An intelligent detection system adapted to safe operation of an elevator

Abstract

This invention provides an intelligent detection system adapted for elevator safety operation, which improves the accuracy of elevator fault detection and the precision of response strategies. First, a dynamic sensing module deploys a movable sensor array, dynamically adjusting based on the elevator's operating and stopping status. Next, an intelligent analysis module constructs three independent evaluation dimensions: a physical analysis unit, a soundprint analysis unit, and a door system analysis unit, generating three risk values. Finally, a hierarchical execution module implements a differentiated three-level response mechanism based on the three risk values, performing early warning monitoring, speed-limiting maintenance, and emergency stop for single anomalies, combined anomalies, and door gap anomalies, respectively. This invention significantly improves the intelligent monitoring level and proactive protection capabilities for elevator safety operation through the coordinated work of dynamic sensing, multi-dimensional evaluation, and hierarchical response.

Description

Technical Field

[0001] This invention relates to the field of elevator operation, specifically an intelligent detection system adapted for the safe operation of elevators. Background Technology

[0002] As an indispensable vertical transportation tool in modern buildings, the safe operation of elevators is directly related to the safety of people's lives and property. However, as a complex electromechanical system, elevators inevitably experience various malfunctions and safety hazards during long-term operation. Traditional elevator safety inspections mainly rely on manual periodic inspections and annual mandatory inspections. However, manual inspections are limited by the experience level and subjective judgment of the inspectors, making it difficult to accurately identify early signs of malfunctions, which can easily lead to missed inspections and misjudgments. Furthermore, the time interval between periodic inspections is relatively long, making it impossible to monitor the elevator's operating status in real time. This causes potential malfunctions to accumulate continuously during the inspection cycle, increasing the risk of sudden accidents.

[0003] Existing technologies include intelligent elevator detection systems that collect vibration data during elevator operation by installing vibration sensors inside the elevator car and using spectrum analysis to identify abnormal vibration patterns, thus achieving automatic detection of elevator mechanical faults. However, existing technologies still have significant shortcomings. First, the sensor deployment locations are fixed and singular, making it impossible to adjust the monitoring focus according to different elevator operating states. Second, existing systems often only focus on single-dimensional fault characteristics, lacking fusion analysis of multi-source heterogeneous data. Finally, existing technologies typically employ a binary judgment mode with fixed thresholds, i.e., an alarm is triggered if the threshold is exceeded, and normal operation is maintained if it is not exceeded. This one-size-fits-all approach cannot distinguish the severity of the fault, nor can it implement differentiated response strategies based on different fault types, easily leading to false alarms or missed alarms, affecting the practicality and reliability of the system. Summary of the Invention

[0004] In order to solve the technical problems mentioned in the background art, the present invention proposes an intelligent detection system adapted to the safe operation of elevators.

[0005] Therefore, the technical solution adopted by the present invention is as follows:

[0006] An intelligent detection system adapted for safe elevator operation, the system comprising:

[0007] M1: Dynamic sensing module, which deploys a movable sensor array in the elevator shaft. The movable sensor array is dynamically deployed according to the elevator status to obtain the elevator's vibration spectrum, acoustic signature matrix and door gap width change curve.

[0008] M2: Intelligent analysis module, including a physical analysis unit, an acoustic signature analysis unit, and a door system analysis unit. The physical analysis unit identifies vibration modes and outputs a first risk value by comparing the vibration spectrum diagram with a preset fault feature library. The acoustic signature analysis unit analyzes the acoustic signature feature matrix and obtains a second risk value by combining it with the elevator's cumulative operating mileage. The door system analysis unit obtains the maximum door gap width and the standard deviation of the door gap width by analyzing the door gap width variation curve, and further outputs a third risk value.

[0009] M3: Tiered execution module, which implements a three-level response mechanism based on the first risk value, the second risk value, and the third risk value.

[0010] Furthermore, the movable sensor array includes a triaxial vibration sensor, an omnidirectional microphone, and a radar sensor.

[0011] The elevator status is divided into running status and stopped status. Based on the running status and stopped status, the following deployments are executed respectively.

[0012] 1) In operation, the triaxial vibration sensor collects the three-dimensional vibration signal of the elevator and transmits it to the edge computing node. The edge computing node obtains the vibration spectrum through wavelet transform and fast Fourier transform, and performs normalization processing on the vibration spectrum.

[0013] The omnidirectional microphone records the acoustic signal of the elevator, and further obtains the voiceprint feature matrix through wavelet transform and short-time Fourier transform;

[0014] 2) In the docked state, the radar sensor emits electromagnetic waves and receives reflected signals, and uses the principle of frequency-modulated continuous wave to measure the width of the gap between the elevator door and the landing door, forming a curve of the change in the gap width.

[0015] Furthermore, the physical analysis unit processes the vibration spectrum using a convolutional neural network, which is trained using a preset fault feature library.

[0016] The vibration spectrum is output to the trained convolutional neural network, which then outputs the probability value of normal vibration. and abnormal vibration probability value And calculate the first risk value, expressed as: in, Indicates the first risk value; Indicates the abnormal vibration score; Indicates a positive number; when the probability value of normal vibration is greater than the probability value of abnormal vibration, the first risk value of the elevator is 0, and the elevator is in normal vibration mode; when the probability value of abnormal vibration is greater than the probability value of normal vibration, the first risk value is greater than 0, and the elevator is in abnormal vibration mode.

[0017] Furthermore, the voiceprint analysis unit analyzes the elevator's operating speed. Calculate the fundamental frequency , is represented as: in, Indicates the spacing between guide rail joints;

[0018] The voiceprint analysis unit extracts the fundamental frequency using a peak detection algorithm. The amplitudes of the harmonic components at integer multiples of frequency constitute the harmonic component vector of the audio waveform. And calculate the harmonic ratio, expressed as: in, Indicates the harmonic ratio; Indicates the fundamental frequency amplitude; Indicates the first The amplitude of each harmonic component;

[0019] The guide rail wear index is calculated using the harmonic ratio and the elevator's cumulative operating mileage, and is expressed as follows: in, Indicates the guide rail wear index; Represents the regression coefficient; Indicates reference mileage;

[0020] A wear threshold is set. If the guide rail wear index is not greater than the wear threshold, there is no second risk value. If the guide rail wear index is greater than the wear threshold, it is determined that abnormal guide rail wear has occurred, and a second risk value is calculated, expressed as: in, Indicates the second risk value; A score indicating the severity of guide rail wear; This indicates the wear threshold.

[0021] Furthermore, the maximum value of the door gap width is the maximum value among all sampling points on the door gap width variation curve;

[0022] The standard deviation of the door gap width is a statistical measure of the dispersion of all data on the door gap width variation curve, expressed as: in, Indicates the standard deviation of the door gap width; Indicates the first The door gap width value corresponding to each sampling point; Indicates the number of sampling points; This represents the average width of the door gap;

[0023] Combining the maximum door gap width and the standard deviation of the door gap width, the third risk value is calculated and expressed as: in, Indicates the third risk value; These represent the weighting coefficients for the maximum door gap width and the standard deviation of the door gap width, respectively. These represent reference values ​​for the maximum door gap width and the standard deviation of the door gap width, respectively.

[0024] A threshold for abnormal door gaps is set. When the third risk value is greater than the preset threshold for abnormal door gaps, it is determined that an abnormal door gap width has occurred.

[0025] Furthermore, the three-level response mechanism includes a first-level response mechanism, a second-level response mechanism, and a third-level response mechanism.

[0026] The first-level response mechanism is implemented when the first risk value is greater than 0, or when a second risk value exists.

[0027] The second-level response mechanism is implemented when the first risk value is greater than 0 and a second risk value exists.

[0028] Furthermore, the third-level response mechanism is the highest level of security response, and its implementation condition is that the third risk value is greater than the preset abnormal threshold for door gaps.

[0029] Compared with the prior art, the advantages of the present invention are as follows:

[0030] 1. This invention employs a movable sensor array that dynamically deploys based on the elevator's operating and stopping states. In operation, the sensor array automatically converges to the top of the car to form a dense detection network, accurately capturing vibration and acoustic signals. In stopping, the sensor array disperses and moves to the landing door area to monitor the door gap width in real time. This achieves an intelligent monitoring mode that focuses on vibration and acoustic patterns during operation and on the door system during stopping, significantly improving the targeting of data acquisition and the ability to capture fault characteristics.

[0031] 2. This invention constructs three independent risk assessment dimensions: a physical analysis unit, an acoustic analysis unit, and a door system analysis unit. These dimensions output three risk values: abnormal vibration, abnormal guide rail wear, and abnormal door gap width, respectively. This overcomes the limitations of the single-dimensional detection in existing technologies, comprehensively reflects the overall health status of the elevator system, and effectively improves the accuracy and comprehensiveness of fault identification.

[0032] 3. This invention designs a differentiated three-level response mechanism, which overcomes the crude mode of fixed threshold binary judgment in existing technologies. It realizes the implementation of precise response strategies based on the type and severity of anomalies, which not only avoids over-response caused by minor anomalies, but also ensures that faults that endanger life safety are controlled in the strictest way, significantly improving the practicality and reliability of the system. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a flowchart of the intelligent detection process of the present invention;

[0035] Figure 2 This is a flowchart of the dynamic sensing module of the present invention;

[0036] Figure 3 This is a flowchart of the intelligent analysis module of the present invention. Detailed Implementation

[0037] To achieve the above objectives, the present invention provides an intelligent detection system adapted for the safe operation of elevators. Please refer to the following technical solution. Figures 1-3 The system includes:

[0038] M1: Dynamic Sensing Module. A movable sensor array is deployed within the elevator shaft. This array dynamically adjusts based on the elevator's status and collects raw elevator data. After processing, it generates vibration spectrum, acoustic signature matrix, and door gap width variation curves.

[0039] This module deploys a movable sensor array within the elevator shaft. This sensor array includes several sensor nodes, each integrating a triaxial vibration sensor, an omnidirectional microphone, and a radar sensor.

[0040] In this embodiment, the triaxial vibration sensor uses a MEMS accelerometer with a measurement range of ±16g and a sampling frequency of 2000Hz to capture vibration signals during elevator operation; the omnidirectional microphone has a frequency response range of 20Hz to 20kHz and a sampling rate of 44.1kHz to collect acoustic signals during elevator operation; the radar sensor operates at a frequency of 77GHz, has a detection distance of 0.5 meters, and a distance resolution of 0.1 millimeters to monitor the door gap width between the landing door and the car door.

[0041] The movable sensor array is dynamically deployed based on the elevator's status, which is divided into running status and stopped status.

[0042] In this embodiment, when the elevator enters the running state, the movable sensor array receives the elevator start signal and moves along the magnetic guide rail of the shaft via a built-in electric slide rail mechanism, converging to the monitoring area on the top of the elevator car. A dedicated sensor docking platform is arranged on the top of the car, and the movable sensor array is fixed to the platform by magnetic adsorption, forming a dense vibration and acoustic joint detection network. At this time, the spatial distribution spacing of the sensor nodes in the movable sensor array is reduced to 0.1 meters, which can accurately capture the local vibration signals and acoustic signals of the car during operation.

[0043] During operation, a triaxial vibration sensor acquires three-dimensional vibration signals, which are then transmitted to edge computing nodes deployed on the shaft wall. These edge computing nodes employ embedded processors. The edge computing nodes first perform noise reduction on the three-dimensional vibration signals using wavelet transform technology. Next, they generate a vibration spectrum using a fast Fourier transform on the denoised three-dimensional vibration signals, and then normalize the vibration spectrum, mapping the amplitude to... interval,

[0044] During operation, the omnidirectional microphone synchronously records acoustic signals. The acoustic signals are first processed by wavelet noise reduction. The same wavelet transform method as the vibration signal is used to obtain the noise-reduced acoustic signal. In order to extract voiceprint features, the noise-reduced acoustic signal is converted into a time-frequency domain representation using short-time Fourier transform. The voiceprint feature matrix is ​​constructed based on the time-frequency domain representation. The rows of the matrix correspond to the time axis, the columns correspond to the frequency axis, and the values ​​of the matrix elements represent the energy intensity of the frequency at that moment.

[0045] When the elevator enters a stopped state, that is, when the elevator stops at a certain floor, the movable sensor array moves to the corresponding floor's door area via magnetic guide rails based on the car's position information. In the door area, radar sensors monitor the width of the gap between the door and the elevator door in real time. The radar measures the door gap width by emitting electromagnetic waves and receiving reflected signals, using the principle of frequency-modulated continuous wave. The radar sensor continuously measures the door gap width at a frequency of 100Hz, forming a door gap width change curve. Under normal circumstances, the door gap width should be kept within a safe threshold.

[0046] M2: Intelligent Analysis Module, including a physical analysis unit, a sound signature analysis unit, and a door system analysis unit. The physical analysis unit identifies abnormal vibration modes and outputs a first risk value by comparing the vibration spectrum with a preset fault feature library. The sound signature analysis unit analyzes the sound signature feature matrix to obtain sound signature harmonic components and combines this with the elevator's cumulative operating mileage to obtain a second risk value. The door system analysis unit obtains the maximum door gap width and the standard deviation of the door gap width through the door gap width variation curve, and further outputs a third risk value.

[0047] The physical analysis unit first uses a convolutional neural network to perform pattern recognition on the vibration spectrum. The convolutional neural network can automatically extract multi-level features from the vibration spectrum, enabling accurate identification of abnormal vibration patterns.

[0048] In this embodiment, the convolutional neural network includes three convolutional layers, two pooling layers, two fully connected layers, and a softmax classification layer. The network input is a vibration spectrum map containing 2048 frequency points, ranging from 0 to 1000 Hz. Through layer-by-layer feature extraction by the three convolutional layers, the network can identify local pattern features in the vibration spectrum map, such as peak values ​​at specific frequencies, the shape of the spectral envelope, and the combination relationship of multiple frequency components, generating feature maps. The pooling layers reduce the dimensionality of the feature maps while enhancing the translation invariance of the features, making the network robust to small frequency shifts. The fully connected layers map the extracted features to two categories: normal vibration and abnormal vibration.

[0049] The convolutional neural network is trained using a pre-set fault feature library. In this embodiment, the feature library contains 10,000 labeled samples, including all historical normal and abnormal vibration samples. The samples are derived from the historical operating data of the actual elevator and laboratory simulation data. The training process adopts a supervised learning approach, using the cross-entropy loss function to measure the difference between the prediction results and the true labels, and continuously optimizing the network parameters through the backpropagation algorithm.

[0050] The vibration spectrum is input into the trained convolutional neural network, and finally, the normal vibration probability value is output through the Softmax classification layer. and abnormal vibration probability value The first risk value is further calculated and expressed as: in, To represent extremely small positive numbers, preventing the denominator from being 0; The abnormal vibration score is set according to the actual situation, so that the output of the first risk value is a score value. In this embodiment, the abnormal vibration score is set to 5. When the normal vibration probability value is greater than the abnormal vibration probability value, the max function outputs 0, and the elevator's first risk value is 0, indicating that the elevator is in normal vibration mode. Only when the abnormal vibration probability value is greater than the normal vibration probability value, that is, when the elevator is in abnormal vibration mode, will the first risk value be greater than 0.

[0051] The function of the acoustic signature analysis unit is to analyze the acoustic signature feature matrix, extract the acoustic signature harmonic components, calculate the guide rail wear index based on the elevator's cumulative operating mileage, and output a second risk value.

[0052] During elevator operation, the friction between the guide rails and guide shoes generates periodic acoustic signals. The frequency characteristics of these signals are closely related to the elevator's speed and the condition of the guide rail surface. When the guide rail surface is smooth, the acoustic signals generated by friction are mainly concentrated on the fundamental frequency and its lower harmonic components. When the guide rail surface is worn, uneven, or locally damaged, the spectral composition of the friction noise becomes richer, and the energy of the higher harmonic components is significantly enhanced.

[0053] The voiceprint analysis unit first determines the elevator's operating speed. Calculate the fundamental frequency , is represented as: in, This indicates the spacing between the guide rail joints; in the acoustic signature feature matrix, besides the energy peak at the fundamental frequency, there are also harmonic components at integer multiples of the fundamental frequency. The acoustic signature analysis unit extracts the preceding harmonic components using a peak detection algorithm. The amplitude of the th harmonic component, specifically, for the th One harmonic component, in The maximum energy value in the range search is used as the first The amplitude of each harmonic component, For frequency tolerance, it is set to 5% of the fundamental frequency to accommodate small fluctuations in speed;

[0054] Before extraction The amplitudes of each harmonic component constitute the harmonic component vector of the acoustic signature. To quantify the degree of guide rail wear, the acoustic signature analysis unit calculates the ratio of higher-order harmonic amplitudes to the fundamental frequency amplitude to obtain the harmonic ratio. , is represented as: in, Indicates the fundamental frequency amplitude; Indicates the first The harmonic ratio reflects the energy proportion of high-frequency components relative to the fundamental frequency. When the guide rail surface is flat, the harmonic ratio is small, usually below 0.2. When the guide rail wear intensifies, the harmonic ratio increases significantly, reaching 0.8 or even higher. The harmonic ratio provides an intuitive acoustic characteristic parameter for guide rail condition assessment.

[0055] However, harmonic ratio alone is insufficient to accurately predict guide rail wear, as guide rail wear is a long-term cumulative process closely related to the elevator's mileage. For example, a newly installed elevator may have a slightly higher harmonic ratio due to temporary factors during the installation and commissioning phase, and does not necessarily indicate that the guide rails are already worn. Conversely, an elevator that has been in operation for many years may have a slightly lower harmonic ratio, but due to accumulated wear, it may already be close to the replacement standard. Therefore, the acoustic analysis unit introduces the elevator's cumulative mileage as an auxiliary parameter. The cumulative mileage is obtained through an odometer and is expressed in kilometers, reflecting the elevator's total operating volume.

[0056] Through harmonic ratio and cumulative operating mileage Based on the calculation of the guide rail wear index, it is expressed as: in, Indicates the guide rail wear index; Represents the regression coefficient; The reference operating mileage is set to 100 kilometers. The formula indicates that the guide rail wear index is jointly determined by acoustic characteristics (harmonic ratio) and usage history (cumulative operating mileage). The regression coefficients are determined through statistical analysis of a large amount of historical data. In this embodiment, a total of 500 sets of historical data were collected. Each set of data includes harmonic ratio, cumulative operating mileage, and actual measured guide rail surface roughness (measured by a surface roughness meter, in micrometers). The least squares method is used to perform multiple linear regression analysis to obtain the regression coefficients.

[0057] The voiceprint analysis unit compares the guide rail wear index with a preset wear threshold. The wear threshold is the criterion for judging abnormal guide rail wear and is determined according to the guide rail's technical specifications and safety requirements. In this embodiment, the wear threshold is preset to 50. When the guide rail wear index is greater than 50, an abnormal guide rail wear is determined to have occurred, and a second risk value is calculated. When the guide rail wear index is not greater than 50, the second risk value is not output.

[0058] The second risk value is calculated based on the guide rail wear index. The second risk value is expressed as a nonlinear function to reflect the nonlinear impact of wear on risk, as follows: in, The severity score of guide rail wear is set according to the actual situation, so that the output of the second risk value is a score value. In this embodiment, the severity score of guide rail wear is set to 5. Indicates the wear threshold; when As the value gradually increases, the exponential term gradually decreases, while the second risk value gradually increases; when Much larger When the index term is close to 0, the second risk value is close to the severity score, indicating that the risk has reached an extremely high level. This non-linear relationship is consistent with the actual law of the impact of guide rail wear on elevator safety, that is, mild wear has little impact on safety, but when the wear reaches a certain level, the safety risk will rise sharply.

[0059] The function of the door system analysis unit is to analyze the door gap width variation curve, extract the maximum door gap width and the standard deviation of the door gap width, comprehensively assess the safety status of the door system, and output a third risk value.

[0060] The door system is one of the most critical safety components of an elevator. The width of the door gap directly affects passenger safety. Under normal circumstances, when the elevator stops at a floor, the door gap width should remain within a safe threshold. If the door gap width is too large, it may indicate problems such as door guide rail deformation, door roller wear, or insufficient door operator torque. This can have a fatal impact on the subsequent safe operation of the elevator. Therefore, the door system analysis unit prioritizes door gap width as the highest priority monitoring indicator.

[0061] The maximum door gap width is defined as the maximum value among all sampling points on the door gap width variation curve. In this embodiment, according to the elevator safety technical specifications, the door gap width safety threshold is set to 3 mm.

[0062] The standard deviation of door gap width is a statistic describing the dispersion of data. It measures the fluctuation range of the door gap width variation curve and is expressed as: in, Indicates the standard deviation of the door gap width; Indicates the first The door gap width value corresponding to each sampling point; Indicates the number of sampling points; The standard deviation of the door gap width represents the mean value of the door gap width. It reflects the stability of the door system operation. When there is a mechanical failure in the door system, it will cause the elevator door to shake, stop or repeat. The door gap width variation curve will show irregular fluctuations and the standard deviation will increase significantly. Therefore, the standard deviation of the door gap width is an important indicator for assessing the health status of the door system.

[0063] After extracting two key feature parameters, the gate system analysis unit comprehensively calculates the third risk value using a weighted summation method. The calculation needs to consider both the absolute size and the fluctuation range of the door gap width, expressed as: in, These represent reference values ​​for the maximum door gap width and the standard deviation of the door gap width, respectively, used for normalization processing; These represent the weighting coefficients for the maximum door gap width and the standard deviation of the door gap width, respectively. In this embodiment, the weighting coefficients are determined based on the severity analysis of door system failures and statistical analysis of a large amount of historical data. The allocation of these weighting coefficients reflects the contribution of characteristic parameters to safety risks. The maximum door gap width accounts for a larger proportion than the standard deviation of the door gap width. These reference values ​​are determined based on the technical specifications and safety standards of the elevator door system.

[0064] The door system analysis unit compares the third risk value with the preset door gap anomaly threshold. When the third risk value is greater than the preset door gap anomaly threshold, it determines that an abnormal door gap width has occurred.

[0065] The tiered execution module implements a three-tiered response mechanism based on the first, second, and third risk values.

[0066] This module serves as the execution terminal of the entire intelligent detection system. Based on the output of the intelligent analysis module, it implements a differentiated three-level response mechanism. The core of the three-level response mechanism is to implement a graded and classified response strategy according to the type, severity, and scope of safety impact of the anomaly, so as to ensure the safe operation of the elevator and avoid operational efficiency loss caused by over-response.

[0067] The first-level response mechanism targets single-dimensional anomalies. Specifically, it is triggered when the first risk value is greater than 0, or a second risk value exists but the two do not occur simultaneously, and the third risk value does not exceed the door gap anomaly threshold, and no door gap width anomaly occurs; in this case, it is classified as a first-level risk state.

[0068] From a risk perspective, both abnormal vibration and abnormal guide rail wear are progressive mechanical failures that will not immediately cause direct harm to passengers. However, if ignored for a long time, they may gradually evolve into serious malfunctions. Therefore, the strategy for a Level 1 response focuses on strengthening early warning and monitoring.

[0069] When a Level 1 response is triggered, this module first records the anomaly information to the system database, including detailed information such as the time of occurrence, anomaly type, risk value, and elevator operating status, forming a complete fault tracing chain. Simultaneously, it automatically generates an early warning report, which is pushed in real-time to the maintenance management platform and the mobile terminals of property management personnel via IoT communication. The early warning report includes an anomaly description, preliminary cause analysis, and suggested inspection items.

[0070] To closely monitor abnormal development trends, this module adjusts the operating mode of the dynamic sensing module, increasing the data acquisition frequency to several times the normal level. In this embodiment, the acquisition cycle of the triaxial vibration sensor and omnidirectional microphone is once every 30 minutes under normal conditions, and is shortened to once every 15 minutes after triggering the first-level response.

[0071] In Level 1 response mode, the elevator continues to operate normally without limiting its speed or the number of floors served, in order to maximize operational efficiency. The tiered execution module continuously monitors the risk value data collected subsequently. If the risk value is normal for three consecutive collection cycles, the Level 1 response mode is automatically lifted, and normal monitoring mode is restored. If the risk value continues to exceed the threshold or shows an upward trend, the system maintains Level 1 response mode and prompts property management personnel to arrange professional technicians for on-site inspection and maintenance as soon as possible.

[0072] The second-level response mechanism addresses situations involving multiple overlapping anomalies. Specifically, it is triggered when the first risk value is greater than 0, a second risk value exists (i.e., vibration and guide rail wear occur simultaneously), and the third risk value does not exceed the door gap anomaly threshold, meaning no door gap width anomaly occurs; in this case, the situation is classified as a second-level risk state.

[0073] From the perspective of failure mechanism analysis, the simultaneous occurrence of abnormal vibration and abnormal guide rail wear often indicates a relatively serious complex fault in the elevator mechanical system. The superposition of these two types of abnormalities will significantly enhance their destructive effects. Therefore, the focus of the second-level response mechanism is to reduce operational risks and accelerate maintenance response.

[0074] When the second-level response mechanism is triggered, this module first sends a speed-limiting command to the elevator control system and increases the data acquisition frequency. In this embodiment, if the elevator's rated speed is 2.5 meters per second, the maximum operating speed after speed limiting is 1.5 meters per second. The speed-limiting measure can effectively reduce the dynamic load and vibration amplitude of the mechanical system, slow down the rate of fault deterioration, and provide passengers with a smoother riding experience.

[0075] Simultaneously, the tiered execution module generates a level-two risk alarm, which simultaneously issues warnings in the elevator car and machine room via audible and visual alarms, reminding passengers and maintenance personnel to pay attention to the elevator's status and automatically sending a high-priority maintenance work order to the maintenance management platform.

[0076] The tiered execution module continuously evaluates the effectiveness of the second-level response mechanism. If either the first or second risk value exceeds the preset severe anomaly threshold after execution, it automatically escalates to the third-level response mechanism and implements more stringent security measures.

[0077] The Level 3 response mechanism is the highest level of security response, with the highest trigger priority. Specifically, it is triggered when the third risk value exceeds the abnormal door gap threshold, resulting in an abnormal door gap width. Regardless of whether the first and second risk values ​​exist and are greater than 0, the situation is immediately classified as a Level 3 risk state, and the emergency response procedure is initiated.

[0078] The priority design of the third-level response mechanism is based on the unique dangers of door system malfunctions. Abnormal door gap width is directly related to passenger safety, and accidents often occur suddenly and rapidly, leaving extremely short response time windows. Therefore, abnormal door gap width must be considered the most urgent safety threat, triggering the most stringent protective measures.

[0079] When the Level 3 response mechanism is triggered, this module immediately sends an emergency stop command to the elevator control system. If the elevator is already in operation, the safety stop procedure is initiated. First, the elevator decelerates at normal speed and smoothly stops at the nearest floor. During the stopping process, if an abnormal increase in door gap width or other danger signals are detected, the safety brake or other emergency stop mechanism is immediately activated. If the elevator is already stopped, the control system prevents the elevator doors from closing and restarting, keeping the doors open.

[0080] Following an emergency elevator stop, the tiered execution module issues a three-level risk alert through multiple channels. Specifically, inside the elevator car, a safety notice is broadcast to passengers via the emergency intercom, informing them that the elevator is temporarily out of service for safety reasons and requesting them to remain calm and follow instructions to evacuate safely. Outside the elevator, a malfunction warning is issued via LED display and an audible and visual alarm to prevent other personnel from continuing to use the elevator. At the remote monitoring terminal, emergency alarm information is simultaneously sent to the maintenance management platform, property management center, and emergency management department, activating the emergency response plan.

[0081] In a Level 3 response state, the elevator is completely shut down and all operation is prohibited. Only after professional technicians have completed a comprehensive inspection, confirmed that the door system fault has been completely eliminated, the door gap width has been restored to a safe range, and multiple no-load and load tests have been conducted to verify that the Level 3 response state has been lifted and the elevator can resume normal operation.

[0082] This invention proposes an intelligent detection system adapted to the safe operation of elevators. Through the collaborative work of a dynamic sensing module, an intelligent analysis module, and a hierarchical execution module, it achieves a complete closed loop from data acquisition and risk assessment to response execution. The dynamic sensing module performs differentiated deployments based on the elevator's operating and stopping status; the intelligent analysis module constructs three independent risk assessment dimensions; and the hierarchical execution module designs a differentiated three-level response mechanism, implementing early warning monitoring, speed limit maintenance, and emergency elevator stop for single anomalies, compound anomalies, and door gap anomalies, respectively, ensuring the accuracy and effectiveness of the response strategy.

[0083] In summary, this invention overcomes the shortcomings of existing technologies, such as fixed sensor positions, single-dimensional detection, and fixed threshold judgment, by organically combining a dynamic deployment strategy of movable sensor arrays, a multi-dimensional independent risk assessment system, and a differentiated three-level response mechanism. It achieves comprehensive perception, accurate assessment, and intelligent response of elevator safety status, significantly improving the active protection capability of elevator safety operation and the practicality and reliability of the system, and providing an innovative technical solution for intelligent elevator monitoring.

[0084] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An intelligent detection system adapted for the safe operation of elevators, characterized in that, The system includes: M1: Dynamic sensing module, which deploys a movable sensor array in the elevator shaft. The movable sensor array is dynamically deployed according to the elevator status to obtain the elevator's vibration spectrum, acoustic signature matrix and door gap width change curve. M2: Intelligent analysis module, including a physical analysis unit, an acoustic signature analysis unit, and a door system analysis unit. The physical analysis unit identifies vibration modes and outputs a first risk value by comparing the vibration spectrum diagram with a preset fault feature library. The acoustic signature analysis unit analyzes the acoustic signature feature matrix and obtains a second risk value by combining it with the elevator's cumulative operating mileage. The door system analysis unit obtains the maximum door gap width and the standard deviation of the door gap width by analyzing the door gap width variation curve, and further outputs a third risk value. M3: Tiered execution module, which implements a three-level response mechanism based on the first risk value, the second risk value, and the third risk value; The elevator status is divided into running status and stopped status; in the running status, the movable sensor array converges to the top of the elevator car; in the stopped status, the movable sensor array disperses and moves to the landing door area of ​​the floor. The first risk value detects the vibration mode; the second risk value determines abnormal wear of the guide rail; the third risk value determines abnormal door gap width. The movable sensor array includes a triaxial vibration sensor, an omnidirectional microphone, and a radar sensor. In the operating state, the triaxial vibration sensor collects the three-dimensional vibration signal of the elevator and transmits it to the edge computing node. The edge computing node obtains the vibration spectrum map through wavelet transform and fast Fourier transform, and at the same time performs normalization processing on the vibration spectrum map. The omnidirectional microphone records the acoustic signal of the elevator, and further obtains the voiceprint feature matrix through wavelet transform and short-time Fourier transform; In the stopped state, the radar sensor emits electromagnetic waves and receives reflected signals, and uses the frequency-modulated continuous wave principle to measure the door gap width between the elevator door and the landing door, forming a door gap width variation curve; The physical analysis unit processes the vibration spectrum using a convolutional neural network, which is trained using a pre-defined fault feature library. The vibration spectrum is output to the trained convolutional neural network, which then outputs the probability value of normal vibration. and abnormal vibration probability value And calculate the first risk value, expressed as: ； in, Indicates the first risk value; Indicates the abnormal vibration score; Represents a positive number; when the probability value of normal vibration is greater than the probability value of abnormal vibration, the first risk value of the elevator is 0, and the elevator is in normal vibration mode; when the probability value of abnormal vibration is greater than the probability value of normal vibration, the first risk value is greater than 0, and the elevator is in abnormal vibration mode. The voiceprint analysis unit analyzes the elevator's operating speed. Calculate the fundamental frequency , is represented as: ； in, Indicates the spacing between guide rail joints; The voiceprint analysis unit extracts the fundamental frequency using a peak detection algorithm. The amplitudes of the harmonic components at integer multiples of frequency constitute the harmonic component vector of the audio waveform. And calculate the harmonic ratio, expressed as: ； in, Indicates the harmonic ratio; Indicates the fundamental frequency amplitude; Indicates the first The amplitude of each harmonic component; The guide rail wear index is calculated using the harmonic ratio and the elevator's cumulative operating mileage, and is expressed as follows: ； in, Indicates the guide rail wear index; , and Represents the regression coefficient; Indicates reference mileage; A wear threshold is set. If the guide rail wear index is not greater than the wear threshold, there is no second risk value. If the guide rail wear index is greater than the wear threshold, it is determined that abnormal guide rail wear has occurred, and a second risk value is calculated, expressed as: ; in, Indicates the second risk value; A score indicating the severity of guide rail wear; Indicates the wear threshold; The maximum value of the door gap width is the maximum value among all sampling points on the door gap width variation curve; The standard deviation of the door gap width is a statistical measure of the dispersion of all data on the door gap width variation curve, expressed as: ； in, Indicates the standard deviation of the door gap width; Indicates the first The door gap width value corresponding to each sampling point; Indicates the number of sampling points; This represents the average width of the door gap; Combining the maximum door gap width and the standard deviation of the door gap width, the third risk value is calculated and expressed as: ； in, Indicates the third risk value; and These represent the weighting coefficients for the maximum door gap width and the standard deviation of the door gap width, respectively. and These represent reference values ​​for the maximum door gap width and the standard deviation of the door gap width, respectively. A threshold for abnormal door gaps is set. When the third risk value is greater than the preset threshold for abnormal door gaps, it is determined that an abnormal door gap width has occurred.

2. The intelligent detection system adapted for elevator safety operation according to claim 1, characterized in that, The three-level response mechanism includes a first-level response mechanism, a second-level response mechanism, and a third-level response mechanism. The first-level response mechanism is implemented when the first risk value is greater than 0, or when a second risk value exists. The second-level response mechanism is implemented when the first risk value is greater than 0 and a second risk value exists.

3. The intelligent detection system adapted for elevator safety operation according to claim 2, characterized in that, The third-level response mechanism is the highest level of security response, and its implementation condition is that the third risk value is greater than the preset abnormal threshold for door gaps.

Images