An intelligent monitoring method and system for the running state of a chemical fiber production power device

Abstract

This invention discloses an intelligent monitoring method and system for the operational status of electrical equipment in chemical fiber production, relating to the fields of fault prediction and health management. Through an importance assessment model, the method accurately classifies the importance levels of electrical equipment and matches them with differentiated monitoring strategy configurations, enabling personalized settings for data acquisition frequency and early warning thresholds. Subsequently, real-time operating parameters are acquired through a sensor network, and statistical characteristic values ​​calculated using a sliding window processing method are compared with preset early warning thresholds to promptly detect abnormal fluctuations, generating primary abnormal events and anomaly level classifications. This application enables the development of flexible monitoring strategies for different equipment importance in chemical fiber production, allowing for early identification of potential faults by capturing subtle changes in operating parameters in real time, thereby improving the efficiency of electrical equipment maintenance and ensuring production continuity.

Description

Technical Field

[0001] This invention relates to the field of fault prediction and health management technology, and in particular to an intelligent monitoring method and system for the operating status of electrical equipment used in chemical fiber production. Background Technology

[0002] In the chemical fiber production sector, the stable operation of electrical equipment plays a crucial role in ensuring the continuity of production lines and product quality. As a high-energy-consuming and highly continuous industrial process, chemical fiber production is susceptible to disruptions caused by sudden power equipment failures, which can lead to significant economic losses. Therefore, researching how to use intelligent methods to monitor and provide early warnings of electrical equipment operation in real time has become a core issue for improving production efficiency and safety. Research in this area not only affects enterprise operating costs but also directly impacts the sustainable development of the entire industry.

[0003] However, current monitoring methods for electrical equipment in chemical fiber production still have significant shortcomings. Many traditional methods tend to focus only on surface parameters during equipment operation, lacking in-depth insights into the equipment's health status and making them ill-suited to the complex and ever-changing working environment of chemical fiber production. Especially when facing hidden problems arising from long-term equipment operation, existing methods often fail to capture subtle changes in a timely manner, significantly reducing the accuracy and timeliness of fault warnings. This limitation allows potential risks in production process control to accumulate continuously, increasing the likelihood of sudden equipment failures.

[0004] A deeper technical challenge lies in how to establish reasonable monitoring standards for different equipment and accurately identify abnormal signals during equipment operation in a highly continuous environment like chemical fiber production. A key difficulty lies in the varying importance of different equipment. For example, spinning machines and winding machines are core equipment, and the stability of their power supply systems directly determines the normal operation of the production line, while the requirements for other auxiliary equipment are relatively lower. If the monitoring focus and warning thresholds cannot be flexibly adjusted according to the importance of the equipment, uneven resource allocation can easily occur, either over-focusing on secondary equipment and wasting resources, or neglecting critical equipment and leading to failure risks. Furthermore, this differentiated monitoring requirement brings another challenge: how to accurately judge subtle changes in equipment operating status in a complex production environment, such as whether fluctuations in current or temperature in the power supply system have reached a level requiring intervention.

[0005] Therefore, how to formulate flexible monitoring strategies for different equipment in chemical fiber production, and how to identify potential faults in advance by capturing subtle changes in operating parameters in real time, has become a key issue in improving the maintenance efficiency of power equipment and ensuring production continuity. Summary of the Invention

[0006] This invention provides an intelligent monitoring method and system for the operating status of electrical equipment in chemical fiber production. This system enables the formulation of flexible monitoring strategies based on the importance of different equipment in the chemical fiber manufacturing process management. It also allows for the early identification of potential faults by capturing subtle changes in operating parameters in real time, thereby improving the maintenance efficiency of electrical equipment and ensuring production continuity.

[0007] This invention provides an intelligent monitoring method for the operating status of electrical equipment in chemical fiber production, executed by a computer, comprising: Based on the static attributes and historical maintenance records of all electrical equipment in the chemical fiber production line, an importance assessment model is used to determine the importance level of each electrical device. Based on the importance level, the monitoring strategy configurations corresponding to power equipment of different importance levels are called from the preset strategy template library; Based on the acquisition frequency set in the monitoring strategy configuration, the real-time operating parameters of the target power equipment are continuously acquired through a sensor network deployed on the device side. The real-time operating parameters are processed using a sliding window method to determine the first target statistical feature value of the real-time operating parameters within each time window; If the first target statistical feature value exceeds the preset early warning threshold in the monitoring strategy configuration, then based on the first target statistical feature value, the primary abnormal event corresponding to the target power equipment and the abnormality level corresponding to the primary abnormal event are determined.

[0008] This invention provides an intelligent monitoring method and system for the operational status of electrical equipment in chemical fiber production. Addressing the comprehensive business scenario of assessing the importance of electrical equipment, real-time monitoring, and dynamically adjusting early warning thresholds in chemical fiber production lines, the method uses an importance assessment model to accurately classify the importance levels of electrical equipment and matches differentiated monitoring strategy configurations, enabling personalized settings for data acquisition frequency and early warning thresholds. Subsequently, real-time operating parameters are acquired through a sensor network, and statistical characteristic values ​​calculated using a sliding window processing method are compared with preset early warning thresholds to promptly detect abnormal fluctuations, generating primary abnormal events and anomaly level classifications. This application enables flexible monitoring strategies to be formulated based on the importance of different equipment in chemical fiber production, allowing for early identification of potential faults by capturing subtle changes in operating parameters in real time, thereby improving the maintenance efficiency of electrical equipment and ensuring production continuity. Attached Figure Description

[0009] Figure 1 This is one of the flowcharts illustrating an intelligent monitoring method for the operating status of electrical equipment in chemical fiber production, provided in an embodiment of the present invention. Figure 2 This is a second schematic flowchart of an intelligent monitoring method for the operating status of electrical equipment in chemical fiber production, provided by an embodiment of the present invention. Figure 3 This is the third flowchart illustrating an intelligent monitoring method for the operating status of electrical equipment in chemical fiber production, provided in an embodiment of the present invention. Figure 4 This is the fourth flowchart illustrating an intelligent monitoring method for the operating status of electrical equipment in chemical fiber production, provided in an embodiment of the present invention. Figure 5 This is the fifth flowchart illustrating an intelligent monitoring method for the operating status of electrical equipment in chemical fiber production, provided in an embodiment of the present invention. Figure 6 This is the sixth flowchart illustrating an intelligent monitoring method for the operating status of electrical equipment in chemical fiber production, provided in an embodiment of the present invention. Figure 7 This is the seventh flowchart of an intelligent monitoring method for the operating status of electrical equipment in chemical fiber production provided by an embodiment of the present invention. Detailed Implementation

[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0011] Reference Figure 1 This invention provides an intelligent monitoring method for the operating status of electrical equipment used in chemical fiber production, comprising the following steps: Step 100: Based on the static attributes and historical maintenance records of all electrical equipment in the chemical fiber production line, an importance assessment model is used to determine the importance level of each electrical device. First, all power equipment within the monitoring scope is quantified by collecting static attribute data for each piece of equipment, such as equipment type, rated power, process position in the production line, and inherent characteristics like the degree of impact on production continuity. Simultaneously, historical maintenance records are combined, including historical data reflecting equipment reliability and maintenance costs, such as failure frequency, repair duration, and spare parts replacement status. By inputting this multi-dimensional information—including static attributes and historical maintenance records—into a pre-built importance assessment model, the model can quantify and classify the comprehensive importance of each piece of equipment, ultimately determining a corresponding importance level for each power device.

[0012] The importance assessment model transforms the previously scattered and independent static attributes of equipment and historical maintenance records into comparable quantitative levels. Essentially, it's a comprehensive weighted assessment and classification model. First, the importance assessment model converts input data into calculable numerical features. Static attributes are quantified first. For inherent attributes of equipment, such as equipment type, rated power, process position in the chemical fiber production line, and impact on production continuity, the importance assessment model assigns a specific quantitative score to each item. Next, historical maintenance records are quantified. The importance assessment model extracts key indicators from these records, such as historical failure frequency, average repair time, spare parts replacement cycle and cost, and downtime caused by failures, converting the various data from historical maintenance records into standardized values. Furthermore, the importance assessment model internally pre-sets a weighting allocation mechanism to reflect the different positions of various dimensions in the importance assessment. For example, in the chemical fiber industry, where continuous production requirements are extremely high, the weights of the impact on production continuity and historical failure frequency may be set higher, while the weight of rated power may be relatively lower. The importance assessment model further multiplies the quantified values ​​from the first step by their corresponding weights and sums them to arrive at an initial score representing the overall importance of the equipment. After calculating the overall score for all equipment, the continuous score results are mapped to discrete, finite importance levels according to preset grading rules, outputting a level label for each piece of power equipment. For example, a common classification method may include a three-level classification of critical equipment, important equipment, and general equipment.

[0013] Step 200: Based on the importance level, retrieve the monitoring strategy configurations corresponding to power equipment of different importance levels from the preset strategy template library; After determining the importance level of each power device, a strategy template library is pre-built using this importance level. This library stores multiple pre-set monitoring strategy configurations for different importance levels. Each monitoring strategy configuration includes the data acquisition frequency, early warning threshold, and corresponding data analysis rules that match the level of the device. Once the importance level of the power device is determined, the corresponding monitoring strategy configuration is automatically retrieved and invoked. This allows for the matching of the most suitable monitoring scheme for each power device based on its importance level, thus establishing a correlation between monitoring resources and device criticality. This enables high-importance devices to receive more intensive data acquisition and more sensitive early warning responses, while low-importance devices adopt a more economical monitoring mode.

[0014] Step 300: Based on the acquisition frequency set in the monitoring strategy configuration, the real-time operating parameters of the target power equipment are continuously acquired through the sensor network deployed on the device side. After configuring a monitoring strategy that matches the equipment's importance level, the real-time data acquisition phase for equipment operation begins. Specifically, based on the invoked monitoring strategy configuration, the acquisition frequency parameter set for the target power equipment is extracted. This acquisition frequency determines the density and real-time requirements of data acquisition. Then, a sensor network pre-deployed on each piece of equipment in the chemical fiber production line is used to collect real-time operating parameters of the power equipment. This sensor network can consist of various types of sensors, such as current, voltage, temperature, and vibration sensors. Therefore, according to the set acquisition frequency, various operating parameters of the target power equipment are continuously acquired, resulting in a stable and uninterrupted data stream to capture instantaneous changes and dynamic trends during equipment operation. Thus, the real-time operating parameters collected by the sensor network ensure real-time perception of the equipment's actual operating condition.

[0015] Real-time operating parameters are the most direct source of raw data for sensing the health status of equipment. The main categories of real-time operating parameters include electrical parameters, thermal parameters, mechanical condition parameters, and operating status and environmental parameters. Electrical parameters reflect the power quality and electrical health of electrical equipment and are core indicators for monitoring equipment such as motors, transformers, and distribution cabinets. Examples include current, voltage, power, and frequency. Thermal parameters reflect the thermal state of equipment, as most electrical faults are accompanied by abnormal temperature rises. Thermal parameters typically include winding temperature, bearing temperature, casing temperature, and cooling medium parameters. Mechanical parameters assess the mechanical integrity of rotating equipment and typically include vibration, speed, and noise. Operating status and environmental parameters monitor the equipment's own operating logic and the impact of the external environment, including switch status, operating time, ambient temperature and humidity, and dust concentration.

[0016] Step 400: Perform sliding window processing on the real-time operating parameters to determine the first target statistical feature value of the real-time operating parameters in each time window; After acquiring real-time operating parameters through a sensor network, these parameters are typically stored as a dynamic sequence that extends over time. To effectively capture subtle changes in the device's operating status, a sliding window processing mechanism is introduced. Specifically, the real-time operating parameters are divided into overlapping or contiguous data segments, using a preset time window length as the unit. As time progresses, the window slides forward continuously to process the real-time operating parameters. For each time window, statistical calculations are performed on the real-time operating parameters contained within it to obtain a first target statistical feature value that represents the operating characteristics of that time period. This first target unified feature value can be a quantitative representation of the central tendency, dispersion, or other specific attributes of the data within the window. For example, the average value can reflect the operating level within that time period, the standard deviation can reflect the data fluctuation, and the peak value can capture instantaneous shock changes.

[0017] In this step, the original messy and dense raw data stream is transformed into an ordered feature sequence with clear physical meaning by using a sliding window processing method. This achieves efficient compression and dimensionality reduction of massive data, while smoothing out occasional noise interference, so that the changing trend and abnormal signs of the device's operating status can be presented in a clearer and more stable way.

[0018] Step 500: If the first target statistical feature value exceeds the preset early warning threshold in the monitoring strategy configuration, then based on the first target statistical feature value, determine the primary abnormal event corresponding to the target power equipment and the abnormality level corresponding to the primary abnormal event.

[0019] After obtaining the first target statistical feature value within each time window through sliding window processing, a threshold-based anomaly detection mechanism is introduced to proactively identify potential problems with power equipment. Specifically, the first target statistical feature value calculated for each time window is compared with a preset warning threshold for the power equipment in the monitoring strategy configuration. This preset warning threshold is pre-set in the monitoring strategy configuration based on the equipment's importance level and its historical operating characteristics. When it is detected that the first target statistical feature value within a time window exceeds the preset warning threshold, it is determined that the power equipment has shown abnormal signs in the current period, and the abnormal event identification process is initiated. Therefore, based on the numerical characteristics and trends of the exceeding first target statistical feature value, the primary abnormal event type faced by the target power equipment can be identified. For example, a continuously rising mean may indicate that the power equipment is overloaded or its performance is deteriorating, while a sudden increase in variance may indicate unstable operation or intermittent faults.

[0020] Furthermore, the severity and severity of the initial anomaly can be assessed by combining the magnitude and duration of the exceeding eigenvalue limits with the importance level of the power equipment itself. This process transforms data fluctuation characteristics into clearly directional operational status information, enabling early warning of potential equipment failures.

[0021] Setting the preset warning threshold is a comprehensive process that combines equipment mechanisms, historical experience, and risk levels. The preset warning threshold is a core parameter in the entire monitoring strategy configuration, determining the sensitivity and accuracy of anomaly detection. Considering factors such as equipment aging, seasonal changes, ambient temperature variations, or process adjustments, static fixed thresholds are often difficult to maintain effectiveness in the long term. Therefore, a dynamic threshold adjustment mechanism can be introduced to automatically recalculate and update the preset warning threshold based on real-time operating parameters over a preset period. Thus, the preset warning threshold can adaptively adjust to the slow aging of equipment or seasonal ambient temperature changes, avoiding frequent false alarms due to natural equipment degradation. Fiber production lines may have different operating modes, such as normal production, shutdown awaiting materials, and startup. Therefore, the preset warning threshold can also be updated based on the operating mode of the fiber production line to identify the current operating condition and call the corresponding warning threshold configuration. For example, the large current at startup is usually not considered a fault, so the current threshold during startup will be temporarily relaxed or disabled.

[0022] This invention provides an intelligent monitoring method for the operating status of electrical equipment in chemical fiber production. It enables the formulation of flexible monitoring strategies based on the importance of different equipment in chemical fiber production, and can identify potential faults in advance by capturing subtle changes in operating parameters in real time, thereby improving the maintenance efficiency of electrical equipment and ensuring production continuity.

[0023] In one embodiment, please refer to Figure 2 Based on the static attributes and historical maintenance records of all electrical equipment in the chemical fiber production line, an importance assessment model is used to determine the importance level of each piece of electrical equipment, including: Step 101: Based on the static attributes and historical maintenance records of each electrical device in the chemical fiber production line, a structured equipment basic dataset is obtained through data cleaning and format processing. Step 102: Based on the structured equipment basic dataset, extract the location distribution, functional redundancy information and historical fault impact range of each power equipment in the production process; Step 103: Based on the location distribution, the functional redundancy information, and the impact range of historical faults, an importance assessment model is used to determine a preliminary importance score; Step 104: Based on the preliminary importance score and combined with the historical fault impact information, determine the fault impact value of each power device. If the fault impact value exceeds the preset impact threshold, then the preliminary importance score is weighted and adjusted to determine the adjusted score result. Step 105: Based on the adjusted scoring results, determine the functional redundancy value involved in the historical data of each power device. If the functional redundancy value of any power device is lower than the preset standard value, then increase the importance level corresponding to the adjusted scoring results to obtain the importance level.

[0024] In this step, the static attributes and historical maintenance records of all electrical equipment on the chemical fiber production line are first preprocessed and standardized. This includes cleaning the raw data to remove errors or invalid information, and integrating the scattered and heterogeneous data into a structured and standardized equipment basic dataset through a unified format conversion. After obtaining the structured equipment basic dataset, more in-depth and relevant feature information is extracted from it. Specifically, by analyzing the structured equipment basic dataset, the specific location distribution of each electrical device in the production process is obtained to determine whether it is in the core process section or the auxiliary link. At the same time, the functional redundancy information of the electrical equipment is obtained, i.e., whether there is a spare device of the same model or an online standby machine. In addition, historical fault records are obtained to quantify the scope of production units affected by each fault and the duration of downtime, thus obtaining the impact range of historical faults.

[0025] Based on the above characteristics, an importance assessment model is introduced to comprehensively calculate factors such as location distribution, functional redundancy, and the impact range of historical failures, resulting in a preliminary importance score for each piece of equipment. This preliminary importance score represents the basic value of the equipment to the production line under normal conditions. Subsequently, a dynamic correction mechanism based on actual impact is introduced. By combining historical failure impact information, a corresponding failure impact value is calculated for each piece of equipment. This failure impact value reflects the frequency of failures, average repair time, or average loss caused by a single failure. When this failure impact value exceeds a preset impact threshold, it means that when the power equipment malfunctions, the consequences are far more severe than under normal circumstances. Therefore, the preliminary importance score needs to be weighted and adjusted to obtain an adjusted score. This adjustment method allows power equipment with average basic attributes but severe actual failure consequences to receive a higher importance rating.

[0026] The preset impact threshold is used to determine whether the actual consequences of a power equipment failure have reached a critical point requiring adjustment of its importance score. In one possible implementation, the preset impact threshold can be determined through statistical analysis of historical failure data. This involves deeply mining historical maintenance records for all equipment to extract historical failure data, and then using statistical analysis of the loss distribution characteristics of this data to determine a reasonable threshold. For example, the actual impact of each power equipment failure can be statistically analyzed to obtain loss distribution characteristics, which may include downtime, production loss, and maintenance costs. Typically, a higher quantile, such as the 75th or 90th percentile, can be used as the preset impact threshold.

[0027] Finally, based on the adjusted scoring results, further analysis is conducted on the functional redundancy values ​​involved in the historical data of the equipment, i.e., whether there are effective backup alternatives for the power equipment in actual operation and maintenance. If it is detected that any power equipment, although scoring high after previous adjustments, has a functional redundancy value lower than the preset standard value—for example, nominally having a backup unit, but the backup unit is frequently in a faulty state or has a high switching failure rate—it means that the actual fault tolerance of the power equipment is far lower than expected. In this case, the importance level corresponding to its adjusted scoring result is increased again to ensure that those seemingly reliable but actually vulnerable devices are included in the monitoring scope.

[0028] The preset standard value serves as a baseline for evaluating the effectiveness of redundancy configurations in power equipment and its sufficiency in handling faults. In one possible implementation, the preset standard value can be set through industry standards or enterprise design standards, or it can be quantified based on historical switchover success rates. For example, in industries with high continuity requirements, such as chemical fiber production, power equipment in different stages has explicit redundancy design requirements. For instance, a critical process section's cooling fan or oil pump may require one unit to operate and one unit to be on hot standby. Therefore, the preset standard value can be quantified as the ratio of standby equipment to operating equipment. If the standard value is set to 1.0, it means there must be 100% fully online standby capability; if it is set to 0.5, it means only 50% redundancy capability is needed to meet the standard.

[0029] In this embodiment, a weighted adjustment mechanism is used to dynamically calibrate the importance level based on the actual impact of the equipment's failures. This mechanism, which corrects future positioning based on historical performance, effectively avoids evaluation bias caused by relying solely on design parameters.

[0030] In one embodiment, please refer to Figure 3 The step of calling the monitoring policy configurations corresponding to power equipment of different importance levels from a preset policy template library based on the importance level includes: Step 201: Based on the importance level, obtain the strategy templates corresponding to power equipment of different importance levels from the preset strategy template library; Step 202: Based on the strategy template and its corresponding template classification identifier, and in combination with the importance level, generate a preliminary monitoring strategy configuration; Step 203: Extract the set of operating parameter types to be monitored from the preliminary monitoring strategy configuration, and collect the operating parameter data corresponding to the set of operating parameter types in real time; Step 204: If any parameter type in the running parameter data exceeds the preset warning threshold initial value, then the corresponding parameter type is marked as an abnormal state. Step 205: Continuously track the parameter categories of the abnormal state, obtain continuous data sequences within a preset time period, and analyze the changing trend of the continuous data sequences to obtain continuity characteristics; Step 206: Based on the continuity feature, dynamically adjust the monitoring parameters or early warning thresholds in the preliminary monitoring strategy configuration to obtain the monitoring strategy configuration.

[0031] After assessing the importance level of power equipment, the adaptive monitoring strategy configuration phase begins. Through initial configuration, data feedback, and trend analysis, a monitoring strategy configuration that can be optimized based on actual conditions is generated. Specifically, firstly, based on the previously determined importance level, strategy templates corresponding to different levels are retrieved from a pre-set strategy template library. These strategy templates include a basic monitoring framework matching the equipment's importance level, such as suggested data collection frequency, recommended warning threshold ranges, and key parameter types to focus on. Then, the template classification identifier attached to the strategy template is further combined with the specific power equipment's importance level to generate a preliminary monitoring strategy configuration. This template classification identifier can be used to distinguish the applicable scenarios or technology versions of the template.

[0032] After obtaining the initial monitoring strategy configuration, specific monitoring tasks are executed. Specifically, the set of operational parameter types requiring focused attention, such as current, temperature, or vibration, is extracted from the initial monitoring strategy configuration. Based on this, real-time data acquisition of the corresponding parameters of the target power equipment is initiated to obtain the operational parameter data corresponding to the parameter type set. This acquisition process must adhere to the frequency and accuracy requirements set in the initial monitoring strategy configuration to ensure that the acquired data accurately reflects the operating status of the equipment.

[0033] After collecting operational parameter data, a first anomaly screening mechanism is introduced, comparing each collected operational parameter data with the initial warning threshold value preset in the preliminary strategy configuration. When the current value of any parameter type in the operational parameter data exceeds the initial warning threshold value, that parameter type is marked as an abnormal state. Once a parameter type is marked as abnormal, the parameter category in the abnormal state is continuously tracked to obtain a continuous data sequence of that parameter within a preset time period. By analyzing the changing trends of this continuous data sequence, much richer information than single-point exceedances is captured, obtaining continuous characteristics. For example, whether the parameter suddenly jumps or slowly climbs, whether it oscillates at a high level continuously or briefly surges and then quickly falls back. Therefore, it is evident that the continuous characteristics obtained from this trend analysis can effectively distinguish between true fault precursors and occasional noise interference.

[0034] Finally, based on the continuity characteristics obtained from the trend analysis above, the initial monitoring strategy configuration is dynamically adjusted. If the trend analysis shows that the parameter is approaching the danger value at a stable rate, the system may appropriately tighten the warning threshold to issue an early warning; if the analysis indicates that the anomaly is intermittent and has returned to normal, then the sensitivity of the parameter is temporarily relaxed, or the sampling frequency is adjusted. After this round of dynamic optimization, a monitoring strategy configuration suitable for the current actual operating conditions of the equipment is finally obtained.

[0035] In this embodiment, since the generated monitoring strategy configuration has been verified and corrected based on trend characteristics, false alarms based on single-point over-limits but lacking continuous trend support are effectively filtered out, enabling maintenance personnel to focus their attention on truly potential abnormal events, thus greatly improving fault response efficiency.

[0036] In one embodiment, please refer to Figure 4 The method of continuously acquiring real-time operating parameters of the target power equipment based on the acquisition frequency set in the monitoring strategy configuration, through a sensor network deployed on the device side, includes: Step 301: Based on the acquisition frequency set in the monitoring strategy configuration, collect the raw operating parameter dataset of the target power equipment in real time through the sensor network; Step 302: Based on the original operating parameter dataset, the missing data or obvious anomalies identified are filled in using an interpolation method to generate a standardized operating parameter dataset. Step 303: Extract time-series data of key parameters, including three-phase current values, winding temperature values, and vibration amplitude values, from the standardized operating parameter dataset; Step 304: Based on the time series data of the key parameters, time series analysis is used to analyze and determine the changing trend and fluctuation characteristics of each parameter in the time dimension; Step 305: If the change trend or the fluctuation characteristic deviates from the preset operating range in the monitoring strategy configuration, mark the parameter as an abnormal event at the corresponding time point, and record the timestamp, parameter type and value deviation range of the abnormal event. Step 306: Remove the operating parameter data corresponding to the abnormal event from the standardized operating parameter dataset to obtain the real-time operating parameters.

[0037] After obtaining the optimized monitoring strategy configuration for the specific equipment, the substantive data acquisition and preprocessing phase begins to transform the raw sensor signals into clean, reliable, and standardized operating parameters suitable for subsequent analysis. Specifically, firstly, according to the acquisition frequency set in the monitoring strategy configuration, the sensor network deployed on the equipment side is activated to continuously acquire raw data on various operating states of the target power equipment, obtaining a raw operating parameter dataset. This raw operating parameter dataset originates from the sensors and contains the most basic information about equipment operation, but it may also contain various interference factors. Therefore, to address this issue, a data cleaning mechanism is introduced to screen the raw operating parameter dataset for quality, identifying missing data caused by transmission interruptions or transient sensor failures, as well as abnormal jump points that clearly exceed the physical possibilities. For these incomplete data points, interpolation methods are used to complete them, for example, by averaging the values ​​of preceding and following normal data or by estimating values ​​based on historical trends, to generate a continuous, complete, and uniformly formatted standardized operating parameter dataset.

[0038] After obtaining a standardized dataset of operating parameters, the most representative key parameters are extracted. Time-series data of these key parameters, including three-phase current values, winding temperature values, and vibration amplitude values, are extracted from the standardized dataset. These three types of parameters characterize the operational health of the power equipment from three dimensions: electrical load, thermal state, and mechanical integrity, forming the core data source for subsequent condition analysis. Then, these key parameter time-series data are deeply mined. Time series analysis methods are used to analyze the evolution of each type of key parameter time-series data over time, analyzing its changing trends and specific fluctuation characteristics. The changing trends include whether the parameter is operating smoothly, slowly increasing, or fluctuating periodically; the fluctuation characteristics include the amplitude, frequency, and statistical distribution of the fluctuations. This trend analysis can effectively capture subtle changes in equipment condition, providing a basis for subsequent anomaly detection.

[0039] Based on trend analysis results, an anomaly detection mechanism based on trend deviation is introduced. This mechanism compares the analyzed trends or fluctuation characteristics with the preset operating range in the monitoring strategy configuration. If a trend or fluctuation characteristic exceeds the preset operating range—for example, a continuous rise in the baseline of vibration amplitude or a significant increase in the imbalance of three-phase current—the corresponding parameter is marked as an anomaly at the corresponding time point. The timestamp of the event, the specific parameter type involved, and the specific magnitude of the deviation from the preset range are recorded in detail. This transforms abstract trend analysis results into event records with clear time indications and quantitative characteristics.

[0040] The preset operating range is a benchmark interval used to determine whether the equipment's operating status has deviated abnormally. This preset operating range focuses on the dynamic monitoring of parameter change trends and fluctuation characteristics, representing the definition of the equipment's normal operating condition boundaries. For example, the preset operating range can be determined by the static boundaries of the equipment's inherent characteristics and design parameters. Chemical fiber production is a continuous process, and the operating range of electrical equipment needs to meet the requirements of downstream processes. For example, the speed fluctuation range of the motor driving the spinning pump needs to be controlled within ±0.5% of the process allowance to ensure uniform fiber fineness. Therefore, this process requirement can be used as the actual constraint boundary of the operating range.

[0041] Finally, the standardized operating parameter dataset undergoes a purification process. To ensure that the data used for subsequent condition assessment and feature extraction accurately reflects the normal operating status of the equipment, operating parameter data corresponding to events marked as abnormal are removed from the dataset. The data obtained after this purification process becomes the real-time operating parameters used for subsequent sliding window processing and statistical feature value calculation. This step, by removing identified abnormal segments, effectively prevents these abnormal data from contaminating the subsequent calculation of the normal state baseline, thereby improving the robustness of the entire monitoring system.

[0042] In this embodiment, by using time-series-based trend deviation detection, early fault symptoms that have not yet reached the absolute threshold but continue to develop in an unfavorable direction can be identified, thus achieving true predictive maintenance capabilities.

[0043] In one embodiment, please refer to Figure 5 The step of performing sliding window processing on the real-time operating parameters to determine the first target statistical feature value of the real-time operating parameters within each time window includes: Step 401: Based on the real-time operating parameters, a sliding window method is used to divide the time into multiple consecutive time windows to obtain a subset of operating parameters within each time window; Step 402: Perform statistical analysis on the subset of operating parameters to calculate the initial feature values ​​corresponding to the time window, wherein the initial feature values ​​include at least the mean and standard deviation; Step 403: Extract historical data segments from the pre-established historical data repository that match the duration and time attributes of each time window, and calculate the historical statistical feature values ​​of the historical data segments. Step 404: Compare the initial feature value with the historical statistical feature value, calculate the deviation between the initial feature value and the historical statistical feature value, and obtain the deviation evaluation index; Step 405: Based on the deviation evaluation index, determine the confidence weight of the time window, wherein the confidence weight is negatively correlated with the deviation evaluation index; Step 406: Based on the confidence weight, the initial feature value is weighted and fused with the historical statistical feature value of the corresponding time window to obtain the first target statistical feature value.

[0044] After collecting, cleaning, and standardizing the raw data, the data analysis phase begins. Through sliding window processing and intelligent fusion computing, highly representative and reliable primary statistical feature values ​​are extracted from the real-time operating parameters. This process, using a mechanism based on historical comparison and dynamic weighting, ensures that the final feature values ​​more accurately and robustly reflect the actual operating status of the equipment. Specifically, based on the real-time operating parameters, a sliding window method is used to divide them into continuous and overlapping or connected time windows. Each window corresponds to a subset of the operating parameters of the power equipment within a specific time period. This division method ensures the continuity and real-time nature of monitoring. Then, statistical analysis is performed on the subset of operating parameters within each time window to calculate the initial feature values ​​corresponding to that window. These initial feature values ​​include at least the mean, reflecting the central tendency of the data, and the standard deviation, reflecting the degree of data dispersion.

[0045] After obtaining the initial feature values, a historical comparison mechanism is introduced. This involves extracting historical data segments from a pre-established historical data repository that match the duration and time attributes of the current time window, such as those from the same season, production period, or operating condition. The historical statistical feature values ​​corresponding to these historical segments are then calculated. Essentially, this step provides a historical reference point for the current operating status of the power equipment, allowing for the determination of whether the current performance is within the normal fluctuation range or deviates from historical norms.

[0046] Next, the initial feature values ​​of the current window are compared and analyzed with the extracted historical statistical feature values. By calculating the deviation between the two, a quantitative deviation evaluation index is obtained. This deviation evaluation index is used to intuitively reflect the degree of abnormal deviation of the current operating state compared with the historical normal state. Based on this deviation evaluation index, the confidence weight of the time window is further determined, and this confidence weight is negatively correlated with the deviation evaluation index. That is, if the initial feature value of the current window deviates more from the historical normal, the confidence level of the feature value is lower, and the weight assigned to it is smaller; conversely, if the deviation is small, the confidence level is high, and the weight is correspondingly increased. It can be seen that the more abnormal the data, the more its influence in the final feature value should be limited to avoid excessive interference from a single drastic fluctuation in the overall judgment.

[0047] Finally, using the confidence weights calculated above, the initial feature values ​​of the current window are weighted and fused with the historical feature values ​​of the corresponding time window. Through this fusion calculation, a first target statistical feature value is obtained that reflects the current real-time state and has been calibrated according to a regular pattern, thus preserving the sensitivity of real-time monitoring while incorporating the smoothness and robustness of historical data.

[0048] In this embodiment, a dynamic risk assessment machine is constructed by using confidence weights and deviation assessment indicators. This avoids overreaction when facing abnormal data and allows for full trust when facing regular data, thus ensuring the sensitivity of monitoring and achieving a dynamic balance in data monitoring.

[0049] In one embodiment, please refer to Figure 6 If the first target statistical feature value exceeds the preset early warning threshold in the monitoring strategy configuration, then based on the first target statistical feature value, the primary abnormal event corresponding to the target power equipment and the abnormality level corresponding to the primary abnormal event are determined, including: Step 501: If the statistical feature value of the first target exceeds the preset warning threshold in the monitoring strategy configuration, calculate the percentage of relative deviation between the statistical feature value of the first target and the preset warning threshold to obtain the over-limit amplitude coefficient of the current time window. Step 502: Count the number of time windows in which the statistical feature value of the first target exceeds the preset warning threshold consecutively, and obtain the abnormal duration coefficient; Step 503: Based on the over-limit amplitude coefficient and the abnormal duration coefficient, the severity score of the current abnormality is obtained by weighted calculation; Step 504: Match the severity score with a preset score range to generate a primary abnormal event and its corresponding abnormality level.

[0050] After obtaining the first target statistical feature value through sliding window processing and weighted fusion, the process enters the anomaly judgment and quantitative grading stage to quantitatively assess the severity of the anomaly from multiple dimensions and transform it into a primary anomaly event with an accurate grade. Specifically, firstly, after confirming that the first target statistical feature value has exceeded the corresponding preset warning threshold in the monitoring strategy configuration, the excess range is quantitatively calculated. The relative percentage deviation between the first target statistical feature value and the preset warning threshold is calculated, that is, the ratio of the excess value to the threshold, thereby obtaining the excess range coefficient for the current time window, providing a quantitative metric for distinguishing between minor exceedances and severe exceedances.

[0051] Next, the anomalies are examined from a time perspective. The number of time windows in which the statistical characteristic value of the first target continuously exceeds the preset warning threshold is counted to obtain the anomaly duration coefficient. This anomaly duration coefficient is used to distinguish between occasional instantaneous fluctuations and continuous deterioration. If a parameter only briefly crosses the threshold and then quickly recovers, its duration coefficient will be low; however, if a parameter remains above the threshold for multiple consecutive windows, it indicates that the device is indeed in an abnormal state and there is no improvement, and the duration coefficient will increase accordingly. This time-based consideration effectively filters out false alarms caused by noise interference.

[0052] Next, the over-limit amplitude coefficient and the abnormal duration coefficient are comprehensively quantified. Based on these coefficients, a severity score for the current anomaly is obtained using a preset weighted calculation rule. In this scoring model, the amplitude coefficient and duration coefficient are assigned corresponding weights; for example, some types of anomalies may be more sensitive to amplitude, while others are more sensitive to duration. Through this weighted fusion, the overall threat level posed by the abnormal event to the safe operation of the equipment can be comprehensively assessed. Finally, the calculated severity score is matched with preset scoring intervals. Several consecutive scoring intervals are pre-defined, each corresponding to a specific anomaly level, such as attention level, warning level, or danger level. When the severity score falls into a certain scoring interval, a primary anomaly event of the corresponding level is generated. The primary anomaly event includes the result of the abnormal situation occurring and also carries a quantified level label.

[0053] In this embodiment, continuous severity scores are mapped to discrete abnormality levels, achieving a smooth transition from quantitative analysis to hierarchical decision-making. This level classification method based on objective scores is more scientific and stable than relying on human experience to set levels, thus improving the accuracy of the level classification results.

[0054] In one embodiment, please refer to Figure 7 It also includes: Step 600: If the anomaly level is high, a high-level warning notification is triggered; Step 700: If the anomaly level is medium or low, then obtain the second target statistical feature value, time domain change feature value, and correlation feature value between parameters of other operating parameters associated with the real-time operating parameters, and perform a comprehensive analysis on the second target statistical feature value, the time domain change feature value, and the correlation feature value to determine the parameter change trend. If the parameter change trend has collaborative anomaly characteristics, then trigger the corresponding level of early warning notification according to the degree of anomaly of the parameter change trend.

[0055] After completing the anomaly identification and grading based on the first objective statistical characteristic value, the decision-making phase begins with differentiated responses based on the equipment anomaly level. For equipment of varying criticality, distinctly different early warning triggering strategies are employed to ensure the safety of core assets while avoiding waste of operational resources due to overreaction. Specifically, a direct early warning triggering mechanism is first established for equipment with high importance. When an anomaly event is identified in such critical equipment, a high-level early warning notification is immediately triggered regardless of its anomaly level. Since highly important equipment is often the core asset on the chemical fiber production line, even if the current severity of an anomaly is not high, it may harbor the risk of rapid deterioration or have a decisive impact on production continuity. Therefore, each anomaly of high-level equipment is treated with care to ensure that the operations and maintenance team is notified and can intervene immediately.

[0056] When an anomaly occurs in equipment at a medium or low anomaly level, a multi-parameter collaborative analysis mechanism is activated. Specifically, this mechanism acquires the second target statistical characteristic value, time-domain variation characteristic value, and correlation characteristic value among other operating parameters associated with the anomaly parameter. The second target statistical characteristic value reflects the overall level of the associated parameters, the time-domain variation characteristic value characterizes the dynamic evolution of the associated parameters over time, and the correlation characteristic value depicts the inherent linkage between different parameters. Through in-depth fusion analysis of these three dimensions, a comprehensive view of the equipment's current state is constructed, allowing for the determination of the overall trend of parameter changes.

[0057] Following in-depth fusion analysis, the system identifies whether there are coordinated anomalies, i.e., whether multiple related parameters simultaneously or successively exhibit abnormal signs, and whether their variation patterns conform to a known fault mode. For example, if only a single-phase current slightly exceeds the limit, but the other two-phase currents, winding temperature, and vibration are normal, it may just be an occasional load fluctuation; however, if the current exceeds the limit while the winding temperature is also continuously rising, and the vibration amplitude is increasing, the coordinated anomalies among these three factors may point to a genuine equipment failure. Based on the identification of such coordinated anomalies, an appropriate level of early warning notification is triggered according to the overall degree of anomaly in the parameter change trend. This mechanism, which determines the triggering of an early warning notification based on the presence of coordinated anomalies in parameter change trends, improves the reliability of early warnings for low-to-medium importance equipment and effectively avoids false alarms caused by occasional fluctuations in a single parameter.

[0058] In this embodiment, the optimized allocation of limited operation and maintenance resources is achieved. For highly important equipment, any anomaly will trigger an immediate response; for medium and low-importance equipment, a more prudent analysis process is used for filtering. This hierarchical response mechanism achieves a precise match between monitoring resource investment and equipment value.

[0059] The following describes an intelligent monitoring system for the operating status of electrical equipment used in chemical fiber production, provided by the present invention. The intelligent monitoring system for the operating status of electrical equipment used in chemical fiber production described below can be referred to in correspondence with the intelligent monitoring method for the operating status of electrical equipment used in chemical fiber production described above.

[0060] This invention also provides an intelligent monitoring system for the operating status of electrical equipment in chemical fiber production, comprising: The assessment module is used to determine the importance level of each electrical device based on the static attributes and historical maintenance records of all electrical equipment in the chemical fiber production line, using an importance assessment model. The calling module is used to call the monitoring strategy configuration corresponding to power equipment of different importance levels from the preset strategy template library based on the importance level; The acquisition module is used to continuously acquire the real-time operating parameters of the target power equipment through a sensor network deployed on the device side, based on the acquisition frequency set in the monitoring strategy configuration. The processing module is used to perform sliding window processing on the real-time operating parameters to determine the first target statistical feature value of the real-time operating parameters in each time window; The analysis module is used to determine the primary abnormal event corresponding to the target power equipment and the abnormality level corresponding to the primary abnormal event based on the first target statistical feature value if the first target statistical feature value exceeds the preset early warning threshold in the monitoring strategy configuration.

[0061] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent monitoring of the operating status of electrical equipment in chemical fiber production, characterized in that, Executed by a computer, including: Based on the static attributes and historical maintenance records of all electrical equipment in the chemical fiber production line, an importance assessment model is used to determine the importance level of each electrical device. Based on the importance level, the monitoring strategy configurations corresponding to power equipment of different importance levels are called from the preset strategy template library; Based on the acquisition frequency set in the monitoring strategy configuration, the real-time operating parameters of the target power equipment are continuously acquired through a sensor network deployed on the device side. The real-time operating parameters are processed using a sliding window method to determine the first target statistical feature value of the real-time operating parameters within each time window; If the first target statistical feature value exceeds the preset early warning threshold in the monitoring strategy configuration, then based on the first target statistical feature value, the primary abnormal event corresponding to the target power equipment and the abnormality level corresponding to the primary abnormal event are determined.

2. The intelligent monitoring method for the operating status of electrical equipment in chemical fiber production according to claim 1, characterized in that, Based on the static attributes and historical maintenance records of all electrical equipment in the chemical fiber production line, an importance assessment model is used to determine the importance level of each piece of electrical equipment, including: Based on the static attributes and historical maintenance records of each electrical device in the chemical fiber production line, a structured equipment basic dataset is obtained through data cleaning and format processing. Based on the structured equipment dataset, the location distribution, functional redundancy information, and historical fault impact range of each power device in the production process are extracted. Based on the location distribution, the functional redundancy information, and the impact range of historical failures, an importance assessment model is used to determine a preliminary importance score. Based on the preliminary importance score and the historical fault impact information, the fault impact value of each power device is determined. If the fault impact value exceeds the preset impact threshold, the preliminary importance score is weighted and adjusted to determine the adjusted score result. Based on the adjusted scoring results, the functional redundancy value involved in the historical data of each power device is determined. If the functional redundancy value of any power device is lower than the preset standard value, the importance level corresponding to the adjusted scoring result is increased to obtain the importance level.

3. The intelligent monitoring method for the operating status of electrical equipment in chemical fiber production according to claim 1, characterized in that, The step of calling the monitoring policy configurations corresponding to power equipment of different importance levels from a preset policy template library based on the importance level includes: Based on the importance level, obtain the strategy templates corresponding to power equipment of different importance levels from the preset strategy template library; Based on the policy template and its corresponding template classification identifier, and combined with the importance level, a preliminary monitoring policy configuration is generated; Extract the set of operating parameter types to be monitored from the initial monitoring strategy configuration, and collect the operating parameter data corresponding to the set of operating parameter types in real time; If any parameter type in the operating parameter data exceeds the preset warning threshold initial value, the corresponding parameter type will be marked as an abnormal state. The parameter categories of the abnormal state are continuously tracked to obtain a continuous data sequence within a preset time period, and the changing trend of the continuous data sequence is analyzed to obtain the continuity characteristics. Based on the aforementioned continuity characteristic, the monitoring parameters or early warning thresholds in the initial monitoring strategy configuration are dynamically adjusted to obtain the monitoring strategy configuration.

4. The intelligent monitoring method for the operating status of electrical equipment in chemical fiber production according to claim 1, characterized in that, The method of continuously acquiring real-time operating parameters of the target power equipment based on the acquisition frequency set in the monitoring strategy configuration, through a sensor network deployed on the device side, includes: Based on the acquisition frequency set in the monitoring strategy configuration, the raw operating parameter dataset of the target power equipment is collected in real time through the sensor network. Based on the original operating parameter dataset, missing data or obvious anomalies are filled in using interpolation methods to generate a standardized operating parameter dataset. From the standardized operating parameter dataset, extract the time-series data of key parameters, including three-phase current values, winding temperature values, and vibration amplitude values; Based on the time series data of the key parameters, time series analysis methods are used to analyze and determine the changing trends and fluctuation characteristics of each parameter in the time dimension. If the trend of change or the fluctuation characteristics deviates from the preset operating range in the monitoring strategy configuration, the parameter is marked as an abnormal event at the corresponding time point, and the timestamp, parameter type and value deviation range of the abnormal event are recorded. The standardized operating parameter dataset is processed by removing the operating parameters corresponding to the abnormal events to obtain the real-time operating parameters.

5. The intelligent monitoring method for the operating status of electrical equipment in chemical fiber production according to claim 1, characterized in that, The step of performing sliding window processing on the real-time operating parameters to determine the first target statistical feature value of the real-time operating parameters within each time window includes: Based on the real-time operating parameters, a sliding window method is used to divide the time window into multiple consecutive time windows, and a subset of operating parameters is obtained in each time window. Statistical analysis is performed on the subset of operating parameters to calculate the initial feature values ​​corresponding to the time window, wherein the initial feature values ​​include at least the mean and standard deviation; Extract historical data segments from a pre-established historical data repository that match the duration and time attributes of each time window, and calculate the historical statistical characteristic values ​​of the historical data segments. The initial feature value is compared with the historical statistical feature value, and the deviation between the initial feature value and the historical statistical feature value is calculated to obtain the deviation evaluation index. Based on the deviation assessment index, the confidence weight of the time window is determined, wherein the confidence weight is negatively correlated with the deviation assessment index; Based on the confidence weight, the initial feature value is weighted and fused with the historical statistical feature value of the corresponding time window to obtain the first target statistical feature value.

6. The intelligent monitoring method for the operating status of electrical equipment in chemical fiber production according to claim 1, characterized in that, If the first target statistical feature value exceeds the preset early warning threshold in the monitoring strategy configuration, then based on the first target statistical feature value, the primary abnormal event corresponding to the target power equipment and the abnormality level corresponding to the primary abnormal event are determined, including: If the statistical feature value of the first target exceeds the preset warning threshold in the monitoring strategy configuration, the percentage of relative deviation between the statistical feature value of the first target and the preset warning threshold is calculated to obtain the over-limit amplitude coefficient of the current time window. The number of time windows in which the statistical feature value of the first target continuously exceeds the preset warning threshold is counted to obtain the abnormal duration coefficient; The severity score of the current anomaly is obtained by weighted calculation based on the excess amplitude coefficient and the anomaly duration coefficient. The severity score is matched with a preset score range to generate a primary abnormal event and its corresponding abnormality level.

7. The intelligent monitoring method for the operating status of electrical equipment in chemical fiber production according to claim 1, characterized in that, Also includes: If the anomaly level is high, a high-level warning notification will be triggered; If the anomaly level is medium or low, then the second target statistical feature value, time domain change feature value, and correlation feature value between parameters of other operating parameters associated with the real-time operating parameters are obtained, and the second target statistical feature value, the time domain change feature value, and the correlation feature value are comprehensively analyzed to determine the parameter change trend. If the parameter change trend has collaborative anomaly characteristics, then the corresponding level of early warning notification is triggered according to the degree of anomaly of the parameter change trend.

8. An intelligent monitoring system for the operating status of electrical equipment in chemical fiber production, characterized in that, include: The assessment module is used to determine the importance level of each electrical device based on the static attributes and historical maintenance records of all electrical equipment in the chemical fiber production line, using an importance assessment model. The calling module is used to call the monitoring strategy configuration corresponding to power equipment of different importance levels from the preset strategy template library based on the importance level; The acquisition module is used to continuously acquire the real-time operating parameters of the target power equipment through a sensor network deployed on the device side, based on the acquisition frequency set in the monitoring strategy configuration. The processing module is used to perform sliding window processing on the real-time operating parameters to determine the first target statistical feature value of the real-time operating parameters in each time window; The analysis module is used to determine the primary abnormal event corresponding to the target power equipment and the abnormality level corresponding to the primary abnormal event based on the first target statistical feature value if the first target statistical feature value exceeds the preset early warning threshold in the monitoring strategy configuration.

Images