilization and queue depth. The consistency of time information within the data packet refers to whether the time difference between multiple levels of timestamps carried by the data packet (e.g., collection timestamp, sending timestamp, application layer timestamp, etc.) matches expectations. Significant deviations between these timestamps may indicate anomalies in the data packet's generation or transmission.

[0037] The proposed solution, by comprehensively evaluating the three key factors mentioned above, can more fully and accurately reflect the actual transmission time of data packets and the reliability of their timestamps. Traditional methods may rely solely on the timestamps inherent in the data packets, neglecting the uncertainties of network transmission and the characteristics of the nodes themselves. By introducing historical transmission delay characteristics, systematic delay deviations of specific nodes can be corrected; by considering real-time network load information, time uncertainties caused by network congestion can be dynamically adjusted; and by checking the consistency of internal time information within the data packets, potential errors or tampering within the data packets themselves can be identified. Therefore, the determined time confidence interval can more realistically reflect the time range of the physical events represented by the data packets.

[0038] The aforementioned technical solution allows for higher accuracy and reliability in determining the data packet time confidence interval. This helps in the subsequent construction of the velocity field model to more accurately determine whether data packets from different distributed velocity measurement nodes belong to the same physical event at the same time, thereby improving the overall accuracy and robustness of velocity field measurement data analysis and effectively avoiding data misjudgment caused by errors from a single time information source or network fluctuations.

[0039] In some embodiments described above, this application proposes a method to determine the time confidence interval of a data packet by comprehensively evaluating the historical transmission delay characteristics of the data packet source node, the real-time network load information of the data packet transmission path, and the consistency of time information within the data packet. However, in real-world industrial environments, the transmission delay characteristics of distributed speed measurement nodes may not be constant but rather dynamically change due to various factors. If only static or simple historical evaluation methods are used, it may not be able to reflect the real-time fluctuations in node transmission delay in a timely and accurate manner, resulting in an insufficiently precise time confidence interval and affecting the accuracy of subsequent physical event judgments.

[0040] In this regard, this application further proposes steps for evaluating the historical transmission delay characteristics of the source node of the aforementioned data packets, including: Maintain a transmission delay sliding window for each distributed speed measurement node, and record the transmission delay of historical data packets within the transmission delay sliding window; The average transmission delay is calculated based on a sliding window of transmission delay. If the average transmission delay exceeds a preset delay deviation value, the confidence level of the data stamp carried in the data packet is reduced and the time confidence interval of the data packet is expanded.

[0041] Specifically, maintaining a transmission delay sliding window for each distributed speed test node refers to allocating a finite-sized storage structure to each node. This structure stores the transmission delay data of data packets sent by the node in chronological order. For example, the sliding window can be a first-in, first-out (FIFO) queue, where the oldest data is removed as new transmission delay data enters. Recording the transmission delay of historical data packets within the transmission delay sliding window aims to provide a dynamically updated dataset reflecting the node's recent transmission performance. Calculating the average transmission delay based on the transmission delay sliding window can be understood as taking the arithmetic mean of the transmission delay values ​​recorded within the sliding window to obtain the node's average transmission performance metric in the recent period. The preset delay deviation value is a configurable threshold used to determine whether the current node's average transmission delay significantly deviates from its normal or expected transmission behavior. For example, this deviation value can be set based on historical statistical data or system operating experience. In practical applications, if the average transmission delay exceeds the preset delay deviation value, the confidence level of the data stamp embedded in the data packet is reduced, and the time confidence interval of the data packet is expanded. Lowering the confidence level of a data stamp means giving less weight to the time information carried by the data packet itself, because the transmission delay of its source node is abnormal. Expanding the time confidence interval, on the other hand, increases the length of the time interval to encompass the actual occurrence time of the data packet with a larger range of uncertainty, thereby improving the robustness of time determination when transmission delays are abnormal.

[0042] Through the above technical solution, this application can dynamically adapt to changes in the transmission delay of distributed velocity measurement nodes, avoiding the problem of inaccurate time confidence intervals caused by fluctuations in node transmission performance. This solution significantly enhances the system's ability to judge the time reliability of data packets by real-time evaluation of the historical transmission delay characteristics of nodes and adjusting the time confidence interval according to the degree of deviation. Therefore, when constructing a velocity field model, it can more accurately determine whether data packets from different distributed velocity measurement nodes belong to the same physical event at the same time, thereby improving the overall accuracy and robustness of velocity field measurement data analysis.

[0043] In some preferred embodiments, it is assumed that the average transmission delay of a distributed speed measurement node during normal operation is 50 milliseconds, and the system sets a preset delay deviation value of 20 milliseconds. The node's transmission delay sliding window is configured to record the transmission delay of the most recent 100 data packets. If, within a certain time period, due to network congestion or a decrease in the node's own processing capacity, the transmission delay of the data packets sent by the node begins to increase, and the average transmission delay calculated within the sliding window rises to 80 milliseconds. At this time, the average transmission delay of 80 milliseconds exceeds the preset delay deviation value of 50 milliseconds plus 20 milliseconds (i.e., 70 milliseconds). The system will immediately determine that the node's transmission delay is abnormal and take measures for the data packets subsequently sent by the node: First, reduce the confidence of the data stamps embedded in these data packets, for example, adjust their weight from 1.0 to 0.7; second, expand the time confidence interval of these data packets from the original ±10 milliseconds to ±25 milliseconds. In this way, even if the node's transmission delay is temporarily unstable, the system can still encompass the actual occurrence time by expanding the time confidence interval, thereby reducing time synchronization errors caused by abnormal transmission delays and ensuring the accuracy of the speed field model construction.

[0044] In some of the embodiments described above in this application, a method is proposed to comprehensively evaluate real-time network load information of the data packet transmission path to determine the time confidence interval. However, in actual industrial environments, network load conditions are complex and variable. If the impact of network load on data transmission latency is not accurately and dynamically evaluated, the determination of the time confidence interval may be inaccurate, thereby affecting the accuracy of subsequent physical event judgments.

[0045] In response, this application further proposes steps for assessing the real-time network load status of the aforementioned data packet transmission path, including: Monitor the operating parameters of key network switching devices along the data packet transmission path; The system determines whether network congestion exists based on operating parameters and adjusts the length of the time confidence interval based on the degree of network congestion.

[0046] Specifically, monitoring the operating parameters of key network switching devices along the data packet transmission path refers to real-time monitoring of critical network devices such as core switches and routers that data packets pass through during transmission from the distributed rate measurement nodes to the data analysis system. These operating parameters may include, but are not limited to, port traffic, CPU utilization, memory usage, packet loss rate, error packet rate, and queue depth. These parameters can be collected periodically using SNMP (Simple Network Management Protocol) or other network monitoring protocols.

[0047] Determining network congestion based on operational parameters can be understood as analyzing the collected operational parameters to identify whether network links are saturated or overloaded. For example, network congestion can be identified when the traffic on a port consistently exceeds a preset percentage of its bandwidth capacity (e.g., 80%), or when the device's CPU utilization or memory utilization remains high for an extended period, or when packet loss rate, error packet rate, or queue depth increases significantly. A series of thresholds and rules can be set, and a comprehensive judgment can be made through expert systems or machine learning models.

[0048] In practical applications, adjusting the length of the time confidence interval based on network congestion levels refers to dynamically expanding or shrinking the time confidence interval for data packets based on the assessed severity of network congestion. For example, when network congestion is low, the time confidence interval can remain short; when network congestion is moderate, the time confidence interval can be appropriately expanded; and when network congestion is severe, the time confidence interval needs to be significantly expanded to reflect the increased uncertainty in data packet transmission delay. This adjustment can be linear, piecewise, or based on a preset functional relationship. Its purpose is to more accurately reflect the delay fluctuations of data packets caused by changes in network conditions during transmission, thereby improving the reliability of the time confidence interval.

[0049] Through the above technical solution, this application overcomes the limitations of traditional methods in assessing network load, which may involve static or coarse evaluations. By real-time monitoring and analysis of the operating parameters of key network devices, network congestion can be dynamically and precisely perceived, and the time confidence interval length of data packets can be flexibly adjusted accordingly. This significantly improves the accuracy and robustness of the time confidence interval determination, making the subsequent judgment on whether data packets from different distributed velocity measurement nodes belong to the same physical event at the same time more reliable when constructing velocity field models. This, in turn, enhances the overall accuracy and adaptability of the entire industrial big data-driven velocity field measurement data analysis method.

[0050] Specifically, in the aforementioned industrial big data-driven velocity field measurement data analysis method, the steps for evaluating the consistency of time information within data packets include: Parse the data packet to retrieve at least two levels of timestamps; Calculate the difference between timestamps of adjacent levels. If the difference exceeds the preset normal range, it is determined that there is inconsistency in the time information within the data packet, and the time confidence interval of the data packet is expanded.

[0051] The at least two-level timestamps carried within the data packet can be understood as time stamps applied by different systems or devices at different stages of the packet's generation, transmission, or processing. For example, one-level timestamp could be the timestamp when the data packet is generated at a distributed speed measurement node, while the other-level timestamp could be the timestamp when the data packet arrives at a network switching device, or the timestamp when the data packet is processed at the application layer. These timestamps aim to reflect the time attributes of the data packet from different dimensions.

[0052] Furthermore, calculating the difference between timestamps at adjacent levels aims to detect any anomalies in the time records of data packets at different processing stages. For example, if the time difference between a data packet's generation and its transmission to an intermediate node is too large or too small, it may indicate that the data packet experienced abnormal delays or out-of-order delivery during transmission, or that the timestamp itself is incorrect. The preset normal range can be empirically set based on typical data transmission delays and network jitter in actual industrial scenarios, or obtained through historical data analysis.

[0053] If the difference exceeds the preset normal range, an inconsistency in the time information within the data packet is determined. This inconsistency may stem from various causes, such as sensor clock drift, network attacks, data packet tampering, or abnormal data packet processing due to severe network congestion. Once this inconsistency is detected, the time confidence interval of the data packet is expanded. Expanding the time confidence interval means having greater uncertainty about the precise time point of the data packet, thereby reducing its contribution weight to the precise time point in subsequent velocity field model construction, to avoid introducing erroneous time information.

[0054] Through the above technical solution, this application can effectively identify inconsistencies in time information within data packets, such as timestamp errors caused by sensor clock drift, data packet tampering, or abnormal internal processing delays. This internal verification mechanism enhances the accuracy and robustness of time confidence interval determination, avoiding errors that may arise from relying solely on external factors for evaluation. Therefore, the time confidence interval for each data packet can be determined more reliably, thereby improving the accuracy of physical event judgments in subsequent velocity field model construction and ultimately enhancing the reliability of the entire velocity field measurement data analysis method.

[0055] Specifically, the step of determining whether data packets from different distributed speed measurement nodes belong to the same physical event at the same time based on the time confidence interval of data packets can be further implemented in the following way.

[0056] Specifically, the steps for determining whether data packets from different distributed speed measurement nodes belong to the same physical event at the same time, based on the time confidence interval of the data packets, include: Set the build time window; Data packets whose time confidence intervals overlap with the preset construction time window are filtered out to form a candidate data packet set; For any two data packets from different distributed speed measurement nodes in the candidate data packet set, if the time confidence intervals of any two data packets meet the overlap condition and the overlap degree reaches the preset overlap degree threshold, they are judged to belong to the same physical event at the same time.

[0057] Setting a construction time window refers to the system pre-defining a time interval to focus on physical events within a specific time range. This time interval defines the time range corresponding to the velocity field model currently being analyzed. The construction time window can be a fixed length, such as 50 milliseconds or 100 milliseconds, or it can be dynamically adjusted according to the actual application scenario. Its purpose is to narrow the data processing scope and improve processing efficiency and relevance.

[0058] Furthermore, filtering data packets whose time confidence intervals overlap with the preset construction time window to form a candidate data packet set involves comparing the time confidence intervals of all received data packets with the currently set construction time window. Only when a data packet's time confidence interval intersects with the construction time window is the data packet considered relevant to the physical event at the current target time and included in a temporary candidate data packet set. This step aims to exclude data packets irrelevant to the current analysis target, ensuring the accuracy and efficiency of subsequent judgments.

[0059] Furthermore, for any two data packets from different distributed rate measurement nodes in the candidate data packet set, if the time confidence intervals of any two data packets meet the overlap condition and the overlap degree reaches a preset overlap threshold, they are determined to belong to the same physical event at the same time. The overlap condition here refers to the intersection of the time confidence intervals of the two data packets on the time axis. The overlap degree is an indicator that measures the degree of this intersection. For example, the overlap degree can be defined as the ratio of the length of the intersection of two time confidence intervals to the length of the longer interval. When this ratio reaches or exceeds the preset overlap threshold, it indicates that the two data packets are highly likely to have recorded a physical event occurring at the same time, and are therefore determined to belong to the same physical event at the same time. The preset overlap threshold can be set according to the actual application's requirements for time synchronization accuracy, such as 0.5, 0.7, or higher, to balance the sensitivity and accuracy of the judgment.

[0060] Specifically, in the process of determining whether data packets from different distributed speed measurement nodes belong to the same physical event at the same time, the overlap conditions and the calculation method of overlap can be further clarified.

[0061] The aforementioned overlap condition is defined as the intersection of the time confidence intervals of any two data packets. This means that when the time confidence intervals of two data packets share a common time period on the time axis, they are considered to satisfy the overlap condition.

[0062] Furthermore, the overlap ratio is calculated by dividing the length of the overlap between the time confidence intervals of any two data packets by the length of the longest time confidence interval between those two data packets. For example, if the time confidence interval of data packet A is [tA_start, tA_end] and the time confidence interval of data packet B is [tB_start, tB_end], then the length of the overlap can be calculated as max(0, min(tA_end, tB_end) - max(tA_start, tB_start)). The length of the longest time confidence interval can be calculated as max(tA_end - tA_start, tB_end - tB_start). Thus, the overlap ratio is determined as the ratio of the length of the overlap to the length of the longest time confidence interval.

[0063] Through the above technical solution, this application can more accurately identify data packets from different distributed velocity measurement nodes that actually correspond to the same physical event. By quantifying the overlap, it can avoid misjudgments that may arise from simply judging interval intersections, improving the accuracy of data synchronization in velocity field model construction, and thus enhancing the overall accuracy and reliability of velocity field measurement. This precise event judgment mechanism is crucial for velocity field analysis driven by industrial big data, and can effectively support high-precision industrial process monitoring and optimization.

[0064] In some embodiments described above, this application proposes assigning weights to data packets in the velocity field model based on the length of the time confidence interval. However, in practical applications, how to accurately and effectively map the length of the time confidence interval to weights to ensure the accuracy and robustness of the velocity field model is a problem that requires further refinement. If the weight allocation mechanism is not reasonable enough, it may lead to inaccurate assessment of data packet reliability, thereby affecting the quality of the velocity field model construction.

[0065] In this regard, this application further proposes that if the data packets are determined to be physical events at the same time, the steps of assigning their weights in the velocity field model based on the length of the time confidence interval of the data packets include: assigning the weights of data packets belonging to physical events at the same time using an inverse proportional relationship or a Gaussian function based on the length of the time confidence interval of the data packets.

[0066] Specifically, using an inverse proportionality relationship to allocate weights means that the shorter the time confidence interval of a data packet, the higher its time reliability, and the greater its weight should be in the velocity field model. This inverse proportionality relationship intuitively links time reliability with weight, ensuring that data packets with higher time accuracy receive greater influence. Specifically, the inverse proportionality relationship can be understood as the weight being directly proportional to the reciprocal of the time confidence interval length, or directly proportional to a negative power of the time confidence interval length.

[0067] Furthermore, using a Gaussian function to allocate weights means using the length of the time confidence interval as the input variable of the Gaussian function, and determining the weight of the data packet through the output value of the Gaussian function. The Gaussian function provides a smooth and continuous weight allocation mechanism, causing the weight to gradually decrease as the time confidence interval length increases. Moreover, the parameters of the Gaussian function (such as mean and standard deviation) can be adjusted according to actual needs to adapt to different time reliability distribution patterns. For example, the peak value of the Gaussian function can be set at the position where the time confidence interval length is zero (ideally). As the length increases, the weight decreases according to the Gaussian curve, thus more precisely reflecting the impact of time reliability on the weight.

[0068] The above technical solution enables refined management and allocation of data packet weights, allowing the velocity field model to make fuller use of time-reliable measurement data during its construction. This not only improves the accuracy and robustness of the velocity field model but also enhances its adaptability to data packets with varying time reliability. Specifically, employing an inverse proportional relationship or a Gaussian function effectively avoids excessive interference from time-reliable data packets due to improper weight allocation, thereby improving the reliability and effectiveness of the entire data analysis method.

[0069] In some preferred embodiments, specifically when using an inverse proportional relationship to allocate weights, the relationship between the weight W and the time confidence interval length L can be set as W = k / L, where k is a positive proportionality constant. For example, if the time confidence interval length of data packet A is 10ms and the time confidence interval length of data packet B is 5ms, then the weight of data packet B will be twice that of data packet A, which intuitively reflects the higher time reliability of data packet B.

[0070] As another specific implementation, when using a Gaussian function to allocate weights, the weight can be set as W = A * exp(-(L^2) / (2 * σ^2)), where L is the length of the time confidence interval, A is the maximum weight value, and σ is the standard deviation, used to control the rate at which the weight decays with the length. For example, when L approaches 0, the weight W approaches A; as L increases, the value of W decreases smoothly according to a Gaussian curve. By adjusting the values ​​of A and σ, the influence of data packets with different time reliability on the velocity field model can be flexibly controlled, thereby better adapting to the different requirements for data accuracy in industrial scenarios.

[0071] In some embodiments described above, a scheme is proposed to determine whether data packets belong to the same physical event at the same time based on their time confidence intervals. This is accomplished by setting a time window, filtering candidate data packet sets, and determining the overlap conditions and degree of overlap of the time confidence intervals. However, in practical applications, even if the time confidence intervals of data packets overlap, there may be cases where the data packets themselves have low time reliability (e.g., the time confidence interval is too long) or the overlap with other data packets is insufficient to support reliable evidence that they are the same physical event. If these situations are not handled specially, it may introduce uncertainty or erroneous information into the velocity field model, affecting the accuracy and reliability of the model.

[0072] In response, this application further proposes a processing mechanism to perform special processing on a data packet when the length of the time confidence interval of the data packet exceeds a preset length threshold, or when the overlap between the time confidence intervals of the data packet and the data packets in the candidate data packet set does not reach a preset overlap threshold, so as to improve the robustness of the velocity field model and the accuracy of data analysis.

[0073] Specifically, when the length of the time confidence interval of a data packet exceeds a preset length threshold, or when the overlap between the time confidence intervals of the data packet and the data packets in the candidate data packet set does not reach a preset overlap threshold, the method further includes: The data packets are marked as having questionable time reliability, and their display priority in the velocity field model is reduced. In the velocity field interpolation calculation in the velocity field model, data packets are assigned weight values ​​lower than the preset normal weights; Trigger an alarm to prompt operations and maintenance personnel to check the distributed speed test node corresponding to the data packet.

[0074] The phrase "the length of the time confidence interval of a data packet exceeds a preset length threshold" means that when the time confidence interval of a data packet is too long, it indicates that the timestamp of the data packet may have significant uncertainty or delay, and its time information is not accurate enough, thus its reliability is low. The preset length threshold can be set according to the time accuracy requirements of the actual application scenario, for example, it can be set to 100 milliseconds, 200 milliseconds, etc. The phrase "the overlap between the time confidence intervals of a data packet and the data packets in the candidate data packet set does not reach a preset overlap threshold" means that although the data packet may overlap with the construction time window and be included in the candidate set, its overlap with other related data packets is insufficient to confirm that they originate from the same physical event. This may mean that the data packet has a weak correlation with the physical event at the current target time. The preset overlap threshold can be adjusted according to the strictness of event synchronization, for example, it can be set to 0.5, 0.7, etc.

[0075] When any of the above conditions are met, the data packet will be marked as a data packet with questionable time reliability. This marking aims to clearly distinguish data whose time information is uncertain or has a weak correlation with other events, so as to facilitate subsequent processing and analysis. At the same time, "reducing the display priority of data packets in the velocity field model" means that in visualization or user interface, these questionable data packets may be displayed with different colors, transparency, or smaller icons, or even not displayed by default, to avoid misleading users' understanding of the velocity field model.

[0076] Furthermore, in the velocity field interpolation calculation within the velocity field model, assigning a weight value lower than the preset normal weight to data packets means that when constructing the velocity field model, such as during spatial interpolation (e.g., Kriging interpolation, inverse distance weighted interpolation, etc.) or time series analysis, the impact of these questionable data packets on the final model results will be weakened. By assigning lower weights, the negative impact of their uncertainty on the overall model accuracy can be effectively reduced, ensuring that the model primarily relies on highly reliable data. The preset normal weight refers to the weight assigned when the data packet is considered reliable, typically 1.

[0077] Furthermore, "triggering alarms to prompt operations and maintenance personnel to check the distributed speed measurement nodes corresponding to the data packets" is a proactive operations and maintenance mechanism. When the system detects data packets with questionable time reliability continuously or frequently, it will automatically generate alarm information to notify operations and maintenance personnel. This helps operations and maintenance personnel to locate the source of the problem in a timely manner, such as checking for sensor failures, network transmission anomalies, or clock synchronization issues in the distributed speed measurement nodes, thereby fundamentally solving data quality problems and improving the stability and reliability of the entire system.

[0078] The above technical solutions significantly improve the velocity field model's resilience to abnormal or low-quality data, resulting in more stable and reliable model output. Furthermore, by reducing the display priority and assigning lower weights to questionable data, user perception of the velocity field model can be optimized, avoiding misjudgments caused by uncertain data. Maintenance personnel can also intervene promptly through alarms, ensuring the healthy operation of the data acquisition chain, thereby improving the monitoring and control level of the entire industrial production process.

[0079] Secondly, referring to Figure 2 This application proposes an industrial big data-driven velocity field measurement data analysis system, the system comprising: The data receiving and processing module 210 is used to receive data packets from distributed speed measurement nodes and determine a time confidence interval for each data packet; The physical event judgment module 220 is used to determine whether data packets from different distributed velocity measurement nodes belong to the same physical event when constructing the velocity field model at the target time, based on the time confidence interval of the data packets. The weight allocation module 230 is used to assign weights to data packets in the velocity field model based on the length of the time confidence interval of the data packets if they are determined to be physical events at the same time.

[0080] Distributed velocity measurement nodes refer to sensor or equipment units deployed in industrial sites to collect equipment vibration data and calculate local velocity information. These nodes are typically distributed across different locations within a factory, transmitting the collected data packets to a central processing system via a network. A data packet is a data unit carrying velocity measurement data and related time information. A time confidence interval is a time range representing the interval within which the actual occurrence time of the physical event recorded by the data packet may fall; it reflects the reliability and uncertainty of the data packet's timestamp. A velocity field model is a mathematical model used to describe the overall velocity distribution of a group of equipment at a specific target time, typically constructed through interpolation calculations.

[0081] The data receiving and processing module 210 is used to receive data packets from the distributed speed measurement nodes and determine a time confidence interval for each data packet. The specific methods for receiving data packets and determining the time confidence interval for each data packet have been described in the above embodiments and will not be repeated here. It should be emphasized that the data receiving and processing module 210 can be configured as a hardware module. For example, it can be a dedicated network interface card (NIC) integrating data preprocessing logic, or a programmable logic controller (PLC) or embedded system whose internal firmware implements the reception of data packets and the preliminary calculation of the time confidence interval. Alternatively, the data receiving and processing module 210 can also be configured as a software module, for example, as a service process running on a server or industrial PC, receiving data by calling the network API provided by the operating system and executing the time confidence interval calculation logic.

[0082] The physical event judgment module 220 is used to determine, based on the time confidence interval of data packets, whether data packets from different distributed velocity measurement nodes belong to the same physical event when constructing the velocity field model at the target time. The specific method for determining whether data packets from different distributed velocity measurement nodes belong to the same physical event has been described in the above embodiments and will not be repeated here. It should be emphasized that the physical event judgment module 220 can be configured as an independent computing unit, for example, an independent service on a high-performance computing server, specifically responsible for receiving data with time confidence intervals from the data receiving and processing module 210 and executing complex interval overlap judgment algorithms. Alternatively, this module can be integrated into the backend of the data receiving and processing module 210 as part of its data processing pipeline. In some embodiments, the physical event judgment module 220 can use rule-based logic for judgment, for example, by determining it through preset thresholds and overlap calculations.

[0083] The weight allocation module 230 is used to assign weights to data packets in the velocity field model based on the length of their time confidence intervals if the packets are determined to belong to the same physical event at the same time. The specific method for assigning weights to data packets in the velocity field model has been described in the above embodiments and will not be repeated here. It is important to emphasize that the weight allocation module 230 can be configured as a component tightly coupled with the physical event judgment module 220, performing weight allocation immediately after the judgment is completed. For example, it can be a software library or function called by the physical event judgment module 220 to calculate and append weight information. Alternatively, the weight allocation module 230 can also be an independent microservice that receives a set of data packets determined to be the same physical event, calculates weights based on their time confidence interval lengths, and then outputs the weighted velocity data to the velocity field model construction module (not shown). In some embodiments, the weight allocation module 230 can use a simple lookup table or predefined function to quickly allocate weights.

[0084] The industrial big data-driven velocity field measurement data analysis system of this application effectively solves the problem of inaccurate data time alignment in complex industrial network environments by introducing the concept of time confidence interval, judging whether data packets belong to the same physical event at the same time, and allocating weights according to the length of the time confidence interval.

[0085] Specifically, the system of this application first receives data packets from distributed speed measurement nodes through the data receiving and processing module 210, and determines a time confidence interval for each data packet. This time confidence interval does not simply rely on the timestamp inherent in the data packet, but comprehensively considers various factors that may cause time deviations, thereby more accurately reflecting the actual occurrence time range of the physical events carried by the data packet. For example, by evaluating the historical transmission delay characteristics of the data packet source node, the real-time network load of the transmission path, and the consistency of the time information within the data packet, the length and position of the time confidence interval can be dynamically adjusted to adapt to the nonlinear and cumulative delay fluctuations in industrial networks.

[0086] Subsequently, when constructing the velocity field model at the target time, the system of this application uses the physical event judgment module 220 to determine whether data packets from different distributed velocity measurement nodes truly belong to the same physical event at the same time, based on the time confidence intervals of these data packets. Traditional systems often rely solely on the comparison of a single timestamp, which is easily affected by network latency fluctuations and can lead to misjudgments. This system, by comparing time confidence intervals, can more robustly identify physical events that truly overlap in time, avoiding the confusion of data from different times. For example, by setting a construction time window and filtering out data packets whose time confidence intervals overlap with that window, a candidate data packet set is formed. For any two data packets from different distributed velocity measurement nodes in the candidate data packet set, they are only judged to belong to the same physical event at the same time when their time confidence intervals meet the overlap condition and the overlap degree reaches a preset threshold. This mechanism significantly improves the accuracy of data alignment.

[0087] Finally, if the data packets are determined to belong to the same physical event at the same time, the system of this application will assign corresponding weights to them in the velocity field model based on the length of the time confidence interval of the data packets through the weight allocation module 230. The shorter the time confidence interval, the higher the time reliability of the data packets, and the greater their contribution to the construction of the velocity field model should be, thus being assigned a higher weight; conversely, the longer the time confidence interval, the lower the time reliability, and the weight will be reduced accordingly. This mechanism of dynamically allocating weights based on time reliability enables the velocity field model to prioritize the use of data with higher time accuracy during the construction process, thereby effectively reducing the errors introduced by inaccurate time alignment and improving the overall accuracy and reliability of the velocity field model.

[0088] Compared with existing technologies, the core innovation of this industrial big data-driven velocity field measurement data analysis system lies in its modular design, which organically combines the determination of time confidence intervals, the judgment of physical events, and the allocation of weights to cope with nonlinear and cumulative delay fluctuations in complex industrial networks. Traditional systems often fail when faced with nonlinear and cumulative delay fluctuations in complex industrial networks, as their synchronization mechanisms based on single timestamps or simple statistical estimations often fail, leading to a decrease in the accuracy of velocity field model construction. This system determines the time confidence interval by comprehensively evaluating multiple factors through the data receiving and processing module 210, and then performs more refined physical event judgments and weight allocations based on this interval through the physical event judgment module 220 and the weight allocation module 230. This effectively addresses the unpredictability of network latency and significantly improves the time alignment accuracy between different velocity measurement data and the accuracy of the velocity field model. This system design not only more accurately depicts the overall motion state of equipment at a given moment but also provides a more reliable data foundation for the identification and early warning of early equipment faults, thereby improving the equipment health management level of smart factories.

[0089] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for analyzing velocity field measurement data driven by industrial big data, characterized in that, include: Receive data packets from distributed speed measurement nodes and determine a time confidence interval for each data packet; When constructing the velocity field model at the target time, based on the time confidence interval of the data packet, it is determined whether data packets from different distributed velocity measurement nodes belong to the same physical event at the same time. If it is determined that the data packet belongs to the same physical event at the same time, then the data packet is assigned a weight in the velocity field model based on the length of the time confidence interval of the data packet.

2. The industrial big data-driven velocity field measurement data analysis method according to claim 1, characterized in that, The step of determining a time confidence interval for each data packet includes: The following factors are considered to determine the time confidence interval: historical transmission delay characteristics of the data packet source node, real-time network load information of the data packet transmission path, and consistency of time information within the data packet.

3. The industrial big data-driven velocity field measurement data analysis method according to claim 2, characterized in that, The steps for evaluating the historical transmission delay characteristics of the source node of the data packet include: Maintain a transmission delay sliding window for each of the distributed speed measurement nodes, and record the transmission delay of historical data packets within the transmission delay sliding window; The average transmission delay is calculated based on the transmission delay sliding window. If the average transmission delay exceeds a preset delay deviation value, the confidence level of the data stamp carried in the data packet is reduced and the time confidence interval of the data packet is expanded.

4. The industrial big data-driven velocity field measurement data analysis method according to claim 2, characterized in that, The steps for assessing the real-time network load status of the packet transmission path include: Monitor the operating parameters of key network switching devices along the data packet transmission path; The system determines whether network congestion exists based on the operating parameters and adjusts the length of the time confidence interval based on the degree of network congestion.

5. The industrial big data-driven velocity field measurement data analysis method according to claim 2, characterized in that, The steps for evaluating the consistency of time information within the data packet include: Parse at least two levels of timestamps carried in the data packet; Calculate the difference between adjacent timestamps of the aforementioned levels. If the difference exceeds a preset normal range, it is determined that there is inconsistency in the time information within the data packet, and the time confidence interval of the data packet is expanded.

6. The industrial big data-driven velocity field measurement data analysis method according to claim 1, characterized in that, The step of determining whether data packets from different distributed speed measurement nodes belong to the same physical event at the same time based on the time confidence interval of the data packets includes: Set the build time window; Data packets whose time confidence intervals overlap with the preset construction time window are selected to form a candidate data packet set; For any two data packets from different distributed speed measurement nodes in the candidate data packet set, if the time confidence intervals of the two data packets meet the overlap condition and the overlap degree reaches the preset overlap degree threshold, they are determined to belong to the same physical event at the same time.

7. The industrial big data-driven velocity field measurement data analysis method according to claim 6, characterized in that, The overlap condition is that the time confidence intervals of any two data packets intersect; The overlap degree is obtained by calculating the length of the overlapping portion of the time confidence intervals of any two data packets, and then calculating the ratio of the length of the overlapping portion to the length of the longest time confidence interval among the two data packets.

8. The industrial big data-driven velocity field measurement data analysis method according to claim 1, characterized in that, If it is determined that the data packet belongs to the same physical event at the same time, the step of assigning the weight of the data packet in the velocity field model based on the length of the time confidence interval of the data packet includes: Based on the length of the time confidence interval of the data packet, the weight of the data packet belonging to the physical event at the same moment is assigned using an inverse proportional relationship or a Gaussian function.

9. The industrial big data-driven velocity field measurement data analysis method according to claim 6, characterized in that, When the length of the time confidence interval of the data packet exceeds a preset length threshold, or when the overlap between the time confidence intervals of the data packet and the data packets in the candidate data packet set does not reach the preset overlap threshold, the method further includes: The data packet is marked as a data packet with questionable time reliability, and the display priority of the data packet in the velocity field model is reduced; In the velocity field interpolation calculation in the velocity field model, the data packet is assigned a weight value lower than the preset normal weight; An alarm is triggered to prompt maintenance personnel to check the distributed speed test node corresponding to the data packet.

10. An industrial big data-driven velocity field measurement data analysis system, characterized in that, The system includes: The data receiving and processing module is used to receive data packets from distributed speed measurement nodes and determine a time confidence interval for each data packet; The physical event judgment module is used to determine, based on the time confidence interval of the data packet, whether data packets from different distributed velocity measurement nodes belong to the same physical event when constructing the velocity field model at the target time. The weight allocation module is used to assign a weight to the data packet in the velocity field model based on the length of the time confidence interval of the data packet if it is determined that the data packet belongs to the physical event at the same time.

Images