Elevator running state evaluation method and system based on multi-source heterogeneous data

Abstract

The invention relates to the technical field of elevator running state evaluation, in particular to an elevator running state evaluation method and system based on multi-source heterogeneous data. Comprising the following steps that elevator car vertical acceleration signals, traction machine output torque signals, traction machine output current signals, motor encoder rotating speed signals, elevator running log time, stopping floor records, running frequency records and door opening and closing frequency records are collected, all data are aligned based on timestamps, and the elevator car vertical acceleration signals and the traction machine output current signals are obtained according to starting and running of elevator running. The stop state segments the data stream to form a period of time, generating a synchronized multi-source operating data stream. Multi-source heterogeneous data such as elevator car vertical acceleration signals, traction machine output torque signals and motor encoder rotating speed signals are integrated, synchronization data streams are generated based on timestamp alignment and running state segmentation, and the problem of feature misjudgment caused by data asynchronization or state mixing is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of elevator operating status evaluation, and in particular to an elevator operating status evaluation method and system based on multi-source heterogeneous data. Background Art

[0002] The technical field of elevator operation status assessment lies at the intersection of elevator safety monitoring and intelligent operation and maintenance. Its core goal is to achieve quantitative evaluation of elevator operation health and abnormal warning through multi-dimensional data collection, dynamic feature analysis and risk modeling.

[0003] Existing technologies use single indicator thresholds to identify anomalies, such as monitoring only acceleration or current signal outages. This ignores the correlations between mechanical and electrical parameters, making it difficult to distinguish between normal operating fluctuations and potential fault characteristics. Furthermore, existing risk assessment models fail to integrate external environmental data. For example, they fail to consider the impact of heavy rain on the friction coefficient of elevator guide rails or the short-term, high-frequency use caused by large-scale events. This results in conventional range settings deviating from operating requirements. Therefore, improvements are needed. Summary of the Invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an elevator operation status evaluation method and system based on multi-source heterogeneous data.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for evaluating elevator operating status based on multi-source heterogeneous data, comprising the following steps:

[0006] Collect elevator car vertical acceleration signals, traction machine output torque signals, traction machine output current signals, motor encoder speed signals, elevator operation log time, stop floor records, operation number records, and door opening and closing number records. Align each data item based on timestamps, and segment the data stream into time periods based on the elevator's start, running, and stop states to generate synchronized multi-source operation data streams.

[0007] Based on the synchronized multi-source operation data stream, a ratio sequence of instantaneous torque values and acceleration values during the elevator start-stop process is calculated, a difference sequence between the motor encoder speed signal and the speed estimate calculated based on the traction machine output torque signal is monitored, and rapid change points in the amplitude of the car vertical acceleration signal are detected to obtain a preliminary signal deviation index set. Based on the preliminary signal deviation index set, abnormal signal points are screened and determined to obtain a set of operation feature points;

[0008] Based on the synchronized multi-source operation data stream, the operation frequency of the time period is calculated to obtain basic operation statistics. Based on the basic operation statistics, the combination of the statistical values and external conditions is compared with weather conditions and large-scale event information to determine whether it deviates from the normal range, and to establish a context-related operation mode profile;

[0009] Based on the frequency of occurrence of abnormal signal points in the operating feature point set and the context-related operating mode portrait, the operating environment-related risk score value is calculated, and the operating environment-related risk score value is compared and judged with the preset attention alarm threshold to generate a comprehensive elevator operating status assessment conclusion.

[0010] Preferably, the steps of acquiring the synchronized multi-source operation data stream are:

[0011] Collect the elevator car vertical acceleration signal, traction machine output torque signal, traction machine output current signal, motor encoder speed signal, elevator operation log time, stop floor record, operation number record and door opening and closing number record, extract the timestamp field of each signal, calculate the absolute value of the difference between each signal timestamp and the reference time of the elevator operation log time, define the absolute value of the difference as the initial transmission delay value, and use cubic spline interpolation to fill the missing time points for signals whose absolute value of the difference exceeds the preset threshold, thus generating a multi-source signal time series data set with completely aligned timestamps;

[0012] Based on the multi-source signal time series data set with completely aligned timestamps, the state switching marks in the elevator operation log time are parsed to extract the starting time point of the start state, the duration of the running state, and the ending time point of the stop state. The data segment boundaries are divided according to the starting time point and the ending time point. The vertical acceleration signal of the car is cut into the start segment, the running segment, and the stop segment according to the boundaries. The traction machine output torque signal, the traction machine output current signal, and the motor encoder speed signal are synchronously divided into corresponding data segments to generate a segmented multi-source signal data set with state labels;

[0013] Based on the segmented multi-source signal data set with state labels, the mapping relationship between floor numbers and timestamps in the stop floor records is extracted, the floor numbers are matched to the corresponding state intervals according to the divided data segment boundaries, and the operation count records and door opening and closing count records are merged to the end of the data segment according to the timestamps to generate a synchronized multi-source operation data stream.

[0014] Preferably, the steps of obtaining the preliminary signal deviation indicator set are:

[0015] Extracting the instantaneous values of the torque signal and the acceleration signal during the elevator start-stop process based on the synchronized multi-source operation data stream, segmenting the data into segments according to time windows, and generating an instantaneous torque-acceleration data set for the start-stop process;

[0016] Calculate the torque to acceleration ratio at each time point based on the instantaneous torque-acceleration data set during the start-stop process , generate the torque-acceleration ratio sequence, the calculation formula is:

[0017] ;

[0018] in, For the The torque value at a time point, For the The acceleration value at a time point;

[0019] Based on the torque-acceleration ratio sequence, the speed signal recorded by the motor encoder is and the speed estimate derived from the motor output torque signal Perform point-by-point difference calculation to obtain the difference value , generate the speed-estimated speed difference value sequence, the formula is:

[0020] ;

[0021] Based on acceleration value , calculate the acceleration difference between adjacent time points ,in is the acceleration value at adjacent time points, and the mutation threshold is set to , for The mean of for The standard deviation of The time point of is the acceleration rapid change point, and the torque-acceleration ratio value sequence, the speed-estimated speed difference value sequence and the acceleration rapid change point are integrated to generate a preliminary signal deviation indicator set.

[0022] Preferably, the steps of obtaining the set of running feature points are:

[0023] Based on the preliminary signal deviation indicator set, the preset thresholds of the torque-acceleration ratio sequence, the preset thresholds of the speed-estimated speed difference sequence, and the amplitude thresholds of the acceleration rapid change points are extracted, the time window range is defined as the interval of 250 milliseconds before and after adjacent time points, and the multi-indicator threshold parameters and time window definition rules are generated;

[0024] According to the multi-index threshold parameters and time window definition rules, each time point in the preliminary signal deviation index set is traversed. If the torque-acceleration ratio corresponding to the current time point exceeds the preset threshold, and the speed-estimated speed difference value exceeds the preset threshold within the 250 millisecond window before and after the current time point, and the amplitude of the acceleration rapid change point within the window exceeds the threshold, then the current time point is marked as a candidate abnormal signal point, and a multi-index synchronous over-limit time point set is generated;

[0025] Based on the multi-index synchronous over-limit time point set, the candidate abnormal signal points with adjacent time intervals less than 300 milliseconds are merged. After merging, the time point with the largest deviation of the torque-acceleration ratio in each cluster is retained as the representative point, and the isolated candidate points without adjacent clusters are eliminated to generate the operation feature point set.

[0026] Preferably, the steps for obtaining the basic running statistics are:

[0027] Based on the elevator operation log time in the synchronized multi-source operation data stream, the complete time axis from 0:00 to 24:00 every day recorded in the log is extracted, and the time axis is divided into 48 continuous intervals with each 30 minutes as a statistical period. Combined with the timestamps in the stop floor records, the operation number records and door opening and closing number records in each period are counted to generate the operation-door opening and closing frequency data set after period division;

[0028] According to the operation-door opening and closing frequency data set divided by the time period, the operation frequency value of each time period is calculated , generate the period operation frequency sequence, the calculation formula is:

[0029] ;

[0030] in, For the The number of runs in each period, For the The number of door openings and closings in a period of time, is the time period length;

[0031] Based on the operating frequency sequence of the time period, the sliding mean and standard deviation of the operating frequency values of 6 consecutive time periods are calculated, and the time period where the sliding mean exceeds 2 times the standard deviation of the historical mean during the same period is marked as a high-frequency operating interval. The operating frequency values of all time periods are integrated to generate basic operating statistics.

[0032] Preferably, the steps for obtaining the context-related operation mode portrait are:

[0033] Based on the basic operation statistics, the operation frequency value and high-frequency operation interval mark of each statistical period are extracted, the rainfall and wind speed in the externally input weather condition data and the event scale and number of participants in the large-scale event information are associated, and the timestamps are aligned and matched to the corresponding statistical period to generate an external condition-operation statistic association data set;

[0034] Based on the external condition-operation statistic association data set, the mean and standard deviation of the operation frequency values in the past three months under the same rainfall and activity scale classification conditions in the historical database are compared. If the operation frequency value of the current period exceeds the range of two standard deviations of the historical mean, the current period is marked as a statistical period data that deviates from the normal range, and a statistical deviation flag set is generated;

[0035] Based on the statistical deviation mark set, the distribution density of abnormal signal points in the operating feature point set in large-scale scenarios is counted, and the percentage of abnormal signal points in the total number of operations in large-scale scenarios per unit time is calculated. Large-scale scenarios refer to rainfall levels of heavy rain or typhoons and large-scale activities with a scale of tens of thousands of people. If the percentage exceeds a preset threshold, it is determined to be a high-density abnormal scenario. All high-density abnormal scenarios and statistical time period data that deviate from the normal range are integrated to generate a scenario-related operating mode portrait.

[0036] Preferably, the steps for obtaining the operating environment associated risk score value are:

[0037] Based on the set of operating feature points, the number of occurrences of abnormal signal points in each statistical period is extracted. Combined with the high-density abnormal scenarios in the scenario-related operating mode portrait, the ratio of the number of occurrences of abnormal signal points in the high-density abnormal scenarios in each period to the total operating time is calculated to generate an abnormal frequency-high-density scenario association data set;

[0038] Based on the abnormal frequency-high-density context association data set, calculate the deviation percentage between the number of abnormal signal points and the average number of abnormal points in the same period of history, define the period when the deviation exceeds 20% as a risk-sensitive period, and generate a risk-sensitive period marker sequence;

[0039] Based on the risk-sensitive period mark sequence, the operating environment associated risk score value is calculated using the following formula:

[0040] ;

[0041] in, Associate a risk score value with the operating environment, For the The number of abnormal signal points in a period, For the The high-density abnormal situation impact factor for each period, here , is the number of abnormal points in the period, The total running time of the time period.

[0042] Preferably, the steps for obtaining the comprehensive elevator operation status evaluation conclusion are:

[0043] Based on the operating environment associated risk score value, extract the operating environment associated risk score value of each statistical period, compare it with the preset attention alarm threshold, define the threshold level interval, and generate a risk score data set with the threshold interval;

[0044] According to the risk score data set with threshold intervals, the risk score values of each statistical period are traversed. If the score value exceeds the severe alarm threshold, it is marked as "emergency state", if it exceeds the moderate alarm threshold but does not reach the severe threshold, it is marked as "high risk state", if it exceeds the mild concern threshold but does not reach the moderate threshold, it is marked as "potential risk state", and the rest are marked as "normal state", thereby generating a risk level state mark set;

[0045] Based on the risk level status mark set, the number of time periods marked as "emergency state" or "high-risk state" in three consecutive time periods is counted. If all three consecutive time periods are high-risk or contain one emergency state, an "immediate maintenance recommendation" conclusion is generated to obtain the comprehensive elevator operation status assessment conclusion.

[0046] The present invention provides an elevator operation status evaluation system, comprising:

[0047] The data synchronization module collects the car vertical acceleration signal, traction machine output torque signal, traction machine output current signal, motor encoder speed signal, operation log time, stop floor record, operation number record and door opening and closing number record, aligns the collection time stamp, divides the start state, operation state and stop state into independent time periods, and combines all signals in the divided time periods in series according to time sequence to establish a multi-source synchronous data stream;

[0048] The signal offset determination module, based on multi-source synchronous data streams, pairs the acceleration signals and torque signals in the start and stop states to obtain a ratio sequence. It also obtains a numerical difference sequence between the motor encoder speed signal and the speed estimate calculated from the traction machine output torque signal. It performs amplitude increment calculations between adjacent values in the sequence of the car vertical acceleration signal and marks the mutation points. It then screens the maximum value points in the ratio sequence, the absolute deviation points in the difference sequence, and the time period containing the mutation points to obtain a set of offset characteristic signals.

[0049] The operation frequency statistics module, based on multi-source synchronous data streams, counts the total number of daily operations, the number of daily changes in the number of floors stopped, the duration of each operation, and the total number of doors opened and closed each day. It extracts the value distribution interval within a 24-hour range on a daily basis, compares the difference between the distribution interval and the common distribution range in the historical operation cycle, and establishes the periodic operation frequency characteristics;

[0050] The situational feature characterization module, based on the periodic operation frequency characteristics, calls the weather condition information of each statistical day and the records of large-scale activities in the area, compares the total number of daily operations with the weather conditions, makes a combined judgment on the number of floor changes and the presence of large-scale activities, and combines the total number of daily door openings and closings with the temperature level. It establishes three types of combination relationships: the combination pattern of operation number and weather, the combination pattern of floor changes and activities, and the combination pattern of door openings and closings and temperature. It marks the combination types whose distribution exceeds the normal pattern interval, summarizes the combination types and the frequency of occurrence within the statistical time interval, and generates a situational combination operation map;

[0051] The risk scoring module assigns a risk level to the combination type whose offset number in the combination relationship is higher than the standard deviation interval based on the signal offset points marked in the offset feature signal set, and obtains the risk distribution scoring result.

[0052] Compared with the prior art, the advantages and positive effects of the present invention are:

[0053] This invention integrates multi-source heterogeneous data, including elevator car vertical acceleration signals, traction machine output torque signals, and motor encoder speed signals, and generates synchronized data streams based on timestamp alignment and operating state segmentation, eliminating feature misjudgment caused by data asynchrony or state congestion. By calculating the ratio sequence of torque and acceleration during elevator start-stop processes, and the difference sequence between the speed signal and the torque-derived speed, combined with the detection of rapid change points in the acceleration signal amplitude, a multidimensional deviation index is constructed from the perspective of the correlation between mechanical dynamic response and electrical control, improving the sensitivity and anti-interference capability of abnormal signal identification. Contextual correlation analysis of weather conditions and large-scale event information is introduced to establish a dynamic matching model between the external environment and operating statistics, enabling risk scores to reflect the comprehensive impact of complex operating conditions. A multi-level comparison mechanism of risk scores and preset thresholds combines discrete anomalies with continuous high-density scenarios, enabling hierarchical assessment from local anomalies to systemic risks, reducing the false alarm rate of single-threshold judgments and providing a priority ranking basis for maintenance decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 Schematic diagram of the steps of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0056] See also Figure 1 The present invention provides a technical solution, a method for evaluating the operation status of an elevator based on multi-source heterogeneous data, comprising the following steps:

[0057] Collect elevator car vertical acceleration signals, traction machine output torque signals, traction machine output current signals, motor encoder speed signals, elevator operation log time, stop floor records, operation number records, and door opening and closing number records. Align each data item based on timestamps, and segment the data stream into time periods based on the elevator's start, running, and stop states to generate synchronized multi-source operation data streams.

[0058] Based on synchronized multi-source operation data streams, the system calculates the ratio sequence of instantaneous torque and acceleration values during the elevator start-stop process, monitors the difference sequence between the motor encoder speed signal and the speed estimate calculated based on the traction machine output torque signal, detects the rapid change points of the car vertical acceleration signal amplitude, and obtains a preliminary signal deviation index set. Based on this preliminary signal deviation index set, it screens and determines abnormal signal points to obtain a set of operation feature points.

[0059] Based on synchronized multi-source operation data streams, the operation frequency of each period is calculated to obtain basic operation statistics. Based on these basic operation statistics, the combination of each statistical value and external conditions is compared with weather conditions and large-scale event information to determine whether it deviates from the normal range, and to establish a context-related operation mode profile;

[0060] Based on the frequency of occurrence of abnormal signal points in the set of operating feature points and the situation-related operating mode portrait, the operating environment-related risk score is calculated, and the operating environment-related risk score is compared with the preset attention alarm threshold to generate a comprehensive elevator operating status assessment conclusion.

[0061] The steps to obtain synchronized multi-source running data streams are:

[0062] Collect the elevator car vertical acceleration signal, traction machine output torque signal, traction machine output current signal, motor encoder speed signal, elevator operation log time, stop floor record, operation number record and door opening and closing number record, extract the timestamp field of each signal, calculate the absolute value of the difference between each signal timestamp and the reference time of the elevator operation log time, define the absolute value of the difference as the initial transmission delay value, and use cubic spline interpolation to fill the missing time points for signals whose absolute value of the difference exceeds the preset threshold, thus generating a multi-source signal time series data set with completely aligned timestamps;

[0063] Based on a multi-source signal time series data set with fully aligned timestamps, the state switching marks in the elevator operation log are parsed to extract the starting time point of the start state, the duration of the running state, and the end time point of the stop state. The data segment boundaries are divided according to the starting and ending time points. The vertical acceleration signal of the car is divided into the start segment, the running segment, and the stop segment according to the boundaries. The traction machine output torque signal, the traction machine output current signal, and the motor encoder speed signal are synchronously divided into corresponding data segments to generate a segmented multi-source signal data set with state labels.

[0064] Based on a segmented multi-source signal dataset with state labels, the mapping relationship between floor numbers and timestamps in the stop floor records is extracted. The floor numbers are matched to the corresponding state intervals according to the divided data segment boundaries. The operation count records and door opening and closing count records are merged to the end of the data segment according to the timestamps to generate a synchronized multi-source operation data stream.

[0065] Specifically, the acquisition of synchronized multi-source operation data streams is based on the collection of multiple original elevator signals, including the car's three-axis acceleration signals (focusing on the vertical direction, in m / s²) collected by the MEMS accelerometer installed on the top or bottom of the car, the output torque signal (in Nm) collected by the torque sensor installed on the traction motor main shaft or the output shaft of the reduction gearbox, the three-phase current signal (in A) collected by the phase current sensor installed at the motor input end via a current transformer, and the motor speed signal (in r / min) output by the rotary encoder installed coaxially with the traction motor. At the same time, the operation log timestamp (accurate to milliseconds) containing information such as operation status and fault codes is read from the elevator control system log database. , the controller records the stop floor events and timestamps, the operation counter value and the corresponding timestamps, and the door opening and closing completion event timestamps recorded by the door machine system. First, extract the timestamp field in each signal data record. For example, read a traction machine torque signal record as {Timestamp: 1678886401500, Torque: 120.5}, its timestamp is 1678886401500 milliseconds, and read a corresponding elevator operation log record as {LogTime: 1678886401450, Event: 'Start closing the door'}, its timestamp is 1678886401450 milliseconds. Calculate the absolute value of the timestamp base time difference between the two, that is, milliseconds, the 50 milliseconds is the initial transmission delay value of this torque signal sampling point relative to the log record. For each sampling point of all collected signals (acceleration, torque, current, speed), the difference calculation is performed on the timestamp of the elevator operation log closest to it in time, and a preset transmission delay threshold is set. The threshold is determined based on the statistical analysis of historical data transmission delays. For example, the mean and standard deviation of the signal delay time in the past 24 hours are statistically analyzed. If the mean is 40 milliseconds and the standard deviation is 15 milliseconds, the preset threshold can be set to the mean plus three times the standard deviation, that is, If the absolute value of the difference calculated at a certain signal sampling point, for example, 88 milliseconds, exceeds the threshold of 85 milliseconds, it is considered that the signal point has a significant delay or timestamp anomaly and needs to be processed. For these signal points with differences exceeding the threshold and the possible missing time points of the data series caused by them, the cubic spline interpolation method is used to fill them. Specifically, the time points marked as invalid due to excessive delays and the time points that may be completely missing due to reasons such as network packet loss are identified, and the nearest valid data points are found before and after them. Based on the values and timestamps of these valid data points, the data curve is fitted using a piecewise cubic polynomial to ensure that the function value, first-order derivative and second-order derivative at the interpolation point are correct. The number of continuous signals is used to calculate the estimated signal value at the missing time point. For example, if acceleration data is missing at time point T, the nearest valid point before T (T1, A1) and the nearest valid point after T (T2, A2), as well as their respective adjacent points, are found, and a cubic spline function is constructed to solve the estimated acceleration value A_interp corresponding to time point T. All signals (acceleration, torque, current, and speed) are processed in this way so that their timestamp sequences are strictly aligned to a unified reference time axis (usually based on the running log time, or one of the high-frequency signals is selected as the reference and then the other signals are aligned), generating a multi-source signal time series data set with completely aligned timestamps.

[0066] Based on a multi-source signal time series data set with fully aligned timestamps, the operation logs recorded by the elevator control system are further analyzed. These log entries contain key information describing the transitions in the elevator's operating state, such as text tags "elevator starts running," "elevator reaches the target floor and begins to decelerate," "elevator stops at the leveling floor," "door opening command issued," and "door closing action completed." By identifying these specific state switching marker texts or corresponding status codes and extracting the precise timestamps accompanying these markers, the key time nodes of the elevator's operating cycle are determined. Specifically, log entries marked with "start" or similar meanings are searched for and their timestamps are recorded as the starting time points of the start state. , find the log entry marked as "stop", "leveling in place" or similar, and record its timestamp as the end time point of the stop state , defined from A short time period at the beginning (for example, set to 1.5 seconds based on experience, i.e. arrive ) is the start acceleration phase, which is defined as a time point before the stop signal appears (for example, according to the typical deceleration time setting )arrive To stop the deceleration phase, then between the start acceleration phase end ( ) and the deceleration phase starts ( ) is the duration of the running state. According to the starting time point determined in this way and end time point (and the middle dividing point and ) as the boundaries of the data segments, and cut the car vertical acceleration signal sequence after the timestamps are aligned according to these boundary points. For example, if a complete operation cycle starts from the timestamp 1678886405000 and stops completely at 1678886415000, the data interval of the start segment is [1678886405000, 1678886406500], the data interval of the operation segment is (1678886406500, 1678886413000), and the data interval of the stop segment is [1678886413000, 1678886415000]. At the same time, using exactly the same edge The bounded timestamps [1678886405000, 1678886406500, 1678886413000, 1678886415000] are used to synchronously segment the motor output torque signal sequence, the motor output current signal sequence, and the motor encoder speed signal sequence with aligned timestamps. This ensures that each signal is divided into start, run, and stop segments corresponding to the same time range. Finally, a corresponding state label, such as "start," "run," or "stop," is assigned to each segmented data segment (whether it is an acceleration, torque, current, or speed signal segment), generating a segmented multi-source signal dataset with state labels.

[0067] Based on the segmented multi-source signal dataset with state labels, the discrete event information is further integrated. First, the stop floor information recorded by the elevator controller is extracted. This information usually contains the floor number (such as the physical floor number "5" or the logical number) and the timestamp of the stop event on that floor (for example, the door opening timestamp 1678886415100). This timestamp 1678886415100 is analyzed and matched with the data segment boundaries divided in the previous step to determine which state interval of the operation cycle the stop event belongs to. Since the stop occurs at the end of the operation, the floor number "5" will be closed. Connect to the stop segment corresponding to the timestamp 1678886415000, or immediately after it, and establish a mapping relationship between the floor number and the corresponding operation cycle state interval. Then, process the operation number record and the door opening and closing number record. These records may be cumulative values or single event records. If it is a single event record (such as a run completion event, timestamp 1678886415050; a door opening and closing cycle completion event, timestamp 1678886418000), then according to its timestamp, merge it into the end data segment of the corresponding operation cycle (usually the stop segment or the door opening and closing action segment). For example, the run completion event timestamp 1678886415050 should be attributed to the stop segment of the run cycle that just ended, and the door opening and closing count record 1678886418000 should also be attributed to the subsequent door operation time associated with the stop segment. If the counter cumulative value at a certain point in time is recorded (for example, the total number of runs at 1678886400000 is 1025, and the total number of runs at 1678886420000 is 1026), the number of occurrences in the time period is determined by calculating the difference between adjacent record points (1026-1025=1 run). The operation is attributed to the end of the stop segment of the operation cycle that occurred within the time period [1678886400000, 1678886420000]. The extracted and mapped floor number, as well as the number of operations and door openings merged by timestamp, are appended to the metadata of the corresponding data segment to form a structured data stream, in which each entry represents a complete operation cycle segment with a status label (start, run, stop), and additional context information (stop floor, operation and door opening count within the cycle), generating a synchronized multi-source operation data stream.

[0068] The steps to obtain the preliminary signal deviation indicator set are:

[0069] Based on synchronized multi-source operation data streams, the instantaneous values of the torque signal and the acceleration signal during the elevator start-stop process are extracted, and the data segments are divided into time windows to generate the instantaneous torque-acceleration data set during the start-stop process;

[0070] Calculate the torque to acceleration ratio at each time point based on the instantaneous torque-acceleration data set during the start-stop process , generate the torque-acceleration ratio sequence, the calculation formula is:

[0071] ;

[0072] in, For the The torque value at a time point, For the The acceleration value at a time point;

[0073] Based on the torque-acceleration ratio sequence, the speed signal recorded by the motor encoder and the speed estimate derived from the motor output torque signal Perform point-by-point difference calculation to obtain the difference value , generate the speed-estimated speed difference value sequence, the formula is:

[0074] ;

[0075] Based on acceleration value , calculate the acceleration difference between adjacent time points ,in is the acceleration value at adjacent time points, and the mutation threshold is set to , for The mean of for The standard deviation of The time point of is the acceleration rapid change point, and the torque-acceleration ratio value sequence, the speed-estimated speed difference value sequence and the acceleration rapid change point are integrated to generate a preliminary signal deviation indicator set.

[0076] Specifically, based on the synchronized multi-source operation data stream, the instantaneous value of the traction motor output torque signal during the two dynamic processes of elevator starting and stopping is extracted. and the instantaneous value of the car vertical acceleration signal , filter out all data segments marked as "start-up segment" and "stop-up segment", and pair the time in these segments with the corresponding torque and acceleration values. For example, for a start-up segment, its data may be [(t1, M1, A1), (t2, M2, A2), ..., (tn, Mn, An)]. Merge all the data points of the start-up segment and the stop segment to form a set, which is recorded as the instantaneous torque-acceleration data set of the start-stop process. Then, based on this data set, calculate each time point The torque to acceleration ratio , apply the formula Calculate, where It's at the time The measured torque value, At the same time The measured acceleration value needs to be processed carefully To avoid division by zero errors, for example, you can set a very small positive number , such as 0.01m / s², when When the ratio of this point is not calculated or marked as invalid, for example, at the moment of startup, When the torque Nm, acceleration m / s², then the ratio Nm / (m / s²), repeat this calculation for all valid time points in the starting and stopping segments to generate a time series of torque-acceleration ratio values Then, based on the complete synchronized data stream (including the running segment), the speed signal directly recorded by the motor encoder is (Unit: m / s, converted from r / min, for example, by ,in is the speed in r / min, The diameter of the traction wheel is 0.5m, is the reduction ratio) and the speed estimate calculated by the motor model using the traction machine output torque signal. Perform point-by-point difference comparison and speed estimation middle ,in is a simplified coefficient, for each time point , calculate the absolute value of the difference between the two , for example at time point , encoder measures speed m / s, model estimated speed m / s, then the difference m / s, generates a sequence of speed-estimated speed difference values , and furthermore, using the acceleration signal (Also a complete data stream), calculate the absolute value of the acceleration difference between adjacent time points , for example at time point , acceleration m / s², next time point , acceleration m / s², then m / s², calculate the entire sequence After the value is set, an acceleration mutation threshold needs to be set To identify fast-changing points, the threshold Defined as ,in yes The mean of the series over a representative period of time (such as the past 24 hours), is the standard deviation of the sequence. The specific calculation process is: collect all the The arithmetic mean of the values is calculated to get , calculate its standard deviation to get , for example, we can calculate m / s², m / s², then the mutation threshold m / s², all calculated The value is compared with this threshold. If m / s², then the time point (i.e. the time point after the change occurs) is marked as the point of rapid acceleration change. Finally, the generated torque-acceleration ratio sequence , speed-estimated speed difference value sequence , and the timestamps of all points marked as rapid acceleration changes are integrated to generate a preliminary signal deviation indicator set.

[0077] formula illustrate:

[0078] This formula is used to calculate the instantaneous torque-acceleration ratio when the elevator starts or stops. Representatives in the The torque-acceleration ratio at a given time point. The torque sensor is used at the time The collected instantaneous value of the traction machine output torque. It is measured by the acceleration sensor at the same time The collected instantaneous value of the vertical acceleration of the car. Indicates the time point index corresponding to the value The calculation logic of this formula is to divide the instantaneous driving force (proportional to torque) by the generated instantaneous acceleration to obtain a quantity reflecting the combined effects of system inertia and load. Its physical meaning is similar to the system's equivalent moment of inertia or mass (depending on the specific relationship between force and motion). This ratio is calculated to monitor the consistency of the relationship between driving force and motion response during the elevator's dynamic process.

[0079] Example: For example, at a certain point in the startup process , the collected torque Nm, acceleration collected at the same time m / s². Substitute into the formula for calculation:

[0080] ;

[0081] The calculation result Nm / (m / s²) is part of the preliminary set of signal deviation indicators.

[0082] formula illustrate:

[0083] This formula is used to calculate the , the actual speed measured by the motor encoder The speed is estimated based on the model (e.g. using the motor output torque signal). The absolute difference between . Representatives in the The speed difference value at each time point. The instantaneous speed of the car is directly measured and converted by the motor encoder. Indicates that this is the first points. It is the estimated speed calculated by using other signals (mainly torque signal) and motor / system model. It also means the first Point. Absolute value symbol The non-negative value of the difference between the two is taken. The calculation logic of this formula quantifies the deviation between the actual measured speed and the speed predicted by the model. This deviation can reflect problems with the motor control performance, model accuracy, or the encoder itself.

[0084] Example: For example, at a certain point in time during operation , the encoder measures the speed m / s. Based on the torque value at that time and a simplified model ,For example Nm, then the estimated speed m / s. Substitute into the formula for calculation:

[0085] ;

[0086] The calculation result m / s is another part of the set of preliminary signal deviation indicators.

[0087] formula and illustrate:

[0088] Calculate two adjacent time points and The absolute change in acceleration between the two accelerations, that is, the absolute value of the acceleration difference, can be regarded as a discrete approximation of the rate of change of acceleration (jerk). It's time The acceleration value, The previous time point This calculation is used to detect sudden changes in acceleration during operation, which may indicate rough operation, vibration, or shock.

[0089] Defines the threshold for determining whether acceleration changes are "rapid" or "abrupt". is the mutation threshold. is calculated from all the history (e.g. the past day) The arithmetic mean of the values. These The standard deviation of the values. This formula adopts the "3 sigma principle" in statistics, that is, if the data roughly follows a normal distribution, then about 99.7% of the data will fall within the range of the mean plus or minus three standard deviations. Points outside this range are considered statistical anomalies or low-probability events. The logic here is that if a certain acceleration change exceeds the normal fluctuation range (given by definition), it is considered a significant and noteworthy rapid change point.

[0090] Example: For example, the mean acceleration difference is calculated using historical data m / s², standard deviation m / s². Then the threshold m / s². Now calculate the current time point The acceleration difference. m / s², m / s², then m / s². Because , so the time point This is marked as a point of rapid acceleration change. This marking information is also part of the preliminary signal deviation indicator set.

[0091] The benefits of the formula are:

[0092] By monitoring the relationship between driving force and motion response, it helps to identify abnormal loads or changes in mechanical resistance; Comparing the measured speed with the model-estimated speed can help detect control system deviations or sensor failures; and The combined use of these indicators can objectively identify bumps or impact events in operation through statistical methods. These indicators together constitute a preliminary quantitative assessment of the signal's deviation from normal behavior patterns.

[0093] This result shows that the calculated ratio , difference value , acceleration difference The specific values (and the mark of whether it exceeds the threshold) constitute the preliminary signal deviation indicator set. These numerical sequences and marking points provide basic data for the subsequent identification of operating characteristic points.

[0094] The steps to obtain the running feature point set are:

[0095] Based on the preliminary signal deviation indicator set, the preset thresholds for the torque-acceleration ratio sequence, the preset thresholds for the speed-estimated speed difference sequence, and the amplitude thresholds for rapid acceleration change points are extracted. The time window range is defined as the interval of 250 milliseconds before and after adjacent time points, and multi-indicator threshold parameters and time window definition rules are generated.

[0096] According to the multi-indicator threshold parameters and time window definition rules, each time point in the preliminary signal deviation indicator set is traversed. If the torque-acceleration ratio corresponding to the current time point exceeds the preset threshold, and the speed-estimated speed difference value exceeds the preset threshold within the 250 millisecond window before and after the current time point, and the amplitude of the acceleration rapid change point within the window exceeds the threshold, then the current time point is marked as a candidate abnormal signal point, and a multi-indicator synchronous over-limit time point set is generated;

[0097] Based on the multi-index synchronous over-limit time point set, the candidate abnormal signal points with adjacent time intervals less than 300 milliseconds are merged. After merging, the time point with the largest deviation of the torque-acceleration ratio in each cluster is retained as the representative point, and the isolated candidate points without adjacent clusters are eliminated to generate the operation feature point set.

[0098] Specifically, based on the preliminary signal deviation index set, it is necessary to first set various thresholds for screening operating characteristic points. These thresholds need to be determined by referring to the historical data distribution under normal operating conditions of the equipment, for example, the torque-acceleration ratio sequence The preset threshold It can be set as the 95th percentile of the ratio distribution under normal operating conditions, for example, by analyzing historical normal data and calculating The 95% quantile of is 195Nm / (m / s²), so set , speed-estimated speed difference value sequence The preset threshold It can be set according to the acceptable control accuracy or model error. For example, if the speed tracking error is required to be no more than 0.1m / s, then set m / s, the amplitude threshold of the acceleration rapid change point It is for those who have passed Further screening of the rapid change points filtered by the threshold is used to focus on shocks with particularly large amplitudes, which can be set to 1.5 times the value or a fixed experience value, such as setting m / s², and define the time window range as the current observation time point The interval of 250 milliseconds before and after, that is , a total duration of 500 milliseconds, these thresholds ( ) and the time window rule (500ms) together constitute the multi-indicator threshold parameters and time window definition rules. Then, according to these rules, traverse each time point in the initial signal deviation indicator set , execute the synchronization over-limit judgment logic: check the current time point The torque-acceleration ratio Whether it exceeds its threshold ,Right now If this condition is met, further check is performed on The central time window Is there at least one time point within , its speed-estimated speed difference Exceeding the threshold ,Right now If this condition is also met, then check the same time window again Is there at least one time point marked as a rapid acceleration change point within (i.e. its corresponding ), and the magnitude of the acceleration change The amplitude threshold must also be exceeded ,Right now , only when these three conditions ( Exceed limit, exist within the window When the limit is exceeded and there is a large rapid acceleration change point in the window, the current time point will be Marked as candidate abnormal signal points, for example, at time point , calculated Nm / (m / s²), because , the first condition is met, and then in the time window Search within and find at the time point hour, m / s, because , the second condition is met, and the search continues within the window. It is found that at the time point is marked as a rapid change point, and its change amplitude m / s², because , satisfies the third condition, so the time point It is marked as a candidate abnormal signal point. All time points that pass the three-condition screening are collected to form a multi-index synchronous over-limit time point set. Then, based on this set, the candidate points that are too close to each other are processed. A time interval threshold is set to 300 milliseconds. The candidate points in the set are sorted by time. If two adjacent candidate points are found and The time difference between If the time is less than 300 milliseconds, they are considered as an abnormal event cluster, and all adjacent points that meet this condition are merged into the same cluster. For each formed cluster, the torque-acceleration ratio values of all candidate points in the cluster are compared. , select that The time point with the largest value is taken as the only representative point of the cluster and retained. Finally, all the remaining representative points after merging and filtering are checked. If a representative point is not incorporated into any cluster (that is, there are no other candidate points within 300ms before and after it), it is regarded as an isolated point and is eliminated. Finally, all the representative points retained constitute the running feature point set.

[0099] The steps to obtain basic running statistics are:

[0100] Based on the elevator operation log time in the synchronized multi-source operation data stream, the complete time axis recorded in the log from 0:00 to 24:00 every day is extracted. The time axis is divided into 48 continuous intervals with each 30 minutes as a statistical period. Combined with the timestamps in the stop floor records, the number of runs and door openings in each period are counted to generate the run-door opening and closing frequency dataset after period division.

[0101] Calculate the operating frequency value for each period based on the operation-door opening and closing frequency data set divided into time periods , generate the period operation frequency sequence, the calculation formula is:

[0102] ;

[0103] in, For the The number of runs in each period, For the The number of door openings and closings in a period of time, is the time period length;

[0104] Based on the period operation frequency sequence, the sliding mean and standard deviation of the operation frequency values of 6 consecutive periods are calculated. The period where the sliding mean exceeds 2 times the standard deviation of the historical mean for the same period is marked as a high-frequency operation interval. The operation frequency values of all periods are integrated to generate basic operation statistics.

[0105] Specifically, based on the elevator operation log time and related counting records in the synchronized multi-source operation data stream, the complete daily time record is first extracted from the log, covering the time range from 0:00:00 to 23:59:59. Then, the time axis of the whole day is divided into a statistical period of fixed length of 30 minutes, and a total of 48 consecutive time intervals are obtained, with interval numbers from j=1 (00:00:00-00:29:59) to j=48 (23:30:00-23:59:59). Combined with the already merged stop floor records, operation number records and door opening and closing number records in the synchronized data stream, the statistics fall in each 30-minute period. The total number of runs within And the total number of door openings and closings For example, for period j = 19 (09:00:00-09:29:59), the number of runs is accumulated by traversing the count information appended to the end of all run cycle records within this time period. times, number of times the door is opened and closed These statistical results are organized into a data set, where each entry contains the time period index j, the corresponding number of runs and door opening and closing times , forming an operation-door opening and closing frequency data set after time period division, as shown in Table 1 below.

[0106] Table 1 Example of operation and door opening and closing frequency during period

[0107]

[0108] As shown in Table 1, the table lists some 30-minute time periods and their corresponding operation times and door opening and closing times. Based on this frequency data set, the following is calculated for each time period: The comprehensive operating frequency value , using the formula Calculate, where It is The number of times the period is run, It is Number of door openings and closings during a period of time, is the time period length, which is fixed at 30 minutes here, so , for example, to calculate the running frequency of period j=19: Events / minute. This calculation is performed for all 48 periods, generating a period-running frequency sequence containing 48 frequency values. ,Afterwards, based on this operating frequency sequence, the high-frequency operating interval is identified, and the sliding window method is used. The window size is set to 6 consecutive periods (i.e., 3 hours). For each window (for example, the window from period j=18 to j=23), the operating frequency values of these 6 periods are calculated ( arrive ) and standard deviation At the same time, it is necessary to query the historical database to obtain the average operating frequency of the elevator in the same time period on the same day of the week in history (for example, 08:30-11:30 on all historical Tuesdays) and standard deviation , set the high frequency judgment condition to the average frequency of the current sliding window Is it greater than the historical mean plus twice its standard deviation? Is it true? For example, for the window from period 18 to 23, the current Times / minute, query the historical period times / minute, times / minute, then compare and ,because , so this time window (08:30-11:30) is marked as the high-frequency operation interval, and the operation frequency values of all time periods are The high-frequency operation interval marking information obtained according to the sliding window comparison result is integrated together to generate the basic operation statistics.

[0109] formula illustrate:

[0110] This formula is used to calculate the The comprehensive operation frequency of elevators within a statistical period. Representative period The running frequency value is usually expressed in "times / minute" or "times / hour". It is during the period The total number of elevator operations recorded in the timer. It is during the period The total number of elevator door opening and closing cycles recorded in the It is the time period The calculation logic of this formula is to add the number of two major elevator activities (running and door opening and closing) in a time period, and then divide it by the total length of the time period to obtain an average activity rate per unit time, which is used to quantify the intensity of elevator usage or busyness during the time period.

[0111] Example: For period j=19 (09:00:00-09:29:59), we know from Table 1 that Second-rate, Times. Period length Minutes. Substitute into the formula to calculate:

[0112] ;

[0113] The calculation result Times / minute is part of the basic operation statistics, which means that during the period from 9:00 to 9:30 in the morning, the elevator has an average of about 1.67 operation or door opening and closing events per minute.

[0114] The results show that the calculated frequency values for each period A basic data sequence reflecting the daily usage pattern of the elevator is formed, which, combined with the markers of the high-frequency interval, constitutes the basic operation statistics, providing a basis for subsequent contextual correlation analysis.

[0115] The steps to obtain the context-related operation mode portrait are as follows:

[0116] Based on basic operating statistics, we extract the operating frequency value and high-frequency operating interval mark for each statistical period. We then correlate the rainfall and wind speed in the external weather data with the event scale and number of participants in the large-scale event information. We then align the timestamps to the corresponding statistical period to generate a dataset associated with external conditions and operating statistics.

[0117] Based on the external condition-operation statistics association data set, the mean and standard deviation of the operation frequency values in the past three months under the same rainfall and activity scale classification conditions in the historical database are compared. If the operation frequency value of the current period exceeds the range of two standard deviations of the historical mean, the current period is marked as a statistical period data that deviates from the normal range, and a statistical deviation flag set is generated;

[0118] Based on the statistical deviation mark set, the distribution density of abnormal signal points in the operation feature point set in large-scale scenarios is counted, and the percentage of abnormal signal points in the total number of operations in large-scale scenarios per unit time is calculated. Large-scale scenarios refer to rainfall levels of heavy rain or typhoons and large-scale activities with a scale of tens of thousands of people. If the percentage exceeds the preset threshold, it is determined to be a high-density abnormal scenario. All high-density abnormal scenarios and statistical period data that deviate from the normal range are integrated to generate a scenario-related operation mode portrait.

[0119] Specifically, based on the running frequency value of each statistical period (48 in total) included in the basic running statistics The context association analysis starts with the introduction of external environment and activity information data, such as hourly rainfall data (in mm) and average wind speed data (in m / s) from the meteorological service interface, and information about large-scale events from urban management or event organizers, including event location, time, estimated scale (such as the number of participants: <1,000, 1,000-5,000, 5,000-10,000, >10,000) and activity information. The dynamic type is to align and match these external data with the 30-minute period of elevator operation statistics according to their recorded timestamps. If there are multiple external data records in a period, the average value or the most relevant value can be taken. For example, for period j = 30 (15:00-15:30), the average rainfall in this period is 5mm and the average wind speed is 3m / s. At the same time, it is found that a venue nearby is holding an exhibition with a scale of 8,000 people. These external conditions (rainfall 5mm, wind speed 3m / s, event scale 8,000 people) are matched with the operation statistics of this period (such as times / minute, not marked as high-frequency intervals) to form an external condition-operation statistics association data set. Then, using this association data set, compare the past operation data stored in the historical database (for example, data from the past three months) to filter out records with the same or similar external conditions as the current period. The similarity here needs to be defined. For example, the rainfall is classified into categories (no rain, light rain, moderate rain, heavy rain, and rainstorm), and the activity scale is classified into categories (as above). Then, find all time period records in the historical database with the same rainfall level of "light rain" and the same activity scale of "5,000-10,000 people" and calculate the operation frequency in these historical records. The mean and standard deviation , and then set the running frequency value of the current period Compared with the statistical range under the same historical situation, the judgment condition is For example, query the history to find that when it is light raining and there are 5k-10k people active, The historical average times / minute, standard deviation times / minute, the calculation comparison threshold is , because currently Therefore, the current period j=30 is marked as a statistical period data that deviates from the normal range, forming a statistical deviation mark set. Then, it is necessary to analyze the distribution density of operating characteristic points (abnormal signal points) under specific "large-scale scenarios". First, define what is a "large-scale scenario". According to the text description, large-scale scenarios include rainfall levels judged as "rainstorm" (for example, hourly rainfall > 16mm) or "typhoon" (for example, average wind speed > 32.7m / s), and event scales judged as "large-scale events with 10,000 people" (number of participants > 10,000 people). Find all 30-minute statistical periods that meet any large-scale scenario conditions and count the total number of operating characteristic points (abnormal signal points) that appear in these large-scale scenario periods. , and count the total number of elevator operations during these same time periods , calculate the occurrence density of abnormal signal points in a large-scale scenario, that is, the proportion of abnormal points in the number of unit runs: , set a density threshold ,The threshold is set according to the risk tolerance, e.g. , which means that if there are more than 5 abnormal signal points per 100 runs in a large scenario, the density is considered too high. , then the current operating state is determined to be in a "high-density abnormal situation". Finally, all time period information marked as "high-density abnormal situation" and the previously generated "statistical deviation mark set" (that is, those time period data with significantly higher operating frequencies under specific external conditions) are integrated to form a situation-related operating mode portrait.

[0120] The steps to obtain the operating environment associated risk score are as follows:

[0121] Based on the set of operating feature points, the number of abnormal signal points occurring in each statistical period is extracted. Combined with the high-density abnormal scenarios in the context-related operating mode portrait, the ratio of the number of abnormal signal points occurring in the high-density abnormal scenarios in each period to the total operating time is calculated to generate an abnormal frequency-high-density scenario association dataset.

[0122] Based on the abnormal frequency-high-density context association data set, the deviation percentage between the number of abnormal signal points and the average number of abnormal points in the same period of history is calculated. The period with a deviation exceeding 20% is defined as a risk-sensitive period, and a risk-sensitive period marker sequence is generated.

[0123] Based on the risk-sensitive period mark sequence, the operating environment associated risk score is calculated using the following formula:

[0124] ;

[0125] in, Associate a risk score value with the operating environment, For the The number of abnormal signal points in a period, For the The high-density abnormal situation impact factor for each period, here , is the number of abnormal points in the period, The total running time of the time period.

[0126] Specifically, based on the set of operating feature points generated in the previous step (recording the timestamp of each abnormal signal) and the situation-related operating mode portrait (identifying the high-density abnormal situation period), first count each 30-minute statistical period The total number of operating characteristic points (abnormal signal points) that occur within Then, combined with the label information about "high-density abnormal situations" in the portrait, the time period marked as high-density abnormal situations is counted. The number of abnormal signal points (here equal , because it is a statistic for a specific period k) and the period Total running time of the elevator , total running time It is obtained by accumulating the duration of the "running segments" of all running cycles within the 30-minute period. For example, period k=35 (17:00-17:30) is marked as a high-density abnormal situation, during which Abnormal signal points, after accumulating the running time of the segment, the total running time of the period is obtained seconds, and convert this information, that is, the number of abnormal points in each period k , whether it is a high-density situation, and the number of abnormal points in a high-density situation and total runtime , associated and stored to form an abnormal frequency-high density situation association data set. Then, based on this data set, calculate each time period The number of abnormal signal points The average number of abnormal points compared to the same period in history (for example, the same day of the week and the same time period in the past month) The percentage deviation is calculated as , set a risk sensitivity threshold, for example 20%, if , then the time period Mark as risk-sensitive period, for example, period k=35 , the average number of abnormal points in the same period of history , then the deviation ,because , so period k=35 is marked as a risk-sensitive period, and a sequence containing risk-sensitive labels of all periods is generated. Finally, based on the risk-sensitive period label sequence and the previously calculated parameters, the risk of each period is calculated. The risk score associated with the operating environment , apply the formula Calculate, where It is The number of abnormal signal points in a period, It is The high-density abnormal situation impact factor for each period is calculated as follows: , that is, the number of abnormal points in the period (In high density scenarios, this is equal to ) divided by the total running time of the period , for non-high-density abnormal situation periods, The value of may be 0 or other reference values, for example, only meaningful calculations are made during high-density scenarios. , for period k=35 (high density abnormal situation), , , Seconds, calculate the impact factor first , and then substitute it into the risk scoring formula:

[0127] ;

[0128] For all 48 periods Repeat this risk score calculation to obtain the operating environment associated risk score value for each period .

[0129] formula and illustrate:

[0130] Calculate the The impact factor of high-density abnormal situations in each period. Represents the impact factor value. It is during the period The number of abnormal signal points that occur within a certain period (especially during periods defined as high-density abnormal situations). It is the time period The total actual operation time of the elevator (in seconds). This factor quantifies the occurrence rate of abnormal signal points per unit operation time in a specific (high-density) scenario.

[0131] Calculate the The risk score value associated with the operating environment in each period. is the final risk score. It is the time period The total number of abnormal signal points within. is the impact factor calculated above. This formula structure is similar to the calculation of vector in two-dimensional space A trigonometric function value of (or related to) , but a more direct understanding is that it combines two risk indicators (one is the number of times) that may have very different scales , one is the ratio ) and normalized by the denominator. is the Euclidean length of the vector. and When one is much larger than the other, the score It will be closer to smaller values (relative to its own scale), playing a balancing role. It ensures that the score value is bounded (theoretically between 0 and Near, depends on the specific value, more precisely, its value range depends on and The formula aims to combine the total number of anomalies and their intensity in a specific context to produce a comprehensive risk measure.

[0132] Example: Taking time period k=35 as an example, this time period is determined to be a high-density abnormal situation. An abnormal point, abnormal points, total running time Second.

[0133] First calculate the impact factor :

[0134] ;

[0135] Then calculate the risk score :

[0136] ;

[0137] ;

[0138] The calculation result It is the risk score associated with the operating environment in period 35, which will be used in the final status assessment.

[0139] The formula is useful in that by combining the total number of abnormal events ( ) and its incidence in specific high-risk situations ( ), and using the normalization method, It provides a comprehensive and scalable assessment of the risk level in a single period, which can more comprehensively reflect the potential risks, especially after taking into account external environmental pressure factors.

[0140] The result It is a key indicator for quantifying the degree of risk in each time period. The value itself directly reflects the level of risk and provides a direct basis for the subsequent classification of risk levels and making assessment conclusions.

[0141] The steps to obtain the conclusion of the comprehensive elevator operation status assessment are as follows:

[0142] Based on the operating environment-related risk score value, extract the operating environment-related risk score value for each statistical period, compare it with the preset attention alarm threshold, define the threshold level interval, and generate a risk score dataset with threshold intervals;

[0143] Based on the risk score dataset with threshold intervals, the risk score values of each statistical period are traversed. If the score value exceeds the severe alarm threshold, it is marked as "emergency state", if it exceeds the moderate alarm threshold but does not reach the severe threshold, it is marked as "high risk state", if it exceeds the mild concern threshold but does not reach the moderate threshold, it is marked as "potential risk state", and the rest are marked as "normal state". The risk level status mark set is generated;

[0144] Based on the risk level status mark set, the number of time periods marked as "emergency state" or "high-risk state" in three consecutive time periods is counted. If all three consecutive time periods are high-risk or contain one emergency state, an "immediate maintenance recommendation" conclusion is generated, and the comprehensive elevator operation status assessment conclusion is obtained.

[0145] Specifically, based on the calculated statistical time periods The risk score associated with the operating environment First, a set of corresponding rules between risk scores and status levels needs to be established. This requires pre-setting thresholds for attention, moderate warning, and severe warning. The setting of these thresholds should be based on the analysis of actual equipment status and fault records corresponding to different risk score values in historical data, as well as industry safety standards and expert experience. For example, the threshold for mild attention can be set as follows: , moderate alarm threshold , severe alarm threshold ,These thresholds define the risk score range: [0, 0.02) is the normal range, [0.02, 0.05) is the potential risk range, [0.05, 0.10) is the high risk range, [0.10, +∞) is the emergency range, and all time periods are Risk score value Compare with these thresholds to form a risk score dataset with threshold intervals. Then, based on this risk score dataset with threshold intervals, traverse each statistical period Risk score value , perform status determination and marking: If (For example ), then mark the period as "emergency"; if (For example ,satisfy ), it is marked as "high risk status"; if (For example, the previously calculated ,satisfy ), it is marked as "potential risk status"; if (For example ), it is marked as "normal state". After all 48 time periods are marked, a set of marks containing the risk level status of each time period is generated. Finally, based on this risk level status mark set, a comprehensive evaluation is performed to generate the final conclusion. A sliding window is used to check the status marks of 3 consecutive time periods (1.5 hours). If all time periods are marked as "high risk state" in 3 consecutive time periods, or at least one "emergency state" period is included in these 3 time periods (regardless of the status of the other two time periods), the system determines that the current elevator status requires immediate attention and intervention, and generates the conclusion of "immediate maintenance recommendation". For example, if The statuses of segments 33, 34, and 35 are "High Risk State," "Emergency State," and "High Risk State," respectively. Since "Emergency State" is included, "Immediate Maintenance Recommendation" is triggered. If segments 40, 41, and 42 are all "High Risk State," "Immediate Maintenance Recommendation" is also triggered. If the risk status combination of three consecutive segments does not meet any of the above conditions, no immediate maintenance recommendation is generated. The comprehensive elevator operating status assessment conclusion may be based on the risk level of the current latest segment (such as "Potential Risk State") or the risk trend over a period of time (such as "Recent Risk Level Increased"), resulting in a comprehensive elevator operating status assessment conclusion.

[0146] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A method for evaluating elevator operation status based on multi-source heterogeneous data, characterized in that: The following steps are involved: Collect elevator car vertical acceleration signals, traction machine output torque signals, traction machine output current signals, motor encoder speed signals, elevator operation log time, stop floor records, operation number records, and door opening and closing number records. Align each data item based on timestamps, and segment the data stream into time periods based on the elevator's start, running, and stop states to generate synchronized multi-source operation data streams. Based on the synchronized multi-source operation data stream, a ratio sequence of instantaneous torque values and acceleration values during the elevator start-stop process is calculated, a difference sequence between the motor encoder speed signal and the speed estimate calculated based on the traction machine output torque signal is monitored, and rapid change points in the amplitude of the car vertical acceleration signal are detected to obtain a preliminary signal deviation index set. Based on the preliminary signal deviation index set, abnormal signal points are screened and determined to obtain a set of operation feature points; Based on the synchronized multi-source operation data stream, the operation frequency of the time period is calculated to obtain basic operation statistics. Based on the basic operation statistics, the combination of the statistical values and external conditions is compared with weather conditions and large-scale event information to determine whether it deviates from the normal range, and to establish a context-related operation mode profile; Based on the frequency of occurrence of abnormal signal points in the operating feature point set and the context-related operating mode portrait, the operating environment-related risk score value is calculated, and the operating environment-related risk score value is compared and judged with the preset attention alarm threshold to generate a comprehensive elevator operating status assessment conclusion.

2. The elevator operation status evaluation method based on multi-source heterogeneous data according to claim 1 is characterized in that: The steps for acquiring the synchronized multi-source operation data stream are: Collect the elevator car vertical acceleration signal, traction machine output torque signal, traction machine output current signal, motor encoder speed signal, elevator operation log time, stop floor record, operation number record and door opening and closing number record, extract the timestamp field of each signal, calculate the absolute value of the difference between each signal timestamp and the reference time of the elevator operation log time, define the absolute value of the difference as the initial transmission delay value, and use cubic spline interpolation to fill the missing time points for signals whose absolute value of the difference exceeds the preset threshold, thus generating a multi-source signal time series data set with completely aligned timestamps; Based on the multi-source signal time series data set with completely aligned timestamps, the state switching marks in the elevator operation log time are parsed to extract the starting time point of the start state, the duration of the running state, and the ending time point of the stop state. The data segment boundaries are divided according to the starting time point and the ending time point. The vertical acceleration signal of the car is cut into the start segment, the running segment, and the stop segment according to the boundaries. The traction machine output torque signal, the traction machine output current signal, and the motor encoder speed signal are synchronously divided into corresponding data segments to generate a segmented multi-source signal data set with state labels; Based on the segmented multi-source signal data set with state labels, the mapping relationship between floor numbers and timestamps in the stop floor records is extracted, the floor numbers are matched to the corresponding state intervals according to the divided data segment boundaries, and the operation count records and door opening and closing count records are merged to the end of the data segment according to the timestamps to generate a synchronized multi-source operation data stream.

3. The elevator operation status evaluation method based on multi-source heterogeneous data according to claim 1 is characterized in that: The steps for obtaining the preliminary signal deviation indicator set are: Extracting the instantaneous values of the torque signal and the acceleration signal during the elevator start-stop process based on the synchronized multi-source operation data stream, segmenting the data into segments according to time windows, and generating an instantaneous torque-acceleration data set for the start-stop process; Calculate the torque to acceleration ratio at each time point based on the instantaneous torque-acceleration data set during the start-stop process , generate the torque-acceleration ratio sequence, the calculation formula is: ； in, For the The torque value at a time point, For the The acceleration value at a time point; Based on the torque-acceleration ratio sequence, the speed signal recorded by the motor encoder is and the speed estimate derived from the motor output torque signal Perform point-by-point difference calculation to obtain the difference value , generate the speed-estimated speed difference value sequence, the formula is: ； Based on acceleration value , calculate the acceleration difference between adjacent time points ,in is the acceleration value at adjacent time points, and the mutation threshold is set to , for The mean of for The standard deviation of The time point of is the acceleration rapid change point, and the torque-acceleration ratio value sequence, the speed-estimated speed difference value sequence and the acceleration rapid change point are integrated to generate a preliminary signal deviation indicator set.

4. The elevator operation status evaluation method based on multi-source heterogeneous data according to claim 1 is characterized in that: The steps for obtaining the running feature point set are: Based on the preliminary signal deviation indicator set, the preset thresholds of the torque-acceleration ratio sequence, the preset thresholds of the speed-estimated speed difference sequence, and the amplitude thresholds of the acceleration rapid change points are extracted, the time window range is defined as the interval of 250 milliseconds before and after adjacent time points, and the multi-indicator threshold parameters and time window definition rules are generated; According to the multi-index threshold parameters and time window definition rules, each time point in the preliminary signal deviation index set is traversed. If the torque-acceleration ratio corresponding to the current time point exceeds the preset threshold, and the speed-estimated speed difference value exceeds the preset threshold within the 250 millisecond window before and after the current time point, and the amplitude of the acceleration rapid change point within the window exceeds the threshold, then the current time point is marked as a candidate abnormal signal point, and a multi-index synchronous over-limit time point set is generated; Based on the multi-index synchronous over-limit time point set, the candidate abnormal signal points with adjacent time intervals less than 300 milliseconds are merged. After merging, the time point with the largest deviation of the torque-acceleration ratio in each cluster is retained as the representative point, and the isolated candidate points without adjacent clusters are eliminated to generate the operation feature point set.

5. The elevator operation status evaluation method based on multi-source heterogeneous data according to claim 1 is characterized in that: The steps for obtaining the basic running statistics are: Based on the elevator operation log time in the synchronized multi-source operation data stream, the complete time axis from 0:00 to 24:00 every day recorded in the log is extracted, and the time axis is divided into 48 continuous intervals with each 30 minutes as a statistical period. Combined with the timestamps in the stop floor records, the operation number records and door opening and closing number records in each period are counted to generate the operation-door opening and closing frequency data set after period division; Calculate the operating frequency value of each period based on the operation-door opening and closing frequency data set divided into the time periods , generate the period operation frequency sequence, the calculation formula is: ； in, For the The number of runs in each period, For the The number of door openings and closings in a period of time, is the time period length; Based on the operating frequency sequence of the time period, the sliding mean and standard deviation of the operating frequency values of 6 consecutive time periods are calculated, and the time period where the sliding mean exceeds 2 times the standard deviation of the historical mean during the same period is marked as a high-frequency operating interval. The operating frequency values of all time periods are integrated to generate basic operating statistics.

6. The elevator operation status evaluation method based on multi-source heterogeneous data according to claim 1 is characterized in that: The steps for obtaining the context-related operation mode portrait are as follows: Based on the basic operation statistics, the operation frequency value and high-frequency operation interval mark of each statistical period are extracted, the rainfall and wind speed in the externally input weather condition data and the event scale and number of participants in the large-scale event information are associated, and the timestamps are aligned and matched to the corresponding statistical period to generate an external condition-operation statistic association data set; Based on the external condition-operation statistic association data set, the mean and standard deviation of the operation frequency values in the past three months under the same rainfall and activity scale classification conditions in the historical database are compared. If the operation frequency value of the current period exceeds the range of two standard deviations of the historical mean, the current period is marked as a statistical period data that deviates from the normal range, and a statistical deviation flag set is generated; Based on the statistical deviation mark set, the distribution density of abnormal signal points in the operating feature point set in large-scale scenarios is counted, and the percentage of abnormal signal points in the total number of operations in large-scale scenarios per unit time is calculated. Large-scale scenarios refer to rainfall levels of heavy rain or typhoons and large-scale activities with a scale of tens of thousands of people. If the percentage exceeds a preset threshold, it is determined to be a high-density abnormal scenario. All high-density abnormal scenarios and statistical time period data that deviate from the normal range are integrated to generate a scenario-related operating mode portrait.

7. The elevator operation status evaluation method based on multi-source heterogeneous data according to claim 1 is characterized in that: The steps for obtaining the operating environment associated risk score value are: Based on the set of operating feature points, the number of occurrences of abnormal signal points in each statistical period is extracted. Combined with the high-density abnormal scenarios in the scenario-related operating mode portrait, the ratio of the number of occurrences of abnormal signal points in the high-density abnormal scenarios in each period to the total operating time is calculated to generate an abnormal frequency-high-density scenario association data set; Based on the abnormal frequency-high-density context association data set, calculate the deviation percentage between the number of abnormal signal points and the average number of abnormal points in the same period of history, define the period when the deviation exceeds 20% as a risk-sensitive period, and generate a risk-sensitive period marker sequence; Based on the risk-sensitive period mark sequence, the operating environment associated risk score is calculated using the following formula: ； in, Associate a risk score value with the operating environment, For the The number of abnormal signal points in a period, For the The high-density abnormal situation impact factor for each period, here , is the number of abnormal points in the period, The total running time of the time period.

8. The elevator operation status evaluation method based on multi-source heterogeneous data according to claim 1 is characterized in that: The steps for obtaining the comprehensive elevator operation status evaluation conclusion are as follows: Based on the operating environment associated risk score value, extract the operating environment associated risk score value of each statistical period, compare it with the preset attention alarm threshold, define the threshold level interval, and generate a risk score data set with the threshold interval; Based on the risk score dataset with threshold intervals, the risk score values of each statistical period are traversed. If the score value exceeds the severe alarm threshold, it is marked as "emergency state", if it exceeds the moderate alarm threshold but does not reach the severe threshold, it is marked as "high risk state", if it exceeds the mild concern threshold but does not reach the moderate threshold, it is marked as "potential risk state", and the rest are marked as "normal state", thereby generating a risk level state mark set; Based on the risk level status mark set, the number of time periods marked as "emergency state" or "high risk state" in three consecutive time periods is counted. If all three consecutive time periods are high risk or contain one emergency state, an "immediate maintenance recommendation" conclusion is generated, and a comprehensive elevator operation status assessment conclusion is obtained.

9. The elevator operation status evaluation system according to any one of claims 1 to 8, characterized in that: include: The data synchronization module collects the car vertical acceleration signal, traction machine output torque signal, traction machine output current signal, motor encoder speed signal, operation log time, stop floor record, operation number record and door opening and closing number record, aligns the collection time stamp, divides the start state, operation state and stop state into independent time periods, and combines all signals in the divided time periods in series according to time sequence to establish a multi-source synchronous data stream; The signal offset determination module, based on multi-source synchronous data streams, pairs the acceleration signals and torque signals in the start and stop states to obtain a ratio sequence. It also obtains a numerical difference sequence between the motor encoder speed signal and the speed estimate calculated from the traction machine output torque signal. It performs amplitude increment calculations between adjacent values in the sequence of the car vertical acceleration signal and marks the mutation points. It then screens the maximum value points in the ratio sequence, the absolute deviation points in the difference sequence, and the time period containing the mutation points to obtain a set of offset characteristic signals. The operation frequency statistics module, based on multi-source synchronous data streams, counts the total number of daily operations, the number of daily changes in the number of floors stopped, the duration of each operation, and the total number of doors opened and closed each day. It extracts the value distribution interval within a 24-hour range on a daily basis, compares the difference between the distribution interval and the common distribution range in the historical operation cycle, and establishes the periodic operation frequency characteristics; The situational feature characterization module, based on the periodic operation frequency characteristics, calls the weather condition information of each statistical day and the records of large-scale activities in the area, compares the total number of daily operations with the weather conditions, makes a combined judgment on the number of floor changes and the presence of large-scale activities, and combines the total number of daily door openings and closings with the temperature level. It establishes three types of combination relationships: the combination pattern of operation number and weather, the combination pattern of floor changes and activities, and the combination pattern of door openings and closings and temperature. It marks the combination types whose distribution exceeds the normal pattern interval, summarizes the combination types and the frequency of occurrence within the statistical time interval, and generates a situational combination operation map; The risk scoring module assigns a risk level to the combination type whose offset number in the combination relationship is higher than the standard deviation interval based on the signal offset points marked in the offset feature signal set, and obtains the risk distribution scoring result.

Images