Elevator brake safety intelligent early warning method based on online monitoring

Abstract

The invention discloses an elevator brake safety intelligent early warning method based on online monitoring, and belongs to the technical field of elevator brake early warning. The method specifically comprises the steps that multi-source operation monitoring data and machine room environment data in the braking process are synchronously collected; extracting a performance index from the operation data and fusing the performance index with the environment data to generate a state feature vector; constructing a dynamic mapping model associated with the environment change and the performance evolution rate based on the historical data; deducing an evolution track and an expected fluctuation range of a future performance index according to current and predicted environment data by using the model, and defining the evolution track and the expected fluctuation range as a collaborative early warning boundary; by comparing the real-time state feature vector with the collaborative early warning boundary, early warning is triggered during continuous multiple times of deviation; according to the method, quantitative correlation between environmental factors and performance degradation is established, and accurate early warning of early performance degradation of the brake is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of elevator brake early warning, in particular to an elevator brake safety intelligent early warning method based on online monitoring. BACKGROUND

[0002] As an indispensable vertical transportation tool in modern buildings, the running safety of the elevator is directly related to the safety of personnel's life and property. The elevator brake is the core component of the elevator safety system, which undertakes the key functions of keeping still when the elevator stops and emergency stopping in abnormal situations. Therefore, real-time monitoring and accurate early warning of the running state of the elevator brake are important technical means to prevent elevator safety accidents and achieve predictive maintenance.

[0003] In the existing technical solutions, in order to improve the accuracy of monitoring, the industry has developed a monitoring method based on multi-source signal analysis. These methods usually involve collecting one or more physical signals such as current, temperature, vibration of the brake, and extracting characteristic indexes such as brake response time, temperature rise, vibration amplitude from these signals. Further, some technologies set fixed thresholds or dynamic thresholds according to historical data during the initial operation of the device, compare the real-time extracted characteristic indexes with these thresholds, and determine whether the brake state is abnormal and whether to issue an early warning. These methods improve the ability to identify explicit brake failures by introducing multi-dimensional signals and certain dynamic adjustment mechanisms.

[0004] However, most of the existing technical solutions focus on analyzing the physical signals of the brake itself, and do not systematically consider the running environment of the brake, such as machine room temperature, humidity, electromagnetic interference, and other factors that have a lasting impact on the performance degradation process of the brake. Due to the lack of modeling of the correlation between environmental factors and performance degradation, the setting of the existing early warning boundary often relies on the statistical fluctuations of historical data, and fails to reflect the real evolution law of performance degradation under specific environmental stress, resulting in a mismatch between the early warning boundary and the actual risk. The existing early warning method has room for improvement in real-time, accuracy and adaptability. SUMMARY

[0005] The purpose of the present application is to provide an elevator brake safety intelligent early warning method based on online monitoring, which solves the following technical problems: The existing elevator brake early warning method fails to model the correlation between environmental factors and performance degradation, resulting in an early warning boundary that does not truly reflect the performance evolution law under specific environmental stress.

[0006] The purpose of the present application can be achieved by the following technical solutions: An elevator brake safety intelligent early warning method based on online monitoring, comprising the following steps: S1, synchronously collecting multi-source operation monitoring data generated by the elevator brake in each braking process and real-time environment data of the elevator machine room where the brake is located; S2, extracting performance indicators representing the current working state of the brake from the multi-source operation monitoring data; S3, fusing the extracted performance indicators with the synchronously collected real-time environment data to generate a state feature vector with fused environment factors; S4, based on historical data, constructing a dynamic mapping model describing the correlation between environment data changes and performance indicator evolution rates; S5, using the dynamic mapping model to deduce the evolution trajectory and expected fluctuation range of the performance indicators in a future set time period according to the current and predicted environment data sequences, and defining the expected fluctuation range as a collaborative warning boundary; S6, comparing the real-time generated fused state feature vector with the collaborative warning boundary, and triggering a safety warning signal when the fused state feature vector deviates from the collaborative warning boundary in consecutive braking events.

[0007] As a further scheme of the present application: in S1, the specific process of synchronously collecting multi-source operation monitoring data and real-time environment data is as follows: After the braking instruction is issued, the instantaneous values of the voltage and current of the brake electromagnet coil are recorded to form a braking response electrical signal waveform, the temperature distribution data of the brake pad friction surface during braking are collected through the temperature sensor embedded in the brake arm, the mechanical impact vibration signals of the brake at the moment of braking closure and release are collected through the high-frequency vibration sensor installed on the brake support, while collecting the operation monitoring data, the air humidity data, the environment temperature data in the elevator machine room, and the electromagnetic interference intensity data near the heat dissipation port of the frequency converter of the elevator control system are collected, and all data are recorded with a unified time scale, and the time stamps are completely consistent.

[0008] As a further scheme of the present application: in S2, the specific process of extracting performance indicators representing the current working state of the brake from the multi-source operation monitoring data is as follows: The braking response electrical signal waveform is analyzed, the time interval experienced by the current from zero to the rated attraction current value is calculated, and this time interval is defined as the electromagnetic response delay, the temperature distribution data of the brake pad friction surface are processed, the highest temperature point is located, and the time interval experienced by the temperature of this highest point to drop to the same as the machine room environment temperature after the end of the braking process is calculated, and this time interval is defined as the heat dissipation period, the mechanical impact vibration signal is envelope analyzed, the peak value of the vibration signal envelope line is extracted, and the peak value is compared with the reference envelope peak value measured when the brake is newly installed, the percentage value is calculated, and the percentage value is defined as the mechanical impact attenuation rate.

[0009] As a further scheme of the present application: in the S3, the specific process of generating the state feature vector fusing environmental factors is: The three performance indicators of the electromagnetic response delay, the thermal dissipation period and the mechanical impact attenuation rate extracted from each braking event are combined into a basic state vector, the synchronously collected air humidity data, environmental temperature data and electromagnetic interference intensity data are combined into an environmental parameter vector, the basic state vector and the environmental parameter vector are spliced to form an extended vector, and the extended vector is sent to a fully connected linear transformation layer. The fully connected linear transformation layer outputs a fixed-dimension vector, which is the state feature vector fusing environmental factors.

[0010] As a further scheme of the present application: in the S4, the specific process of constructing the dynamic mapping model is: The sequence of fused state feature vectors continuously recorded in a historical period is collected, the environmental parameter vector part and the performance indicator part are separated from the sequence of fused state feature vectors, the environmental parameter vector is subjected to combined condition analysis, and multiple typical environmental combination modes are identified. The typical environmental combination modes include high temperature and high humidity combination, high temperature and strong electromagnetic interference combination, and normal temperature, high humidity and strong electromagnetic interference combination. The performance indicator data under each typical environmental combination mode is subjected to time series analysis, the recursive least squares method with a forgetting factor is used to calculate the change rate coefficient of the electromagnetic response delay, the change rate coefficient of the thermal dissipation period and the change rate coefficient of the mechanical impact attenuation rate under each environmental combination mode, a mapping relationship library is established with the environmental combination mode as the index and the change rate coefficients of the performance indicators as the content, and the mapping relationship library is the dynamic mapping model.

[0011] As a further scheme of the present application: the specific process of calculating the change rate coefficient of each performance indicator under each environmental combination mode is: For each environmental combination mode, the corresponding performance indicator data points are arranged in time sequence, a weight coefficient based on time decay is assigned to each newly arrived data point, the weight coefficient of the data point closer to the current time is larger, the slope estimate value of the performance indicator changing with time is updated through a recursive formula, the recursive formula contains a forgetting factor parameter, which controls the influence degree of historical data in the estimation process, and the convergence state of the slope estimate value is continuously monitored in the calculation process. When the fluctuation range of the slope estimate value of the continuous multiple data points is less than a set threshold, the current slope estimate value is output as the change rate coefficient of the performance indicator under the corresponding environmental combination mode.

[0012] As a further scheme of the present application: in the S5, the specific process of deducing the performance indicator evolution track and the collaborative early warning boundary is: Obtaining predicted environment data of the elevator machine room in a future period of time, the predicted environment data including humidity and temperature data of a weather forecast and an elevator operation schedule, inputting a sequence of the predicted environment data into the dynamic mapping model, and the dynamic mapping model outputting expected change slopes of performance indicators at different time points in the future, the performance indicators including electromagnetic response delay, heat dissipation period and mechanical shock attenuation rate, taking a measured value of the performance indicator at a current time as a starting point, accumulating the expected change slopes, and generating an evolution track of the performance indicator in a future set period of time; Meanwhile, an environment combination mode type is determined according to the predicted environment data, the environment combination mode type including a high-temperature and high-humidity combination, a high-temperature and strong electromagnetic interference combination, and a normal-temperature, high-humidity and strong electromagnetic interference combination, and based on the determined environment combination mode type and natural fluctuation data of the performance indicators in the corresponding environment combination mode in historical data, upper and lower boundaries around the evolution track are calculated, and the upper and lower boundaries are the cooperative early warning boundaries.

[0013] As a further scheme of the present application, the specific process of calculating the upper and lower boundaries around the evolution track is as follows: From the historical data, a historical time period of the same type as the currently predicted environment combination mode type is screened out, deviation values of the performance indicators in these historical time periods relative to their respective evolution tracks are extracted, corresponding boundary calculation parameters are selected according to the determined environment combination mode type, a first set of boundary coefficients is used for the high-temperature and high-humidity combination mode, a second set of boundary coefficients is used for the high-temperature and strong electromagnetic interference combination mode, and a third set of boundary coefficients is used for the normal-temperature, high-humidity and strong electromagnetic interference combination mode, the selected boundary coefficients are multiplied by the standard deviation of the deviation values to obtain boundary adjustment amounts, and the current evolution track value is taken as a reference, the boundary adjustment amounts are added to the current evolution track value as upper boundaries and subtracted from the current evolution track value as lower boundaries, and the upper and lower boundaries jointly constitute the cooperative early warning boundaries for the current environment combination mode.

[0014] As a further scheme of the present application, in S6, the specific process of triggering the safety early warning signal is as follows: For the fusion state feature vector generated by each new braking event, the basic state vector part is extracted, the measured value of the current performance indicator is compared with the upper and lower boundary values of the cooperative early warning boundary at the same time point, and the current environment combination mode type is identified, different counting thresholds are used according to different environment combination mode types, a first counting threshold is used in the high-temperature and high-humidity combination mode, a second counting threshold is used in the high-temperature and strong electromagnetic interference combination mode, and a third counting threshold is used in the normal-temperature, high-humidity and strong electromagnetic interference combination mode, and when the measured value of the performance indicator continuously exceeds the cooperative early warning boundary for a number of times reaching the counting threshold set for the corresponding environment combination mode, a safety early warning signal corresponding to the environment combination mode is triggered.

[0015] The present application has the following beneficial effects: The application quantitatively characterizes the influence of environmental stress on performance degradation by constructing a dynamic mapping model between the environmental combination mode and the change rate of the brake performance index, so that the setting of the early warning boundary is no longer based on the static statistics of historical fluctuations, but based on the dynamic deduction of the performance evolution trajectory under specific future environmental conditions, thereby significantly improving the matching degree of the early warning boundary and the actual risk state. By using the recursive least squares method with a forgetting factor to update the change rate coefficient of the performance index under different environmental combination modes, the model can continuously track the changes of the performance degradation trajectory of the brake itself and the environmental response, ensuring the long-term accuracy of the evolution trajectory prediction. Further, by adaptively selecting boundary calculation parameters according to the identified environmental combination mode, a cooperative early warning boundary with environmental pertinence is generated, and a multi-mode counting threshold triggering mechanism associated with the environmental combination mode is adopted, realizing the fundamental change of the early warning logic from single judgment to environmental coupling degradation speed adaptation, and finally realizing more accurate perception and more timely warning of the early trend of brake performance degradation in complex and variable actual operating environment. BRIEF DESCRIPTION OF DRAWINGS

[0016] The application will be further described below with reference to the accompanying drawings.

[0017] Figure 1 is a flowchart of the application. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the application, not all. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the application.

[0019] Please refer to Figure 1 The application is an elevator brake safety intelligent early warning method based on online monitoring, comprising the following steps: S1, synchronously collecting multi-source operation monitoring data generated by the elevator brake in each braking process, and real-time environmental data of the elevator machine room where the brake is located; S2, extracting performance indicators representing the current working state of the brake from the multi-source operation monitoring data; S3, fusing the extracted performance indicators with the synchronously collected real-time environmental data to generate a state feature vector with fused environmental factors; S4, based on historical data, constructing a dynamic mapping model describing the correlation between environmental data changes and performance evolution rate; S5, according to the current and predicted environment data sequence, deducing the evolution track and expected fluctuation range of the performance index in the future setting time period by using the dynamic mapping model, and defining the expected fluctuation range as a cooperative early warning boundary; S6, comparing the real-time generated fusion state feature vector with the cooperative early warning boundary, and triggering a safety warning signal when the fusion state feature vector deviates from the cooperative early warning boundary in continuous multiple braking events.

[0020] In a preferred embodiment of the present application, in S1, the specific process of synchronously collecting multi-source operation monitoring data and real-time environment data is as follows: At the moment when the brake instruction is issued by the elevator control system, the data acquisition system immediately starts a synchronous trigger mechanism, which is hard-wired connected with the signal bus of the elevator main controller, ensuring that the acquisition timing is completely synchronized with the brake action. For the voltage and current signals of the brake electromagnet coil, high-precision data acquisition cards are used for instantaneous value capture. The sampling rate of the acquisition card is set to not less than 1MHz to meet the high-frequency detail restoration requirement of the electrical signal waveform, and the resolution of the acquisition channel is not less than 16 bits to ensure accurate identification of weak signal changes. The voltage signal is obtained through a differential probe connected in parallel across the electromagnet coil, and the current signal is collected through a Hall current sensor connected in series in the power supply circuit. The output signals of the two sensors are processed by a signal conditioning module for noise reduction and amplification, and then transmitted to the analog input channel of the data acquisition card. The acquisition card records the continuous instantaneous value sequence as the brake response electrical signal waveform according to the preset storage format, and the waveform data is written into the cache unit in binary format in real time.

[0021] The temperature sensor uses a miniature patch-type platinum resistance sensor, which is selected according to the structure size and working temperature range of the brake arm. The measurement accuracy of the sensor is ±0.1℃, and the response time is not more than 50ms, which can quickly capture the temperature dynamic changes in the braking process. During installation, the sensor is embedded in the pre-designed mounting groove inside the brake arm through laser welding process, and the distance between the sensor detection end face and the friction surface of the brake pad is controlled within 2mm to ensure direct sensing of the temperature field distribution of the friction area. The signal transmission of the sensor uses shielded wire, which is routed along the inside of the brake arm and led out through a sealed joint to avoid wire wear or mechanical stress during braking. During the whole braking process, the sensor continuously outputs temperature data at a sampling frequency of 100Hz, forming a complete temperature distribution time sequence.

[0022] The high-frequency vibration sensor selects a piezoelectric acceleration sensor, which has a measurement range matching the mechanical impact strength at the brake closing and releasing moment, and a resonance frequency higher than 10 kHz, so as to effectively capture the high-frequency component in the impact vibration. The sensor is installed on the rigid connection part of the brake support by screw fastening, the installation surface is machined to ensure flatness, coupling agent is applied between the sensor and the installation surface to eliminate the contact gap, and the vibration signal is effectively transmitted. The charge signal output by the sensor is converted into a voltage signal by a charge amplifier, and then the signal is purified by a low-pass filter, and the filter cutoff frequency is set to 5 kHz to filter high-frequency noise interference. During the collection process, the vibration sensor records the vibration signals at the two key moments of brake closing and releasing at a sampling frequency of 2 kHz, and the sampling time covers the complete period from the brake command triggering to the stable mechanical action, so as to ensure the complete capture of the impact signal.

[0023] The same time sequence synchronization system is used for environmental data collection and operation monitoring data collection, a unified time reference is provided by a GPS time grant module, the clock signals of all collection devices are calibrated in real time with the reference clock, the time stamps of all types of data are completely consistent, and the time synchronization accuracy is controlled within 1 ms. The air humidity data is collected by a capacitive humidity sensor installed in a well-ventilated area of the elevator machine room, the measurement range of the sensor is 0%-100%RH, the accuracy is ±2%RH, and the sampling frequency is 10 Hz, which can reflect the dynamic change of the humidity in the machine room in real time. The environmental temperature data is obtained by a platinum resistance temperature sensor, which has the same model as the brake pad temperature sensor and is installed away from heat sources such as brakes and frequency converters to avoid local temperature interference and ensure that the measurement result reflects the overall environmental temperature of the machine room. The electromagnetic interference intensity data is collected by an electromagnetic coupling probe deployed near the heat dissipation port of the elevator control system frequency converter, the detection frequency range of the probe covers the main interference frequency range generated by the frequency converter during operation, and the distance between the probe and the heat dissipation port of the frequency converter is controlled within 30 cm to ensure effective reception of the interference signal. The collected electromagnetic interference signal is converted into a quantitative interference intensity value by a spectrum analyzer, and the sampling frequency is consistent with that of the humidity and temperature sensors. All environmental data and operation monitoring data are stored in the local storage unit of the data collection system in chronological order, and are uploaded to the elevator safety monitoring platform in real time through the Ethernet interface, providing complete raw data support for subsequent data processing.

[0024] In another preferred embodiment of the application, in S2, the specific process of extracting performance indicators representing the current working state of the brake from the multi-source operation monitoring data is as follows: For the brake response electrical signal waveform, first, the pre-processing operation is carried out to eliminate the power frequency interference and noise components in the signal. The pre-processing process adopts an adaptive notch filter to suppress 50Hz power frequency interference, and at the same time, the random noise is smoothed by a sliding average filter, and the filter window size is dynamically adjusted according to the signal sampling rate, ensuring that the signal quality is improved on the premise of retaining the key features of the signal. The pre-processed electrical signal waveform is identified by the data analysis module, which first locates the starting zero point of the current signal, that is, the initial time when the current starts to rise after the brake command is issued, and then identifies the stable time when the current signal reaches the rated inrush current value. The rated inrush current value is a standard value specified in the brake design parameters and is pre-stored in the system parameter library. By calculating the time interval between the two time points, the electromagnetic response delay index is obtained, which directly reflects the response speed of the electromagnet from power-on to generating enough suction force to drive the brake mechanism to act. In the extraction process, the electrical signal waveform of each brake period is analyzed independently to ensure the real-time and independence of the index.

[0025] The temperature distribution data of the brake pad friction surface is first subjected to spatial interpolation processing. The discrete sensor data is expanded in space by the Kriging interpolation algorithm to construct a complete temperature field distribution matrix, which can clearly show the temperature differences of each region of the brake pad friction surface. Based on the temperature field distribution matrix, the maximum value search algorithm is used to locate the coordinates and corresponding temperature values of the highest temperature point. The search process traverses all elements of the matrix to ensure that no local temperature peaks are missed. Then, the system extracts the end time of the braking process, which is triggered by the brake completion signal issued by the elevator control system. Taking this time as the starting point, the temperature change trend of the highest temperature point is continuously monitored until its temperature value and the ambient temperature value of the machine room collected at the same period tend to be consistent, with the judgment standard being that the temperature difference in the last three sampling periods is less than 0.5℃. By calculating the time span from the end time of braking to the time when the temperature tends to be consistent, the heat dissipation period index is obtained, which reflects the heat dissipation performance of the brake pad and is directly related to the material properties and wear state of the brake pad.

[0026] The processing of mechanical impact vibration signals first involves signal preprocessing, including DC component removal and trend term elimination. DC component removal is achieved by subtracting the signal mean, while trend term elimination employs a least-squares fitting method to ensure that the vibration signal contains only the dynamic components generated by the mechanical impact. The preprocessed signal then enters the envelope analysis process. Hilbert transform is used to analyze the vibration signal, obtaining the analytic signal. The envelope is then extracted by modulo operation on the analytic signal. The envelope highlights the amplitude variation characteristics of the vibration signal and effectively reflects the intensity of the mechanical impact. Subsequently, the peak value of the envelope is extracted; this peak value is the maximum value of the envelope throughout the entire sampling period. When the brake is newly installed, the benchmark envelope peak value is measured under standard test conditions, including rated load, standard braking frequency, and normal temperature and humidity. The measured benchmark envelope peak value is stored in the system benchmark database as a reference benchmark for subsequent comparative analysis. The percentage value obtained by comparing the peak value of the current braking event with the peak value of the baseline envelope is the mechanical shock attenuation rate. This indicator can quantitatively reflect the wear, loosening and other state changes of the mechanical transmission components of the brake. The higher the attenuation rate, the more serious the performance degradation of the mechanical components.

[0027] In another preferred embodiment of the present invention, the specific process of generating the state feature vector that integrates environmental factors in step S3 is as follows: After each braking event, the specific values ​​of three performance indicators—electromagnetic response delay, heat dissipation period, and mechanical impact attenuation rate—are extracted from the data processing results. These three values ​​are arranged in a preset order to form a basic state vector. The basic state vector is three-dimensional, with each dimension corresponding to a performance indicator. The values ​​of each element in the vector are retained to three decimal places to ensure data accuracy. Simultaneously, air humidity data, ambient temperature data, and electromagnetic interference intensity data collected synchronously with the braking event are extracted and arranged in a fixed order to form an environmental parameter vector. The environmental parameter vector is also three-dimensional, and the numerical format of each element is consistent with that of the basic state vector to ensure data format uniformity.

[0028] The splicing process of the basic state vector and the environmental parameter vector is realized by dimension expansion, and three elements of the environmental parameter vector are sequentially appended to the end of the basic state vector to form an extended vector with a dimension of 6. Before splicing, the timestamps of the two vectors need to be checked twice to confirm that the corresponding braking events are completely consistent, so as to avoid vector mismatch caused by data transmission delay. After the extended vector is generated, it is first standardized. The Z-score standardization method is used to convert the values of each dimension to standard values with a mean of 0 and a variance of 1. Standardization can eliminate the influence of dimension differences between different indicators and improve the effectiveness of subsequent feature transformation. The standardized extended vector is transmitted to the fully connected linear transformation layer as input data. The transformation layer is a basic structure in the deep learning model, and its core function is to realize linear mapping and dimension unification of features.

[0029] The input dimension of the fully connected linear transformation layer is set to 6, which is completely matched with the dimension of the extended vector. The output dimension is preset according to the input requirements of the subsequent dynamic mapping model, usually set to fixed 8 or 16. The dimension design needs to consider the feature representation ability and the calculation efficiency. The transformation layer includes input layer, hidden layer and output layer. The number of input layer neurons is consistent with the input dimension, and the number of output layer neurons is consistent with the preset output dimension. The number of hidden layer neurons is determined according to the empirical formula to ensure the sufficiency of feature transformation. The weight parameters of the transformation layer are obtained through offline training. The training data uses the extended vector sample set collected historically, and the training target is to minimize the error between the output vector and the labeled feature vector. After training, the weight parameters are fixed in the system and are not adjusted in actual operation. After the extended vector enters the transformation layer through the input layer, it is linearly weighted and calculated by the hidden layer and processed by the activation function. Finally, the fixed dimension vector is output by the output layer, which is the state feature vector fused with environmental factors. After the generation of the feature vector, the system stores it together with the corresponding timestamp and environment combination mode label to provide standardized feature input for the construction of the subsequent dynamic mapping model and the deduction of the early warning boundary.

[0030] In another preferred embodiment of the application, in the S4, the specific process of constructing the dynamic mapping model is: Collect the sequence of fused state feature vectors recorded continuously in the historical period. The sequence needs to cover at least three complete seasonal cycles to ensure that it contains environmental changes and performance evolution data under different climate conditions. The data points in the sequence need to meet the time continuity requirement. For missing data caused by device failure or communication interruption, linear interpolation method based on adjacent valid data points is used for completion. The completed sequence is sorted by timestamp to form a complete historical data set.

[0031] The operation of separating the environment parameter vector part and the performance index part from the fusion state feature vector sequence is realized based on the dimension structure of the feature vector. The first three dimensions of the fusion state feature vector correspond to the electromagnetic response delay, the heat dissipation period, and the mechanical impact attenuation rate, constituting the performance index part; the last three dimensions correspond to the air humidity, the environmental temperature, and the electromagnetic interference intensity, constituting the environment parameter vector part. The two types of data are directly split through dimension indexing and stored in the performance index database and the environment parameter database respectively. The timestamp correspondence of the two types of data needs to be verified in the splitting process to ensure that each set of environment parameters and performance indexes come from the same braking event.

[0032] The combined condition analysis of the environment parameter vector is realized by combining multi-dimensional threshold division and clustering algorithm. First, statistical analysis is performed on the historical data in the environment parameter database to determine the distribution range of each parameter: the division of the environmental temperature is based on the design standard of the computer room ventilation, and the high temperature threshold is set to be the upper limit value exceeding the normal working temperature range of the equipment; the high humidity threshold of the air humidity refers to the humidity warning value in the elevator machine room environment requirement; the strong interference threshold of the electromagnetic interference intensity is determined based on the electromagnetic compatibility standard of the frequency converter under rated working condition. Based on the above threshold, the environment parameter vector is preliminarily classified, and then the K-means clustering algorithm is used to optimize the preliminary classification result, and the optimal clustering number is determined by calculating the contour coefficient, and finally the typical environment combination modes such as high temperature and high humidity combination, high temperature and strong electromagnetic interference combination, normal temperature and high humidity and strong electromagnetic interference combination are identified. Each mode corresponds to a specific parameter range, for example, the high temperature and high humidity combination needs to meet the conditions of the environmental temperature being higher than the high temperature threshold and the air humidity being higher than the high humidity threshold, and the classification result is stored in the environment mode label library to provide classification basis for subsequent performance index analysis.

[0033] For the performance index data under each typical environment combination mode, time series analysis is carried out. First, all data points matching a certain environment mode label are selected from the performance index database and arranged in ascending order of timestamp to form the performance index time series under the mode. The sequence is preprocessed, and abnormal values are identified and removed by 3σ rule, and the replacement of abnormal values adopts the sliding average calculation result of adjacent normal data points. The preprocessed sequence enters the change rate coefficient calculation process, which is realized based on the recursive least squares method with forgetting factor.

[0034] During the calculation process, a time-decay-based weight coefficient is assigned to each newly arrived data point. The weight coefficient exponentially decays with the increase of the time interval between the data point and the current time, ensuring that the data points closer to the current time have a greater impact on the coefficient estimation. In the initial state, the system loads the initial slope estimate value trained by the environmental mode historical data, and updates the slope estimate value by recursion when new data points are input. A forgetting factor parameter is introduced during the recursion process, which has a value range of 0 to 1. By reducing the weight proportion of historical data, the influence of obsolete data on the current estimation result is weakened, so that the model can adapt to the progressive degradation of the brake performance.

[0035] The convergence state of the slope estimate value is achieved by continuous monitoring. The system sets a slope fluctuation threshold, which is determined based on the historical fluctuation amplitude of the performance index. Each time the slope estimate value is updated, the difference from the previous estimate value is calculated. When the difference of a plurality of consecutive data points is less than the fluctuation threshold, it is determined that the slope estimate value reaches a stable state, and the output slope estimate value is the change rate coefficient of the performance index under the corresponding environmental combination mode.

[0036] After completing the calculation of the change rate coefficients of each performance index under all typical environmental combination modes, a mapping relationship library is established. The library takes the environmental combination mode as the first index and the performance index type as the second index, and stores the corresponding change rate coefficients. For example, under the high temperature and high humidity combination index, the electromagnetic response delay change rate coefficient, the heat dissipation period change rate coefficient, and the mechanical impact attenuation rate change rate coefficient are stored respectively. The mapping relationship library supports real-time updating. When new historical data accumulates to a certain amount, the change rate coefficients under each mode are recalculated and the content in the library is updated, ensuring that the model can continuously reflect the correlation between the environment and the performance evolution. The mapping relationship library is the dynamic mapping model.

[0037] In another preferred embodiment of the present application, the specific process of deducing the performance index evolution track and the collaborative early warning boundary in S5 is: The predicted environment data of the elevator machine room in the future period of time is obtained through multi-source information fusion, wherein the humidity and temperature data are derived from a third-party meteorological service interface, the interface provides fine prediction data for the next 72 hours in units of hours, and the data includes the average air humidity, the maximum and minimum environmental temperature of the geographical location of the machine room. The system performs spatial adaptation processing on the meteorological data, and corrects the deviation between the historical data of the environmental sensors installed in the machine room and the data of the weather station, to convert the regional weather forecast into the environmental prediction value of the specific area of the machine room. The elevator operation schedule is derived from the elevator management system, including the expected number of operations, peak period distribution and maintenance downtime plan in the future period of time, the schedule data is used to infer the working frequency of the frequency converter, and then estimate the trend of the electromagnetic interference intensity, the higher the operation frequency, the greater the electromagnetic interference intensity near the heat dissipation port of the frequency converter, and the correlation between the two is determined based on statistical analysis of historical operation data and electromagnetic interference monitoring values.

[0038] The processed predicted environment data is arranged in time sequence to form a predicted environment data sequence, the time granularity of the sequence matches the average occurrence interval of the braking event, and is usually set to 15 minutes per data point. The data sequence includes the predicted air humidity value, the predicted environmental temperature value and the predicted electromagnetic interference intensity value at each time point, which together constitute a predicted environment parameter vector. After the sequence is input into the dynamic mapping model, the model first performs pattern recognition on the predicted environment parameter vector at each time point, and determines the environment combination mode type to which the time point belongs by comparing the parameter range with the typical environment combination mode. Subsequently, the model calls the corresponding performance index change rate coefficient in the mapping relationship library according to the environment combination mode type, i.e. matches the corresponding change rate coefficient for the electromagnetic response delay, heat dissipation period and mechanical impact attenuation rate respectively, to obtain the expected change slope of the performance index at different time points in the future.

[0039] The generation of the performance index evolution trajectory takes the current measured value of the performance index as the starting point. The current measured value is obtained through multi-source data processing of the latest braking event, including the real-time measurement value of the electromagnetic response delay, the real-time calculation value of the heat dissipation period and the real-time comparison value of the mechanical impact attenuation rate. Taking the time axis as the horizontal axis, the expected change slope at each future time point is calculated, and the trajectory value at each time point is equal to the product of the trajectory value at the previous time point and the current slope (the product factor is the time interval). The complete evolution trajectory in the future set period of time is gradually generated. The length of the time period can be adjusted according to the application requirements, and is usually set to 24 hours to balance the timeliness of short-term warning and the efficiency of data processing. During the trajectory generation process, if the environment combination mode in the predicted environment data sequence changes, the system automatically updates the corresponding change rate coefficient to ensure that the trajectory can reflect the influence of the environment mode change on the performance evolution.

[0040] The calculation of the cooperative early warning boundary and the deduction of the evolution track are synchronized, and the core is to determine the dynamic fluctuation range based on the environmental combination mode. First, according to the mode recognition result of each time point in the predicted environmental data sequence, the dominant environmental combination mode type is determined, and if there are multiple modes in the sequence, the mode with the highest time proportion is taken as the current calculation reference. From the historical database, the historical time period completely matched with the mode type is screened out, and the screening conditions include that the environmental temperature, humidity and electromagnetic interference intensity all fall within the parameter range of the corresponding mode, and the duration is not less than 24 hours, so as to ensure the representativeness of the historical data.

[0041] The measured value of the performance index in the screened historical time period and the historical evolution track value at the corresponding moment are extracted, and the difference between the two is the deviation value. The deviation value is preprocessed, and the extreme abnormal value is identified and removed by the box plot method, and the effective deviation value reflecting the normal fluctuation is retained. The standard deviation is calculated based on the effective deviation value, which quantifies the natural fluctuation amplitude of the performance index around the track in the historical data.

[0042] According to the determined environmental combination mode type, the preset boundary coefficient is called, and the boundary coefficient of different modes is optimized through historical early warning effect verification: in the high temperature and high humidity combination mode, the thermal aging and humidity corrosion of the brake pad have a synergistic effect, and the performance fluctuation is larger, so the first set of boundary coefficients is adopted; in the high temperature and strong electromagnetic interference combination mode, the electromagnetic interference has a more significant influence on the response of the electromagnet, so the second set of boundary coefficients is adopted; the comprehensive influence of the normal temperature, high humidity and strong electromagnetic interference combination mode is relatively moderate, so the third set of boundary coefficients is adopted. The selected boundary coefficient is multiplied by the standard deviation of the deviation value to obtain the boundary adjustment amount.

[0043] Taking the track value of each time point on the evolution track as the reference, the upper boundary is obtained by adding the boundary adjustment amount, and the lower boundary is obtained by subtracting the boundary adjustment amount. The upper and lower boundaries jointly constitute the cooperative early warning boundary, and the boundary changes synchronously with the evolution track, and the boundary coefficient is automatically adjusted when the environmental combination mode is switched, so that the boundary always matches the normal performance fluctuation range under the current environmental condition. The finally generated evolution track and cooperative early warning boundary are stored in the form of time sequence, which provides a dynamic reference benchmark for subsequent real-time early warning judgment.

[0044] In another preferred embodiment of the application, the specific process of triggering the safety early warning signal in S6 is: After each new braking event is completed, the fusion state feature vector generated by the event is received in real time, and the basic state vector part is extracted therefrom through dimension analysis. The basic state vector includes the measured values of three performance indicators: electromagnetic response delay, heat dissipation period and mechanical impact attenuation rate, and the extraction process is realized through fixed dimension index, so as to ensure that the data strictly corresponds to the timestamp of the current braking event and avoid cross-event data confusion.

[0045] The pre-generated cooperative warning boundary sequence is called, the time point corresponding to the current braking event is matched accurately through the time stamp, and the upper boundary and lower boundary values of the time point are located. The measured values of the three performance indicators in the basic state vector are compared with the corresponding boundary values respectively, and it is judged whether the measured values exceed the boundary range. The exceeding determination criteria are that the measured value of the electromagnetic response delay is greater than the upper boundary or less than the lower boundary, the measured value of the heat dissipation period is greater than the upper boundary or less than the lower boundary, and the measured value of the mechanical impact attenuation rate is greater than the upper boundary or less than the lower boundary. If any indicator exceeds, it is recorded as a boundary deviation event.

[0046] At the same time, the real-time environmental data of the current elevator machine room is collected, including air humidity, environmental temperature and electromagnetic interference intensity, and by comparing with the parameter range of the typical environmental combination mode, the type of the current environmental combination mode is identified. The comparison process adopts interval matching method, and the real-time environmental parameters are matched with the threshold ranges of the high temperature and high humidity combination, the high temperature and strong electromagnetic interference combination, the normal temperature and high humidity and strong electromagnetic interference combination one by one. If all the parameter conditions of a mode are met, it is determined as the mode.

[0047] According to the identified environmental combination mode type, the system calls the preset counting threshold: the first counting threshold is enabled under the high temperature and high humidity combination mode, which is set according to the characteristics of performance degradation under hot and humid environment; the second counting threshold is enabled under the high temperature and strong electromagnetic interference combination mode, which is adapted to the performance fluctuation scene dominated by electromagnetic interference; and the third counting threshold is enabled under the normal temperature and high humidity and strong electromagnetic interference combination mode, which matches the relatively moderate comprehensive environmental stress.

[0048] A continuous counting mechanism is built in, and the counter value is increased by one each time a boundary deviation event is detected. If all performance indicators in a braking event do not exceed the boundary, the counter is reset to zero. When the counter value accumulates to the counting threshold corresponding to the current environmental combination mode, a safety warning signal is triggered. The warning signal contains key information such as environmental combination mode identification, performance indicator overrun type and continuous overrun times, which is pushed to the maintenance terminal through the elevator safety monitoring platform, and an audible and visual alarm is generated in the local control cabinet, realizing timely response to the abnormal state of the brake.

[0049] The above describes one embodiment of the application in detail, but the content described is only the preferred embodiment of the application and cannot be considered as limiting the scope of the application. Any equivalent changes and improvements made within the scope of the application should still belong to the patent coverage of the application.

Claims

1. An online monitoring-based intelligent early warning method for elevator brake safety, characterized in that, The method comprises the following steps: S1, synchronously collecting multi-source operation monitoring data generated by the elevator brake in each braking process and real-time environmental data of the elevator machine room where the brake is located; S2, extracting performance indicators representing the current working state of the brake from the multi-source operation monitoring data; S3, fusing the extracted performance indicators with the synchronously collected real-time environmental data to generate a state feature vector fused with environmental factors; S4, based on historical data, constructing a dynamic mapping model describing the correlation between environmental data changes and performance indicator evolution rates; S5, using the dynamic mapping model to deduce the evolution trajectory and expected fluctuation range of the performance indicators in a future set time period according to the current and predicted environmental data sequences, and defining the expected fluctuation range as a collaborative warning boundary; S6, comparing the real-time generated fused state feature vector with the collaborative warning boundary, and triggering a safety warning signal when the fused state feature vector deviates from the collaborative warning boundary in continuous multiple braking events.

2. The online monitoring based intelligent early warning method for elevator brake safety according to claim 1, characterized in that, In S1, the specific process of synchronously collecting multi-source operation monitoring data and real-time environmental data is as follows: After the brake command is issued, the instantaneous values of the voltage and current of the brake electromagnet coil are recorded to form a brake response electrical signal waveform, the temperature distribution data of the brake pad friction surface during braking is collected through the temperature sensor embedded in the brake arm, the mechanical impact vibration signal of the brake at the moment of brake closing and releasing is collected through the high-frequency vibration sensor installed on the brake support, and the air humidity data, environmental temperature data in the elevator machine room, and electromagnetic interference intensity data near the heat dissipation port of the frequency converter of the elevator control system are collected at the same time as the operation monitoring data is collected. All data are recorded with a unified time scale, and the time stamps are completely consistent.

3. The online monitoring based intelligent early warning method for elevator brake safety according to claim 1, characterized in that, In S2, the specific process of extracting performance indicators representing the current working state of the brake from the multi-source operation monitoring data is as follows: The brake response electrical signal waveform is analyzed, the time interval experienced by the current from zero to the rated attraction current value is calculated, and this time interval is defined as the electromagnetic response delay. The temperature distribution data of the brake pad friction surface is processed, the highest temperature point is located, and the time interval experienced by the temperature of this highest point to drop to the same as the machine room environmental temperature after the braking process is over is calculated, which is defined as the heat dissipation period. The mechanical impact vibration signal is envelope analyzed, the peak value of the vibration signal envelope line is extracted, and the peak value is compared with the reference envelope peak value measured when the brake is newly installed to calculate a percentage value, which is defined as the mechanical impact attenuation rate.

4. The online monitoring based intelligent early warning method for elevator brake safety according to claim 1, characterized in that, In S3, the specific process of generating a state feature vector fused with environmental factors is as follows: The three performance indicators of the electromagnetic response delay, the heat dissipation period and the mechanical impact attenuation rate extracted from each braking event are combined into a basic state vector, the synchronously collected air humidity data, the environmental temperature data and the electromagnetic interference intensity data are combined into an environmental parameter vector, the basic state vector and the environmental parameter vector are spliced to form an extended vector, and the extended vector is input into a fully connected linear transformation layer.

5. The online monitoring based intelligent early warning method for elevator brake safety according to claim 1, characterized in that, In the S4, the specific process of constructing the dynamic mapping model is as follows: The sequence of the fusion state feature vectors recorded continuously in the historical period is collected, the environmental parameter vector part and the performance indicator part are separated from the sequence of the fusion state feature vectors, the environmental parameter vector is subjected to combination condition analysis, and multiple typical environmental combination modes are identified, including a high-temperature and high-humidity combination, a high-temperature and strong electromagnetic interference combination, and a normal-temperature, high-humidity and strong electromagnetic interference combination; The performance indicator data under each typical environmental combination mode is subjected to time series analysis, the recursive least squares method with a forgetting factor is used to calculate the change rate coefficient of the electromagnetic response delay, the change rate coefficient of the heat dissipation period and the change rate coefficient of the mechanical impact attenuation rate under each environmental combination mode, a mapping relationship library with the environmental combination mode as the index and the change rate coefficients of the performance indicators as the content is established, and the mapping relationship library is the dynamic mapping model.

6. The online monitoring based intelligent early warning method for elevator brake safety according to claim 5, characterized in that, The specific process of calculating the change rate coefficient of each performance indicator under each environmental combination mode is as follows: For each environmental combination mode, the corresponding performance indicator data points are arranged in time sequence, a weight coefficient based on time attenuation is assigned to each newly arrived data point, the weight coefficient of the data point closer to the current time is larger, the slope estimation value of the performance indicator changing with time is updated through a recursive formula, the recursive formula contains a forgetting factor parameter, the forgetting factor parameter controls the influence degree of the historical data in the estimation process, and the convergence state of the slope estimation value is continuously monitored in the calculation process, and when the fluctuation range of the slope estimation value of the continuous multiple data points is less than a set threshold, the current slope estimation value is output as the change rate coefficient of the performance indicator under the corresponding environmental combination mode.

7. The online monitoring based intelligent early warning method for elevator brake safety according to claim 5, characterized in that, In the S5, the specific process of deducing the performance indicator evolution track and the collaborative early warning boundary is as follows: Predicted environmental data of the elevator machine room in a future period of time are acquired, the predicted environmental data include humidity and temperature data of weather forecast and an elevator operation schedule, the sequence of the predicted environmental data is input into the dynamic mapping model, the dynamic mapping model outputs the expected change slope of the performance indicators at different time points in the future, the performance indicators include the electromagnetic response delay, the heat dissipation period and the mechanical impact attenuation rate, the measured value of the performance indicator at the current time is taken as a starting point, the expected change slope is accumulated, and the evolution track of the performance indicator in a set period of time in the future is generated. Meanwhile, the environment combination mode type is determined according to the predicted environment data, the environment combination mode type includes a high-temperature and high-humidity combination, a high-temperature and strong electromagnetic interference combination, and a normal-temperature and high-humidity and strong electromagnetic interference combination, and based on the determined environment combination mode type and natural fluctuation data of the performance index under the corresponding environment combination mode in historical data, upper and lower boundaries around the evolution track are calculated, and the upper and lower boundaries are the collaborative early warning boundaries.

8. The online monitoring based intelligent early warning method for elevator brake safety according to claim 7, characterized in that, The specific process of calculating the upper and lower boundaries around the evolution track is as follows: The historical time periods with the same environment combination mode type as the currently predicted environment combination mode type are filtered out from the historical data, the deviation values of the performance index relative to the respective evolution tracks in the historical time periods are extracted, the corresponding boundary calculation parameters are selected according to the determined environment combination mode type, the first set of boundary coefficients is used for the high-temperature and high-humidity combination mode, the second set of boundary coefficients is used for the high-temperature and strong electromagnetic interference combination mode, and the third set of boundary coefficients is used for the normal-temperature and high-humidity and strong electromagnetic interference combination mode, the selected boundary coefficients are multiplied by the standard deviation of the deviation values to obtain boundary adjustment amounts, and the current evolution track value is taken as a reference, the boundary adjustment amounts are added to the upper boundary and subtracted from the lower boundary, and the upper boundary and the lower boundary jointly constitute the collaborative early warning boundaries for the current environment combination mode.

9. The online monitoring based intelligent early warning method for elevator brake safety according to claim 1, characterized in that, In S6, the specific process of triggering the safety early warning signal is as follows: For the fusion state feature vector generated by each new braking event, the basic state vector part is extracted, the current performance index measured value is compared with the upper and lower boundary values of the collaborative early warning boundary at the same time point, and the current environment combination mode type is identified, different counting thresholds are used according to different environment combination mode types, the first counting threshold is used for the high-temperature and high-humidity combination mode, the second counting threshold is used for the high-temperature and strong electromagnetic interference combination mode, and the third counting threshold is used for the normal-temperature and high-humidity and strong electromagnetic interference combination mode, and when the performance index measured value continuously exceeds the collaborative early warning boundary for a number of times reaching the counting threshold set for the corresponding environment combination mode, the safety early warning signal corresponding to the environment combination mode is triggered.

Images