An industrial fan operating condition data processing method and system

Abstract

The application relates to the technical field of industrial fans, and particularly discloses an industrial fan operation condition data processing method and system, which comprises the following steps: obtaining data packets from an industrial fan and transmitted through at least two heterogeneous networks; obtaining the receiving time stamp and the transmission link type of the data packets, and generating transmission state information of the data packets; distinguishing the data packets into one of directly mergable samples, samples needing relocation, and samples needing isolation and not participating in merging; performing correlation matching on the samples needing relocation and historical data packets that have completed merging, so as to determine the target physical time of the samples needing relocation; merging the directly mergable samples and the samples needing relocation after relocation according to the corresponding physical time, to form a data frame; and outputting the data frame, so as to perform real-time control or energy efficiency evaluation on the industrial fan. The application improves the accuracy and timeliness of real-time control instructions, and avoids misjudgment and control delay caused by data mismatch.

Description

Technical Field

[0001] This application relates to the field of industrial fan technology, and in particular to a method and system for processing operating condition data of industrial fans. Background Technology

[0002] In the field of industrial fan operation data processing, to meet the needs of real-time control and energy efficiency assessment, it is usually necessary to comprehensively process multi-dimensional data such as vibration and pressure. In actual industrial sites, due to limitations in wiring conditions, sensors in different physical locations often use a hybrid wired and wireless networking method for data transmission. For example, vibration data from the fan bearing housing is uploaded at fixed intervals via a CAN bus, while pressure data from the outlet duct is connected to the network via an industrial Wi-Fi node. Each data source is timestamped upon completion of acquisition and then enters the transmission queue. After receiving the data, the central processing unit aligns it according to the timestamps and combines it into a data frame at the same time point for subsequent analysis.

[0003] However, when industrial fans undergo rapid frequency conversion, such as a sudden transition from normal operation to the boundary of slight surge, the instantaneous changes in vibration and pressure need to be precisely corresponded within sub-second timeframes. At this time, if nearby high-power equipment starts or stops, generating broadband electromagnetic interference, the wireless network signal quality will drastically degrade, causing random jitter in the pressure data transmission delay and even triggering a retransmission mechanism. During this process, the pressure data packets arrive at the central processing unit delayed due to retransmission, but they still retain the original local timestamp from when the data acquisition was completed, while the vibration data packets arrive quasi-periodically and normally.

[0004] Because the arrival times of the two data packets are misaligned, the central processing unit (CPU) forcibly concatenates the newly arrived pressure data with the vibration data already in the cache, according to a predetermined original local timestamp matching rule. This simple alignment method based on the original local timestamp ignores the impact of drastic fluctuations in transmission delay, resulting in a physical time mismatch of hundreds of milliseconds between the data groups pointed to by the same original local timestamp after aggregation. This timing mismatch severely distorts the multidimensional transient correlation characteristics required for surge warning, making it difficult for existing conventional methods to detect the physical time disconnect caused by data packet delays in a timely manner. Ultimately, this leads to delays in real-time control command issuance and deviations in energy efficiency assessment results. Therefore, in the case of variable frequency operation of industrial wind turbines, when interference in some links of the heterogeneous network causes drastic fluctuations in transmission delay, the fact that the two data packets are still aligned using the original local timestamp at the time of acquisition causes a severe physical time mismatch between the data groups pointed to by the same original local timestamp after aggregation, resulting in distortion of multidimensional transient correlation characteristics and affecting synchronization requirements. Summary of the Invention

[0005] This application proposes a method and system for processing industrial fan operating condition data, aiming to solve the technical problem that when some links in a heterogeneous network are interfered with and the transmission delay fluctuates drastically under the variable frequency operation of industrial fans, the two data streams still use the original local timestamps at the time of acquisition for alignment, resulting in a serious mismatch in physical time for the data groups pointed to by the same original local timestamp after aggregation, which leads to the distortion of multidimensional transient correlation characteristics and affects the synchronization requirements.

[0006] In a first aspect, this application provides a method for processing industrial fan operating condition data, comprising the following steps: Acquire data packets from an industrial fan transmitted through at least two heterogeneous networks; wherein the data packets include original local timestamps and payload information including operating condition data such as vibration and pressure; Upon receiving the data packet, the receiving timestamp and transmission link type of the data packet are obtained, and the transmission status information of the data packet is generated. Based on the transmission status information, the availability of the original local timestamp in the data packet is evaluated to classify the data packet into one of the following: samples that can be directly merged, samples that need to be relocated, and samples that do not participate in merging and need to be isolated. Based on the payload information, the sample to be relocated is associated and matched with the historical data packets that have been merged to determine the target physical time of the sample to be relocated, thereby obtaining the sample to be relocated after relocation. The directly mergeable samples and the repositioned samples that need to be repositioned are merged according to the corresponding physical time to form data frames corresponding to different physical time segments; wherein, the output level of the data frame is distinguished into normal alignment frames or correction frames according to the type of merging. The data frame is output to enable real-time control or energy efficiency assessment of the industrial fan.

[0007] As some embodiments of this application, the transmission status information includes: time interval variance, retransmission flag, and degree of deviation from historical benchmark; The time interval variance is obtained by calculating the arrival time intervals of multiple consecutive data packets corresponding to the data packet data source within a historical time window, forming an interval sequence, and then calculating the variance of the interval sequence. The retransmission flag is extracted from the underlying network protocol stack of the central controller associated with the industrial fan. The degree of deviation is obtained by comparing the arrival time interval of the current data packets with the corresponding historical transmission records.

[0008] As some embodiments of this application, the step of evaluating the availability of the original local timestamp in the data packet based on the transmission status information, in order to classify the data packet into one of the following: samples that can be directly merged, samples that need to be relocated, and samples that do not participate in merging and need to be isolated: When the time interval variance is not higher than the preset stable variance threshold, and the deviation is not higher than the preset stable deviation threshold and the retransmission flag indicates normal, the availability status of the original local timestamp in the data packet is determined to be stable, and the data packet is classified into samples that can be directly merged. When the time interval variance is higher than the stable variance threshold, and the deviation is not higher than the preset stable deviation threshold and the retransmission flag indicates normal, the availability status of the original local timestamp in the data packet is determined to be fluctuating, and the data packet is classified as a sample that needs to be relocated. When the deviation exceeds the stable deviation threshold, or the retransmission flag indicates an anomaly, the availability status of the original local timestamp in the data packet is determined to be abnormal, and the data packet is classified as an isolated sample that does not participate in the merging.

[0009] As some embodiments of this application, the step of associating and matching the sample to be relocated with historically merged data packets based on the payload information to determine the target physical time of the sample to be relocated, thereby obtaining the relocated sample to be relocated, includes: Determine a candidate time window for the data packets in the sample that needs to be relocated; the start time of the candidate time window is the physical time of the last successfully merged data packet on the data source corresponding to the data packet, and the end time is the current system time; Extract the pressure change trend from the payload information of the data packets in the sample that needs to be relocated; Within the candidate time window, the vibration variation trend is extracted from the payload information of one or more data packets with established time sequences. The pressure change trend and the vibration change trend are used to perform sliding matching calculation within the candidate time window to find the physical time point with the highest matching degree, and this physical time point is used as the target physical time of the corresponding data packet in the sample to be relocated. All data packets that have regained the target physical time are used to form the relocation sample.

[0010] As some embodiments of this application, in the process of extracting the changing trend of the vibration: If the sampling frequency of the vibration is higher than the sampling frequency of the pressure, then the candidate time window is divided into multiple consecutive sub-time intervals according to the sampling frequency of the pressure. Calculate the statistical characteristic values ​​of the raw data of the vibration in each of the sub-time intervals; The statistical feature values ​​are used to generate a feature sequence aligned with the sampling frequency of the pressure; The characteristic sequence is used as the trend of the vibration.

[0011] As some embodiments of this application, before the step of performing a sliding match calculation on the pressure change trend and the vibration change trend within the candidate time window to find the physical time point with the highest matching degree, the method further includes: Based on a preset delay constant that characterizes the inherent physical transmission delay between the pressure and the vibration, phase compensation is performed on the vibration variation trend to obtain the vibration variation trend after phase compensation, which is then used for subsequent sliding matching calculations.

[0012] As some embodiments of this application, the step of performing a sliding match calculation between the pressure change trend and the vibration change trend within the candidate time window to find the physical time point with the highest matching degree includes: Calculate the cross-correlation coefficients between the pressure variation trend and the vibration variation trend at different time offsets within the candidate time window; Determine the time offset that maximizes the cross-correlation coefficient; Based on the start time of the candidate time window and the time offset, the physical time point with the highest matching degree is calculated.

[0013] As some embodiments of this application, the step of performing a sliding match calculation between the pressure change trend and the vibration change trend within the candidate time window to find the physical time point with the highest matching degree further includes: If the maximum value of the cross-correlation coefficient is not higher than the preset matching rejection threshold, the association matching is determined to be unsuccessful, and a matching failure mark is generated for the corresponding data packet and removed from the samples that need to be relocated; wherein, the data packet carrying the matching failure mark is used for network quality analysis or fault tracing of industrial fans.

[0014] As some embodiments of this application, the step of outputting the data frame for real-time control or energy efficiency assessment of the industrial fan includes: The normal alignment frame and the correction frame are output to control the industrial fan in real time; wherein, the normal alignment frame is fully weighted, and the weight of the correction frame in the control decision is dynamically adjusted based on the maximum value of the cross-correlation coefficient calculated in the association matching step of generating the correction frame. The information of the normal alignment frame, the correction frame, and the sample to be isolated is output to perform an energy efficiency assessment of the industrial fan.

[0015] Secondly, this application also provides an industrial fan operating condition data processing system, including: A data receiving module is used to acquire data packets transmitted from an industrial fan through at least two heterogeneous networks; wherein the data packets include an original local timestamp and payload information including operating condition data such as vibration and pressure; The evaluation module is used to obtain the reception timestamp and transmission link type of the data packet when it is received, and to generate the transmission status information of the data packet. The classification module is used to evaluate the availability of the original local timestamp in the data packet based on the transmission status information, so as to classify the data packet into one of the following: samples that can be directly merged, samples that need to be relocated, and samples that do not participate in merging and need to be isolated. The time determination module is used to associate and match the sample to be relocated with the historical data packets that have been merged based on the payload information, so as to determine the target physical time of the sample to be relocated, thereby obtaining the sample to be relocated after relocation. The hierarchical merging module is used to merge the directly merging samples and the repositioned samples that need to be repositioned according to the corresponding physical time, forming data frames corresponding to different physical time segments; wherein, the output level of the data frame is distinguished into normal alignment frames or correction frames according to the type of merging. The output module is used to output the data frame for real-time control or energy efficiency assessment of the industrial fan.

[0016] The technical solution according to the embodiments of this application has at least the following beneficial effects: This application avoids the erroneous splicing of delayed data caused by wireless link interference with earlier data from other dimensions, thereby preventing distortion of multidimensional transient correlation characteristics and ensuring that the data upon which key control functions such as surge warning rely maintain physical time synchronization. As a result, this application improves the accuracy and timeliness of real-time control command issuance, avoids misjudgments and control delays caused by data mismatch, and also provides an accurate and reliable data foundation for energy efficiency assessment, enhancing the robustness and intelligence of industrial fan control systems in complex electromagnetic environments.

[0017] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0018] The accompanying drawings are used to provide a further understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0019] Figure 1 This is a flowchart illustrating a method for processing industrial fan operating condition data, provided in an embodiment of this application.

[0020] Figure 2 This is a flowchart illustrating step S130 in an embodiment of this application.

[0021] Figure 3 This application embodiment provides a flowchart of step S140.

[0022] Figure 4 This is a schematic diagram of the architecture of an industrial fan operation condition data processing system provided in an embodiment of this application. Detailed Implementation

[0023] 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.

[0024] Traditional methods for processing industrial wind turbine operating data often simply align data from different networks based on the original local timestamps. However, when network links are disrupted, causing significant fluctuations in transmission latency, this simplistic alignment method ignores the impact of these latency fluctuations. This results in a mismatch of hundreds of milliseconds in physical time between data groups pointing to the same original local timestamp after aggregation. This timing mismatch severely distorts the multidimensional transient correlation characteristics required for surge warning, making it difficult for existing methods to detect the physical time discrepancies caused by data packet delays. Ultimately, this leads to delays in real-time control command issuance and deviations in energy efficiency assessment results.

[0025] In this regard, such as Figure 1 As shown, this application discloses a method for processing industrial fan operating condition data, including the following steps: S110, acquire data packets from the industrial fan transmitted through at least two heterogeneous networks; wherein the data packets include the original local timestamp and payload information including operating condition data of vibration and pressure; S120: Upon receiving a data packet, obtain the data packet's reception timestamp and transmission link type, and generate the data packet's transmission status information. S130, based on the transmission status information, assess the availability of the original local timestamp in the data packet to classify the data packet into one of the following: samples that can be directly merged, samples that need to be relocated, and samples that do not participate in merging and need to be isolated. S140, Based on the payload information, the sample to be relocated is associated and matched with the historical data packets that have been merged to determine the target physical time of the sample to be relocated, thereby obtaining the sample to be relocated after relocation. S150, the samples that can be directly merged and the samples that need to be relocated after relocation are merged according to the corresponding physical time to form data frames corresponding to different physical time sections; wherein, the output level of the data frame is divided into normal alignment frame or correction frame according to the type of merging. S160 outputs data frames for real-time control or energy efficiency assessment of industrial fans.

[0026] To better understand the technical solution proposed in this application, some key terms involved will be explained first.

[0027] A data packet is the basic unit that carries operational data during network transmission. It contains the original local timestamp at the time of data acquisition and the actual operational data (i.e., "payload information").

[0028] Heterogeneous networks refer to network environments composed of two or more different types of networks (such as wired networks like CAN bus and Ethernet, and wireless networks like industrial Wi-Fi and 5G), which may have different transmission characteristics and latency performance.

[0029] Transmission status information is an indicator used to characterize the network status of data packets during transmission. It may include, for example, the arrival time interval of data packets and retransmission status.

[0030] A sample that can be directly merged refers to a data packet whose original local timestamp is determined to be reliable and can be directly used for data merging.

[0031] A sample that needs to be relocated refers to a data packet whose original local timestamp has been determined to be unreliable and whose physical time needs to be re-determined through subsequent association and matching steps.

[0032] The isolated samples that are not included in the merging process refer to data packets whose data quality or transmission status is abnormal and are not suitable for data merging.

[0033] The target physical time refers to the physical time that best reflects the actual acquisition time of a sample that needs to be relocated, determined through methods such as association matching.

[0034] A data frame is a collection of multi-dimensional operational data, which is formed by merging multiple data packets at a specific physical time point.

[0035] Normally aligned frames are data frames that are mainly composed of directly mergeable samples and have high temporal accuracy.

[0036] A correction frame is a data frame that contains the repositioned sample that needs to be repositioned, and its time accuracy has been corrected.

[0037] The implementation environment of this application is typically an industrial control system, which includes industrial fans, various sensors, heterogeneous network infrastructure, and a central processing unit or data processing unit.

[0038] The core of the industrial fan operation condition data processing method proposed in this application lies in the intelligent timestamp availability assessment and physical time relocation of data packets from heterogeneous networks.

[0039] The data packets carry critical operating condition data of the industrial fan, such as vibration and pressure. They are tagged with a raw local timestamp during acquisition and include payload information for this operating condition data. For example, a data acquisition agent can be configured and deployed near the industrial fan to acquire data from various sensors connected to the fan (such as vibration and pressure sensors). These sensors may connect to the agent via different network interfaces; for example, vibration sensors might connect via a CAN bus, and pressure sensors via industrial Wi-Fi. The agent encapsulates the raw data acquired from these sensors into data packets, records the local timestamp at the time of acquisition, and then sends the data packets out through the corresponding network interface.

[0040] The receive timestamp records the time when the data packet arrived at the processing system, while the transmission link type indicates which heterogeneous network (e.g., wired or wireless) the data packet was transmitted through. Transmission status information can be easily generated by recording the arrival order and intervals of data packets; for example, the arrival time of each data packet can be recorded, and the arrival time interval between consecutive data packets can be calculated.

[0041] Data packets are categorized into three types: directly mergeable samples, samples requiring relocation, and isolated samples that are not included in the merging process. This avoids blindly using potentially biased original local timestamps. For example, a simple rule can be set: if the arrival time intervals of multiple consecutively arriving data packets are very stable and no retransmissions are detected, their original local timestamps are considered usable, and they are marked as directly mergeable samples. If the arrival time intervals fluctuate significantly but there are no obvious anomalies, the original local timestamps are considered potentially biased and require relocation, and they are marked as samples requiring relocation. If serious transmission anomalies are detected, such as packet loss or multiple retransmissions, the data packet quality is considered unreliable, and it is marked as an isolated sample that is not included in the merging process.

[0042] The goal is to determine the target physical time of samples requiring relocation, thereby obtaining the relocated samples. This aims to compensate for the inadequacy of timestamps by leveraging the inherent correlations within the data content itself. For example, the inherent correlations between different physical quantities in industrial fan operating data can be utilized. When a pressure data packet is marked as a sample requiring relocation, the pressure change trend within its payload can be extracted. Simultaneously, the vibration change trend of vibration data packets with a time window close to that of the pressure data packet can be extracted from the merged historical data. Since pressure and vibration typically have a physical coupling relationship during fan operation, the true physical time of the pressure data packet can be inferred by comparing the similarity of these two change trends. For instance, a sliding window approach can be used to compare the pressure change trend with the vibration change trend, finding the moment with the highest matching degree; this moment is then determined as the target physical time of the pressure data packet.

[0043] The output layers of data frames are categorized into normal alignment frames or correction frames based on the type of data they participate in merging. This step integrates the processed data packets to form unified time-series data. For example, all data packets determined to be directly merging samples, as well as all samples requiring relocation after relocation processing, are sorted and combined according to their final determined physical time. If a data frame consists of directly merging samples, it is marked as a normal alignment frame, indicating high time accuracy. If a data frame contains samples requiring relocation after relocation, it is marked as a correction frame, indicating that its time information has been corrected.

[0044] Time-aligned and corrected data frames provide accurate and reliable operating condition information for upper-layer applications. For example, normally aligned and corrected frames can be directly used in the real-time control system of industrial fans to guide operations such as frequency conversion adjustment and fault warning. Simultaneously, these data frames can also be used in energy efficiency assessment systems to evaluate fan operating efficiency and optimize operating strategies by analyzing long-term series of operating condition data.

[0045] The industrial fan operation data processing method proposed in this application fundamentally changes the traditional approach of unconditionally trusting the original timestamp by introducing transmission status information to dynamically evaluate the availability of the original local timestamp of the data packet. When interference in some links of a heterogeneous network causes drastic fluctuations in transmission delay, the system first generates transmission status information based on information such as the received timestamp and transmission link type, and classifies the data packets into samples that can be directly merged, samples that need relocation, and samples that do not participate in merging and need to be isolated. For late data packets marked as samples that need relocation (such as pressure data delayed due to wireless interference), the system does not forcibly splice them according to their distorted original local timestamps. Instead, it analyzes their payload information (such as pressure change trends) and performs sliding matching calculations based on physical association with historical data packets that have been merged (such as vibration data with established time series), thereby accurately determining the true target physical time of the data packet and effectively correcting the deviation of the original timestamp. Finally, the directly mergeable samples and the relocated samples are merged according to their actual physical time to form normal alignment frames or correction frames with different confidence labels, which are used for subsequent real-time control or energy efficiency assessment.

[0046] In summary, this application effectively solves the physical time mismatch problem caused by drastic fluctuations in transmission delay in heterogeneous networks under variable frequency operation of industrial fans by employing a collaborative mechanism of transmission status assessment and payload association matching. This application avoids the erroneous splicing of delayed data caused by wireless link interference with earlier data from other dimensions, thereby preventing distortion of multi-dimensional transient correlation characteristics and ensuring that the data upon which key control functions such as surge warning rely maintain physical time synchronization. Therefore, this application improves the accuracy and timeliness of real-time control command issuance, avoids misjudgments and control delays caused by data mismatch, and also provides an accurate and reliable data foundation for energy efficiency assessment, enhancing the robustness and intelligence level of the industrial fan control system in complex electromagnetic environments.

[0047] It should be noted that, in the embodiments of this application, the transmission status information preferably includes time interval variance, retransmission flag, and the degree of deviation from historical benchmark.

[0048] The transmission status information includes: time interval variance, retransmission flag, and degree of deviation from historical benchmark; The time interval variance is obtained by calculating the arrival time intervals of multiple consecutive data packets corresponding to the data packet data source within a historical time window, forming an interval sequence, and then calculating the variance of the interval sequence. The retransmission flag is extracted from the underlying network protocol stack of the central controller associated with the industrial fan. The degree of deviation is obtained by comparing the arrival time interval of the current data packets with the corresponding historical transmission records.

[0049] Time interval variance can be understood as a metric measuring the stability of data packet arrival times. A smaller variance indicates more stable data packet arrival times and less network jitter. The retransmission flag indicates whether a data packet has been retransmitted. The presence of a retransmission flag usually signifies packet loss or congestion in the network, thus affecting the real-time performance and reliability of data transmission. Deviation quantifies the difference between the current data packet arrival time interval and historical normal transmission patterns. A large deviation may indicate a significant change in network conditions, such as increased network latency or bandwidth fluctuations.

[0050] This application's solution effectively identifies the impact of network jitter on timestamp stability by introducing time interval variance; it directly reflects transmission anomalies caused by network packet loss or retransmission by acquiring retransmission flags; and it promptly detects abnormal changes in network behavior by calculating the degree of deviation from historical benchmarks. This multi-dimensional transmission status information enables the system to more precisely judge the quality of data packets, thereby more accurately distinguishing data packets into samples that can be directly merged, samples requiring relocation, or samples that do not participate in merging and need to be isolated. This provides a more reliable and accurate data foundation for subsequent data merging and real-time control or energy efficiency assessment of industrial fans.

[0051] In the above-described embodiments of this application, such as Figure 2 As shown, the preferred step of evaluating the availability of the original local timestamp in the data packet based on the transmission status information to classify the data packet into one of the following: samples that can be directly merged, samples that need to be relocated, and samples that do not participate in the merging and need to be isolated: S131, when the time interval variance is not higher than the preset stable variance threshold, and the deviation is not higher than the preset stable deviation threshold and the retransmission flag indicates normal, the availability status of the original local timestamp in the data packet is determined to be stable, and the data packet is divided into samples that can be directly merged. S132, when the time interval variance is higher than the stable variance threshold, and the deviation is not higher than the preset stable deviation threshold and the retransmission flag indicates normal, the availability status of the original local timestamp in the data packet is determined to be fluctuating, and the data packet is classified into samples that need to be relocated. S133, when the deviation is higher than the stable deviation threshold, or the retransmission flag indicates an abnormality, the availability status of the original local timestamp in the data packet is determined to be abnormal, and the data packet is classified into isolated samples that do not participate in the merging.

[0052] The stability variance threshold and stability deviation threshold are pre-set parameters used to define the stable range of data packet transmission status. The stability variance threshold can be understood as the maximum acceptable value of the variance of the data packet arrival time interval. When the actually calculated time interval variance is lower than this threshold, it indicates that the data packet transmission interval is relatively stable. The stability deviation threshold refers to the maximum acceptable deviation of the data packet arrival time interval from historical transmission records. When the actual deviation is lower than this threshold, it indicates that the data packet transmission behavior is basically consistent with historical norms.

[0053] A normal retransmission flag indicates that the retransmission flag extracted from the underlying network protocol stack does not show any abnormalities such as retransmission or packet loss of the data packet; conversely, an abnormal retransmission flag indicates that the data packet encountered a problem during transmission, such as retransmission.

[0054] Samples that can be directly merged are those whose original local timestamps are considered highly reliable and can be directly used for data merging without additional time relocation processing. Samples requiring relocation are those whose original local timestamps, while exhibiting some fluctuation, still possess payload information with the potential for correlation and matching, requiring subsequent correlation and matching steps to determine their target physical time. Samples that should not participate in merging and require isolation are those whose transmission status is abnormal, whose original local timestamps have extremely low reliability, and are unsuitable for data merging; they should be isolated to avoid contaminating data frames, but their information can be used for network quality analysis or fault tracing.

[0055] This application's solution, by comprehensively considering time interval variance, deviation, and retransmission flags, can effectively identify data packets with different transmission qualities and perform intelligent classification accordingly. This not only avoids unnecessary relocation processing for high-quality data packets, improving data processing efficiency, but also enables timely detection and isolation of low-quality or abnormal data packets, preventing them from negatively impacting subsequent real-time control and energy efficiency assessment. Furthermore, by relocating data packets with fluctuating states, this application maximizes the use of effective data while ensuring data accuracy, significantly improving the robustness and reliability of industrial fan operating condition data processing.

[0056] The following is a specific example to illustrate this.

[0057] Suppose that an industrial fan transmits its operating condition data via a wireless network.

[0058] Scenario 1: The system continuously receives data packets from the wind turbine, with the variance of their arrival time intervals consistently remaining below 0.01 seconds (below the preset stable variance threshold of 0.05 seconds), and the deviation from the historical baseline is also extremely small (below the preset stable deviation threshold of 0.1 seconds), and the network protocol stack does not report any retransmissions. In this case, according to the above judgment logic, the availability status of the original local timestamps of these data packets is determined to be stable, and they are classified as samples that can be directly merged, directly entering the data merging process.

[0059] Scenario 2: During a certain period, due to slight network congestion, the variance of the arrival time interval of data packets increases to 0.08 seconds (above the stable variance threshold of 0.05 seconds), but the deviation from the historical baseline remains at 0.05 seconds (below the stable deviation threshold of 0.1 seconds), and the retransmission flag indicates normal operation. In this case, the availability status of the original local timestamp of these data packets is determined to be fluctuating and they are classified as samples requiring relocation. These samples will enter the subsequent association and matching step, where time calibration is performed using payload information.

[0060] Scenario 3: In another time period, due to severe wireless signal interference, the arrival time interval of data packets deviates from the historical baseline by as much as 0.2 seconds (higher than the stable deviation threshold of 0.1 seconds), while the network protocol stack reports a large number of retransmissions (the retransmission flag indicates anomalies). In this case, the availability status of the original local timestamps of these data packets is judged as abnormal and classified as isolated samples that do not participate in the merging process. The information of these isolated data packets can be used to analyze the causes of network failures, but will not be included in the data merging for real-time control or energy efficiency assessment to avoid introducing errors.

[0061] Through the aforementioned specific classification mechanism, this application can intelligently select the most suitable processing path based on the actual transmission quality of the data packets, thereby ensuring the accuracy and reliability of industrial fan operating condition data processing.

[0062] In some embodiments of this application, such as Figure 3 As shown, the step of associating and matching the sample to be relocated with historical data packets that have been merged based on the payload information to determine the target physical time of the sample to be relocated, thereby obtaining the relocated sample to be relocated, preferably includes: S141, determine the candidate time window for the data packets in the sample to be relocated; the start time of the candidate time window is the physical time of the last successfully merged data packet on the data source corresponding to the data packet, and the end time is the current system time; S142, extract the pressure change trend from the payload information of the data packets in the sample to be relocated; S143, within the candidate time window, extract the vibration variation trend from the payload information of one or more data packets with established timing sequences; S144, The pressure change trend and the vibration change trend are used to perform sliding matching calculation within the candidate time window to find the physical time point with the highest matching degree, and this physical time is used as the target physical time of the corresponding data packet in the sample to be relocated. S145, all data packets that have regained the target physical time are used to form the relocation sample after relocation.

[0063] Determining the candidate time window for data packets in the sample that need to be relocated means setting a possible time range for the data packets to be relocated. The aim is to limit the search range to a reasonable and physically meaningful interval, avoid invalid matching within an excessively large time range, and thus improve matching efficiency and accuracy.

[0064] Extracting the pressure change trend can be understood as identifying and quantifying the pattern or characteristics of pressure parameter changes over time from the operating condition data carried by the data packet. For example, a sequence of pressure values, the rate of pressure change, or the statistical characteristics of pressure over a period of time (such as mean, variance, peak value, etc.) can be extracted to form a trend sequence that can represent the dynamic characteristics of pressure.

[0065] Extracting the vibration trend specifically refers to extracting the pattern of vibration parameter changes over time from historical data that has been time-aligned and has reliable physical timestamps. Similar to pressure trends, vibration trends can also be represented by a sequence of vibration values, vibration frequency, vibration amplitude, or their statistical characteristics. Since vibration and pressure are often physically correlated in the operation of industrial fans, their trend changes are, to some extent, synchronous or have a predictable phase relationship.

[0066] The proposed solution utilizes the inherent correlation between different physical quantities for cross-validation and matching, enabling precise determination of the physical time of data packets even when the original timestamp is unreliable. This leads to more accurate subsequent data merging, resulting in data frames that more realistically reflect the operating status of industrial fans. Consequently, it provides a more reliable and accurate data foundation for real-time control and energy efficiency assessment, effectively enhancing the robustness of the entire data processing system and the effectiveness of decision-making.

[0067] The following is a specific example to illustrate this.

[0068] Suppose that during the operation of an industrial fan, its pressure sensor data packets are classified as "fluctuating" due to network fluctuations, causing the original local timestamp to be determined as "fluctuating," thus classifying them as samples requiring relocation. The system first determines a reasonable candidate time window based on the physical time of the last successfully merged data packet from the pressure sensor and the current system time. For example, if the physical time of the last successfully merged pressure data packet is 10:00:00 and the current system time is 10:00:10, the candidate time window might be set to [10:00:00, 10:00:10]. Next, the pressure change trend is extracted from the payload information of the sample requiring relocation; for example, this could be a sequence containing the most recent pressure measurements. Simultaneously, within the aforementioned candidate time window, the vibration change trend is extracted from the merged historical data; for example, this could also be a sequence containing vibration measurements within the corresponding time period. Subsequently, a sliding match calculation is performed between the pressure and vibration change trends, for example, using a cross-correlation algorithm, to find the physical time point with the highest correlation between the two. Assuming the cross-correlation coefficient between the two reaches its maximum at 10:00:05, then 10:00:05 is determined as the target physical time of the sample requiring relocation. Ultimately, all data packets whose target physical time has been redefined in this way constitute the relocated sample requiring relocation, which can be accurately merged into the corresponding time segment.

[0069] It should be noted that in the process of extracting the trend of the vibration: If the sampling frequency of the vibration is higher than the sampling frequency of the pressure, then the candidate time window is divided into multiple consecutive sub-time intervals according to the sampling frequency of the pressure. Calculate the statistical characteristic values ​​of the raw data of the vibration in each of the sub-time intervals; The statistical feature values ​​are used to generate a feature sequence aligned with the sampling frequency of the pressure; The characteristic sequence is used as the trend of the vibration.

[0070] In the acquisition of operating data for industrial fans, different types of sensors may have different data acquisition rates. For example, vibration sensors typically sample at a higher frequency to capture subtle mechanical changes, while pressure sensors may sample at a relatively lower frequency. When it is necessary to correlate and match data from these two different sampling frequencies, directly using the raw data may lead to data volume mismatch or difficulties in time alignment.

[0071] Dividing the candidate time window into multiple consecutive sub-time intervals based on the pressure sampling frequency means that, in order to keep the vibration data and pressure data consistent in the time dimension, the entire candidate time window is subdivided into smaller, continuous time periods corresponding to the pressure sampling period, using pressure data with a lower sampling frequency as a benchmark. For example, if the pressure is sampled once per second, the candidate time window will be divided into sub-time intervals in seconds.

[0072] Within each defined sub-time interval, statistical analysis is performed on all raw vibration data contained within that interval to extract numerical values ​​that represent the vibration characteristics of that interval. These statistical features may include, but are not limited to, the average, maximum, minimum, variance, standard deviation, root mean square (RMS), or peak value. Selecting appropriate statistical features can effectively summarize the vibration behavior within that time period.

[0073] The vibration statistical characteristic values ​​calculated for each sub-time interval are arranged in chronological order to form a new sequence. The sampling rate of this sequence is the same as that of the pressure data, thus achieving alignment of data with different sampling rates.

[0074] The vibration feature sequence obtained after the above processing, which is aligned with the sampling frequency of the pressure, can accurately reflect the change trend of vibration within the corresponding time period and can be directly used for matching calculation with the pressure change trend.

[0075] The proposed solution significantly improves the accuracy and robustness of data association matching by downsampling high-frequency vibration data and extracting statistical features to align it with the sampling rate of low-frequency pressure data. This avoids matching failures or time relocation errors caused by sampling rate mismatch, thereby ensuring that the target physical time of samples requiring relocation can be determined more accurately. This, in turn, enhances the reliability of the entire data processing method and the accuracy of real-time control or energy efficiency assessment of industrial fans.

[0076] The following is a specific example to illustrate this.

[0077] Assume that the vibration sensor installed on the industrial fan collects data at a sampling frequency of 1000Hz, while the pressure sensor collects data at a sampling frequency of 100Hz. When time relocation of the sample to be relocated is required, a candidate time window is first determined. When extracting the vibration trend, since the vibration sampling frequency (1000Hz) is higher than the pressure sampling frequency (100Hz), the system divides the candidate time window into multiple consecutive 10-millisecond (1 / 100Hz) sub-time intervals based on the pressure sampling frequency (100Hz). Within each 10-millisecond sub-time interval, the statistical characteristic values ​​of the 10 raw vibration data points (1000Hz * 0.01s = 10 points) are calculated, such as the average or root mean square value. Arranging these statistical characteristic values ​​in chronological order forms a vibration characteristic sequence aligned with the 100Hz sampling rate. This characteristic sequence is then used as the vibration trend and subjected to sliding match calculation with the pressure trend to accurately determine the target physical time of the sample to be relocated.

[0078] In a specific embodiment of this application, before the step of performing a sliding match calculation on the pressure change trend and the vibration change trend within a candidate time window to find the physical time point with the highest matching degree, it is preferable to further include: Based on a preset delay constant that characterizes the inherent physical transmission delay between the pressure and the vibration, phase compensation is performed on the vibration variation trend to obtain the vibration variation trend after phase compensation, which is then used for subsequent sliding matching calculations.

[0079] The delay constant is a pre-set time value that quantifies the physical time interval between a pressure change and the corresponding vibration response in an industrial fan system. For example, when the pressure inside the fan changes, this change requires a certain amount of time to be transmitted through the mechanical structure and cause measurable vibration. The purpose is to accurately capture the intrinsic temporal relationship between the two physical quantities of pressure and vibration. The delay constant can be obtained in the following ways: For example, before the industrial fan leaves the factory, a standard dynamic test is performed. A step speed command is applied at the rated speed, and the response data of the outlet duct pressure sensor and the bearing housing acceleration sensor are recorded simultaneously. By analyzing the time difference between the start of pressure increase and the start of vibration amplitude increase, the average value is obtained after multiple tests to obtain the pressure-vibration delay constant of this model of fan. This constant is recorded in the factory dynamic test report for direct reading during system initialization. Alternatively, during the stable operation phase after the fan is put into operation at the site, the system automatically collects high-reliability pressure and vibration data marked as "directly mergeable samples" over a continuous period. The time offset that causes the cross-correlation coefficient between the two to reach its maximum value is calculated through sliding cross-correlation analysis, and the average value of the calculation results for multiple stable periods is taken as the real-time delay constant, which can be periodically updated as the fan ages or undergoes structural changes. For example, for a typical centrifugal industrial fan, at rated speed, the inherent physical delay constant from the change in outlet duct pressure to the vibration response of the bearing housing is usually between 20 milliseconds and 80 milliseconds.

[0080] Based on the aforementioned delay constant, the time series of vibration data is subjected to a comprehensive time shift or offset. For example, if the delay constant indicates that vibration lags behind pressure, the time axis of the vibration data can be shifted forward by an amount equal to the delay constant, aligning the vibration data with the pressure data in time. In this way, the time misalignment caused by physical transmission delay can be eliminated or reduced, allowing the pressure change trend and vibration change trend to more accurately reflect the performance of the same physical event on different sensors in subsequent sliding matching calculations.

[0081] The obtained vibration trend after phase compensation refers to the vibration data sequence after time shifting. This new sequence has been calibrated in time with the pressure data sequence, thus providing a more synchronized input for subsequent sliding matching calculations.

[0082] Through the above technical solution, this application can improve the accuracy of determining the target physical time of samples requiring relocation. By considering and compensating for the inherent physical transmission delay between pressure and vibration, the sliding matching calculation can more accurately identify the true physical time point of the data packet, avoiding matching errors caused by time misalignment. This makes the time information of the samples requiring relocation more reliable after relocation, thereby improving the quality of data merging and ensuring that the data frame can more accurately reflect the operating conditions of the industrial fan at a specific physical time section. Ultimately, this helps to improve the accuracy of real-time control of industrial fans and the reliability of energy efficiency assessment, providing a more solid data foundation for the intelligent operation and maintenance of industrial fans.

[0083] The following is a specific example to illustrate this.

[0084] Suppose that during the operation of an industrial fan, there is a fixed physical transmission delay, for example, 50 milliseconds, between the pressure change detected by the internal pressure sensor and the vibration response detected by the vibration sensor mounted on the fan casing. This 50-millisecond delay constant is predetermined based on the structural characteristics of this type of fan and the sensor installation location, through prior experimental testing and physical modeling.

[0085] When the system receives a data packet containing the trends of pressure and vibration changes, before performing sliding matching calculations, it first performs phase compensation on the vibration trend based on a preset 50-millisecond delay constant. Specifically, if the vibration data's timestamp is T_v, then during matching, the vibration data is shifted forward 50 milliseconds on the time axis, logically aligning it with the pressure data whose timestamp is T_v - 50 milliseconds. In other words, the vibration data is treated as occurring 50 milliseconds earlier than its original recording time.

[0086] After phase compensation, the vibration trend, which originally lagged behind the pressure trend, is "advanced" by 50 milliseconds, thus achieving a closer temporal alignment with the pressure trend. Subsequently, the system performs a sliding match calculation between the phase-compensated vibration trend and the pressure trend within a candidate time window, for example, using a cross-correlation algorithm. Because the inherent physical delay is eliminated, the cross-correlation function is more likely to reach its maximum value at the correct physical time point, thus more accurately determining the target physical time for the sample requiring relocation. For example, if the original vibration data reaches its peak at time T_v, while the pressure data reaches its peak at time T_p, and T_v - T_p ≈ 50ms, then after 50ms of phase compensation, the vibration data will align with the pressure data at time T_v' = T_v - 50ms at time T_p, allowing the matching algorithm to accurately identify T_p as the optimal matching time point.

[0087] In a further embodiment of this application, the step of performing a sliding matching calculation between the pressure change trend and the vibration change trend within a candidate time window to find the physical time point with the highest matching degree preferably includes: Calculate the cross-correlation coefficients between the pressure variation trend and the vibration variation trend at different time offsets within the candidate time window; Determine the time offset that maximizes the cross-correlation coefficient; Based on the start time of the candidate time window and the time offset, the physical time point with the highest matching degree is calculated.

[0088] The cross-correlation coefficient is a statistic that measures the similarity between two signals at different time offsets. In practical applications, it can be obtained by mathematically analyzing the trend sequences of pressure and vibration. Specifically, for discrete signals, the cross-correlation coefficient can be calculated by summing the products of one signal and a shifted copy of the other. By systematically changing the time offset, a series of cross-correlation values ​​can be generated, reflecting the similarity between the two trends under different time alignments.

[0089] When the cross-correlation coefficient reaches its maximum value, the corresponding offset indicates the strongest linear correlation or the optimal time alignment point between the pressure change trend and the vibration change trend. For example, this can be achieved by iterating through all the calculated cross-correlation values ​​and finding the maximum value and its corresponding time shift.

[0090] Once the time offset that maximizes the cross-correlation coefficient is determined, the physical time point with the highest matching degree can be calculated based on the start time of the candidate time window and this time offset. This physical time point is then determined as the target physical time of the corresponding data packet in the sample that needs to be relocated.

[0091] This application's solution, through the introduction of cross-correlation coefficients for sliding matching calculations, effectively captures the inherent physical correlation and time delay between pressure and vibration trends in industrial fan operating data. By systematically calculating cross-correlation coefficients at different time offsets and identifying the offset corresponding to the maximum cross-correlation coefficient, the optimal time alignment point between two signals can be accurately determined. This method utilizes the statistical properties of the signals themselves, overcoming potential fluctuations or anomalies in the original local timestamps, thus providing an accurate and reliable physical timestamp for samples requiring relocation.

[0092] It should be noted that the step of performing a sliding match calculation between the pressure change trend and the vibration change trend within the candidate time window to find the physical time point with the highest matching degree also preferably includes: If the maximum value of the cross-correlation coefficient is not higher than the preset matching rejection threshold, the association matching is determined to be unsuccessful, and a matching failure mark is generated for the corresponding data packet and removed from the samples that need to be relocated; wherein, the data packet carrying the matching failure mark is used for network quality analysis or fault tracing of industrial fans.

[0093] A match rejection threshold is a pre-set value used to measure the minimum acceptable level of cross-correlation coefficients. When the maximum calculated cross-correlation coefficient is lower than this threshold, it indicates that the similarity between the two data trends is insufficient to support reliable correlation matching. The match rejection threshold can be set in the following ways: For example, during system development, a set of historical data samples with known correct time alignment can be selected, and correlation matching calculations can be performed on the data packets marked as needing relocation. The distribution of cross-correlation coefficients for successfully matched samples can be statistically analyzed, and the correlation coefficient value corresponding to the lower percentile (e.g., 5% or 10%) of the distribution can be used as the initial threshold. Alternatively, different levels of transmission delay interference can be simulated in a laboratory environment to generate a large amount of test data with real physical time labels. By iterating through the matching accuracy and false negative rate under different thresholds, an ROC curve can be plotted, and the threshold point that optimizes overall performance can be selected. For example, for a typical industrial fan pressure-vibration correlation matching scenario, the match rejection threshold is usually set between 0.5 and 0.7. Below this value, the correlation between the two trends is considered insufficient to support reliable time positioning, and the system will determine that the match has failed and proceed to the exception handling process. This threshold can be periodically calibrated and dynamically adjusted based on the statistical characteristics of the field data.

[0094] A match failure flag is an identifier attached to a data packet to indicate that the packet failed to find a reliable physical time during time relocation. This flag can be a specific field in the packet's metadata, such as a boolean value or an error code. Once a packet is marked as a match failure, it is removed from the relocation sample and no longer participates in the subsequent merging process, thus preventing the introduction of inaccurate data into the data frame.

[0095] Network quality analysis refers to evaluating the performance and stability of industrial fan data transmission networks using data packets carrying match failure flags. For example, if a large number of data packets are flagged for match failures, it may indicate severe data transmission delays, jitter, or packet loss, leading to data trend distortion and affecting the success rate of association matching. By analyzing the distribution and frequency of these failure flags, network problems can be diagnosed.

[0096] Fault tracing refers to using data packets carrying match failure flags to help identify potential faults in industrial fans or their sensor systems. For example, if data packets from a particular sensor frequently fail to match, it may indicate a fault in the sensor itself, causing abnormal output data that cannot be effectively correlated with data from other sensors. This helps to quickly locate the source of the fault and perform maintenance or replacement.

[0097] This application's solution introduces a match rejection threshold to verify the validity of the maximum value of the cross-correlation coefficient calculated by sliding matching. When the maximum value of the cross-correlation coefficient is lower than the preset match rejection threshold, it indicates that although an "optimal" match point was found within the candidate time window, the confidence level of the match is low and insufficient to ensure the accuracy of time relocation. In this case, by determining that the association match has failed and generating a match failure flag for the corresponding data packet, these low-confidence data packets can be effectively prevented from participating in subsequent physical time merging, thereby avoiding the introduction of inaccurate time information into the final data frame. Furthermore, using these rejected data packets with match failure flags for network quality analysis or fault tracing can provide valuable diagnostic information for system operation and maintenance, helping to identify potential problems in the data transmission link or the sensor itself.

[0098] The following is a specific example to illustrate this.

[0099] Suppose that during time relocation of data packets in a sample requiring relocation, a sequence of cross-correlation coefficients between pressure and vibration trends at different time offsets is obtained through sliding matching calculation. Further, the maximum cross-correlation coefficient in this sequence is determined to be 0.45. The system then compares this maximum value with a preset match rejection threshold (e.g., set to 0.6). Since 0.45 is not higher than 0.6, the system determines that the association match has failed. Subsequently, a match failure flag is generated for this data packet, and it is removed from the sample requiring relocation, no longer participating in subsequent physical time merging. This data packet with the match failure flag is then sent to a dedicated analysis module. This analysis module can statistically analyze the number and source of data packets that failed to match within a specific time period; for example, it may find that data packets from a specific sensor frequently experience match failures. Based on this analysis, maintenance personnel can infer that the sensor may be faulty or that its data transmission link is unstable, thus enabling timely inspection and maintenance to avoid control decision errors or energy efficiency assessment deviations due to data quality issues.

[0100] In a preferred embodiment of this application, the step of outputting the data frame for real-time control or energy efficiency assessment of the industrial fan includes: The normal alignment frame and the correction frame are output to control the industrial fan in real time; wherein, the normal alignment frame is fully weighted, and the weight of the correction frame in the control decision is dynamically adjusted based on the maximum value of the cross-correlation coefficient calculated in the association matching step of generating the correction frame. The information of the normal alignment frame, the correction frame, and the sample to be isolated is output to perform an energy efficiency assessment of the industrial fan.

[0101] A normal alignment frame refers to a data frame formed from samples whose original local timestamps are determined to be stable and directly merging. Its time accuracy is high, therefore it is given full weight in control decisions; its data is completely trusted and used for control calculations. A correction frame refers to a data frame formed after the samples requiring relocation have undergone correlation matching and the target physical time has been determined. Its timestamps have been corrected. To reflect the reliability of the correction frame, this application dynamically adjusts the weight of the correction frame in control decisions based on the maximum value of the cross-correlation coefficient calculated in the correlation matching step of generating the correction frame. The maximum value of the cross-correlation coefficient can be understood as a quantitative indicator of the degree of matching between the pressure change trend and the vibration change trend. The higher the value, the better the matching degree, and the more reliable the corrected timestamp. Therefore, by using this maximum value as a weighting factor, the control system can intelligently adjust according to the reliability of the correction frame data, avoiding control deviations introduced by low-reliability data.

[0102] When evaluating the energy efficiency of industrial fans, in addition to outputting normal alignment frames and correction frames, information on samples requiring isolation is also output. Samples requiring isolation refer to data packets whose original local timestamps are deemed abnormal and not included in the merging process. Although these samples are not directly used for real-time control, the anomaly information they contain is of significant value for energy efficiency evaluation and fault diagnosis. For example, by analyzing the transmission status information of the samples requiring isolation (such as deviation degree and retransmission flags), network transmission quality can be assessed, potential network faults or sensor anomalies can be identified, thus providing more comprehensive data support for the energy efficiency optimization and maintenance of industrial fans.

[0103] This application's solution introduces a dynamic weight adjustment mechanism for correction frames into real-time control, enabling the control system to intelligently utilize corrected data based on data reliability, thus avoiding a decrease in control accuracy due to data uncertainty. Simultaneously, by incorporating information from samples requiring isolation into energy efficiency assessment, the assessment process becomes more comprehensive, focusing not only on normal operating data but also uncovering potential fault information and network performance issues from abnormal data. It is precisely this refined data utilization and comprehensive information output that makes the real-time control of industrial fans more stable and reliable, and the energy efficiency assessment more in-depth and effective.

[0104] Through the above technical solutions, in terms of real-time control, the weight of the calibration frame in the control decision can be dynamically adjusted according to the matching reliability, thereby improving the robustness and accuracy of the control system and avoiding potential risks introduced by uncertain data. In terms of energy efficiency assessment, by outputting the information of the isolated samples, the assessment is no longer limited to normal aligned data, but can comprehensively analyze the operating status of industrial fans, network quality, and potential faults, providing a data foundation for more accurate energy efficiency optimization and fault tracing.

[0105] The following is a specific example to illustrate this.

[0106] Suppose that during the operation of an industrial fan, the data packets from its pressure sensor experience fluctuations in their original local timestamps due to network volatility, and are classified as samples requiring relocation. The system will perform correlation matching between the payload information of this data packet (such as pressure change trends) and established time-series vibration data. During the sliding matching calculation, if the maximum cross-correlation coefficient calculated is 0.92, indicating a high degree of matching between pressure and vibration trends, this correction frame will be assigned a weight of 0.92 in the real-time control decision. This means that the data from this correction frame will participate in control algorithms such as fan speed or valve opening adjustment with a 92% confidence level. If the maximum cross-correlation coefficient of another correction frame is only 0.65, its weight in the control decision will be reduced accordingly to 0.65, thereby reducing its impact on the control results and avoiding excessive deviations introduced by data with low matching degrees. In addition, if a data packet deviates too much due to transmission link anomalies and is determined to be a sample requiring isolation, its original local timestamp, receiving timestamp, transmission link type, and degree of deviation will be recorded and output. In subsequent energy efficiency assessment reports, the information of these isolated samples can be analyzed. For example, if a large number of data packets are found to be isolated within a certain period, it may indicate that there is severe network congestion or intermittent sensor failure during that period, which in turn affects the normal operation and energy efficiency of the wind turbine, providing targeted diagnostic basis for operation and maintenance personnel.

[0107] like Figure 4 As shown, this application also discloses an industrial fan operating condition data processing system, including: The data receiving module 210 is used to acquire data packets transmitted from an industrial fan through at least two heterogeneous networks; wherein the data packets include an original local timestamp and payload information including operating condition data such as vibration and pressure. The evaluation module 220 is used to obtain the reception timestamp and transmission link type of the data packet when the data packet is received, and to generate the transmission status information of the data packet. The classification module 230 is used to evaluate the availability of the original local timestamp in the data packet based on the transmission status information, so as to classify the data packet into one of the following: samples that can be directly merged, samples that need to be relocated, and samples that do not participate in merging and need to be isolated. The time determination module 240 is used to perform association matching between the sample to be relocated and the historical data packets that have been merged based on the payload information, so as to determine the target physical time of the sample to be relocated, thereby obtaining the sample to be relocated after relocation. The hierarchical merging module 250 is used to merge the directly mergeable samples and the repositioned samples that need to be repositioned according to the corresponding physical time, forming data frames corresponding to different physical time segments; wherein, the output level of the data frame is distinguished into normal alignment frames or correction frames according to the type of merging. Output module 260 is used to output the data frame for real-time control or energy efficiency assessment of the industrial fan.

[0108] The data receiving module 210 can be configured to have multiple network interfaces, such as wired network interfaces (e.g., CAN bus interface, Ethernet interface) and wireless network interfaces (e.g., industrial Wi-Fi module, 5G communication module), to adapt to the complexity of heterogeneous networks in industrial environments. This module is responsible for the initial encapsulation and buffering of data packets received from different network interfaces, providing a data source for subsequent processing.

[0109] The evaluation module 220 can integrate a high-precision clock unit for accurately recording the arrival time of data packets. The transmission link type can be determined by parsing the network layer header information of the data packets or by using the interface identifier of the data receiving module. Transmission status information can be generated into a series of indicators, such as the sequence of data packet arrival time intervals and the statistics of data packet retransmission counts, which can reflect the stability and reliability of the current network link.

[0110] The classification module 230 can have a built-in set of rule-based or machine learning-based decision logic. For example, a threshold can be set so that when certain indicators in the transmission status information (such as fluctuations in arrival time intervals or the number of retransmissions) exceed a preset range, the original local timestamp is determined to be unusable. The output of the classification module is a data packet with classification tags, which guide subsequent modules to perform different processing on the data packet.

[0111] The time determination module 240 may include a data analysis engine capable of extracting key features from the data packet payload (e.g., the rate of change of pressure data, the spectral characteristics of vibration data). This module also features a sliding matching algorithm that finds the optimal temporal correspondence between the features of the sample to be relocated and the features of historically merged data. For example, cross-correlation algorithms or dynamic time warping (DTW) algorithms can be used to calculate the matching degree at different time offsets, thereby determining the target physical time.

[0112] The hierarchical merging module 250 can maintain a time-series database, inserting or updating all processed data packets according to their final determined physical time. This module can determine the output level of data frames based on the packet classification label (directly merging or relocated). For example, data frames primarily composed of directly merging samples are labeled as normally aligned frames, while data frames containing at least one relocated sample are labeled as corrected frames.

[0113] Output module 260 can provide various data interfaces, such as real-time data stream interfaces (e.g., MQTT, Kafka) for interfacing with industrial control systems, or historical data storage interfaces (e.g., database interfaces) for offline analysis by energy efficiency assessment systems. This module can output normal alignment frames and correction frames in different formats and frequencies according to the needs of downstream applications, ensuring that data can be used in a timely and accurate manner.

[0114] It should be noted that the information interaction and execution process between the above modules are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, which will not be repeated here.

[0115] The preferred embodiments of this application have been described in detail above, but this application is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application.

Claims

1. A method for processing operating condition data of an industrial fan, characterized in that, Includes the following steps: Acquire data packets from an industrial fan transmitted through at least two heterogeneous networks; wherein the data packets include original local timestamps and payload information including operating condition data such as vibration and pressure; Upon receiving the data packet, the receiving timestamp and transmission link type of the data packet are obtained, and the transmission status information of the data packet is generated. Based on the transmission status information, the availability of the original local timestamp in the data packet is evaluated to classify the data packet into one of the following: samples that can be directly merged, samples that need to be relocated, and samples that do not participate in merging and need to be isolated. Based on the payload information, the sample to be relocated is associated and matched with the historical data packets that have been merged to determine the target physical time of the sample to be relocated, thereby obtaining the sample to be relocated after relocation. The directly mergeable samples and the repositioned samples that need to be repositioned are merged according to the corresponding physical time to form data frames corresponding to different physical time segments; wherein, the output level of the data frame is distinguished into normal alignment frames or correction frames according to the type of merging. The data frame is output to enable real-time control or energy efficiency assessment of the industrial fan.

2. The method for processing industrial fan operating condition data according to claim 1, characterized in that, The transmission status information includes: time interval variance, retransmission flag, and degree of deviation from historical benchmark; The time interval variance is obtained by calculating the arrival time intervals of multiple consecutive data packets corresponding to the data packet data source within a historical time window, forming an interval sequence, and then calculating the variance of the interval sequence. The retransmission flag is extracted from the underlying network protocol stack of the central controller associated with the industrial fan. The degree of deviation is obtained by comparing the arrival time interval of the current data packets with the corresponding historical transmission records.

3. The method for processing industrial fan operating condition data according to claim 2, characterized in that, The step of assessing the availability of the original local timestamp in the data packet based on the transmission status information, in order to classify the data packet into one of the following categories: samples that can be directly merged, samples that need to be relocated, and samples that do not participate in merging and need to be isolated: When the time interval variance is not higher than the preset stable variance threshold, and the deviation is not higher than the preset stable deviation threshold and the retransmission flag indicates normal, the availability status of the original local timestamp in the data packet is determined to be stable, and the data packet is classified into samples that can be directly merged. When the time interval variance is higher than the stable variance threshold, and the deviation is not higher than the preset stable deviation threshold and the retransmission flag indicates normal, the availability status of the original local timestamp in the data packet is determined to be fluctuating, and the data packet is classified as a sample that needs to be relocated. When the deviation exceeds the stable deviation threshold, or the retransmission flag indicates an anomaly, the availability status of the original local timestamp in the data packet is determined to be abnormal, and the data packet is classified as an isolated sample that does not participate in the merging.

4. The method for processing industrial fan operating condition data according to claim 1, characterized in that, The step of associating and matching the sample to be relocated with historically merged data packets based on the payload information to determine the target physical time of the sample to be relocated, thereby obtaining the relocated sample to be relocated, includes: Determine candidate time windows for data packets in the samples requiring relocation; the start time of the candidate time window is the physical time of the last successfully merged data packet on the data source corresponding to the data packet, and the end time is the current system time. Extract the pressure change trend from the payload information of the data packets in the sample that needs to be relocated; Within the candidate time window, the vibration variation trend is extracted from the payload information of one or more data packets with established time sequences. The pressure change trend and the vibration change trend are used to perform sliding matching calculation within the candidate time window to find the physical time point with the highest matching degree, and this physical time point is used as the target physical time of the corresponding data packet in the sample to be relocated. All data packets that have regained the target physical time are used to form the relocation sample.

5. The method for processing industrial fan operating condition data according to claim 4, characterized in that, In the process of extracting the trend of the vibration: If the sampling frequency of the vibration is higher than the sampling frequency of the pressure, then the candidate time window is divided into multiple consecutive sub-time intervals according to the sampling frequency of the pressure. Calculate the statistical characteristic values ​​of the raw data of the vibration in each of the sub-time intervals; The statistical feature values ​​are used to generate a feature sequence aligned with the sampling frequency of the pressure; The characteristic sequence is used as the trend of the vibration.

6. The method for processing industrial fan operating condition data according to claim 4, characterized in that, Before the step of performing a sliding match calculation between the pressure change trend and the vibration change trend within the candidate time window to find the physical time point with the highest matching degree, the method further includes: Based on a preset delay constant that characterizes the inherent physical transmission delay between the pressure and the vibration, phase compensation is performed on the vibration variation trend to obtain the vibration variation trend after phase compensation, which is then used for subsequent sliding matching calculations.

7. The method for processing industrial fan operating condition data according to claim 4, characterized in that, The step of performing a sliding match calculation between the pressure change trend and the vibration change trend within the candidate time window to find the physical time point with the highest matching degree includes: Calculate the cross-correlation coefficients between the pressure variation trend and the vibration variation trend at different time offsets within the candidate time window; Determine the time offset that maximizes the cross-correlation coefficient; Based on the start time of the candidate time window and the time offset, the physical time point with the highest matching degree is calculated.

8. The method for processing industrial fan operating condition data according to claim 7, characterized in that, The step of performing a sliding match calculation between the pressure change trend and the vibration change trend within the candidate time window to find the physical time point with the highest matching degree also includes: If the maximum value of the cross-correlation coefficient is not higher than the preset matching rejection threshold, the association matching is determined to be unsuccessful, and a matching failure mark is generated for the corresponding data packet and removed from the samples that need to be relocated; wherein, the data packet carrying the matching failure mark is used for network quality analysis or fault tracing of industrial fans.

9. The method for processing industrial fan operating condition data according to claim 7, characterized in that, The step of outputting the data frame for real-time control or energy efficiency assessment of the industrial fan includes: The normal alignment frame and the correction frame are output to control the industrial fan in real time; wherein, the normal alignment frame is fully weighted, and the weight of the correction frame in the control decision is dynamically adjusted based on the maximum value of the cross-correlation coefficient calculated in the association matching step of generating the correction frame. The information of the normal alignment frame, the correction frame, and the sample to be isolated is output to perform an energy efficiency assessment of the industrial fan.

10. An industrial fan operating condition data processing system, characterized in that, include: A data receiving module is used to acquire data packets transmitted from an industrial fan through at least two heterogeneous networks; wherein the data packets include an original local timestamp and payload information including operating condition data such as vibration and pressure; The evaluation module is used to obtain the reception timestamp and transmission link type of the data packet when it is received, and to generate the transmission status information of the data packet. The classification module is used to evaluate the availability of the original local timestamp in the data packet based on the transmission status information, so as to classify the data packet into one of the following: samples that can be directly merged, samples that need to be relocated, and samples that do not participate in merging and need to be isolated. The time determination module is used to associate and match the sample to be relocated with the historical data packets that have been merged based on the payload information, so as to determine the target physical time of the sample to be relocated, thereby obtaining the sample to be relocated after relocation. The hierarchical merging module is used to merge the directly merging samples and the repositioned samples that need to be repositioned according to the corresponding physical time, forming data frames corresponding to different physical time segments; wherein, the output level of the data frame is distinguished into normal alignment frames or correction frames according to the type of merging. The output module is used to output the data frame for real-time control or energy efficiency assessment of the industrial fan.

Images