Elevator running state monitoring method and system integrating multiple sensors

Abstract

The invention relates to the technical field of elevator running state monitoring, and discloses an elevator running state monitoring method and system integrating multiple sensors. The method comprises the following steps: acquiring a vibration signal and a current signal of an elevator door track, and generating an initial multi-source signal set; performing time synchronization processing on the initial multi-source signal set to obtain a time synchronization signal sequence, extracting features of various signals from the time synchronization signal sequence, and generating a signal feature fragment set; analyzing an interaction mode between the signals according to the signal feature fragment set, determining an interaction abnormal feature group, fusing the interaction abnormal feature group and historical fault data of the elevator door track, and generating corrected door track operation state description; and if the abnormal component in the corrected door track operation state description exceeds a preset threshold value, abnormal data are optimized, optimized state description is obtained, an elevator fault mode is matched according to the optimized state description, and a fault diagnosis report for the elevator door track system is generated. The fault monitoring precision of the elevator door rail is improved.

Description

Technical Field

[0001] This application relates to the field of elevator operation status monitoring technology, and in particular to a method and system for elevator operation status monitoring that integrates multiple sensors. Background Technology

[0002] With the acceleration of urbanization and the increase in building height, elevators have become an indispensable vertical transportation tool in modern buildings. The safe operation of elevators is directly related to the safety of passengers. Therefore, how to monitor the operating status of elevators in real time and detect potential malfunctions in advance has become an important issue in elevator maintenance and management.

[0003] Currently, elevator fault monitoring methods mainly rely on single sensors or periodic inspections based on human experience. This traditional monitoring approach has several problems, such as difficulty in real-time monitoring of various elevator anomalies, inability to promptly detect potential fault modes in the system, and low accuracy in fault diagnosis. Elevator faults often occur suddenly without warning, especially when there are faults such as wear or jamming in the door rail system, which can easily lead to elevator shutdown or more serious safety accidents. In existing elevator monitoring technologies, multi-sensor data fusion and intelligent analysis technologies have not been fully utilized. While traditional vibration monitoring methods can provide some fault indications, they cannot comprehensively assess the overall operating status of the elevator and struggle to provide accurate diagnostic results when faced with complex elevator fault modes. To address these issues, in recent years, increasing research has focused on intelligent diagnostic technologies that integrate multiple sensors. By combining signals such as vibration, temperature, and current, comprehensive monitoring of the elevator system can be achieved, and modern data analysis methods can be used to improve the accuracy of fault diagnosis. Nevertheless, existing technologies still lack effective algorithmic and physical model support, resulting in significant room for improvement in the real-time performance, accuracy, and reliability of elevator operation status monitoring systems.

[0004] Therefore, this application proposes an elevator operation status monitoring method that integrates multi-sensor data, combines physical models and dynamic optimization, which can better diagnose potential faults in elevator systems and ensure safe elevator operation. Summary of the Invention

[0005] To address the aforementioned technical issues, this application provides a method and system for monitoring elevator operating status by integrating multiple sensors, which can improve the accuracy of elevator door rail fault monitoring.

[0006] In a first aspect, this application provides a method for monitoring the operating status of an elevator by integrating multiple sensors, the method comprising: Step S1: Acquire vibration and current signals of the elevator door rails to generate an initial multi-source signal set; Step S2: Perform time synchronization processing on the initial multi-source signal set to obtain a time synchronization signal sequence, extract features of various signals from the time synchronization signal sequence, and generate a signal feature segment set; Step S3: Analyze the interaction patterns between signals based on the set of signal feature segments, determine the interaction anomaly feature group, fuse the interaction anomaly feature group with the elevator door rail historical fault data, and generate a corrected description of the door rail operating status. Step S4: If the abnormal component in the corrected door rail operating status description exceeds the preset threshold, the abnormal data is optimized to obtain an optimized status description. Based on the optimized status description, the elevator fault mode is matched to generate a fault diagnosis report for the elevator door rail system.

[0007] Secondly, this application provides an elevator operation status monitoring system integrating multiple sensors, the system comprising: The acquisition module is used to acquire vibration and current signals of the elevator door rail and generate an initial multi-source signal set; The synchronization module is used to perform time synchronization processing on the initial multi-source signal set to obtain a time synchronization signal sequence, extract features of various signals from the time synchronization signal sequence, and generate a set of signal feature segments. The analysis module is used to analyze the interaction patterns between signals based on the set of signal feature segments, determine the interaction anomaly feature group, fuse the interaction anomaly feature group with the elevator door rail historical fault data, and generate a corrected description of the door rail operating status. The diagnostic module is used to optimize the abnormal data if the abnormal components in the corrected door rail operating status description exceed a preset threshold, obtain an optimized status description, match the elevator fault mode according to the optimized status description, and generate a fault diagnosis report for the elevator door rail system.

[0008] Compared with the prior art, the beneficial effects of the present invention are at least as follows: This application significantly improves the accuracy and reliability of elevator door rail system condition monitoring through multi-sensor signal fusion and multi-level intelligent analysis. First, by accurately synchronizing and extracting features from vibration and current signals, it effectively overcomes the analysis obstacles of multi-source heterogeneous signals, laying a high-quality data foundation for subsequent diagnosis. Second, by analyzing the interaction patterns between signals and fusing historical fault data, it generates an iteratively corrected description of the operating state, greatly improving the stability and historical consistency of the condition assessment. When a significant anomaly is detected, deep optimization based on a physical dynamics model is initiated, mapping the abnormal phenomenon to specific physical parameter changes, achieving precise location and quantitative diagnosis of the fault root cause. Finally, by combining real-time wear data and runtime prediction, a fault diagnosis report with risk level ranking is generated, providing direct and reliable action guidelines for maintenance decisions. This application effectively reduces the false alarm and missed alarm rates, realizing a shift from passive maintenance to predictive maintenance, and comprehensively ensuring the safety and efficiency of elevator operation. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of one embodiment of the elevator operation status monitoring method integrating multiple sensors in this application. Figure 2 This is a schematic diagram of the parameterized dynamic model of the elevator door rail system in the embodiments of this application; Figure 3 This is a graph showing the gain effect of the iterative optimization index in the embodiments of this application; Figure 4 This is a schematic diagram of one embodiment of the elevator operation status monitoring system integrating multiple sensors in this application. Detailed Implementation

[0011] This application provides a method and system for monitoring elevator operation status by integrating multiple sensors. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0012] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the elevator operation status monitoring method integrating multiple sensors in this application includes: Step S1: Obtain the vibration signal and current signal of the elevator door rail to generate an initial multi-source signal set.

[0013] Specifically, with the continuous development of elevator operation status monitoring technology, traditional elevator fault diagnosis methods are gradually failing to meet the high-precision requirements of modern elevator operation. To ensure accurate monitoring of the elevator door rail operation status, this application first deploys sensor arrays at key locations on the elevator door rail, such as the door rail track and motor drive components, to capture vibration and current signals of the elevator door rail in real time. The vibration signals are acquired through accelerometers, which can detect minute vibration changes in the door rail system. These vibrations usually originate from changes in the operating status of the elevator door rail, wear, or jamming. The current signals are monitored by current sensors to detect changes in the motor's drive current. The current signals reflect the motor's load, start-up, acceleration, and other operating states. If a fault occurs in the elevator door rail system, changes in the motor's load will cause fluctuations in the current, thus providing important clues for fault diagnosis. The sensor array setup ensures comprehensive signal acquisition, covering the operating status of all components of the elevator door rail.

[0014] The acquired vibration and current signals undergo signal preprocessing to remove high-frequency noise and environmental interference, ensuring signal validity. For example, the raw vibration signal data may contain high-frequency noise components; low-pass filters are used during preprocessing to remove these interferences. The preprocessed signal data is then normalized to eliminate differences in amplitude between different signals, making the signals more comparable in subsequent analyses. These preprocessed signal data form the initial multi-source signal set, which becomes the basis for subsequent signal analysis and fault diagnosis. The generation of the initial multi-source signal set is the starting point of the entire elevator operation status monitoring system, providing necessary data support for subsequent steps such as time synchronization processing, signal feature extraction, and fault mode identification.

[0015] Step S2: Perform time synchronization processing on the initial multi-source signal set to obtain a time synchronization signal sequence. Extract features of various signals from the time synchronization signal sequence to generate a signal feature segment set.

[0016] In step S2, time synchronization processing of the initial multi-source signal set to obtain a time synchronization signal sequence includes: processing the time series segments of vibration and current signals using timestamp alignment technology based on the initial multi-source signal set; if the timestamp deviation exceeds a preset threshold, adjusting the sampling frequency through signal synchronization verification to eliminate the deviation; performing data interpolation on the time series segments of vibration and current signals in conjunction with signal abrupt change intervals to generate continuous time series data; determining the time synchronization signal sequence based on the interpolated time series data; and performing consistency verification on the time synchronization signal sequence to ensure the alignment accuracy of vibration and current signals on the time axis.

[0017] Specifically, in the monitoring of elevator system operation status, due to factors such as the sampling frequency and time delay of different sensors, these signals often cannot be synchronized, leading to deviations in subsequent data analysis and fault diagnosis. To ensure that different signals can be effectively analyzed together, time synchronization processing must be performed on the different signals. Time synchronization processing eliminates time deviations between signals, ensuring that different signals are compared and analyzed at the same point in time. Specifically, it executes the following sub-steps: The time series of vibration and current signals are aligned according to the timestamp of each data point in the initial multi-source signal set. However, in practical applications, timestamp deviations may occur due to factors such as environmental noise and differences in sensor performance. If the timestamp deviation exceeds a preset threshold, such as 0.5 seconds, signal synchronization verification is triggered. By calculating the correlation coefficient between the vibration and current signals, the degree of matching between the two at critical moments, such as when signal peaks occur, can be determined. The correlation coefficient is a statistic used to measure the strength and direction of the linear relationship between two signals. The commonly used correlation coefficient is the Pearson correlation coefficient, whose value ranges from -1 to 1. If the correlation coefficient is lower than a preset threshold, such as 0.8, it indicates poor synchronization between signals. In this case, the sampling frequency needs to be adjusted to eliminate the deviation. For example, the sampling frequency of the vibration signal can be increased from 100Hz to 120Hz to keep it consistent with the current signal and ensure that their time series are more accurately synchronized.

[0018] After adjusting the sampling frequency, data interpolation is needed to ensure signal continuity and integrity during abrupt signal changes. For example, in an elevator door track system, if vibration signals are missing at certain times, it may be due to sensor malfunction or environmental interference. Linear interpolation inserts estimated values ​​between missing data points, allowing the signal to transition smoothly within abrupt changes and maintaining signal continuity. Interpolation generates a continuous time-series data, ensuring perfect alignment of the vibration and current signals on the time axis. Finally, the interpolated and synchronized signal sequence enters the consistency verification stage. This stage calculates the correlation coefficient of the verification sequence. If the correlation coefficient between the vibration and current signals is greater than 0.9, it indicates high alignment accuracy on the time axis, meeting the signal synchronization requirements for further analysis and feature extraction. Through this precise synchronization and verification, we ensure the accuracy of subsequent steps such as feature extraction and fault mode recognition.

[0019] Step S2, which involves extracting features of various signals from the time-synchronized signal sequence to generate a set of signal feature segments, includes: extracting time-domain features of vibration signals and frequency-domain features of current signals from the time-synchronized signal sequence; identifying and locating abnormal trigger points based on the time-domain and frequency-domain features; segmenting the time-synchronized signal sequence based on the abnormal trigger points to obtain multiple signal segments; calculating the correlation coefficient between vibration signals and current signals in each signal segment; identifying the proportion of abnormal components in each signal segment based on the correlation coefficient; classifying the signal segments according to the proportion of abnormal components; generating a set of signal feature segments based on the classification results; standardizing the feature data in the set of signal feature segments; analyzing the correlation between different signal segments based on the standardized feature data; and adjusting the structure of the set of signal feature segments based on the correlation analysis results to obtain an optimized set of signal feature segments.

[0020] Step S3: Analyze the interaction patterns between signals based on the signal feature segment set, determine the interaction anomaly feature group, fuse the interaction anomaly feature group with the elevator door rail historical fault data, and generate a corrected door rail operating status description.

[0021] In step S3, analyzing the interaction patterns between signals based on the set of signal feature segments and determining the interaction anomaly feature group includes: analyzing the interaction patterns between vibration signals and current signals for the set of signal feature segments, and quantifying the interaction patterns using deviation distribution interval mapping technology; constructing interaction feature vectors based on the quantization results, and determining the proportion and deviation range of abnormal components based on the interaction feature vectors to determine the interaction anomaly feature group; prioritizing the abnormal components in the interaction anomaly feature group, filtering key anomaly features based on the sorting results, and adjusting the interaction anomaly feature group based on the filtered key anomaly features.

[0022] Specifically, to improve the accuracy of fault diagnosis in elevator door rail systems, traditional methods typically analyze vibration or current signals separately, lacking in-depth analysis of the interaction between these two types of signals. To address this issue, a method is proposed that analyzes the interaction patterns between signals based on sets of signal feature segments to determine groups of interactive abnormal features. By analyzing the relationship between vibration and current signals, potential fault modes can be identified more accurately, thereby improving the accuracy and efficiency of diagnosis.

[0023] Specifically, the interaction pattern between vibration and current signals is first analyzed based on the set of signal feature segments collected from the elevator door track system. The interaction pattern refers to the correspondence between the time-domain characteristics of the vibration signal and the frequency-domain characteristics of the current signal, such as the synchronicity between vibration peaks and current fluctuations. To accurately analyze the interaction between these signals, this application employs a deviation distribution interval mapping technique to quantify the interaction pattern. Specifically, the deviation distribution interval mapping technique maps the deviation values ​​in the interaction pattern to a preset distribution interval. For example, the deviation values ​​are divided into three intervals: low, medium, and high, each corresponding to a different quantization coefficient. The overall pattern is quantified by calculating the distribution density of the deviation in each interval. First, the deviation values ​​between the vibration and current signals in the interaction pattern are calculated. For example, the deviation value is obtained by the absolute value of the difference between their characteristics. The deviation values ​​are then mapped to distribution intervals. For example, the interval boundaries are set as 0 to 0.5 for low deviation, 0.5 to 1 for medium deviation, and above 1 for high deviation. A quantization index is calculated based on the mapping results. For example, the quantization index is the weighted average of the deviation values ​​in each interval. For example, in an elevator door track system, when the vibration signal shows high-frequency oscillations while the current signal shows low-frequency fluctuations, deviation distribution interval mapping can quantify the high deviation ratio up to 60%, which helps identify potential jamming risks. This quantification can improve the accuracy of anomaly detection and reduce false positives. In this case, deviation distribution interval mapping can quantify these changes and determine their degree of influence, thus providing a basis for subsequent anomaly feature identification. Based on the quantization results, an interactive feature vector is constructed. This application uses interactive feature vector construction technology to generate interactive feature vectors. Specifically, the interactive feature vector construction technology converts the quantized interactive mode into a vector form to obtain, for example, an interactive feature vector. The vector components include dimensions such as the proportion of abnormal components and the deviation range. To further analyze these features, the proportion of abnormal components and the deviation range are determined based on the results of the interactive feature vector, and the interactive anomaly feature group is determined. The specific implementation process is as follows.

[0024] The process of determining interactive anomaly feature groups includes: performing multi-dimensional interactive analysis on a set of signal feature segments and constructing an interactive feature matrix based on the results of the multi-dimensional interactive analysis; calculating the proportion of abnormal components based on the interactive feature matrix; dividing the proportion of abnormal components into intervals using a deviation distribution interval mapping technique; determining the deviation range based on the interval division results; generating interactive anomaly feature groups based on the deviation range and the proportion of abnormal components; performing feature clustering on the interactive anomaly feature groups; analyzing the distribution pattern of abnormal features based on the clustering results; and adjusting the weight allocation of the interactive anomaly feature groups based on the distribution pattern.

[0025] Specifically, the detailed implementation process for determining the interactive anomaly feature group includes the following steps: First, a multi-dimensional interactive analysis is performed on the set of signal feature segments. This means that it is necessary not only to consider the simple relationship between vibration signals and current signals, but also to analyze their interaction in multiple dimensions. For example, the time-domain characteristics of vibration signals, such as peak value, mean, and variance, and the frequency-domain characteristics of current signals, such as dominant frequency components and spectral distribution, are two different types of features. Interactive analysis can reveal the complex relationship between these two signals under different operating conditions. By constructing these analysis results into an interactive feature matrix, which contains the interaction between vibration signals and current signals at different time points and in different feature dimensions, for example, a certain element of the matrix may represent the correlation strength between the peak value of the vibration signal and the dominant frequency component of the current signal at a certain time point. The correlation strength is usually represented by the Pearson correlation coefficient. This matrix provides a comprehensive data foundation for subsequent fault diagnosis. The proportion of anomalous components in each signal segment is calculated using the interaction feature matrix. Anomalous components refer to the parts of the signal that deviate from the normal operating state. Assuming the vibration signal and current signal are [1.2, 1.5, 3.0, 1.3, 1.7, 2.1] and [0.8, 1.1, 2.3, 1.0, 1.2, 1.5] respectively, the correlation strength between the vibration signal and the current signal is first extracted from the interaction feature matrix. Assuming the Pearson correlation coefficient r is 0.4, this indicates a weak linear correlation between the vibration signal and the current signal. A threshold of 0.5 is set. If the correlation coefficient is lower than 0.5, it indicates that the signal segment may contain anomalies. The abnormal points in the abnormal signal segments are further identified. Under normal circumstances, the vibration amplitude should not exceed about 2.0g and the current should not exceed about 1.5A. The third data point of the vibration signal, 3.0g, is significantly deviated from the normal range, and the third data point of the current signal, 2.3A, is also higher than the normal value. Therefore, these two points can be considered to be abnormal trigger points. There are a total of 6 sampling points in the entire signal sequence, so the abnormal points account for 1 / 6, and the proportion of abnormal components is 16.7%.

[0026] After obtaining the proportion of anomalous components, a deviation distribution interval mapping technique is used to divide the proportion of anomalous components into intervals. Specifically, this technique maps the proportion of anomalous components to preset intervals to quantify the severity of the anomalies. For example, the proportion of anomalous components can be divided into several intervals: low deviation interval (0%-10%): indicating that the signal change is not significant and the proportion of anomalous components is low; medium deviation interval (10%-30%): indicating that the signal has some anomalies, but not enough to affect the normal operation of the system; high deviation interval (above 30%): indicating that the signal anomaly is obvious and may correspond to a more serious fault mode. If the proportion of anomalous components is 16.7%, it will be mapped to the medium deviation interval, indicating that the degree of signal anomaly is moderate. Next, based on these deviation intervals and anomalous component proportions, interactive anomaly feature groups are generated. For example, if the proportion of anomalous components in a certain signal segment is higher than 30%, it is classified as a high deviation interval.

[0027] Subsequently, feature clustering is performed on all generated interactive anomaly feature groups. Cluster analysis allows for the grouping of segments with similar features into one category, thus helping to identify common characteristics among different anomaly signal segments and discover their potential corresponding fault modes. For example, if the vibration peaks and current fluctuations of multiple signal segments increase simultaneously at the same time point, and their anomalous component proportions are high, these segments are likely to indicate a fault in the elevator door rail system, such as jamming. This clustering analysis can summarize the potential causes of the fault and provide more evidence for fault diagnosis. Analyzing the distribution patterns of anomaly features can help identify the distribution trends of fault modes. For example, during the operation of the elevator door rail, certain anomaly features may frequently appear within a short period, which may indicate that the elevator is susceptible to a certain fault mode under specific operating conditions. Based on these distribution patterns, the weight allocation of interactive anomaly feature groups can be adjusted, assigning higher weights to features with greater influence and lower weights to features with less influence. For example, when the elevator door rail system operates for more than 500 hours, certain abnormal signals, such as abnormal current fluctuations, become more frequent. The weight of these signals may need to be increased. Through this dynamic weight adjustment, it is possible to better adapt to changes in different operating times and environmental conditions, thereby improving the diagnostic capability for complex faults in the elevator door rail system.

[0028] After determining the interactive anomaly feature group, the anomaly components within the group are sorted, prioritizing features that have the greatest impact on elevator door rail faults. For example, if the correlation between vibration and current signals is high and the proportion of anomaly components is large, these are given higher priority. This step aims to ensure that the most critical anomaly features are processed first, thereby improving fault diagnosis efficiency. Following sorting, the interactive anomaly feature group is adjusted based on the selected key anomaly features. This adjustment process ensures that only features that significantly improve fault diagnosis accuracy are retained. By prioritizing these anomaly features, the elevator fault diagnosis system can identify door rail problems earlier, allowing for timely maintenance. Subsequently, historical elevator door rail fault data is acquired and fused with the adjusted interactive anomaly feature group to generate a corrected description of the door rail operating status. The detailed implementation process will be described later. Through dynamic updates to the weight distribution and optimization of the interactive feature group, accurate description of the elevator operating status and efficient fault diagnosis can be achieved.

[0029] Step S3, generating the corrected door track operating status description, includes: fusing the interactive anomaly feature group with historical elevator door track fault data, calculating the rate of change of the current anomaly component proportion relative to the historical fault data, dynamically updating the weight distribution of each anomaly component based on the rate of change, using the updated weight distribution to prioritize and weight the anomaly components in the interactive anomaly feature group, forming a weighted feature vector; using the weighted feature vector as input, generating the current door track operating status description through a preset status description model, performing a correction iteration convergence judgment on the current door track operating status description to obtain a preliminary corrected door track operating status description; performing a consistency check on the preliminary corrected door track operating status description, adjusting the parameters of the weight distribution based on the check results, and using the adjusted parameters to optimize the status description, generating the final corrected door track operating status description.

[0030] Specifically, the process of generating the corrected description of the elevator door rail's operating status is as follows: First, the interactive anomaly feature group is fused with the historical fault data of the elevator door rail. This historical fault data includes past fault records of the elevator door rail system, such as instances of vibration anomalies and localized jamming. By comparing the characteristics of the current signal with the historical data, the rate of change of the current anomaly component proportion relative to the historical fault data is calculated. For example, if the current vibration signal's anomaly component proportion is 15%, while this proportion is typically 10% in the historical data, then the rate of change is 50%. Based on this rate of change, the weight distribution of each anomaly component is dynamically updated. For instance, if the anomaly component of the vibration signal in the elevator system's historical data is highly correlated with faults, and an increase in the current anomaly component proportion will increase the weight of the vibration signal, for example, from 0.6 to 0.7, to better reflect the current fault mode. Next, the updated weight distribution is used to prioritize and weight the abnormal components in the interactive abnormal feature group. This step is to ensure that the system can focus on the most critical abnormal features. For example, if the frequency fluctuation of the current signal changes significantly during a fault, while the change of the vibration signal is relatively small, the weight of the current signal may be given priority, reflecting its greater impact on the current fault mode.

[0031] Then, the weighted feature vector is input into a pre-defined state description model, which outputs a risk level related to the proportion of abnormal components. This state description model can be a regression model, a machine learning model, or a statistical model, depending on the complexity of the elevator system and the characteristics of the data. It uses the weighted feature vector as input, which contains information about the abnormal components of the current elevator door track system. Based on these feature vectors, it calculates the elevator's risk level, which is generated according to the correlation between the proportion of abnormal components and the failure mode. For example, if the proportion of abnormal components is high, such as exceeding 30%, the model may classify the elevator as high-risk; if the proportion is low, such as below 10%, it will be classified as low-risk. After generating the initial description, the model iteratively corrects itself by calculating the difference between the current state and the target state until the difference is below a preset threshold, such as 0.01, indicating that the model has converged to the optimal state. This method can accurately reflect the operating status of the elevator system, detect faults in a timely manner, and ensure the safe operation of the system. After the initial correction and generation of the state description, a consistency check is performed to ensure that the current state description is consistent with historical data. This check compares the proportion of abnormal components with the range of changes in historical records to determine consistency. For example, if the proportion of abnormal components in historical data fluctuates within 5%, while the proportion in the current description changes within ±5%, then consistency is considered acceptable. If inconsistent, the weight distribution parameters are adjusted, for example, the weight of the vibration signal is adjusted from 0.7 to 0.75 to improve accuracy. Finally, the adjusted state description is optimized to further improve the accuracy of representing the proportion of abnormal components. For example, starting from the initial description "abnormal component 10%", it converges to "abnormal component 9.5%" after 15 iterations. Through the optimized description, the elevator system can more accurately reflect the fault state, helping maintenance personnel to quickly take necessary measures. The optimized door rail operating state description is saved to a database for subsequent analysis.

[0032] Step S4: If the abnormal component in the corrected door rail operating status description exceeds the preset threshold, the abnormal data is optimized to obtain an optimized status description. The elevator fault mode is matched according to the optimized status description to generate a fault diagnosis report for the elevator door rail system.

[0033] The optimized state description obtained in step S4 includes: if the abnormal components in the corrected gate rail operation state description exceed a preset threshold, the abnormal data is divided into a high-frequency interference layer and a low-frequency offset layer according to the signal source and frequency domain characteristics; the abnormal data within the preset deviation range boundary is optimized based on the different characteristics of the high-frequency interference layer and the low-frequency offset layer; a preliminary optimized state description is generated based on the optimization results; error analysis is performed on the preliminary optimized state description; the correction cycle parameter is adjusted based on the analysis results; the iterative correction process is started with the adjusted correction cycle parameter and the optimized abnormal data until the residual converges below the preset threshold, and the final optimized state description is output.

[0034] Specifically, the corrected door rail operating status description includes the proportion of abnormal components and the risk level. If the abnormal components in the corrected door rail operating status description exceed a preset threshold, such as 0.15, it means that the operating status of the elevator door rail system has deviated significantly, and the elevator may be in a medium to high-risk state with potential malfunctions. Further optimization and fault diagnosis are required to ensure the safe operation of the elevator. This application classifies abnormal data using anomaly data layering processing technology. Specifically, based on the source of the abnormal data, such as vibration signals or current signals, it is layered into a high-frequency interference layer and a low-frequency offset layer. The high-frequency interference layer is usually related to sudden noise or short-term fluctuations in the elevator door rail system. For example, during the door rail opening and closing process, the instantaneous vibration of the elevator door rail may cause high-frequency noise. The low-frequency offset layer is related to long-term wear, aging, or environmental changes in the elevator door rail system, and usually manifests as a relatively stable but persistent signal deviation. Through this layering, the classified data can be directly used for subsequent processing, forming clear abnormal data groups. Based on the different characteristics of the high-frequency interference layer and the low-frequency offset layer, the abnormal data within the preset deviation range boundary is optimized, and a preliminary optimization state description is generated based on the optimization results. The specific implementation process is as follows.

[0035] The preliminary optimization state description generated based on the optimization results includes: based on the frequency domain characteristics of abnormal data, matching physical parameter models related to instantaneous impact for the high-frequency interference layer and physical parameter models related to long-term wear or deformation for the low-frequency offset layer from a pre-established elevator door rail parameterized dynamic model library; using abnormal data within the preset deviation range boundary as the target signal, adjusting the key parameters in the physical parameter model in reverse through an iterative algorithm to minimize the residual between the simulation signal output by the model and the target signal; when the residual converges to a preset threshold, using the optimal parameter value of the physical parameter model at this time, along with the fault mode and confidence level corresponding to the optimal parameter value in the historical fault database, together as the preliminary optimization state description.

[0036] Specifically, based on the frequency domain characteristics of abnormal data, suitable physical parameter models are selected from a pre-established dynamic model library of elevator door rail systems for matching. Each physical parameter model in the library corresponds to a specific type of fault mode in the elevator door rail system and includes related key parameters. For example, the high-frequency interference layer is usually related to short-term sudden noise or instantaneous impact in the elevator door rail system; therefore, the corresponding physical parameter model may focus on the characteristics of instantaneous impact, such as the vibration mode when the elevator door rail starts or stops. For the low-frequency offset layer, it is usually related to long-term wear, aging, or deformation; in this case, the selected physical parameter model will focus on the long-term changes in the elevator door rail system, such as motor load fluctuations or increased vibration caused by wear. Figure 2The diagram shows a parameterized dynamic model of an elevator door rail system. This model includes key physical parameters such as mass *m*, stiffness *k*, damping *c*, and friction coefficient *μ*. These parameters reflect the inertia, elasticity, energy consumption characteristics, and contact surface friction state of the door rail system, respectively. Once a suitable physical parameter model is selected, abnormal data within a preset deviation range is used as the target signal. An iterative algorithm is used to adjust key parameters in the physical parameter model, such as stiffness *k*, friction coefficient *μ*, mass *m*, and damping *c*, to minimize the residual between the simulated signal and the target signal. For example, in the high-frequency interference layer of the door rail, if there is a sudden change in the vibration signal, such as a short-term impact during door opening and closing, the impact response parameters in the physical parameter model, such as the friction coefficient or elastic modulus, are adjusted to better match the change in the simulated signal with the abrupt change characteristics of the target signal. Similarly, in the low-frequency offset layer, if the current signal exhibits continuous fluctuations, such as wear caused by long-term operation of the door rail, the long-term variation parameters in the model, such as the damping coefficient or stiffness parameter, are adjusted to better fit the long-term trend of the target signal. When the residual of the model adjustment reaches a preset threshold, such as 0.01, it indicates that the simulated signal is highly consistent with the target signal. At this point, the model parameters are considered optimal. The optimal parameter values ​​are then matched with its corresponding historical fault database to find fault types similar to the current signal pattern. For example, if the optimal parameters match the characteristics of local jamming faults in the historical database, the confidence level of that fault mode, such as 90%, is extracted, representing the probability of that fault mode occurring. Ultimately, the preliminary optimized state description will include: the optimal physical parameter model and its key parameter values, the fault mode corresponding to the abnormal data, and the confidence level of the fault mode. This information, combined, forms a preliminary description of the current operating state of the elevator door rail system, providing strong support for subsequent fault diagnosis and risk assessment. By combining optimized anomaly data with the physical dynamics model of the elevator door rails, the actual fault modes in the elevator system can be reflected more accurately. The key parameters of the physical parameter model are adjusted using iterative algorithms, and potential fault modes, such as door rail jamming or wear, can be identified at an early stage by comparing them with historical fault data. By calculating the confidence level of the fault mode, the probability assessment of the occurrence of the fault is provided. This makes elevator fault diagnosis more reliable, provides early warning of potential problems, reduces the risk of sudden failures, and improves the operating efficiency of the elevator system.

[0037] After generating the preliminary optimized state description, its accuracy needs to be further verified to avoid judgment bias caused by model fitting errors, environmental noise, or sensor fluctuations. Therefore, error analysis is first performed on the preliminary optimized state description. The core of the error analysis is to compare the optimal parameter values ​​in the preliminary description and the generated simulation signal with the original time synchronization signal sequence. The reliability of the optimized description is quantified by calculating the residual between the two. The residual can be calculated using the root mean square error. For example, if the simulation signal is [1.2, 1.5, 1.8] in a certain period and the actual synchronization signal is [1.1, 1.6, 1.9], then the residual is approximately 0.1. When the residual exceeds a preset threshold, it indicates that the preliminary optimized description still needs further correction. Based on the error analysis results, the correction cycle parameter is adjusted. The correction cycle parameter refers to the step size and update frequency of the system during iterative optimization. For example, when the residual is large, it indicates that the signal fluctuates rapidly or there are obvious abnormal changes. In this case, the correction cycle is shortened from 10 seconds to 8 seconds to improve the model's response speed. Conversely, the cycle is appropriately extended to avoid overcorrection. The adjusted correction cycle parameter can adapt to the gate rail's operating state, making the iterative process more closely match the actual working conditions. After the parameter adjustment is completed, the iterative correction process is started using the optimized abnormal data and the new correction cycle parameter. The iterative process updates the key parameters in the physical parameter model multiple times, making the simulated signal gradually approach the actual signal. For example, the residual of the first iteration is 0.12, the second decreases to 0.06, and the third decreases to 0.03. When the residual converges to below the preset threshold, such as 0.01, the iteration terminates. At this time, the final optimized state description is output. The final optimized state description usually includes: the system physical parameters after residual convergence, such as the friction coefficient 0.45 and the damping coefficient 0.12; the proportion of abnormal components (such as 12%); and the system risk level (low, medium, high). Next, the system matches historical fault modes with the optimized state description. For example, when the friction coefficient increases significantly and the low-frequency offset layer increases abnormally, the similarity with the "local jamming of the door rail" mode in the historical database is 0.88, and this fault mode is identified as the most likely anomaly. Finally, by combining factors such as similarity, the proportion of abnormal components, and the running time of the door rail, a fault diagnosis report for the elevator door rail system is generated. The report includes the fault mode category, risk level, potential causes (such as long-term wear leading to an increased damping coefficient), recommended maintenance measures (such as lubrication, rail adjustment, or component replacement), and includes risk trend prediction to help maintenance personnel develop more effective maintenance plans. Through residual analysis, adaptive adjustment of the correction cycle, and iterative correction, the final optimized state description is made closer to the actual physical state, thereby significantly reducing the misdiagnosis rate. By matching the optimized physical parameters with historical fault modes, the cause of the fault can be accurately located, improving the confidence of fault identification. The final diagnostic report has a clear structure and accurate risk assessment, providing a highly reliable basis for elevator maintenance decisions and improving the safety of door rail operation.

[0038] The step S4, generating a fault diagnosis report for the elevator door rail system, includes: based on the optimized state description, combined with the local wear degree of the door rail and historical operating data, matching preliminary fault modes using state description error correction technology, analyzing the impact of door rail operating time on abnormal components, and generating a fault diagnosis report for the elevator door rail system by integrating the preliminary fault mode and operating time impact analysis results; conducting a risk level assessment on the fault diagnosis report, determining the door rail operating risk level based on the assessment results; classifying fault modes according to risk levels, prioritizing the classified fault modes, and forming the final fault diagnosis report.

[0039] Specifically, the process of generating a fault diagnosis report for an elevator door rail system involves multiple steps. The aim is to accurately diagnose the fault status of the elevator door rail system through a comprehensive analysis of optimized state descriptions, local wear levels of the door rails, and historical operating data, providing maintenance personnel with a reliable basis for decision-making. First, based on the optimized state description, combined with the local wear levels of the door rails and historical operating data, fault modes are matched. Local wear levels can be obtained through methods such as laser scanning; for example, a wear depth of 0.5 mm obtained through laser scanning. Combined with historical elevator operating data, such as a cumulative operating time of 5000 hours, this information forms a comprehensive dataset, serving as the basis for subsequent fault mode matching. Next, state description error correction techniques are used to further correct deviations in the optimized state description. Using error correction techniques such as least squares, the difference between the optimized state description and historical fault data is calculated and iteratively adjusted until the error is minimized. Assuming the proportion of abnormal components in the optimized state description is 15%, after correction, it may match the "local jamming fault" mode, and the risk of this mode is assessed as medium. This reduces misjudgments caused by model bias and provides more accurate fault mode matching. After matching the fault modes, the impact of elevator operating time on abnormal components is further analyzed. For example, using least squares linear regression to analyze the relationship between runtime and anomaly components, if the regression analysis shows that the anomaly component increases by 2% for every 1000-hour increase in runtime, then the failure mode assessment can be adjusted based on this regression coefficient. For instance, if the runtime is 6000 hours and the anomaly component reaches 15%, then according to the regression model, the anomaly component may rise to 17% in the next 1000 hours. This information will help in long-term failure prediction and maintenance planning for elevator door rail systems.

[0040] Subsequently, the fault diagnosis report will assess the risk level based on the optimized fault modes and regression analysis results. For example, if the anomaly component exceeds 10%, it will be classified as medium risk. The assessed risk level can be: low risk (anomaly component less than 10%), medium risk (anomaly component between 10% and 30%), or high risk (anomaly component exceeding 30%). Based on the risk level, the fault modes will be categorized; for example, "jamming fault" might be classified as medium risk. The categorized fault modes will then be prioritized based on risk level and frequency of occurrence, using a weighted ranking method. Higher-priority fault modes, such as high-risk modes, will be ranked first. For instance, if jamming faults are frequent in the elevator door track system and have a significant impact on operation, they will be prioritized. The final fault diagnosis report will adjust its content structure based on the ranking results. High-priority fault modes will be placed at the beginning of the report, providing detailed fault descriptions and handling recommendations, such as immediate shutdown or regular inspection. This report will help maintenance personnel take preventative measures as early as possible to reduce the risk of fault occurrence. Finally, all fault diagnosis reports will be stored in a cloud database for easy retrieval and reference later.

[0041] By combining optimized condition descriptions, wear levels, and historical operating data, the generated fault diagnosis report provides accurate fault modes, risk assessments, and handling recommendations. The application of error correction techniques and regression analysis effectively improves the accuracy of fault mode matching, while prioritization ensures that high-risk faults are addressed first. This method can provide early warnings of potential faults, helping maintenance personnel develop reasonable maintenance plans, reduce equipment failure rates, extend the service life of the elevator door rail system, and thus improve the safety and reliability of the elevator.

[0042] In summary, this application presents a multi-sensor-integrated elevator operation status monitoring method, which achieves accurate perception and intelligent diagnosis of elevator door rail operation status. First, based on the features extracted from multi-sensor signals, a pre-defined state description model is used to generate a description of the current door rail operation status, completing a preliminary quantitative assessment of the system status. Subsequently, this preliminary description is optimized into a final corrected door rail operation status description through fusion with historical fault data and iterative convergence judgment. This step significantly improves the stability of the status assessment and its consistency with historical experience. If the abnormal components in this corrected description exceed a preset threshold, a deeper analysis mechanism is triggered. This involves inverting and fitting the abnormal data with a parameterized physical dynamics model to generate a preliminary optimized state description, thereby mapping the fault phenomenon to specific physical parameter changes and achieving preliminary fault root cause localization. This preliminary description then undergoes an iterative correction process based on residual error analysis and adaptive correction cycle, ultimately outputting a final optimized state description containing accurate physical parameters and high-confidence fault modes. Finally, by integrating this optimized state description, real-time wear data, and runtime prediction model, a multi-dimensional risk assessment and prioritization are performed to generate a final fault diagnosis report that can directly guide maintenance decisions. This series of technical steps, from raw signal acquisition to fault diagnosis report generation, through gradual optimization and correction, has enabled the fault diagnosis of the elevator door rail system to evolve from a preliminary, rough description to a precise and efficient system state assessment. To visually verify the continuous improvement in diagnostic accuracy and information completeness of this method, such as... Figure 3 As shown, this is a graph illustrating the performance gains of the iterative optimization metrics. This graph quantitatively demonstrates the performance gains across the five core diagnostic metrics as the process evolves from the "current state description" to the "final optimized description".

[0043] The above describes a method for monitoring elevator operation status by fusing multiple sensors according to an embodiment of this application. The following describes a system for monitoring elevator operation status by fusing multiple sensors according to an embodiment of this application. Please refer to [link / reference]. Figure 4 One embodiment of the elevator operation status monitoring system integrating multiple sensors in this application includes: The acquisition module is used to acquire vibration and current signals of the elevator door rail and generate an initial multi-source signal set; The synchronization module is used to perform time synchronization processing on the initial multi-source signal set to obtain a time-synchronized signal sequence, extract the features of various signals from the time-synchronized signal sequence, and generate a set of signal feature segments. The analysis module is used to analyze the interaction patterns between signals based on the set of signal feature segments, determine the interaction anomaly feature groups, fuse the interaction anomaly feature groups with historical fault data of elevator door rails, and generate a corrected description of the door rail operating status. The diagnostic module is used to optimize the abnormal data if the abnormal components in the corrected door rail operating status description exceed a preset threshold, obtain an optimized status description, match the elevator fault mode according to the optimized status description, and generate a fault diagnosis report for the elevator door rail system.

[0044] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0045] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0046] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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 this application.

Claims

1. A method for monitoring the operating status of an elevator by integrating multiple sensors, characterized in that, The method includes: Step S1: Acquire vibration and current signals of the elevator door rails to generate an initial multi-source signal set; Step S2: Perform time synchronization processing on the initial multi-source signal set to obtain a time synchronization signal sequence, extract features of various signals from the time synchronization signal sequence, and generate a signal feature segment set; Step S3: Analyze the interaction patterns between signals based on the set of signal feature segments, determine the interaction anomaly feature group, fuse the interaction anomaly feature group with the elevator door rail historical fault data, and generate a corrected description of the door rail operating status. Step S4: If the abnormal component in the corrected door rail operating status description exceeds the preset threshold, the abnormal data is optimized to obtain an optimized status description. Based on the optimized status description, the elevator fault mode is matched to generate a fault diagnosis report for the elevator door rail system.

2. The method according to claim 1, characterized in that, Step S2 involves performing time synchronization processing on the initial multi-source signal set to obtain a time synchronization signal sequence, including: Based on the initial multi-source signal set, time series segments of the vibration signal and current signal are processed using timestamp alignment technology. If the timestamp deviation exceeds a preset threshold, the sampling frequency is adjusted to eliminate the deviation through signal synchronization verification. Data interpolation is performed on the time series segments of the vibration signal and current signal in conjunction with signal abrupt change intervals to generate continuous time series data. The time synchronization signal sequence is determined based on the interpolated time series data. Consistency verification is performed on the time synchronization signal sequence to ensure the alignment accuracy of the vibration signal and current signal on the time axis.

3. The method according to claim 2, characterized in that, Step S2 involves extracting features of various signals from the time synchronization signal sequence to generate a set of signal feature segments, including: The time-domain features of the vibration signal and the frequency-domain features of the current signal are extracted from the time-synchronization signal sequence. Based on the time-domain features and the frequency-domain features, abnormal trigger points are identified and located. The time-synchronization signal sequence is segmented based on the abnormal trigger points to obtain multiple signal segments. The correlation coefficient between the vibration signal and the current signal in each signal segment is calculated. Based on the correlation coefficient, the proportion of abnormal components in each signal segment is identified. The signal segments are classified according to the proportion of the abnormal components. A set of signal feature segments is generated based on the classification results. The feature data in the set of signal feature segments is standardized. Based on the standardized feature data, the correlation between different signal segments is analyzed. Based on the correlation analysis results, the structure of the set of signal feature segments is adjusted to obtain an optimized set of signal feature segments.

4. The method according to claim 1, characterized in that, In step S3, the interaction patterns between signals are analyzed based on the set of signal feature segments, and the abnormal interaction feature groups are determined to include: For the set of signal feature segments, the interaction pattern between vibration signal and current signal is analyzed, and the interaction pattern is quantized using deviation distribution interval mapping technology; an interaction feature vector is constructed based on the quantization result, and the proportion and deviation range of abnormal components are determined according to the interaction feature vector to determine the interaction abnormal feature group; the abnormal components in the interaction abnormal feature group are prioritized, and key abnormal features are screened according to the ranking result; the interaction abnormal feature group is adjusted based on the screened key abnormal features.

5. The method according to claim 4, characterized in that, The group of identified interaction anomaly features includes: A multidimensional interactive analysis is performed on the set of signal feature segments, and an interactive feature matrix is ​​constructed based on the results of the multidimensional interactive analysis. The proportion of abnormal components is calculated based on the interactive feature matrix. The proportion of abnormal components is divided into intervals using a deviation distribution interval mapping technique. The deviation range is determined based on the interval division results. The interactive abnormal feature group is generated based on the deviation range and the proportion of abnormal components. The interactive abnormal feature group is clustered. The distribution pattern of abnormal features is analyzed based on the clustering results. The weight allocation of the interactive abnormal feature group is adjusted based on the distribution pattern.

6. The method according to claim 4, characterized in that, Step S3 generates the corrected description of the gate rail's operating status, including: The interactive anomaly feature group is fused with the elevator door track historical fault data. The rate of change of the current anomaly component ratio relative to the historical fault data is calculated. The weight distribution of each anomaly component is dynamically updated based on the rate of change. The updated weight distribution is used to prioritize and adjust the anomaly components in the interactive anomaly feature group to form a weighted feature vector. Using the weighted feature vector as input, a current gate rail operation state description is generated through a preset state description model. The current gate rail operation state description is then corrected and iterated to converge, resulting in a preliminary corrected gate rail operation state description. A consistency check is performed on the preliminary corrected gate rail operation state description. Based on the check results, the parameters of the weight distribution are adjusted, and the state description is optimized using the adjusted parameters to generate the final corrected gate rail operation state description.

7. The method according to claim 1, characterized in that, The optimized state description obtained in step S4 includes: If the abnormal components in the corrected gate rail operation status description exceed the preset threshold, the abnormal data will be divided into a high-frequency interference layer and a low-frequency offset layer according to the signal source and frequency domain characteristics of the abnormal data. Based on the different characteristics of the high-frequency interference layer and the low-frequency offset layer, the abnormal data within the preset deviation range boundary is optimized. A preliminary optimized state description is generated based on the optimization results. Error analysis is performed on the preliminary optimized state description, and the correction cycle parameter is adjusted based on the analysis results. An iterative correction process is started using the adjusted correction cycle parameter and the optimized abnormal data until the residual converges below the preset threshold, and the final optimized state description is output.

8. The method according to claim 7, characterized in that, The preliminary optimization state description generated based on the optimization results includes: Based on the frequency domain characteristics of the abnormal data, from the pre-established elevator door rail parameterized dynamic model library, a physical parameter model related to instantaneous impact is matched for the high-frequency interference layer, and a physical parameter model related to long-term wear or deformation is matched for the low-frequency offset layer. Using abnormal data within the preset deviation range boundary as the target signal, the key parameters in the physical parameter model are adjusted in reverse through an iterative algorithm to minimize the residual between the simulation signal output by the model and the target signal. When the residual converges to a preset threshold, the optimal parameter value of the physical parameter model at this time, along with the corresponding fault mode and confidence level in the historical fault database, are used together as the preliminary optimization state description.

9. The method according to claim 1, characterized in that, Step S4 generates a fault diagnosis report for the elevator door rail system, including: Based on the optimized state description, combined with the local wear degree of the door rail and historical operating data, a preliminary fault mode is matched using state description error correction technology. The impact of door rail operating time on abnormal components is analyzed. By integrating the preliminary fault mode and operating time impact analysis results, a fault diagnosis report for the elevator door rail system is generated. The fault diagnosis report is then subjected to a risk level assessment, and the door rail operating risk level is determined based on the assessment results. The fault modes are classified according to the risk level, and the classified fault modes are prioritized to form the final fault diagnosis report.

10. A multi-sensor integrated elevator operation status monitoring system, used to implement the multi-sensor integrated elevator operation status monitoring method as described in any one of claims 1-9, characterized in that, The system includes: The acquisition module is used to acquire vibration and current signals of the elevator door rail and generate an initial multi-source signal set; The synchronization module is used to perform time synchronization processing on the initial multi-source signal set to obtain a time synchronization signal sequence, extract features of various signals from the time synchronization signal sequence, and generate a set of signal feature segments. The analysis module is used to analyze the interaction patterns between signals based on the set of signal feature segments, determine the interaction anomaly feature group, fuse the interaction anomaly feature group with the elevator door rail historical fault data, and generate a corrected description of the door rail operating status. The diagnostic module is used to optimize the abnormal data if the abnormal components in the corrected door rail operating status description exceed a preset threshold, obtain an optimized status description, match the elevator fault mode according to the optimized status description, and generate a fault diagnosis report for the elevator door rail system.

Images