Multi-source data fusion method and system for elevator fault positioning

Abstract

The invention relates to the technical field of elevator operation and maintenance, and discloses a multi-source data fusion method and system for elevator fault localization, and the method comprises the steps: carrying out the noise filtering and data alignment of multi-source monitoring data of an elevator during operation, and obtaining a monitoring data set; performing time-frequency analysis on acceleration sensor data in the monitoring data set to obtain vibration mode characteristics; performing harmonic analysis on the motor current data to obtain current distortion characteristics; performing space-time correlation on the vibration mode characteristics, the current distortion characteristics and the safety loop state in the monitoring data set to obtain a comprehensive fault characteristic vector; according to the comprehensive fault feature vector, specific components and types when the elevator breaks down are determined, so that fault positioning information is obtained; the fault positioning information is transmitted to an elevator monitoring terminal, and the elevator is maintained according to the fault positioning information; according to the invention, the fault positioning efficiency can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the elevator operation and maintenance technical field, and more particularly to a multi-source data fusion method and system for elevator fault positioning. BACKGROUND

[0002] The existing elevator fault positioning technology has obvious defects in the preprocessing stage of multi-source monitoring data. Most schemes only filter the noise of acceleration sensor data for specific frequency bands, which cannot completely remove the high-frequency interference components unrelated to elevator vibration, resulting in residual noise in the processed data, affecting the subsequent vibration feature extraction. The fluctuation suppression of motor current data does not combine the characteristics of load changes in elevator operation, and uses a fixed threshold to filter fluctuations, which is easy to misjudge normal load fluctuations as abnormal signals, and also cannot effectively suppress current fluctuations caused by random interference. The data alignment process is only realized by simple timestamp comparison, without interpolating the filtered acceleration data, smoothed current data and safety circuit state data, so that different types of data have deviations in the time dimension, and the generated monitoring data set cannot meet the requirements of data consistency for fault analysis, which poses a hidden danger for subsequent fault feature extraction.

[0003] In the fault feature extraction and positioning link, the limitations of existing technologies are more prominent. The time-frequency analysis of acceleration sensor data lacks multi-scale processing, and only a single time window is used to calculate the time-frequency distribution, which cannot capture the vibration differences in different stages such as elevator start, operation and stop, and the extracted vibration modal features cannot accurately correspond to the actual state of mechanical components. The harmonic analysis of motor current data only identifies the numerical values of fundamental and harmonic components, without associating elevator operating conditions and load changes, and cannot explore the internal relationship between harmonic variation and faults, resulting in insufficient completeness of current distortion features. More importantly, existing technologies do not correlate vibration modal features, current distortion features and safety circuit states in time and space, and only rely on a single feature to determine faults, which cannot construct a comprehensive fault feature vector, resulting in fault positioning only determining that there is a fault, and cannot accurately identify specific fault components and fault types, greatly increasing the workload of fault troubleshooting and prolonging the elevator downtime. Therefore, how to improve the efficiency of fault positioning has become a problem to be solved. SUMMARY

[0004] The present application provides a multi-source data fusion method and system for elevator fault positioning to solve the problems raised in the background technology.

[0005] To achieve the above purpose, the present application provides a multi-source data fusion method for elevator fault positioning, comprising: S1, filtering noise and data alignment of multi-source monitoring data of the running elevator to obtain a monitoring data set of the elevator; S2, time-frequency analysis is performed on the acceleration sensor data in the monitoring data set to obtain vibration modal characteristics of the elevator; S3, harmonic analysis is performed on the motor current data in the monitoring data set to obtain current distortion characteristics of the elevator; S4, spatiotemporal correlation is performed on the vibration modal characteristics, the current distortion characteristics, and the safety circuit state in the monitoring data set to obtain a comprehensive fault feature vector of the elevator; S5, according to the comprehensive fault feature vector, specific components and types of the elevator when a fault occurs are determined to obtain fault positioning information of the elevator; S6, the fault positioning information is transmitted to an elevator monitoring terminal, and the elevator is maintained according to the fault positioning information.

[0006] In a preferred embodiment, the noise filtering and data alignment of the multi-source monitoring data of the elevator in operation are performed to obtain a monitoring data set of the elevator, including: high-frequency noise components of the acceleration sensor data in the multi-source monitoring data are removed to obtain filtered acceleration data of the elevator; random fluctuations of the motor current data in the multi-source monitoring data are suppressed to obtain smoothed current data of the elevator; the filtered acceleration data and the smoothed current data are interpolated and synchronized with the safety circuit state data of the elevator based on timestamps to obtain the monitoring data set of the elevator.

[0007] In a preferred embodiment, the time-frequency analysis is performed on the acceleration sensor data in the monitoring data set to obtain vibration modal characteristics of the elevator, including: the acceleration sensor data is simultaneously divided into multiple time windows of different scales to obtain a multi-scale data segment set of the elevator; time-frequency distribution calculation is respectively performed on each data segment in the multi-scale data segment set to obtain a time-frequency distribution map of the elevator; a feature frequency combination reflecting a specific operating state of the elevator is extracted from the time-frequency distribution map to obtain a multi-scale feature set of the elevator; the multi-scale feature set is cross-verified with elevator operating state parameters to screen out stable frequency components related to faults; vibration modal characteristics representing vibration characteristics of the elevator are constructed based on the stable frequency components.

[0008] In a preferred embodiment, the multi-scale feature set is cross-verified with elevator operating state parameters to screen out stable frequency components related to faults, including: The elevator operating status parameters are classified according to the elevator operating stages to obtain the elevator status classification results; Correlation analysis is performed between the feature frequencies in the multi-scale feature set and the state classification results to identify feature frequencies that are strongly correlated with specific operating stages, thereby obtaining candidate frequency components of the elevator when a malfunction occurs. The stability of the candidate frequency components is evaluated over time, and the consistency of their occurrence in different time intervals is calculated. Based on the consistency of occurrence, frequency components that appear stably in the corresponding operating phase of the elevator are selected to obtain stable frequency components related to the fault.

[0009] In a preferred embodiment, the formula for calculating the consistency is as follows: ; In the formula, The quantization value that indicates consistency is obtained. In the first The occurrence status of the candidate frequency components in a time interval, wherein an occurrence status of 1 indicates occurrence and 0 indicates non-occurrence. The total number of intervals in the time dimension. Let be the interval ordinal number of the time dimension. For the first The occurrence status of the candidate frequency components in each time interval.

[0010] In a preferred embodiment, the step of performing harmonic analysis on the motor current data in the monitoring dataset to obtain the current distortion characteristics of the elevator includes: The motor current data is segmented according to the elevator operating conditions to obtain multiple operating condition current data segments of the elevator. Identify the fundamental and harmonic components in the operating current data segment to obtain the set of harmonic components of the elevator; By analyzing the variation patterns of each harmonic in the harmonic component set under different operating conditions, the harmonic variation mode of the elevator is obtained. The harmonic variation patterns are correlated with the load state of the elevator to identify harmonic characteristics related to load changes. Based on the correlation between the harmonic characteristics and load changes, the current distortion characteristics of the elevator are constructed.

[0011] In a preferred embodiment, the step of spatiotemporally correlating the vibration modal characteristics, the current distortion characteristics, and the safety loop status in the monitoring dataset to obtain the comprehensive fault feature vector of the elevator includes: The vibration mode characteristics are mapped to mechanical transmission chain status indicators, and the current distortion characteristics are mapped to electric drive chain status indicators. Using the safety loop state as the system safety benchmark, the state consistency of the mechanical transmission chain state index and the electric drive chain state index is verified, and the verified indexes are collected into the elevator state index set. Based on the time series of elevator operation, the spatiotemporal evolution trajectory of the state index set is constructed to identify the state anomaly propagation mode of the elevator. Based on the state anomaly propagation mode, the multidimensional fault feature space of the elevator is reconstructed to obtain the comprehensive fault feature vector of the elevator.

[0012] In a preferred embodiment, the step of constructing the spatiotemporal evolution trajectory of the state indicator set based on the elevator's operating time series, and identifying the state anomaly propagation pattern of the elevator, includes: Arrange the state indicators in the set of state indicators according to the time series to obtain the state evolution sequence of the elevator; Align the state change trajectories under different operating cycles in the state evolution sequence to obtain a standardized state evolution pattern; In the standardized state evolution model, the causal relationship between state indicators is identified to obtain the state propagation network of the elevator; By analyzing the transmission path and diffusion pattern of abnormal states in the state propagation network, the state anomaly propagation mode of the elevator is obtained.

[0013] In a preferred embodiment, identifying the causal relationships between state indicators in the standardized state evolution model to obtain the elevator's state propagation network includes: The causal influence strength among various state indicators in the standardized state evolution model is determined, wherein the calculation formula for the causal influence strength is as follows: ; In the formula, To represent state indicators Status indicators The strength of causal influence The total length of the running cycle. As a time factor, It is a logarithmic function. Status indicators exist The value at time, Status indicators exist The value at time, Status indicators exist the value of the time, is and under the condition conditional probability, is under the condition conditional probability, is marginal probability, is the conditional probability under the condition ; Introduce time delay factor in the causal relationship strength, analyze the time sequence characteristics of the causal relationship, obtain the causal relationship of the elevator; Hierarchical organization is carried out according to the influence strength and time sequence characteristics, and the causal network diagram of the elevator is constructed; The community discovery analysis is carried out on the causal network diagram, and the state propagation network of the elevator is obtained.

[0014] In order to solve the above problems, the application also provides a multi-source data fusion system for elevator fault positioning, the system comprises: A data preprocessing module is used for noise filtering and data alignment of the multi-source monitoring data of the running elevator, and a monitoring data set of the elevator is obtained; A vibration modal analysis module is used for time-frequency analysis of the acceleration sensor data in the monitoring data set, and vibration modal characteristics of the elevator are obtained; A current distortion analysis module is used for harmonic analysis of the motor current data in the monitoring data set, and current distortion characteristics of the elevator are obtained; A feature fusion module is used for spatio-temporal correlation of the vibration modal characteristics, the current distortion characteristics and the safety circuit state in the monitoring data set, and a comprehensive fault feature vector of the elevator is obtained; A fault diagnosis module is used for determining the specific components and types when the elevator fails according to the comprehensive fault feature vector, so as to obtain the fault positioning information of the elevator; A monitoring and maintenance module is used for transmitting the fault positioning information to the elevator monitoring terminal and maintaining the elevator according to the fault positioning information.

[0015] Compared with the prior art, the application has the following beneficial effects: 1.The application removes high-frequency noise of acceleration sensor data and suppresses random fluctuations of motor current data, and then synchronizes multiple types of data based on timestamps to obtain a high-quality monitoring data set. Subsequently, multi-scale time-frequency analysis is performed on the acceleration data to capture vibration characteristics in different operating stages to generate accurate vibration modal characteristics, and harmonic analysis is performed on the motor current data according to the working condition and the load state is associated to obtain complete current distortion characteristics, which provide reliable data and feature support for subsequent fault analysis.

[0016] 2.The application associates vibration modal characteristics, current distortion characteristics and safety circuit state in space and time, constructs a comprehensive fault feature vector, accurately determines the specific components and types of faults, and generates accurate fault positioning information. Then, the information is transmitted to the elevator monitoring terminal to guide maintenance, which not only ensures the accuracy of fault positioning, but also improves the timeliness of maintenance, effectively reduces the fault troubleshooting time and elevator downtime, and significantly improves the efficiency of elevator fault handling and operation and maintenance reliability. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 A flowchart of a multi-source data fusion method for elevator fault positioning provided by an embodiment of the application is shown. Figure 2 A functional module diagram of a multi-source data fusion system for elevator fault positioning provided by an embodiment of the application is shown. The implementation, functional characteristics and advantages of the application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0018] It should be understood that the specific embodiments described herein are only used to explain the application and do not limit the application.

[0019] The embodiment of the application provides a multi-source data fusion method for elevator fault positioning. The execution subject of the multi-source data fusion method for elevator fault positioning includes but is not limited to at least one of the electronic devices capable of being configured to execute the method provided by the embodiment of the application, such as a server, a terminal and the like. In other words, the multi-source data fusion method for elevator fault positioning can be executed by software or hardware installed in a terminal device or a server device. The server includes but is not limited to a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be a stand-alone server, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content distribution networks (CDN), and big data and artificial intelligence platforms, etc. Basic cloud computing services.

[0020] ReferenceFigure 1 Fig. 1 shows a flowchart of a multi-source data fusion method for elevator fault positioning according to an embodiment of the present application. In this embodiment, the multi-source data fusion method for elevator fault positioning comprises: S1, filtering noise and aligning data of multi-source monitoring data of the running elevator to obtain a monitoring data set of the elevator; In this embodiment, filtering noise and aligning data of multi-source monitoring data of the running elevator to obtain a monitoring data set of the elevator comprises: removing high-frequency noise components of acceleration sensor data in the multi-source monitoring data to obtain filtered acceleration data of the elevator; suppressing random fluctuations of motor current data in the multi-source monitoring data to obtain smoothed current data of the elevator; interpolating and synchronizing the filtered acceleration data and the smoothed current data with the safety circuit state data of the elevator based on timestamps to obtain the monitoring data set of the elevator.

[0021] Through the built-in acceleration data filtering unit of the elevator, the frequency range of the vibration signal during normal operation of the elevator is first analyzed, and the frequency interval corresponding to the high-frequency noise unrelated to the mechanical movement of the elevator is determined. Then, the original data collected by the acceleration sensor is input into the filtering unit, which completely blocks the high-frequency components exceeding the normal vibration frequency range according to the preset frequency interception rule, and only retains the effective vibration signals reflecting the running state of the mechanical parts of the elevator. The output signal after this processing is the filtered acceleration data of the elevator.

[0022] Through the motor current data smoothing processing unit, the correlation between different load states of the elevator and normal current fluctuations is first established, and the reasonable fluctuation interval of the current under each load is determined. Then, the original motor current data is received in real time, each current data point is compared with the reasonable fluctuation interval under the corresponding load, and the random fluctuation signals exceeding the interval are identified. Through gradient buffer adjustment of the fluctuation signals, irregular random interference is eliminated, while the normal current change trend caused by load changes is completely retained. The processed current data is the smoothed current data of the elevator.

[0023] Through the data synchronization processing unit, the timestamps of the filtered acceleration data, the smoothed current data, and the safety circuit state data are first extracted. Based on the time axis of the elevator operation, the time coincidence nodes of the three types of data are determined. For the data points with unmatched timestamps, interpolation calculation is performed according to the change law of the adjacent effective data before and after them to supplement the synchronization data points corresponding to the reference time axis. This forms a one-to-one corresponding relationship among the three types of data in the same time dimension. Finally, the structured data set obtained by integration is the monitoring data set of the elevator.

[0024] The beneficial effect is that the implementation process effectively eliminates invalid interference components in multi-source monitoring data through targeted noise filtering and fluctuation suppression method, and ensures the consistency and accuracy of filtered acceleration data, smoothed current data and safety circuit state data through timestamp-based interpolation synchronization processing, so that the generated monitoring data set can truly reflect the elevator operation state, provide high-quality data support for subsequent time-frequency analysis, harmonic analysis and fault feature extraction, and ensure the accuracy of fault positioning.

[0025] S2, time-frequency analysis is performed on the acceleration sensor data in the monitoring data set to obtain vibration modal characteristics of the elevator; In the embodiment of the application, the time-frequency analysis of the acceleration sensor data in the monitoring data set to obtain the vibration modal characteristics of the elevator comprises: The acceleration sensor data is divided into a plurality of time windows of different scales at the same time to obtain a multi-scale data segment set of the elevator; Time-frequency distribution calculation is performed on each data segment in the multi-scale data segment set to obtain a time-frequency distribution graph of the elevator; From the time-frequency distribution graph, a characteristic frequency combination reflecting a specific operation state of the elevator is extracted to obtain a multi-scale feature set of the elevator; The multi-scale feature set and the elevator operation state parameters are cross-verified to screen out stable frequency components related to faults; Based on the stable frequency components, a vibration modal characteristic representing the vibration characteristics of the elevator is constructed.

[0026] The cross-verification of the multi-scale feature set and the elevator operation state parameters to screen out stable frequency components related to faults comprises: The elevator operation state parameters are classified according to the elevator operation stage to obtain a state classification result of the elevator; The correlation analysis is performed on the characteristic frequencies in the multi-scale feature set and the state classification result to identify characteristic frequencies strongly correlated with a specific operation stage to obtain candidate frequency components of the elevator when a fault occurs; The stability of the candidate frequency components in the time dimension is evaluated, and the consistency of the candidate frequency components in different time intervals is calculated; Based on the consistency, frequency components that stably appear in the corresponding operation stage of the elevator are screened out to obtain stable frequency components related to faults.

[0027] The calculation formula of the consistency is as follows: ; In the formula, a quantity of occurrences of the candidate frequency component in the time interval, a quantity of occurrences of the candidate frequency component in the time interval, a quantity of occurrences of the candidate frequency component in the time interval, a total number of intervals of the time dimension, an interval ordinal number of the time dimension, a quantity of occurrences of the candidate frequency component in the time interval, a quantity of occurrences of the candidate frequency component in the time interval.

[0028] Through the elevator data processing unit, first, the time window scale range covering the starting, uniform speed, deceleration, and stopping of the whole running process is determined in combination with the elevator running cycle and mechanical component response characteristics, and the time windows of three different fixed lengths, short, medium, and long, are accurately set. The short window length matches the response cycle of transient vibration, is used to capture the millisecond-level transient vibration change, the medium window length covers the key segment of a single running stage, is used to reflect the vibration trend of the stage, and the long window length is equivalent to the complete running cycle period, is used to cover a complete running segment. Subsequently, the original data collected by the acceleration sensor are simultaneously divided in sliding mode according to the three window scales, and the sliding step length is set to half of the short window length, so as to ensure that the window of each scale has neither data omission nor repeated coverage. The vibration data segment set under different scales obtained after the division is the multi-scale data segment set of the elevator.

[0029] Each data segment in the multi-scale data segment set is input into the time-frequency analysis unit one by one. The unit first performs time axis subdivision on each data segment, divides the data segment into a plurality of micro time slices with equal length, and then analyzes the frequency component composition in each micro time slice. The strength value is obtained by calculating the amplitude of each frequency component, so as to convert the vibration signal of a single time dimension into three-dimensional information of time-frequency-strength. Subsequently, the three-dimensional information is presented in a two-dimensional image rendering mode. The horizontal axis strictly corresponds to the sequence of the time axis, the vertical axis is arranged in order from low to high according to the frequency, the pixel brightness is positively correlated with the frequency strength, the higher the strength, the brighter the pixel, and each data segment generates an independent two-dimensional image after complete processing. The two-dimensional images corresponding to all data segments are arranged in order of scale and time, and together constitute the time-frequency distribution diagram of the elevator.

[0030] The characteristic frequency extraction unit performs a comprehensive scan on each time-frequency distribution row by row and column by column, a preset frequency intensity standard is determined based on a vibration intensity threshold value in the normal operation of the elevator, the threshold value is obtained by a large number of normal operation data statistics, peak frequencies with a frequency intensity higher than the standard in each graph are identified, and the peak frequencies are main contribution components of the vibration of the elevator. According to the vibration law under different operating states of the elevator, the peak frequencies reflecting the same operating state under the same scale or different scales are combined, each combination contains a group of characteristic frequencies associated with each other and collectively representing a vibration phenomenon. After all such characteristic frequency combinations are classified according to the operating stage, a multi-scale characteristic set of the elevator is integrated.

[0031] Through the operating state classification unit, the speed, load, and running time state parameters during operation are collected in real time through the elevator control system. The speed data comes from the speed sensor of the elevator traction machine, the load data comes from the pressure sensor at the bottom of the car, and the speed is divided into the starting stage from 0 to 90% of the rated speed, the uniform speed stage with the speed maintained within ±5% of the rated speed, the deceleration stage from 90% of the rated speed to 0, and the stopping stage with the speed of 0 and no external operating instructions. Subsequently, all operating state parameters are classified according to their corresponding time nodes, and the state classification result of the elevator is obtained.

[0032] Each characteristic frequency in the multi-scale characteristic set is associated and compared with the state classification result, the occurrence number, duration, and intensity proportion of each characteristic frequency in each operating stage are counted, the occurrence number is the total number of occurrences of the frequency in all data segments in the stage, the duration is the cumulative value of the duration of the frequency in each data segment, and the intensity proportion is the ratio of the average intensity of the frequency to the average intensity of all frequencies in the stage. The average value of the three indicators is calculated to measure the close degree of the characteristic frequency and each stage, and the characteristic frequency with an occurrence number proportion, duration proportion, and intensity proportion higher than 80% in a specific operating stage and lower than 20% in other stages is selected. These characteristic frequencies strongly bound to a specific operating stage are candidate frequency components when the elevator fails.

[0033] The time dimension interval division is performed on each candidate frequency component. According to the total length of continuous elevator operation and the data stability requirement, the continuous elevator operation time is uniformly divided into a plurality of time intervals with equal length. The length of each time interval is set to 30 seconds. The occurrence of the candidate frequency component is detected in each interval. If the candidate frequency component occurs, 1 is recorded. If the candidate frequency component does not occur, 0 is recorded. The occurrence state sequence of each candidate frequency component in all time intervals is obtained. The number of 1 in the sequence is counted and divided by the total number of intervals in the time dimension to obtain the occurrence proportion. The absolute value of the difference between the occurrence states of adjacent two time intervals is calculated. The absolute values of the differences between all adjacent intervals are accumulated and divided by the total number of adjacent intervals to obtain the state fluctuation coefficient. The state continuity coefficient is obtained by subtracting the state fluctuation coefficient from 1. The occurrence proportion and the state continuity coefficient are multiplied to obtain the quantitative value of the occurrence consistency of the candidate frequency component in the time dimension.

[0034] Based on statistical analysis of a large number of elevator fault case data, a qualified standard of occurrence consistency is set. The standard is the minimum occurrence consistency quantitative value of the fault-related frequency component. The occurrence consistency result of each candidate frequency component is compared with the qualified standard. The frequency components with an occurrence proportion higher than 70%, a state continuity coefficient higher than 85%, an overall occurrence consistency quantitative value reaching the qualified standard, and a continuous and stable occurrence in the corresponding specific operation phase are selected. The frequency components with a scattered occurrence, a large state fluctuation, and an occurrence consistency not meeting the standard are removed. The finally retained frequency components are the stable frequency components related to the fault.

[0035] Through the vibration modal feature construction unit, the core information of all stable frequency components is integrated first, including the specific numerical value of each frequency, the average occurrence intensity in the corresponding operation phase, and the cooperative occurrence law with other stable frequencies. Then, the corresponding relationship between the frequency and the mechanical component is established by combining the structural characteristics of the elevator mechanical transmission chain. The stable frequency in a specific frequency range is corresponded to different mechanical components such as the traction machine, the guide rail, the car, and the counterweight. The vibration mode corresponding to each frequency component is determined. Finally, the feature combination that can accurately represent the vibration state of different mechanical components of the elevator is formed. The feature combination is the vibration modal feature representing the vibration characteristics of the elevator.

[0036] The beneficial effect is that the implementation process accurately covers the vibration data of the entire operation phase of the elevator through a multi-scale time window. The multi-dimensional conversion of the vibration signal is realized through time-frequency analysis. The interference components are removed through multi-dimensional screening logic. The quantitative calculation of the formula ensures the accuracy of the occurrence consistency evaluation. The finally constructed vibration modal feature is strongly associated with the specific operation phase and can accurately correspond to the vibration state of the mechanical component. It provides high-credibility mechanical-side fault feature support for subsequent fault positioning combined with the current distortion feature and the safety circuit state. The accuracy and reliability of the fault component and type judgment are greatly improved.

[0037] S3, performing harmonic analysis on the motor current data in the monitoring data set to obtain current distortion characteristics of the elevator; In the embodiment of the application, the harmonic analysis on the motor current data in the monitoring data set to obtain current distortion characteristics of the elevator comprises: segmenting the motor current data according to elevator operating conditions to obtain multiple operating condition current data segments of the elevator; identifying fundamental wave and harmonic components in the operating condition current data segments to obtain a harmonic component set of the elevator; analyzing variation laws of each harmonic in the harmonic component set under different operating conditions to obtain a harmonic variation mode of the elevator; performing correlation analysis on the harmonic variation mode and the load state of the elevator to identify a harmonic feature related to load variation; based on the correlation between the harmonic feature and load variation, constructing current distortion characteristics of the elevator.

[0038] Through the elevator condition identification unit, load data and speed data in the elevator running process are first collected, the elevator operating conditions are clearly divided into six types of empty load start, full load start, empty load uniform speed, full load uniform speed, empty load deceleration and full load deceleration according to the preset condition division rule, and the time nodes corresponding to each condition are extracted. The continuous motor current data is segmented by taking the time node as the segmentation boundary, so that each data segment completely covers a single operating condition. The current data segments corresponding to different conditions obtained after segmentation are the multiple operating condition current data segments of the elevator.

[0039] Each operating condition current data segment is input into a frequency component analysis unit. The unit first performs frequency spectrum analysis on the current data to identify the frequency component with the maximum amplitude and the same frequency as the rated frequency of the motor. This component is the fundamental wave. Then, all frequency components with an integer multiple of the fundamental wave frequency are selected as harmonic components. The identification of the fundamental wave and harmonic components is completed for each operating condition current data segment. The harmonic components under all conditions are sorted and integrated according to the frequency to form a set containing the frequency and amplitude information of each harmonic. This set is the harmonic component set of the elevator.

[0040] The harmonic change analysis unit compares each harmonic in the harmonic component set across operating conditions, and statistically records the amplitude values, occurrence duration and change range of each harmonic under different operating conditions, analyzes the change trend of each harmonic with operating condition switching, for example, judges whether the amplitude of a certain harmonic increases with load increase, whether it maintains a stable amplitude under constant speed operating condition, whether it decreases sharply under deceleration operating condition, and the like, classifies and organizes the change rules of each harmonic across operating conditions, and the change trend description corresponding to each harmonic obtained is the harmonic change mode of the elevator.

[0041] The load correlation analysis unit extracts the specific load state values corresponding to each operating condition, and one-to-one corresponds the change mode of each harmonic to the load values of the corresponding operating condition, calculates the correlation degree between the harmonic amplitude change and the load change, and screens out the harmonics with obvious synchronization relationship between the harmonic amplitude change and the load change, and the information such as the frequency and amplitude change rule of these harmonics directly related to the load change is the harmonic feature of the elevator.

[0042] The current distortion feature construction unit integrates all load change related harmonic features, clearly defines the harmonic order corresponding to each feature, the amplitude range under different loads, and the quantitative relationship between load change and amplitude change, and combines the amplitude ratio information of the fundamental wave and the harmonic wave to form a feature combination that can comprehensively represent the distortion state of the motor current caused by factors such as load change and mechanical failure, and the feature combination is the current distortion feature of the elevator.

[0043] The implementation process accurately captures the motor current features under different operating conditions by segmenting according to operating conditions, completely identifies the fundamental wave and harmonic components and analyzes their change rules, and finally constructs the current distortion feature that can accurately reflect the operating state of the elevator power drive chain, which is complementary to the vibration modal feature, and provides a reliable power side fault basis for the construction of subsequent comprehensive fault feature vectors, effectively improving the comprehensiveness and accuracy of elevator fault positioning.

[0044] S4, spatiotemporal correlation is performed on the vibration modal feature, the current distortion feature and the safety circuit state in the monitoring data set to obtain a comprehensive fault feature vector of the elevator; In the embodiment of the present application, the spatiotemporal correlation of the vibration modal feature, the current distortion feature and the safety circuit state in the monitoring data set to obtain a comprehensive fault feature vector of the elevator comprises: The vibration modal feature is mapped to a mechanical transmission chain state indicator, and the current distortion feature is mapped to an electric power drive chain state indicator; The safety circuit state is taken as a system safety reference, and the mechanical transmission chain state indicator and the electric power drive chain state indicator are subjected to state consistency verification, and the indicators that pass the verification are collected as a state indicator set of the elevator. based on a time sequence of elevator operation, constructing a space-time evolution track of the state indicator set, identifying a state abnormal propagation mode of the elevator; According to the state abnormal propagation mode, reconstructing a multi-dimensional fault feature space of the elevator, obtaining a comprehensive fault feature vector of the elevator.

[0045] The state abnormal propagation mode based on the time sequence of elevator operation, constructing the space-time evolution track of the state indicator set, identifying the state abnormal propagation mode of the elevator, comprising: Arranging each state indicator in the state indicator set according to the time sequence, obtaining the state evolution sequence of the elevator; Aligning the state change track under different operation cycles in the state evolution sequence, obtaining the standardized state evolution mode; In the standardized state evolution mode, identifying the causal relationship between state indicators, obtaining the state propagation network of the elevator; Analyze the transmission path and diffusion law of abnormal state in the state propagation network, obtain the state abnormal propagation mode of the elevator.

[0046] The causal relationship between state indicators in the standardized state evolution mode, obtaining the state propagation network of the elevator, comprising: Determine the causal influence strength between each state indicator in the standardized state evolution mode, wherein the calculation formula of the causal influence strength is as follows: ; In the formula, is the value of state indicator at time , the causal influence strength of state indicator , is the total length of the operation cycle, is the time factor, is the logarithmic function, is the value of state indicator at time , is the value of state indicator at time , is the value of state indicator at time , is and the conditional probability of is the conditional probability of , is an edge probability, conditional probability of conditional probability of introducing time delay factor in the causal influence intensity, analyzing time sequence characteristics of the causal relationship, obtaining the causal relationship of the elevator; hierarchically organizing the causal relationship according to the influence intensity and the time sequence characteristics, and constructing the causal network diagram of the elevator; performing community discovery analysis on the causal network diagram, and obtaining the state propagation network of the elevator.

[0047] Through the state index mapping unit, a one-to-one correspondence between vibration modal characteristics and mechanical transmission chain components is established in advance based on the structural principle and vibration propagation characteristics of the elevator mechanical transmission chain. The components in different frequency intervals in the vibration modal characteristics are respectively accurately associated with core mechanical components such as traction machines, guide rails, cars, counterweights, guide wheels, etc. For example, the low-frequency component corresponds to the torsional vibration of the traction machine, and the medium-frequency component corresponds to the friction vibration of the guide rail. The amplitude values, time dimension stability, and cross-condition change rate of each frequency are converted into quantitative indicators representing the operating state of the corresponding components through a linear conversion method. These quantitative indicators are the mechanical transmission chain state indicators. At the same time, based on the electrical principle and harmonic generation mechanism of the power drive chain, a corresponding relationship between current distortion characteristics and power drive chain components is established. The 2nd, 3rd, 5th, etc. harmonic parameters in the current distortion characteristics are respectively associated with the frequency converter, motor winding, power supply module, contactor, etc. The harmonic amplitude peak value, total harmonic distortion rate, and harmonic load change rate are converted into quantitative indicators representing the operating state of the corresponding components through standardization. These quantitative indicators are the power drive chain state indicators.

[0048] Through the state consistency verification unit, the safety circuit state data is extracted from the monitoring data set in real time, including the circuit on-off state, circuit voltage stability value, key contact contact state, etc. Based on the normal working parameters of the safety circuit set by the elevator manufacturer, the state in which the safety circuit is continuously turned on, the voltage fluctuation range does not exceed ±5%, and the contact is in good contact is taken as the system safety benchmark. The mechanical transmission chain state indicators and the power drive chain state indicators are compared with the corresponding normal value range of the benchmark one by one to check whether the indicator values fall within the safety benchmark allowed range. At the same time, the trend correlation analysis is used to check whether the change trend of the mechanical side and the power side indicators is consistent, and the abnormal indicators that exceed the normal range, have inconsistent trends, or are irrelevant are removed. The mechanical transmission chain state indicators and the power drive chain state indicators that pass the verification are classified and integrated according to the function modules to form a complete and consistent indicator set, which is the state indicator set of the elevator.

[0049] By using time series arrangement units, the precise collection timestamps corresponding to each state indicator in the state indicator set are extracted. Using the chronological order of the timestamps as the absolute benchmark, all state indicators are arranged sequentially along a unified time axis. The equal time interval interpolation method is used to supplement the indicator data of a small number of missing time nodes, ensuring that each time node corresponds to a complete set of indicator values. Moreover, the indicator data of the same time node can comprehensively and synchronously reflect the mechanical and electrical integrated status of the elevator at the same moment. The ordered indicator sequence formed after arranging according to this rule is the state evolution sequence of the elevator.

[0050] Through the trajectory alignment unit, the state evolution sequences of at least 10 complete operating cycles of the elevator are collected. Taking the standard operating cycle under the rated load of the elevator as the base cycle, the state evolution sequences of different operating cycles are time-normalized. Each cycle is divided into 20 time segments of equal length. The arithmetic mean of the state indicators in the corresponding time segments of each cycle is calculated. Using the average value of the indicators in each time segment of the base cycle as a reference standard, the distribution of indicator values ​​in the corresponding segments of other cycles is adjusted by the proportional adjustment method. This eliminates the systematic errors caused by load fluctuations and changes in ambient temperature in different cycles, so that the state change trajectories under different operating cycles are highly consistent in terms of time dimension and numerical trend. The unified and standardized trajectory pattern obtained after alignment is the state evolution pattern of the elevator.

[0051] Through the causal influence intensity analysis unit, each state indicator in the standardized state evolution model is subjected to pairwise correlation analysis with all other indicators. The total length of the operating cycle is the total number of time segments divided in the standardized evolution model. Time factors are arranged sequentially from 1 to the total length according to the time segments. First, the state indicators are statistically analyzed. exist Values ​​at time and state indicators exist The value at time and exist The correlation frequency of the value at time point is calculated in Take a certain value and Given the value from the previous moment, The percentage of times the corresponding value is taken at the current moment is obtained. and under conditions Conditional probability; and statistically only Given the value from the previous moment, The percentage of times the corresponding value is taken at the current moment is obtained. under conditions Conditional probability; statistics exist The total number of times a certain value is taken at any given time and the total number of times it is taken in all time segments. The percentage of the total number of values ​​taken is obtained. marginal probability; statistics At a certain value at a certain time, the proportion of times that the corresponding value is taken at a certain time, to obtain the conditional probability under the condition that the ratio of the two conditional probabilities and the ratio of the two marginal probabilities and conditional probabilities are multiplied, and the logarithm of the product is taken to amplify the effective difference and unify the data scale. After accumulating the calculation results of all time segments, divide by the total length of the running period to obtain the average value, which is the index The causal influence strength of the index Through this comprehensive statistical analysis method, the causal influence degree of each index on all other indexes is quantified, that is, the causal influence strength between state indexes.

[0052] Through the time sequence characteristic analysis unit, introduce the time delay factor in the determined causal influence strength. For each pair of indexes with potential causal relationship, test the causal influence strength in different delay intervals of =1, 2, 3 time segments respectively, draw the change curve of influence strength with delay interval, find the delay interval corresponding to the peak value of the curve, which is the optimal delay interval with the maximum influence strength, so as to determine the specific time difference of causal relationship and clarify the sequence of influence between indexes. Combined with the quantitative causal influence strength value and the optimal time delay, the association result of clear time sequence relationship and strength level is the causal relationship of the elevator.

[0053] Through the causal network graph construction unit, each state index is taken as an independent node in the network, and the determined causal relationship is taken as the directed connection edge between nodes. The weight of the connection edge directly corresponds to the quantitative value of the causal influence strength. According to the influence strength, three levels are divided, and the direction of the connection edge is strictly determined according to the time sequence characteristic. From the index node that occurs first to the index node that is affected later, avoid reverse or undirected connection. Organize all nodes and weighted and directional connection edges in hierarchical structure from top to bottom and from left to right according to the time sequence, and form a structured network with clear hierarchy and explicit relationship, which is the causal network graph of the elevator.

[0054] By the community discovery analysis unit, the nodes in the causal network graph are grouped by using a density clustering method, a clustering threshold is set as a causal influence intensity ≥ 0.5, the index nodes with a causal influence intensity higher than the threshold, closely related to each other and similar in functional attributes are classified into the same community, each community corresponds to a functional module of the elevator, the rationality of the community division is verified by calculating the ratio of the node connection density within the community to the node connection density between the communities, the distribution of the connection edges between different communities is analyzed, the key associated nodes connecting the two communities are identified, and a network clearly reflecting the state transmission relationship within and between the functional modules of the elevator is constructed, which is the state propagation network of the elevator.

[0055] By the abnormality propagation analysis unit, the generation of an abnormal state is simulated based on the historical fault data of the elevator, or an abnormal node with an index value exceeding the normal range in the state propagation network is monitored in real time, the abnormal node is taken as an initial propagation source, the abnormal state is tracked from the initial node, a complete path of the abnormal state spreading from the initial node to other nodes in the same community, the key nodes and peripheral nodes across the communities through the weighted directed connection edges is calculated, the residence time, the diffusion speed and the influence range of the abnormal state at each node are counted, the diffusion law of the abnormal state in the network is summarized, and a complete description including the abnormal initial source, the propagation path, the diffusion speed and the influence range is obtained, which is the state abnormality propagation mode of the elevator.

[0056] By the fault feature space reconstruction unit, the state abnormality propagation mode is taken as a core framework, each index value in the state index set, the index abnormality degree, the causal influence intensity, the time sequence delay characteristic, the community attribution information and the like are taken as independent dimensions, a multi-dimensional fault feature space is constructed, each dimension corresponds to a key information directly related to the fault, and the values of each dimension are standardized to ensure the magnitude uniformity, the current abnormality propagation mode of the elevator, the real-time state of each index, the causal correlation data and the like are mapped into the space according to the dimensions one by one, and a vector capable of comprehensively and accurately representing the comprehensive fault state of the elevator is formed, which is the comprehensive fault feature vector of the elevator.

[0057] The implementation process has the beneficial effects that the comprehensive fault feature vector is constructed by accurately mapping the features based on mechanical and electrical principles, verifying the consistency of the safety benchmark constraints, aligning the trajectories by multi-cycle normalization, and deeply mining the correlation rules and propagation paths between the state indexes in combination with the causal influence intensity quantitative calculation containing the time sequence characteristics, the comprehensive fault feature vector comprehensively integrates the fault related information of the mechanical side, the power side and the safety system, completely represents the occurrence source, the development process and the diffusion path of the elevator fault, provides comprehensive, deep and high-credibility feature support for subsequent accurate positioning of the fault components and types, and greatly improves the accuracy, completeness and timeliness of the elevator fault positioning.

[0058] In the embodiment of the present application, S5, according to the comprehensive fault feature vector, the specific component and type of the elevator when the fault occurs are determined to obtain the fault positioning information of the elevator; Through the fault feature library construction unit, a large number of historical fault data of similar elevators in the elevator industry are collected in advance, covering various fault cases of mechanical transmission chain and electric drive chain, and combined with the experience summary of elevator maintenance experts, the corresponding standard comprehensive fault feature vector of each type of fault is extracted, the correspondence between each dimension information in the standard vector and the specific fault component and fault type is determined, and these standard vectors, corresponding component names and fault types are stored according to functional modules to form a structured and detailed elevator fault feature library, which provides a unified reference benchmark for subsequent matching.

[0059] Through the comprehensive feature analysis unit, the comprehensive fault feature vector of the current elevator is fully dimensionally disassembled, and the key information contained in the vector is extracted: the specific value and abnormal degree of the abnormal state index, the initial abnormal node in the state abnormal propagation mode, the core path of abnormal propagation, the causal influence intensity level between key indicators, and the correlation between mechanical side and electric side abnormal features in the vector is sorted, and the key information after analysis is sorted according to the logical order of "abnormal source- propagation path- impact range" to form a structured feature analysis result.

[0060] Through the feature matching unit, the structured feature analysis result is compared with the standard comprehensive fault feature vector in the fault feature library, and the matching degree of the abnormal source index, the core propagation path and the key dimension feature is focused on in the comparison process. The number and degree of coincidence of the current feature and the standard feature in the same dimension are counted, the coincidence degree is divided into four levels of "complete coincidence, high coincidence, medium coincidence and low coincidence", the standard comprehensive fault feature vector with the highest coincidence degree and reaching the high coincidence level or above is selected, and if there are multiple high coincidence standard vectors, the unique most matched standard feature vector is determined by further screening combined with the running condition of the current elevator.

[0061] Through the fault component positioning unit, according to the fault component information corresponding to the matched standard comprehensive fault feature vector, combined with the initial abnormal node of the abnormal propagation mode in the feature analysis result, the elevator component associated with the initial abnormal node is the core source component of the fault, and the components corresponding to the subsequent indicators affected in the abnormal propagation path are referred to determine whether the fault has spread to other associated components, and the core fault component and the affected component are clearly marked. If only the initial node is abnormal, it is determined as single component fault, and if the subsequent associated nodes are abnormal at the same time, it is determined as multi-component associated fault, and finally the specific component of the elevator fault is locked.

[0062] By the fault type determination unit, based on the matched standard feature vector corresponding to the fault type, combined with the abnormal mode details in the feature analysis result: if the abnormal features mainly show that the vibration modal characteristics exceed the standard, the mechanical transmission chain index is abnormal, and the propagation path is concentrated in the related nodes of mechanical components, it is determined as a mechanical fault; if the abnormal features mainly show that the current distortion characteristics exceed the standard, the electric power drive chain index is abnormal, and the propagation path is concentrated in the related nodes of electric power components, it is determined as an electrical fault; if the mechanical and electric power side indexes are abnormal at the same time and there is a clear causal relationship, it is determined as a mechanical and electrical linkage fault, and the fault severity level is quantified according to the abnormal degree, and the fault type and severity are determined.

[0063] By the fault location information integration unit, the determined specific fault components, fault types, fault severity levels, abnormal propagation paths, key abnormal indexes and other information are structured and integrated, and the logical organization language of 'fault component-fault type-severity-impact range' is formed. Text information with clear structure and complete information is formed, which is the fault location information of the elevator, and ensures that the maintenance personnel can quickly obtain the core fault information.

[0064] The beneficial effects are that the implementation process can peel off key information from the comprehensive fault feature vector layer by layer, accurately lock the specific fault components, determine the fault type and severity, and restore the fault propagation path by establishing a standardized fault feature library, combining full-dimensional feature analysis and accurate matching, avoiding the limitations of single feature judgment, and generating fault location information that is comprehensive, accurate and guiding. The difficulty of troubleshooting for maintenance personnel is greatly reduced, and the accuracy and maintenance efficiency of elevator fault location are improved.

[0065] In the embodiment of the application, S6, the fault location information is transmitted to the elevator monitoring terminal, and the elevator is maintained according to the fault location information.

[0066] By the information format standardization unit, the fault location information of the elevator is structured and formatted, and the core information of the fault components, fault types, severity levels, propagation paths and key abnormal indexes is arranged as fixed field text data according to the preset transmission protocol specification. Each field clearly identifies the corresponding information category to ensure uniform data format and no ambiguity, and the text data is compressed to reduce transmission bandwidth occupation and ensure efficient information transmission. The processed standardized fault location information has the conditions for direct transmission and terminal analysis.

[0067] The wireless communication transmission unit adopts an industrial-grade wireless communication module dedicated to elevators to establish a stable communication link between the elevator data processing system and the elevator monitoring terminal, and to transmit the standardized fault positioning information after encryption to prevent tampering or leakage during transmission. The data transmission process is then started, and the encrypted fault positioning information is sent to the designated elevator monitoring terminal in real time. A three-time retransmission mechanism is set during transmission, and if the monitoring terminal does not timely feedback a successful reception signal, it will automatically resend to ensure no loss of information.

[0068] The monitoring terminal information presentation unit first decrypts and decompresses the received encrypted data, then parses the structured fields according to the preset display rules, and presents the fault positioning information in a classified manner. The fault severity level, core fault components, and fault types are highlighted in a pop-up window on the terminal main interface, while the abnormal propagation path, key abnormal indicator data, and impact range are displayed in detail on the detail page. The terminal automatically records the fault positioning information in the local database to form a fault archive, facilitating subsequent traceability and query.

[0069] The maintenance task generation and distribution unit automatically matches the preset maintenance process template based on the severity level and fault type in the fault positioning information, generates a maintenance task sheet containing fault details, required tool list, recommended maintenance plan, and priority sorting, with priority set according to severity level. The task sheet is then pushed to the maintenance personnel's mobile terminal in the corresponding area through the maintenance management system, and a short message is sent for reminder to ensure that the maintenance personnel receive the task notification in a timely manner.

[0070] The on-site maintenance execution unit allows the maintenance personnel to accurately find the fault location based on the specific fault components in the fault positioning information after checking the maintenance task sheet on the mobile terminal against the actual situation of the elevator. The recommended maintenance plan is executed: mechanical faults are handled by disassembly, cleaning, replacement, or calibration, electrical faults are handled by line detection, contact cleaning, component replacement, and mechanical and electrical faults are handled by first eliminating the core fault source and then processing the associated affected components. The maintenance operation steps and key data are recorded throughout the process.

[0071] The maintenance result feedback and verification unit allows the maintenance personnel to fill in the maintenance results on the mobile terminal and upload them to the elevator monitoring terminal after completing the maintenance operation. The monitoring terminal triggers the elevator data processing system to re-collect the elevator operation data, generates a new comprehensive fault feature vector according to the complete process, and compares it with the standard feature vector of a normal elevator. If all abnormal indicators have returned to the normal range, it is determined that the fault has been eliminated, and the monitoring terminal records the maintenance completion status. If there are still abnormal indicators, it is fed back to the maintenance personnel for further investigation and treatment to ensure that the elevator fully recovers to normal operation.

[0072] The beneficial effect is that the implementation process realizes the whole-process closed-loop management of fault positioning information from transmission to maintenance through standardized transmission, accurate pushing, template maintenance task generation and result verification, which not only ensures that the maintenance personnel can quickly obtain accurate fault information and efficiently carry out targeted maintenance, but also guarantees the maintenance quality through the result verification mechanism, greatly shortens the fault response time and troubleshooting time, reduces the elevator downtime, significantly improves the timeliness, accuracy and reliability of elevator operation and maintenance, and reduces the operation and maintenance cost.

[0073] As shown in Figure 2 FIG. 1 is a functional module diagram of a multi-source data fusion system for elevator fault positioning according to an embodiment of the present application.

[0074] The multi-source data fusion system 100 for elevator fault positioning can be installed in an electronic device. According to the functions to be implemented, the multi-source data fusion system 100 for elevator fault positioning can include a data preprocessing module 101, a vibration modal analysis module 102, a current distortion analysis module 103, a feature fusion module 104, a fault diagnosis module 105, and a monitoring and maintenance module 106. The modules of the present application can also be referred to as units, which refer to a series of computer program segments that can be executed by an electronic device processor and can complete a fixed function, which are stored in the memory of the electronic device.

[0075] In the present embodiment, the functions of each module / unit are as follows: The data preprocessing module 101 is configured to filter noise and align data for the multi-source monitoring data of the elevator in operation, to obtain a monitoring data set of the elevator; The vibration modal analysis module 102 is configured to perform time-frequency analysis on the acceleration sensor data in the monitoring data set, to obtain vibration modal features of the elevator; The current distortion analysis module 103 is configured to perform harmonic analysis on the motor current data in the monitoring data set, to obtain current distortion features of the elevator; The feature fusion module 104 is configured to correlate the vibration modal features, the current distortion features, and the safety circuit state in the monitoring data set in time and space, to obtain a comprehensive fault feature vector of the elevator; The fault diagnosis module 105 is configured to determine the specific components and types of the elevator when a fault occurs, according to the comprehensive fault feature vector, to obtain fault positioning information of the elevator; The monitoring and maintenance module 106 is configured to transmit the fault positioning information to an elevator monitoring terminal, and to maintain the elevator according to the fault positioning information.

[0076] In several embodiments provided by the present application, it should be understood that the disclosed method and system can be implemented in other manners. For example, the division of the system embodiments described above is merely illustrative, and the division of the modules can be other division manners in actual implementation.

[0077] The modules described as separated components can or can not be physically separated, and the components displayed as modules can or can not be physical units, i.e., can be located in one place, or can be distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiment.

[0078] In addition, each functional module in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of hardware plus software functional modules.

[0079] It is obvious for those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and the present application can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application.

[0080] The embodiments of the present application can acquire and process related data based on artificial intelligence technology. Artificial intelligence is a theory, method, technology and application system for simulating, extending and expanding human intelligence by using a digital computer or a machine controlled by a digital computer, perceiving an environment, acquiring knowledge and using the knowledge to obtain optimal results.

[0081] Finally, it should be noted that the above embodiments are merely used to illustrate the technical solutions of the present application but not limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present application.

Claims

1. A multi-source data fusion method for elevator fault location, characterized in that, The method comprises: S1, filtering noise and aligning data of multi-source monitoring data of the running elevator to obtain a monitoring data set of the elevator; S2, time-frequency analysis of acceleration sensor data in the monitoring data set to obtain vibration modal characteristics of the elevator; S3, harmonic analysis of motor current data in the monitoring data set to obtain current distortion characteristics of the elevator; S4, spatiotemporal correlation of the vibration modal characteristics, the current distortion characteristics and the safety loop state in the monitoring data set to obtain a comprehensive fault feature vector of the elevator; S5, determining specific components and types when the elevator fails according to the comprehensive fault feature vector to obtain fault positioning information of the elevator; S6, transmitting the fault positioning information to an elevator monitoring terminal and maintaining the elevator according to the fault positioning information.

2. The multi-source data fusion method for elevator fault location of claim 1, wherein, The filtering noise and aligning data of multi-source monitoring data of the running elevator to obtain a monitoring data set of the elevator comprises: removing high-frequency noise components of acceleration sensor data in the multi-source monitoring data to obtain filtered acceleration data of the elevator; suppressing random fluctuations of motor current data in the multi-source monitoring data to obtain smoothed current data of the elevator; interpolating and synchronizing the filtered acceleration data and the smoothed current data with the safety loop state data of the elevator based on timestamps to obtain the monitoring data set of the elevator.

3. The multi-source data fusion method for elevator fault location of claim 1, wherein, The time-frequency analysis of acceleration sensor data in the monitoring data set to obtain vibration modal characteristics of the elevator comprises: dividing the acceleration sensor data into multiple time windows of different scales at the same time to obtain a multi-scale data segment set of the elevator; performing time-frequency distribution calculation on each data segment in the multi-scale data segment set to obtain a time-frequency distribution graph of the elevator; extracting a feature frequency combination reflecting a specific running state of the elevator from the time-frequency distribution graph to obtain a multi-scale feature set of the elevator; cross-verifying the multi-scale feature set with elevator running state parameters to screen out stable frequency components related to faults; constructing vibration modal characteristics representing elevator vibration characteristics based on the stable frequency components.

4. The multi-source data fusion method for elevator fault location of claim 3, wherein, The cross-verifying the multi-scale feature set with elevator running state parameters to screen out stable frequency components related to faults comprises: classifying the elevator running state parameters according to elevator running stages to obtain a state classification result of the elevator; performing correlation analysis on feature frequencies in the multi-scale feature set and the state classification result to identify feature frequencies strongly correlated with a specific running stage to obtain candidate frequency components when the elevator fails; performing stability evaluation of the candidate frequency components in the time dimension to calculate the consistency of the candidate frequency components in different time intervals; based on the consistency, screening out frequency components that stably appear in the corresponding running stage of the elevator to obtain stable frequency components related to faults.

5. The multi-source data fusion method for elevator fault location of claim 4, wherein, The calculation formula of the consistency is as follows: ； wherein is a quantized value of the occurrence consistency, is an occurrence state of the candidate frequency component in the time interval, wherein the occurrence state is 1 when the candidate frequency component occurs and is 0 when the candidate frequency component does not occur, is a total number of intervals of the time dimension, is an ordinal number of an interval of the time dimension, is an occurrence state of the candidate frequency component in the time interval.

6. The multi-source data fusion method for elevator fault location of claim 1, wherein, The harmonic analysis on the motor current data in the monitoring data set obtains the current distortion characteristics of the elevator, including: The motor current data is segmented according to the elevator operating conditions to obtain multiple operating condition current data segments of the elevator; The fundamental wave and harmonic components in the operating condition current data segment are identified to obtain a harmonic component set of the elevator; The change law of each harmonic in the harmonic component set under different operating conditions is analyzed to obtain a harmonic change mode of the elevator; The harmonic change mode and the load state of the elevator are associated and analyzed to identify a harmonic feature related to load change; Based on the association between the harmonic feature and the load change, the current distortion characteristics of the elevator are constructed.

7. The multi-source data fusion method for elevator fault location of claim 1, wherein, The spatio-temporal correlation of the vibration modal characteristics, the current distortion characteristics and the safety loop state in the monitoring data set obtains the comprehensive fault feature vector of the elevator, including: The vibration modal characteristics are mapped to mechanical transmission chain state indicators, and the current distortion characteristics are mapped to electric power drive chain state indicators; The mechanical transmission chain state indicators and the electric power drive chain state indicators are subjected to state consistency verification with the safety loop state as the system safety benchmark, and the indicators that pass the verification are collected as a state indicator set of the elevator; Based on the time sequence of elevator operation, a spatio-temporal evolution trajectory of the state indicator set is constructed to identify a state abnormal propagation mode of the elevator; According to the state abnormal propagation mode, a multi-dimensional fault feature space of the elevator is reconstructed to obtain a comprehensive fault feature vector of the elevator.

8. The multi-source data fusion method for elevator fault location of claim 7, wherein, Based on the time sequence of elevator operation, a spatio-temporal evolution trajectory of the state indicator set is constructed to identify a state abnormal propagation mode of the elevator, including: Each state indicator in the state indicator set is arranged in a time sequence to obtain a state evolution sequence of the elevator; The state change trajectories under different operating cycles in the state evolution sequence are aligned to obtain a standardized state evolution mode; In the standardized state evolution mode, the causal relationship between state indicators is identified to obtain a state propagation network of the elevator; The transmission path and diffusion law of abnormal states in the state propagation network are analyzed to obtain a state abnormal propagation mode of the elevator.

9. The multi-source data fusion method for elevator fault location of claim 8, wherein, In the standardized state evolution mode, the causal relationship between state indicators is identified to obtain a state propagation network of the elevator, including: The causal influence strength between each state indicator in the standardized state evolution mode is determined, wherein the calculation formula of the causal influence strength is as follows: ； wherein is the state indicator is the causal influence strength of the state indicator is the total length of the operating cycle, is the time factor, is the logarithm function, is the state indicator at the time instant , and is the state indicator at the time instant , and is the state indicator at the time instant , and is the conditional probability and the conditional probability under the condition is the conditional probability under the condition , and is the marginal probability of is the conditional probability under the condition The time delay factor is introduced into the causal influence strength to analyze the time sequence characteristics of the causal relationship to obtain the causal relationship of the elevator; The causal relationship is hierarchically organized according to the influence strength and time sequence characteristics to construct a causal network diagram of the elevator; The causal network diagram is subjected to community discovery analysis to obtain a state propagation network of the elevator.

10. An elevator fault positioning multi-source data fusion system for implementing the elevator fault positioning multi-source data fusion method of claim 1, the system comprising: A data preprocessing module is configured to perform noise filtering and data alignment on multi-source monitoring data of the elevator during operation to obtain a monitoring data set of the elevator; A vibration modal analysis module is configured to perform time-frequency analysis on acceleration sensor data in the monitoring data set to obtain vibration modal characteristics of the elevator; A current distortion analysis module is configured to perform harmonic analysis on motor current data in the monitoring data set to obtain current distortion characteristics of the elevator; A feature fusion module is configured to perform time-space correlation on the vibration modal characteristics, the current distortion characteristics, and a safety circuit state in the monitoring data set to obtain a comprehensive fault feature vector of the elevator; A fault diagnosis module is configured to determine specific components and types of the elevator when a fault occurs according to the comprehensive fault feature vector to obtain fault positioning information of the elevator; A monitoring and maintenance module is configured to transmit the fault positioning information to an elevator monitoring terminal and perform maintenance on the elevator according to the fault positioning information.

Images