A method, medium and device for analyzing the running state based on host vibration noise

Abstract

The present application relates to the technical field of elevator control, in particular to a running state analysis method based on host vibration noise, medium and equipment, the method comprises: obtaining the current state of the target elevator, the current state is running state or stagnation state;If the current state of the target elevator is stagnation state, obtain the noise data set and vibration data set of the target elevator in the first preset time period;If the current state of the target elevator is running state, obtain the noise data set and vibration data set of the target elevator in the second preset time period;Obtain the characteristic parameters of the noise data set and the vibration data set, generate the corresponding state value based on the obtained characteristic parameters, and compare the generated state value with the corresponding state threshold value, and determine whether the current state of the target elevator is an abnormal state based on the comparison result.The present application can accurately analyze the running state of the elevator.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of elevator control, in particular to a running state analysis method based on host vibration noise, medium and equipment. BACKGROUND

[0002] In the daily operation and maintenance work of the elevator, accurate and reliable analysis of the running state of the elevator is the core link to ensure the safe and stable operation of the elevator. In the actual operation process of the elevator, there are mainly two states, i.e. running state and stagnation state, and the vibration and noise characteristics generated by the host of the elevator in these two states are significantly different.

[0003] The existing elevator running state analysis method, such as patent CN202311581367.2, discloses an elevator energy-saving control method and system, which has obvious deficiencies in accuracy. The main problem is that it fails to fully consider the essential difference between the running and stagnation states of the elevator, and lacks targeted time setting strategy in the data collection stage, resulting in that the vibration and noise data obtained cannot accurately reflect the real characteristics under the corresponding state. At the same time, in the data analysis process, the analysis indicators and judgment thresholds for data under different states lack distinction, making the identification of abnormal conditions of the running state of the elevator not accurate enough, and prone to misjudgment or omission, which is difficult to meet the actual needs of safe operation and efficient maintenance of the elevator. SUMMARY

[0004] In view of the above technical problems, the technical scheme adopted by the present application is as follows:

[0005] According to the first aspect of the present application, a running state analysis method based on host vibration noise is provided, which comprises the following steps:

[0006] S100, obtaining the current state of the target elevator, the current state being running state or stagnation state.

[0007] S200, if the current state of the target elevator is the stagnation state, obtaining the noise data set A1 and the vibration data set B1 of the target elevator within a first preset time period; performing S400.

[0008] S300, if the current state of the target elevator is the running state, obtaining the noise data set A2 and the vibration data set B2 of the target elevator within a second preset time period; performing S500.

[0009] S400, obtaining the characteristic parameters of A1 and B1, generating a stagnation state value based on the obtained characteristic parameters, and comparing the generated stagnation state value with a preset stagnation state threshold value, and determining whether the current state of the target elevator is an abnormal state based on the comparison result.

[0010] S500, obtain characteristic parameters of A2 and B2, generate an in-operation state value based on the obtained characteristic parameters, compare the generated in-operation state value with a preset in-operation state threshold value, and determine whether the current state of the target elevator is an abnormal state based on a comparison result.

[0011] According to the second aspect of the present application, an electronic device is provided, comprising a processor and a memory; the processor is configured to execute the steps of the method according to the first aspect of the present application by invoking programs or instructions stored in the memory.

[0012] According to the third aspect of the present application, a computer readable storage medium is provided, which stores programs or instructions for causing a computer to execute the steps of the method according to the first aspect of the present application.

[0013] The present application has at least the following beneficial effects:

[0014] The running state analysis method based on host vibration noise provided by the embodiment of the present application comprises the following steps: firstly, obtaining the current state of the target elevator to determine whether it is in an in-operation state or a stagnation state. Differentiated data acquisition strategies are adopted for different states: if it is in a stagnation state, noise data sets and vibration data sets in a first preset time period are obtained; if it is in an in-operation state, noise data sets and vibration data sets in a second preset time period are obtained. Then, corresponding state values are generated by weighted integration of characteristic parameters, and compared with preset threshold values set independently. Finally, whether the current state is abnormal is determined according to the comparison result: if the state value is greater than the corresponding threshold value, it is determined to be abnormal; otherwise, it is normal. Through the cooperative design of state division, accurate sampling, characteristic adaptation and differential threshold, the present application fully adapts to the essential difference between the static and dynamic states of the elevator, effectively improves the pertinence of vibration noise characteristic capture and the accuracy of abnormality judgment, and provides scientific and reliable technical support for the safe operation and maintenance of the whole life cycle of the elevator.

[0015] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become apparent through the following description. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0017] Figure 1A flow chart of a running state analysis method based on host vibration noise is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be clearly and completely described with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the description of the application herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0020] It is noted that some of the example embodiments are described as a process or method depicted as a flowchart. Although the flowchart describes the steps of the process as sequential, many of the steps can be performed in parallel, concurrently or even simultaneously. In addition, the order of the steps can be re-arranged. The process can be terminated when its operations are completed, but can also have additional steps not included in the figure. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0021] The embodiment of the present application provides a kind of based on the running state analysis method of host vibration noise, to be in through capturing the vibration noise characteristics of elevator core driving component (host), realize the accurate monitoring and abnormal judgment of elevator running state.Host as the core driving component of elevator, its vibration noise characteristics can be directly mapped the running state of elevator.The normal operation of elevator relies on the stable work of host such as traction machine, driving motor and other core components, and the vibration noise signal of host will change dynamically with elevator state: in the running state such as start, acceleration, uniform speed, deceleration stage, host is in running state, and dynamic vibration and noise will be generated due to motor work, transmission component friction, etc.;In the stagnation state, host stops running, but characteristic vibration noise will be generated due to static stress, component gap, etc.: on the one hand, the weight of car and counterweight will form continuous static load on host through traction rope, traction wheel and other components, leading to small friction vibration of host bearing, brake and other components under static pressure, such as low-frequency microvibration caused by static friction;On the other hand, the assembly gap between components, material stress release or environmental interference such as slight vibration of building transmitted to host can also cause weak structural vibration and convert into monitorable noise signal.These signals contain the state information of host and elevator as a whole, and once elevator fails, such as host bearing wear, guide rail blockage, brake abnormality, etc., key characteristic parameters of host vibration noise, such as variance reflecting dispersion degree, mean value reflecting concentration trend, extreme difference reflecting fluctuation range, etc., will change abnormally.Based on this, the method provided by the present application takes host vibration noise as monitoring object, and realizes accurate analysis and abnormal judgment of the overall running state of elevator by capturing the state signal of core component.

[0022] As Figure 1 Indicated, the method can include the following steps:

[0023] S100, obtain the current state of target elevator, and the current state is running state or stagnation state.

[0024] In the embodiment of the present application, the target elevator refers to a specific elevator equipment to be monitored, including but not limited to building number unique traction type, hydraulic type and other types of elevators, and the running state thereof needs to be analyzed through vibration noise characteristics.

[0025] In the embodiment of the present application, the running state refers to the stage in which the elevator is in the process of car movement, covering the whole process from starting, accelerating, uniform speed, decelerating to stopping before the target floor in response to the call signal; at this time, the main machine (such as traction machine, driving motor) is in running state, and the car and the guide rail, counterweight and other components exist dynamic interaction. The static state refers to the stage in which the elevator is in the process of car stopping and the main machine stopping running, including the standby state after completing the door opening and closing operation after the car stops at the target floor, or the state of keeping still due to not receiving the call signal; at this time, the main machine and the core transmission components have no power output, and the car has no significant relative motion with the surrounding components.

[0026] In S200, if the current state of the target elevator is the static state, the noise data set A1 and the vibration data set B1 of the target elevator in the first preset time period are obtained; S400 is executed.

[0027] In the embodiment of the present application, the noise data is collected by an acoustic sensor, including sound pressure level (quantifying noise intensity in decibels), frequency distribution (reflecting the proportion of energy at each frequency through spectrum analysis), time domain waveform (recording the change and mutation characteristics of sound over time); the vibration data is collected by a vibration sensor, including vibration acceleration (reflecting the degree of vibration, unit m / s²), vibration speed (measuring dynamic vibration characteristics, unit mm / s), vibration displacement (reflecting vibration amplitude, unit μm or mm) and frequency characteristics (obtained by Fourier transform, used to identify the vibration source).

[0028] In the embodiment of the present application, the determination method of the first preset time period is as follows: calculating the total duration T of the time period from the end time of the previous running state to the current time, when T is greater than or equal to the preset threshold, obtaining the key time period in the time period from the end time of the previous running state to the current time, and dividing the key time period into a plurality of time slices according to the preset division method, the set of the plurality of time slices constitutes the first preset time period; when T is less than the preset threshold, the first preset time period is directly the time period from the end time of the previous running state to the current time, that is, the total duration of the first preset time period is T.

[0029] In the embodiment of the present application, the core role of the preset threshold is to distinguish between the regular static duration and the excessively long static duration, and its setting needs to be combined with the use scene of the elevator, the characteristic change law of the static state and the data collection efficiency.

[0030] In an exemplary embodiment, the core is based on: the "typical idle period" and "stagnation state feature stability duration" of the elevator. The preset threshold is equal to max (k times the typical idle period, the feature stability duration). Wherein, k is a proportionality coefficient, for example, it can be 0.7. The typical idle period refers to the regular stationary duration of the elevator without operation instruction, such as about 8 hours of idle at night for residential elevators, about 2 hours of idle during lunch break for office building elevators, and about 1 hour of idle during off-peak for shopping mall elevators. The feature stability duration refers to the time for the vibration and noise features to transition from dynamic to static stability in the stagnation state, usually 30 minutes to 1 hour, after which the feature fluctuation tends to be flat.

[0031] Example 1: The typical idle period of a residential elevator at night is 8 hours, the feature stability duration is 1 hour, and the preset threshold is 8*70%=5.6 hours (about 5 hours and 36 minutes);

[0032] Example 2: The typical idle period of an office building elevator during lunch break is 2 hours, the feature stability duration is 40 minutes, and the preset threshold is 2*70%=1.4 hours (84 minutes);

[0033] Through the threshold set by this rule, it can cover most of the stable stage of the regular idle period (70% of the typical period), and avoid collecting redundant data when the feature has stabilized. When T≥preset threshold (stagnation time is too long), the key time period needs to be extracted from the total time T to focus on the stage where the feature is most likely to expose abnormalities, and to eliminate the redundant period without significant features. Specific embodiments include the following:

[0034] Embodiment 1: Key time period based on state stage division

[0035] The key time period includes the initial transition period, the middle stable period, and the late approaching period, which are divided as follows:

[0036] Initial transition period: t1 period after the end of the previous operation state (exemplary value 0-1 hour), in this stage, the host cools down from the running state to room temperature, and the component stress is released (such as temperature change of traction sheave and brake causing dynamic change of vibration and noise features), which can capture residual abnormalities after operation such as component overheating caused by friction during operation, and abnormal vibration in the initial stagnation period;

[0037] Middle stable period: middle 50% period of total stagnation time (such as T=10 hours, take 3-7 hours), in this stage, the feature tends to be stable, which can reflect the component state under static load such as continuous micro-vibration caused by uneven tension of traction rope;

[0038] Late approaching period: t2 period before the current time (exemplary value within 1 hour before), in this stage, the feature may mutate due to environmental changes such as sudden temperature rise, external vibration interference, or component state deterioration such as slow loosening of brake, which can capture recent abnormalities.

[0039] Example: T=10 hours (previous run ended at 0:00, current time 10:00), critical time period is 0:00-1:00 (initial), 3:00-7:00 (middle), 9:00-10:00 (late).

[0040] Embodiment 2: Critical time period based on cumulative area ratio of vibration noise curve

[0041] This method quantifies the energy distribution of vibration noise signals, focusing on the core period with the highest energy contribution. It is suitable for scenarios where the feature gradually changes over time, such as slow energy decay or accumulation. The steps are as follows:

[0042] Data preprocessing: Continuously sample the vibration dataset B1 and noise dataset A1 within T to generate the vibration noise amplitude-time curve f(t). The vibration noise amplitude is the sum of the normalized vibration amplitude and noise amplitude, which is used to eliminate dimensional differences.

[0043] Cumulative energy calculation: Calculate the total area S (total energy) enclosed by the curve f(t) and the horizontal axis, and the cumulative area S(t) from the stagnation starting time t0 to any time t. .

[0044] Threshold point determination: Obtain the time t1 with a cumulative area ratio of k1 and the time t2 with a cumulative area ratio of k2, t1 satisfies S(t1) / S=25%, t2 satisfies S(t2) / S=75%, in an illustrative embodiment, k1=25%, k2=75%. Set the critical time period as [t1, t2], i.e. the period covering the middle 50% energy contribution, excluding the initial and decay segments with low energy.

[0045] Example: T=10 hours, total energy S=1000, t1=2 hours (S(t1)=250), t2=7 hours (S(t2)=750), critical time period is 2-7 hours. If the curve has multiple energy peaks (such as sudden vibration), take the smallest time period that contains all the peaks.

[0046] Embodiment 3: Critical time period based on feature fluctuation significance

[0047] This method focuses on the period with the most dramatic fluctuation of vibration noise features (corresponding to the highest risk of anomalies), which is suitable for scenarios with sudden feature changes such as sudden loosening of components leading to sudden increase in vibration. The steps are as follows:

[0048] Fluctuation quantification: Divide T into several time windows (e.g. 10 minutes per window), calculate the feature fluctuation value of each window. Use the absolute value of the variance within the window or the difference between adjacent windows to represent, the larger the value, the more significant the fluctuation.

[0049] Threshold setting: determine a significant fluctuation threshold based on historical normal data, such as 1.5 times the normal fluctuation value.

[0050] Period extraction: merge windows with fluctuation values greater than the threshold into continuous key time periods; if the fluctuation is dispersed, take the top 30% of windows with fluctuation values.

[0051] Example: T=10 hours divided into 60 10-minute windows, the 5-6, 20-21 windows (corresponding to 1:40-2:00, 6:40-7:00) have fluctuation values exceeding the threshold, and the key time period is the combination of the two periods.

[0052] Implementation 4: Key time period based on feature similarity clustering

[0053] This method divides the features within T into similar subgroups through clustering algorithms, suitable for complex feature change patterns such as alternating stable and fluctuating stages, with the following steps:

[0054] Feature vector construction: divide T into several sub-periods (such as 30 minutes per sub-period), and extract the combined feature vector such as variance, mean value, and low-frequency energy proportion of vibration and noise in each sub-period.

[0055] Clustering analysis: use K-means algorithm to cluster, such as into stable class, fluctuation class, and transition class.

[0056] Representative period selection: select 1-2 representative sub-periods from each cluster, where the stable class selects a typical period, the fluctuation class selects a period with the largest outlier value, and the combination forms the key time period, ensuring coverage of all feature patterns.

[0057] The key time period may be a long continuous period (such as 5 hours), and directly calculating the global feature may dilute local anomalies, such as a 10-minute sudden vibration peak being masked by the overall average or dynamic changes being smoothed out, such as a gradual trend in vibration amplitude over time. Therefore, the key time period needs to be divided into several time slices according to a pre-set division method, which includes:

[0058] Average duration division: uniformly split by fixed duration (such as dividing a 5-hour key period into 30 10-minute time slices);

[0059] Clustering division based on historical data: use the feature clustering results of elevator historical stop data to determine a typical division pattern (such as analyzing 100 historical data and finding that they generally present a "initial 5 minutes, middle 8 minutes, and late 2 minutes" three stages, then divide according to this rule), which fits the individual characteristics of the equipment;

[0060] Minimum time unit combination division: set the minimum time unit (such as 1 minute), flexibly combine according to the total length of the key time period, and ensure that the length of each time slice is an integer multiple of the minimum unit (for example, a 70-minute key period is divided into 7 10-minute slices).

[0061] The first preset time period determined in the above manner can accurately cover the core feature period of the stagnation state, and can capture local abnormalities and dynamic changes through time slice refinement, thereby providing a high-quality data basis for subsequent feature extraction and state evaluation.

[0062] In S300, if the current state of the target elevator is the running state, noise data set A2 and vibration data set B2 of the target elevator in the second preset time period are obtained; S500 is executed.

[0063] In the embodiment of the application, the second preset time period is a time window for completely capturing the dynamic characteristics of the running state, and the determination manner is as follows: in the total time period from the end time of the previous stagnation state to the current time, after excluding the start-up stage time period corresponding to each running in the total time period, the remaining time period set covering the full process of each running (including uniform speed, deceleration, stop stage and stable running stage after the start-up stage) is obtained.

[0064] The previous stagnation state refers to an ultra-long time stationary state of the elevator due to idling, no running instruction or maintenance, etc., and the stationary duration is greater than a preset stagnation threshold (such as > 2 hours); the end time of the state is the time before the elevator switches from the stagnation state to the running state for the first time (i.e. the last moment before the first start-up).

[0065] Each running refers to all independent running cycles occurring from the end of the previous stagnation state to the current time, and each running cycle is a complete process of start-up → acceleration → uniform speed → deceleration → stop, wherein the stop is a short stationary state in running, such as a short stationary state after stopping at a target floor, which is different from the ultra-long stationary state.

[0066] The start-up stage time period refers to the continuous process of the elevator switching from the stationary state (including the previous stagnation state or the stop stage after the last running) to the stable running state, and the core duration is from the start-up initial time (the moment when the car starts to move and the main machine start-up instruction is triggered) to the time when the elevator reaches the rated running speed.

[0067] The significant feature of this stage is to include the full transition process from static to dynamic: the motor starts from standby state, the car starts to accelerate from stationary state, and the transmission system (dragging wheel, steel wire rope, guide rail, etc.) switches from static load to dynamic running state, accompanied by typical transition characteristics such as impact vibration at the start-up moment, high-frequency noise during motor loading, and increasing vibration amplitude during acceleration process, etc. These characteristics are essentially different from the normal characteristics of the stable running stage.

[0068] It is worth noting that the starting phase time period of each operation can not be fixed, and the difference rule can be determined based on historical operation data statistics: for example, the first operation after the end of the previous stagnation state, because the components may be in a long static state such as brake slight lock, high viscosity of lubricating oil, the starting phase time period is usually the longest (may reach 5 seconds); in subsequent operations, as the components enter the dynamic working state such as the fully lubricated brake contact surface, the preheating of the transmission system is completed, the starting phase time period will gradually decrease and tend to be stable (usually stable at 3-4 seconds). The difference in length essentially reflects the process characteristics of the elevator from static to dynamic, so the actual starting phase time period of each operation needs to be accurately defined during actual collection, rather than using a fixed value.

[0069] The time range of the second preset time period takes the end time of the previous stagnation state as the starting point and the current time as the ending point, covering all operation-related time periods within the total time period. Only the "starting phase time period" (3-5 seconds of transition process) corresponding to each operation in the total time period is excluded, and the exclusion logic is based on: the characteristics of the starting phase mainly reflect the transition characteristics from static to dynamic, which are significantly different from the normal characteristics of the stable operation phase, and need to be analyzed separately (or selectively excluded according to the scene requirements). The remaining time period needs to completely include the core phase of each operation, i.e.: the stable operation phase after the end of the starting phase to the start of deceleration (uniform speed process, reflecting normal operation characteristics); the deceleration phase near the target floor (braking system effect, reflecting deceleration braking characteristics); the stopping phase after stopping (short-term static in operation, reflecting stopping stability characteristics). Ensure that the core dynamic characteristics of the running state are collected (such as vibration stability during uniform speed, brake noise during deceleration, car micro-motion during stopping, etc.).

[0070] For example, if the end time of the previous stagnation state is 10:00 (the initial time of the first starting phase), the current time is 12:00, and the total time period is 10:00-12:00; during which 6 independent operations occur, and the starting phase time period of each operation is:

[0071] 10:00-10:03 (3 seconds, first start), 10:15-10:18, 10:30-10:33, 11:00-11:03, 11:20-11:23, 11:45-11:48.

[0072] Then the second preset time period is 10:00-12:00, excluding the above 6 starting phase time periods, and includes the following set of time periods:

[0073] 10:03-10:15 (from the first start to the second start, including the stable, deceleration, and stopping phases of the first operation);

[0074] 10:18-10:30 (after the second start to before the third start);

[0075] …

[0076] 11:48-12:00 (after the sixth start to the current time, including the current running stable stage).

[0077] Through this setting, the second preset time period can completely cover the dynamic process of each running, providing full-cycle data support for subsequent extraction of characteristic parameters (such as acceleration vibration peak, uniform speed noise stability, deceleration braking characteristics, etc.) in the running state.

[0078] S400, acquiring the characteristic parameters of A1 and B1, generating a stagnation state value based on the acquired characteristic parameters, and comparing the generated stagnation state value with a preset stagnation state threshold, and determining whether the current state of the target elevator is an abnormal state based on the comparison result.

[0079] In the embodiments of the application, the feature extraction method can be dynamically selected based on the physical characteristics of the stagnation state, and the characteristic parameters of A1 and B1 are acquired, which are quantitative descriptions of the essential characteristics of the vibration noise signal in the stagnation state.

[0080] The core physical characteristics of the stagnation state are static dominance and dynamic weakness: the host stops running, the car is stationary, the core components (such as the brake, the traction wheel, and the bearing) are in a static load state, there is mainly low-frequency micro-vibration (such as micro-deformation vibration of the host caused by uneven tension of the traction rope), static friction (such as slight relative sliding of the brake contact surface), and environmental interference isolation (no strong dynamic noise during running, which highlights the inherent abnormal signal of the component). Based on these characteristics, the logic of dynamically selecting the feature extraction method is as follows:

[0081] (1) Dispersed degree extraction for stability monitoring

[0082] Under the stagnation state, normal vibration noise should show low fluctuation characteristics (such as stable vibration when the bearing has no gap). If there is looseness, gap, or other abnormalities in the component, it will cause the vibration noise to fluctuate more. Therefore, the variance (or standard deviation) is selected as the dispersion degree index, and the stability is quantified by calculating the data fluctuation amplitude, and the larger the variance, the worse the static stability of the component (such as irregular fluctuation of micro-vibration caused by bearing gap).

[0083] (2) Average value extraction for overall intensity evaluation

[0084] The continuous static friction (such as abnormal contact of the brake and insufficient lubrication of the shaft sleeve) during stagnation will produce stable vibration noise signals. The arithmetic mean is selected to represent the overall intensity, which can reflect the duration of static friction, and an abnormal increase in the average value usually corresponds to continuous friction between components (such as friction noise caused by the brake not being completely released).

[0085] (3) Low-frequency vibration energy ratio extraction for low-frequency feature capture

[0086] The micro-vibration and static friction energy in the stagnation state are mainly concentrated in the low-frequency band (usually < 50 Hz, such as 5-20 Hz corresponding to bearing static friction, 20-50 Hz corresponding to tension fluctuation of the traction system), and the high-frequency signal is mostly environmental interference. Therefore, the wavelet transform is used to extract the energy ratio in the preset low-frequency band (such as < 50 Hz), to accurately separate the inherent characteristics of the component and the environmental interference. The abnormally high low-frequency energy ratio may imply low-frequency abnormalities such as shaft sleeve wear and brake jamming.

[0087] Through the above dynamic selection logic, the feature extraction method is deeply adapted to the physical essence of the stagnation state, ensuring that the extracted parameters can truly reflect the core health status in the static phase. That is, the feature parameters of A1 include the dispersion degree and the average value, the feature parameters of B1 include the dispersion degree, the average value, and the low-frequency vibration energy ratio, and the low-frequency vibration energy ratio is obtained by extracting the energy ratio in the preset frequency band below hertz through wavelet transform.

[0088] When the first preset time period is a single continuous period (T < preset threshold), the stagnation time is short (such as < 5.6 hours) at this time, the feature fluctuation is stable, and there is no significant stage difference. The overall feature parameters in this period are directly calculated.

[0089] When the first preset time period contains several time slices (T ≥ preset threshold), the stagnation time is too long at this time, and the features have stage differences. The sub-feature parameters of each time slice need to be calculated first, and then the overall feature is obtained through fusion, wherein the sub-feature parameters include the dispersion degree and the average value of the noise sub-data set in each time slice, the dispersion degree, the average value, and the low-frequency energy ratio of the vibration sub-data set. The fusion methods can include:

[0090] Weighted average fusion: different weights are given to different time slices (such as higher weight for the component cooling stage in the early stagnation stage, because the features in this stage are more likely to reflect the residual abnormalities after running), which is suitable for scenarios where the features change regularly over time;

[0091] Mean fusion: taking the arithmetic mean of all time slice features, which is suitable for scenarios where the features in each time slice fluctuate uniformly and have no significant stage difference;

[0092] Maximum fusion: selecting the maximum value of all time slice features, which is suitable for capturing instantaneous abnormalities (such as a sudden peak value of component friction in a time slice), and avoiding that local abnormalities are covered by overall average.

[0093] In the embodiment of the present application, the stagnation state value Y1 = w1 × σ a1 + w2 × μ a1 + w3 × σ b1+w4 x mu b1 +w5 x E b1 wherein sigma a1 is the dispersion of A1, mu a1 is the mean of A1, sigma b1 is the dispersion of B1, mu b1 is the mean of B1, E b1 is the low frequency energy ratio of B1, w1 to w5 are preset weight coefficients satisfying (w1+w2+w3+w4+w5)=1, the values of which are set based on the sensitivity of the features to the anomaly, such as the low frequency energy ratio being more sensitive to bearing wear, and a higher weight w5 can be assigned.

[0094] The preset stationary state threshold is a criterion for judging the stationary anomaly, and the value thereof is determined based on the Y1 distribution of a large amount of historical normal stationary data, such as taking 1.2 times the maximum value of the normal data Y1 as the threshold, which avoids misjudgment caused by normal fluctuations and can sensitively capture anomalies. In specific judgment: if Y1 is greater than the preset stationary state threshold, it indicates that the vibration noise feature of the current stationary state deviates from the normal range (such as a too large variance possibly implying component loosening, and an abnormal mean possibly reflecting continuous friction), and it is determined as an abnormal state; if Y1 is less than or equal to the preset stationary state threshold, it is determined as a normal state. Through this logic of feature parameter quantization → multi-dimensional fusion → threshold comparison, accurate evaluation of the stationary state is realized, which not only retains the feature details of different stages, but also improves the reliability of anomaly recognition through comprehensive calculation.

[0095] S500, the feature parameters of A2 and B2 are obtained, the running state value is generated based on the obtained feature parameters, and the generated running state value is compared with a preset running state threshold, and whether the current state of the target elevator is an abnormal state is determined based on the comparison result.

[0096] In the embodiments of the application, the feature extraction method can be dynamically selected based on the physical characteristics of the running state to obtain the feature parameters of A2 and B2.

[0097] The core physical characteristics of the running state are dynamic dominance and multi-stage dramatic changes: the elevator goes through the whole process of starting acceleration, uniform running and deceleration braking, the host machine runs at high speed, the car moves along the guide rail, and the core components (such as the traction machine, the brake and the guide rail) are in a dynamic load state, accompanied by instantaneous impact vibration (at the moment of starting / braking), continuous friction noise (in the uniform speed stage) and dynamic changes in frequency characteristics (resonance characteristics at different speeds). Based on these characteristics, the logic of dynamically selecting the feature extraction method is as follows:

[0098] (1) Dispersion degree extraction for dynamic stability monitoring

[0099] The vibration noise of the running state should show a stable stage, and the overall controllable fluctuation characteristics such as uniform speed stage vibration stability. If the parts are abnormal, such as guide rail unevenness, traction wheel eccentricity, it will cause the fluctuation to intensify and be irregular. Therefore, the variance or standard deviation is selected as the dispersion degree index to quantify the fluctuation amplitude of different stages, and the abnormal increase of the variance reflects the decrease of dynamic stability, such as the irregular vibration fluctuation of the uniform speed stage caused by guide rail jamming.

[0100] (2) Average value extraction for overall load intensity evaluation

[0101] During operation, the motor load and component friction will generate continuous vibration noise, such as traction machine noise in the uniform speed stage and friction vibration between the car and the guide rail. The arithmetic mean is selected to represent the overall intensity, which can reflect the energy level under dynamic load. The abnormal increase of the average value usually corresponds to excessive friction (such as insufficient guide rail lubrication) or motor overload (such as increased running noise caused by abnormal load).

[0102] (3) Extreme value difference extraction for capturing instantaneous drastic change

[0103] There are drastic physical changes in the starting acceleration and deceleration braking stages (such as motor torque impact at start-up and brake pad clamping during braking), which are prone to produce instantaneous peak signals. The extreme value difference (the difference between the maximum and minimum values) is selected to capture such drastic changes. The abnormal increase of the extreme value difference may indicate sudden failures such as brake jamming (vibration peak rising sharply at the moment of braking), excessive start-up impact (abnormal motor output torque), etc.

[0104] (4) Resonance frequency offset extraction for structure resonance monitoring

[0105] Each component of the elevator (such as the main machine and the car) has a natural resonance frequency. The real-time vibration frequency should fluctuate around the natural frequency within a small range during normal operation. If the components are loose or deformed, such as uneven traction rope tension and car frame deformation, it will cause the resonance frequency to shift. By Fourier transform, the deviation between the real-time vibration frequency and the natural frequency (resonance frequency offset) can be identified, which can accurately capture structural abnormalities. If the offset is greater than the threshold, it may reflect that the component connection is loose or the structure is deformed.

[0106] The characteristic parameters of the running state need to cover the dynamic characteristics of the whole process, and the acquisition method needs to adapt to the characteristics of the multi-stage drastic change. The specific acquisition method is:

[0107] Step 1: For each running in the second preset time period, divide the sub-periods according to the stable stage after starting → uniform speed stage → deceleration stage, and extract the sub-period characteristic parameters (dispersion degree, average value, extreme value difference, resonance frequency offset), and then fuse them into the full cycle characteristics of this running, that is, the full cycle characteristics of each running is equal to the weighted fusion of the characteristic parameters of each period.

[0108] For the three sub-period characteristics of a single operation, stage-sensitive weights are used to fuse into the full-cycle characteristics of the operation. The weights are set based on the contribution degree of the high-fault stage (determined by historical fault data statistics), wherein, the deceleration stage is most prone to expose brake faults (such as blockage and wear) in this stage, and is given the highest weight (such as 40%), the stable stage after starting is related to the abnormality of the starting impact (such as uneven motor loading), and is given the second highest weight (such as 30%), and the constant speed stage reflects the stability of normal operation, and is given the basic weight (such as 30%). Taking the dispersion degree of the vibration data set B2 as an example, the dispersion degree of a single operation is equal to the weighted sum of the dispersion degrees of the three sub-periods.

[0109] Step 2: Aggregate the full-cycle characteristics of each operation in the second preset time period to obtain the overall characteristics in the second preset time period.

[0110] The second preset time period contains multiple independent operations (such as 6 operations from the end of the previous stagnation state to the current time), and the full-cycle characteristics of each operation need to be further aggregated to form the overall characteristics covering all operations, wherein, the normal characteristics (dispersion degree, average value) are aggregated by weighted average, and the recent operations (such as the last 2 times) are given higher weights (such as 30%), because the recent characteristics can better reflect the current state; the extreme characteristics (extreme value difference, resonance frequency offset) are aggregated by maximum value, to ensure that the most significant abnormality such as sudden increase of brake extreme value difference in a certain operation is captured.

[0111] Example:

[0112] If the second preset time period contains 3 operations, the vibration extreme value differences are 5, 8 and 6 (unit: m / s²) respectively, and the aggregated overall vibration extreme value difference is 8 (taking the maximum value);

[0113] The noise average values are 65, 68 and 70 (unit: dB) respectively, the recent operation (the third time) is given a weight of 0.3, and the first two times are each given a weight of 0.35, so the overall noise average value is 65*0.35+68*0.35+70*0.3=67.3 (dB).

[0114] In the embodiment of the application, the state value Y2 in operation is generated by weighted fusion of the multi-dimensional characteristics of noise and vibration, and the formula is: Y2=w6×σ a2 +w7×μ a2 +w8×△ a2 +w9×σ b2 +w10×μ b2 +w11×△ b2 +w12×F, wherein, σ a2 is the dispersion degree of A2, μ a2 is the average value of A2, σ b2 is the dispersion degree of B2, μb2 is an average value of B2, Δ a2 is a maximum difference of A2, Δ b2 is a maximum difference of B2, F is a resonance frequency offset of B2, and w6 to w12 are preset weight coefficients, and the sum of w6 to w12 is 1, which sets the adaptive dynamic characteristics, wherein the maximum difference (Δa2, Δb2) and the resonance frequency offset (Foffset) are given higher weights (such as w8, w11, w12), because the start-stop impact and the structural resonance are the core signals of the running fault; the dispersion degree (σ a2 , σ b2 ) is given a medium weight (such as w5, w8), which reflects the stability in the uniform speed stage; and the average value (μ a2 , μ b2 ) weight can be adjusted according to the equipment characteristics (such as w6, w9), which reflects the overall load intensity. All the weights are trained through historical fault data, such as the contribution of each feature in the statistical fault case or the expert experience calibration, to ensure that Y2 can comprehensively represent the health status in the whole running cycle.

[0115] In the embodiment of the application, the extreme difference value is not used as a feature index for the stagnation state value, and the core reason is determined by the vibration and noise characteristics of the elevator in the stagnation state and the analysis target, which is as follows:

[0116] 1. The vibration and noise characteristics of the stagnation state are more stable, and the fluctuation range has no significant analysis value

[0117] When the elevator is in the stagnation state, the main machine and the core components are in a static or low-activity state (no dynamic process such as traction machine operation and car movement), and the vibration and noise mainly come from environmental interference (such as the surrounding environment sound) and the slight friction of static components, and the overall characteristics are stable and low fluctuation. At this time, the difference between the maximum value and the minimum value of the vibration and noise data is usually small, and the difference is mainly caused by random environmental interference (rather than equipment abnormalities), which is difficult to effectively reflect the real abnormal state of the equipment (such as core problems such as component loosening and static friction abnormalities).

[0118] 2. The variance can fully reflect the abnormal fluctuation of the stagnation state

[0119] The core analysis target of the stagnation state is to identify the abnormal deviation under the stable benchmark, and the variance as an index to measure the dispersion degree of the data can accurately capture this deviation. If the equipment has an abnormality such as a slight resonance of the components during the stagnation, the dispersion degree of the data will increase significantly, and the variance can directly quantify this change. At this time, if the difference between the maximum value and the minimum value is additionally introduced, it cannot increase the effective information, but it may cause abnormality of the difference due to accidental environmental interference such as instantaneous external noise, which interferes with the analysis accuracy.

[0120] 3. The difference in characteristics from the running state forms a targeted design

[0121] In the running state, the elevator has dynamic processes such as starting acceleration, uniform running, deceleration and stopping, and the vibration and noise will fluctuate significantly with the running stage, such as impact vibration at start, friction noise peak in running, and the range can effectively capture dynamic extreme abnormalities such as vibration peak caused by sudden blockage and noise surge caused by abnormal friction. The stagnation state does not have such dynamic fluctuations, so it does not need to rely on this index, and only through variance and average value can meet the analysis requirements, avoiding the introduction of redundant indexes affecting the judgment accuracy.

[0122] The preset running state threshold is the basis for judging the running abnormality, and its value is determined based on the Y2 distribution of a large number of historical normal running data, such as taking 1.2 times of the maximum value of normal data Y2, which is not related to the preset stagnation state threshold. Because the overall intensity and fluctuation range of the vibration and noise in the running state are significantly greater than those in the stagnation state, its threshold is usually much higher than the stagnation state threshold. In specific judgment: if Y2 is greater than the preset running state threshold, it means that the vibration and noise characteristics of the current running state deviate from the normal range, such as a2 Too large may reflect the start impact abnormality, and b2 Too high may indicate that the transmission component is worn out, and is determined as an abnormal state; if Y2 is less than or equal to the preset running state threshold, it is determined as a normal state. Through this parameter design and integration logic for the dynamic characteristics of the running state, the abnormal signal in the whole process of starting, running and stopping can be accurately captured, and the evaluation system of the stagnation state is complementary, and reliable monitoring of the elevator in all states is realized.

[0123] In summary, the running state analysis method based on host vibration and noise provided by the embodiment of the application realizes accurate monitoring and abnormal judgment of the elevator running state through targeted state division, data acquisition and analysis strategy, and the core technical effect is embodied in:

[0124] (1) The running state and the stagnation state of the elevator are clearly distinguished, and independent analysis paths are designed for the two states. Because the vibration and noise characteristics of the elevator in the two states are essentially different, such as dynamic friction in running and start-stop impact, and static characteristics in stagnation, this differentiated design avoids the misjudgment problem caused by the mixed analysis of different state characteristics in the traditional method, and makes the abnormal identification more in line with the actual state characteristics;

[0125] (2) By setting the first preset time period and the second preset time period, it is ensured that the collected noise data set (A1, A2) and the vibration data set (B1, B2) can accurately reflect the core characteristics of the corresponding state, wherein the first preset time period in the static state focuses on the key period after the end of the previous running state, and the second preset time period in the running state covers the key nodes of start and stop, effectively filtering invalid data interference, providing a high-quality data basis for subsequent feature analysis, and solving the feature ambiguity problem caused by the lack of pertinence of data collection in the traditional method;

[0126] (3) By independently setting the preset static state threshold and the preset running state threshold, the feature fluctuation law in the two states is fully adapted, the abnormality judgment in the static state is based on the static feature reference, and the abnormality judgment in the running state matches the dynamic feature fluctuation range, avoiding the problem of insufficient adaptability of a single threshold to different state characteristics, significantly reducing the misjudgment probability, and improving the reliability of state evaluation;

[0127] (4) Through the standardized process of feature parameter extraction, state value generation and threshold comparison, a quantifiable and reproducible analysis system is formed, which does not need to rely on complex experience judgment, realizes state evaluation through data-driven mode, is convenient for engineering landing and large-scale application, can effectively assist the elevator maintenance personnel to quickly locate the abnormality, reduces the maintenance cost, and provides a scientific monitoring basis for the safe operation of the elevator.

[0128] The embodiment of the application also provides an electronic device, comprising: at least one processor; and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are arranged to execute the method described in the embodiment of the application.

[0129] The embodiment of the application also provides a computer readable storage medium, which stores computer executable instructions, and the computer instructions are used to execute the method described in the embodiment of the application.

[0130] It should be understood that the various forms of processes shown above can be reordered, added to, or deleted from. For example, the steps described in the present application can be executed in parallel, in sequence, or in different orders, as long as the desired results of the technical solutions disclosed in the present application can be achieved, which is not limited herein.

[0131] The above specific embodiments do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent replacement and improvement made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method of analyzing an operating condition based on host vibration noise, characterized by, The method comprises the following steps: S100, acquiring a current state of a target elevator, the current state being an in-operation state or a standstill state; S200, if the current state of the target elevator is the standstill state, acquiring a noise data set A1 and a vibration data set B1 of the target elevator in a first preset time period; performing S400; S300, if the current state of the target elevator is the in-operation state, acquiring a noise data set A2 and a vibration data set B2 of the target elevator in a second preset time period; performing S500; S400, acquiring feature parameters of A1 and B1, generating a standstill state value based on the acquired feature parameters, and comparing the generated standstill state value with a preset standstill state threshold value, and determining whether the current state of the target elevator is an abnormal state based on a comparison result; S500, acquiring feature parameters of A2 and B2, generating an in-operation state value based on the acquired feature parameters, and comparing the generated in-operation state value with a preset in-operation state threshold value, and determining whether the current state of the target elevator is an abnormal state based on a comparison result; The determination manner of the first preset time period is that: a total time length T of a time period from an end time of a previous in-operation state to a current time is calculated, when T is greater than or equal to a preset threshold value, a key time period in the time period from the end time of the previous in-operation state to the current time is acquired, and the key time period is divided into a plurality of time slices in a preset division manner, and a set of the plurality of time slices constitutes the first preset time period; when T is less than the preset threshold value, the first preset time period is the time period from the end time of the previous in-operation state to the current time. The determination manner of the second preset time period is that: in a total time period from an end time of a previous standstill state to a current time, after excluding a start stage time period corresponding to each operation included in the total time period, a time period set covering a whole process of each operation is obtained.

2. The method of claim 1, wherein, The feature parameters of A1 include a discrete degree and an average value, the feature parameters of B1 include a discrete degree, an average value and a low-frequency vibration energy proportion, the low-frequency vibration energy proportion is obtained by extracting an energy proportion of a preset frequency band below a frequency of a preset hertz through wavelet transform, the feature parameters of A2 include a discrete degree, an average value and an extreme value difference, and the feature parameters of B2 include a discrete degree, an average value, an extreme value difference and a resonance frequency offset, the resonance frequency offset is obtained by identifying a deviation between a host inherent frequency and a real-time vibration frequency through Fourier transform.

3. The method of claim 1, wherein, In S400, if the standstill state value is greater than the preset standstill state threshold value, it is determined that the current state is an abnormal state; otherwise, it is a normal state; in S500, if the in-operation state value is greater than the preset in-operation state threshold value, it is determined that the current state is an abnormal state; otherwise, it is a normal state.

4. The method of claim 2, wherein, If the first preset time period includes several time slices, the characteristic parameters of the noise data set A1 are obtained by fusing the dispersion degree and the average value of the noise sub-data sets of all the time slices, and the characteristic parameters of the vibration data set B1 are obtained by fusing the dispersion degree, the average value and the low-frequency vibration energy proportion of the vibration sub-data sets of all the time slices; wherein the fusion manner includes weighted average, mean fusion or taking the maximum value of the characteristic parameters of each time slice.

5. The method of claim 4, wherein, Stagnation state value Y1 = w1 x σ a1 + w2 x μ a1 + w3 x σ b1 + w4 x μ b1 + w5 x E b1 wherein σ a1 is the dispersion degree of A1, μ a1 is the average value of A1, σ b1 is the dispersion degree of B1, μ b1 is the average value of B1, E b1 is the low-frequency energy proportion of B1, and w1 to w5 are preset weight coefficients.

6. The method of claim 4, wherein, Running state value Y2 = w6 x σ a2 + w7 x μ a2 + w8 x Δ a2 + w9 x σ b2 + w10 x μ b2 + w11 x Δ b2 + w12 x F, wherein σ a2 is a dispersion degree of A2, μ a2 is an average value of A2, σ b2 is a dispersion degree of B2, μ b2 is an average value of B2, Δ a2 is a maximum difference of A2, Δ b2 is a maximum difference of B2, F is a resonance frequency offset of B2, and w6 to w12 are preset weight coefficients.

7. An electronic device, comprising: comprising a processor and a memory; the processor is configured to execute the steps of the method according to any one of claims 1 to 6 by invoking programs or instructions stored in the memory.

8. A computer-readable storage medium, characterized in that, the computer readable storage medium is configured to store programs or instructions, which enable the computer to execute the steps of the method according to any one of claims 1 to 6. the computer readable storage medium is configured to store programs or instructions, which enable the computer to execute the steps of the method according to any one of claims 1 to 6.

Images