Intelligent detection method and system for elevator braking performance

Abstract

The present application belongs to the technical field of escalator braking performance detection, and specifically discloses an elevator braking performance intelligent detection method and system, which constructs a parameterized simulation model based on the mechanical topology and dynamic characteristics of an escalator. First, the initial calibration of the model is completed by using the no-load braking test. Then, a gradient simulation load scheme is designed in combination with the actual running load distribution. The model is further calibrated through physical test data under multiple loads. Finally, the high-fidelity calibration model is used to realize accurate prediction of the full-load braking performance. This method not only eliminates the need for full-load weight measurement, significantly improving detection safety and efficiency, but also effectively overcomes the model mismatch caused by factors such as equipment aging and wear through the data-driven iterative calibration mechanism, ensuring long-term prediction accuracy and providing reliable technical support for the normalization and intelligent evaluation of braking performance.

Description

Technical Field

[0001] This invention belongs to the field of escalator braking performance testing technology, and specifically discloses an intelligent testing method and system for elevator braking performance. Background Technology

[0002] As a transportation device operating frequently and with high passenger flow in public places, the performance of the braking system of escalators is directly related to public safety. To ensure safety, national standards mandate regular testing of the full-load braking performance of escalators to verify their stopping capability under the most unfavorable operating conditions.

[0003] Traditional full-load braking performance testing generally relies on stacking physical weights on the steps to simulate full-load conditions. This method has a series of drawbacks, including cumbersome operation, the need to interrupt normal equipment operation, high safety risks associated with handling heavy objects at heights, and low testing efficiency. As a result, such tests can often only be carried out in a discrete, long-term manner, making it impossible to achieve routine and trend-based monitoring of braking performance.

[0004] To overcome the aforementioned shortcomings, existing technologies have proposed some detection methods that do not require full-load weights. For example, Chinese invention patent CN112320550B discloses a method for measuring the stopping distance of an escalator. This method measures multiple parameters, including the escalator's unloaded running speed, unloaded running power, and the average deceleration when stopping under both unloaded and light-load conditions. These parameters are then substituted into a fixed conversion formula derived from dynamic theory to calculate the predicted stopping distance under full-load conditions.

[0005] While this method avoids the use of full-load weights, significantly improving the convenience and efficiency of testing, its calculation of full-load braking relies on a pre-derived fixed physical formula. The parameters in this formula are treated as constant values, failing to account for dynamic characteristic drift caused by factors such as component wear, lubrication degradation, and brake pad aging during long-term service. Therefore, as the equipment's operating time accumulates, the deviation between the model and the actual system may continue to widen, making it difficult to guarantee the accuracy of the full-load braking performance prediction results and limiting its long-term applicability. Summary of the Invention

[0006] Therefore, one objective of this application is to provide an intelligent detection method and system for elevator braking performance. By constructing a parameterized simulation model of escalator braking and performing advanced calibration on it using physical test data under multiple operating conditions, the calibrated high-fidelity model is finally used to predict full-load braking, effectively solving the problems existing in the prior art.

[0007] The objective of this invention can be achieved through the following technical solution: The first aspect of this invention proposes an intelligent detection method for elevator braking performance, comprising the following steps: S1, constructing a parameterized simulation model of the escalator braking system, and calibrating the initial parameters of the simulation model based on a single no-load physical braking test data.

[0008] S2. Extract load distribution based on historical operating data of escalators, and design a gradient simulation load scheme covering the load range from no load to the upper limit of safe load.

[0009] S3. According to the gradient simulation load scheme, apply each simulated load sequentially on the escalator.

[0010] S4. Trigger a braking command after each application of simulated load, synchronously collect multi-signal time data of the entire braking process, identify key feature points and decouple the braking process stages, and extract braking performance characteristics in each braking stage.

[0011] S5. Using the braking performance characteristics extracted from multiple gradient load points at each braking stage, the initially calibrated parametric simulation model is further calibrated.

[0012] S6. Input the standard full-load conditions into the parameterized simulation model after advanced calibration to perform simulation, obtain the predicted value of full-load braking performance, and make a braking safety judgment based on it.

[0013] The second aspect of the present invention proposes an intelligent detection system for elevator braking performance, comprising the following modules: a simulation calibration module: constructing a parameterized simulation model of the escalator braking system, and calibrating the initial parameters of the simulation model based on a single no-load physical braking test.

[0014] Test planning module: Extract load distribution based on historical operation data of escalators, and design a gradient simulation load scheme covering from no load to the upper limit of safe load.

[0015] Simulation Implementation Module: Based on the gradient simulation load scheme, simulated loads are applied sequentially on the escalator. After each simulated load is applied, a braking command is triggered. Simultaneously, multi-dimensional signal time-series data including motor three-phase current, brake action displacement, and main drive shaft speed are collected throughout the braking process. Key feature points are identified and the braking process stages are decoupled. At the same time, braking performance characteristics are extracted in each braking stage.

[0016] Advanced calibration module: Utilizes braking performance characteristics extracted from multiple gradient load points at each braking stage to perform advanced calibration on the initially calibrated parametric simulation model.

[0017] Simulation analysis module: Input the standard full-load conditions into the parameterized simulation model after advanced calibration to perform simulation, obtain the predicted value of full-load braking performance, and make a braking safety judgment based on it.

[0018] Combining all the above technical solutions, the positive effects of this invention are as follows: 1. After constructing a parameterized simulation model of the escalator, this invention first completes the initial calibration of the model using an unloaded braking test. Then, it designs a gradient simulation load scheme based on the actual operating load distribution, and performs advanced calibration of the model using physical test data under multiple load levels. Finally, it achieves accurate prediction of full-load braking performance based on the calibrated model. This not only eliminates the need for actual full-load weight testing, significantly improving detection safety and efficiency, but also effectively overcomes model mismatch problems caused by factors such as equipment aging and wear through data-driven iterative calibration, ensuring long-term prediction accuracy.

[0019] 2. When using physical test data under multi-level loads to perform advanced calibration of the model, the data collected in this invention is not limited to final state indicators such as total braking time and total distance. Furthermore, by decoupling the braking process in stages, performance characteristics with clear physical meaning are extracted at each stage, which significantly enhances the richness and discriminative power of the calibration information. This expands the data on which the model calibration is based from a single final state result to a multi-dimensional full-process feature set, which is beneficial to improving the model fidelity and prediction accuracy. Attached Figure Description

[0020] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0021] Figure 1 This is a diagram illustrating the implementation steps of an intelligent detection method for elevator braking performance according to the present invention.

[0022] Figure 2 This is a flowchart illustrating the implementation of S1 in this invention.

[0023] Figure 3 This is a module connection diagram of an intelligent detection system for elevator braking performance according to the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Example 1

[0026] See Figure 1 As shown, the present invention proposes an intelligent detection method for elevator braking performance, including the following steps: S1, constructing a parameterized simulation model of the escalator braking system, and calibrating the initial parameters of the simulation model based on a single no-load physical braking test data.

[0027] Given that escalators are electromechanical systems, their braking performance is affected by various time-varying and nonlinear factors, and full-load physical tests cannot be frequently conducted in actual operation, it is necessary to construct a calibrable and evolvable parametric simulation model to achieve accurate mapping of braking behavior and performance prediction.

[0028] See Figure 2 As shown, in a specific implementation of this invention, the parametric simulation model is constructed as follows: based on the mechanical topology and dynamic characteristics of the escalator, a parametric simulation model of the escalator braking system is established. This model is essentially a digital twin of the escalator braking system.

[0029] Specifically, modeling begins with understanding the mechanical topology of the escalator, specifically the connections and relationships between components such as the drive chain, step chain, handrail belt, brakes, and tensioning devices. This structure determines how forces are transmitted within the system and how the components move in tandem. Only by realistically recreating these connections can the simulation accurately reflect the forces and motion states of the actual equipment during braking, avoiding distortion of simulation results due to oversimplification of the structure.

[0030] Meanwhile, the braking process involves various physical behaviors such as speed reduction, friction, and inertial effects. Therefore, the model also needs to incorporate basic dynamic laws and describe these behaviors through adjustable fundamental physical parameters such as the system's equivalent moment of inertia and friction coefficient. In this way, the model can not only simulate the entire process from issuing the braking command to complete stop, but also reproduce key signals such as speed changes, motor current, and brake displacement.

[0031] Compared to methods that rely solely on empirical formulas or pure data fitting, this mechanism-based modeling approach has internal parameters with clear physical meanings. These parameters can be calibrated and corrected using measured data, thereby continuously improving model accuracy and effectively avoiding prediction deviations caused by equipment aging, wear, or changes in operating conditions.

[0032] Therefore, the mechanical topology of escalators provides geometric and connection constraints in the construction of parametric simulation models, while the dynamic characteristics provide the evolution laws of motion and force. Together, they form the basis of parametric simulation models.

[0033] Although the parametric simulation model of the escalator braking system is based on mechanical structure and dynamics principles, the parameters used in the model can often only be estimated based on drawings or experience during the design phase, making it difficult to accurately reflect the actual physical state of a specific escalator.

[0034] Therefore, after the model is constructed, this invention uses data from a single no-load physical braking test to calibrate the model's initial parameters. This step is equivalent to adjusting a general template into a digital twin specific to this escalator. The no-load test was chosen because the escalator has no passenger load at this time, the system is under the simplest stress, and the collected speed, current, displacement, and other signals have less interference, which can truly reflect the dynamic behavior of the drive system, brake, and transmission mechanism itself, making it suitable as a benchmark for initial calibration.

[0035] Only by ensuring that the simulation results of the model under no-load conditions are basically consistent with the measured data can we ensure that the basic parameters in the model are within a reasonable range, providing a reliable starting point for more refined calibration under different simulated loads, and avoiding the failure of the optimization process due to excessive initial deviation or deviation from the actual situation.

[0036] Applying the above explanation, the initial parameter calibration process is as follows: a sensor network is deployed on the output shaft of the drive motor, the brake actuator, and the transmission nodes of the escalator.

[0037] When the escalator is in an unloaded state, a braking command is triggered, and the timing data of motor current, brake displacement and spindle speed reflecting the braking process are collected synchronously using a sensor network.

[0038] In a preferred embodiment of the above scheme, the sensor network includes a speed sensor, a displacement sensor, and a current sensor.

[0039] A speed sensor is mounted on the output shaft of the drive motor to record the speed change of the main drive shaft from its operating speed to complete stop. Since the braking effect is most directly reflected in the deceleration process, this data can directly reflect the braking time, the trend of deceleration change, and the final stopping time. It is mainly used to calibrate parameters such as the system's equivalent moment of inertia and friction coefficient in the model.

[0040] A displacement sensor is installed on the brake's actuator to monitor displacement changes during brake actuation. Because it takes time for the brake to actually press against the brake disc after receiving a command, its action involves a delay and a gradual pressure build-up process. This signal allows for accurate determination of when the brake begins to actuate, the speed of the actuation, and whether it has reached its target position, thereby calibrating parameters in the model related to the brake's mechanical response characteristics, such as actuation delay and stroke resistance.

[0041] A current sensor is installed on the power supply circuit of the drive motor, which consists of transmission nodes, to collect changes in the three-phase current of the motor. When a braking command is issued, the control system cuts off the power supply or applies a reverse / energy-dissipating braking current, causing a significant change in current. This signal reflects the response speed and braking intervention method of the electronic control system and is used to calibrate relevant parameters such as electrical response delay and motor braking characteristics in the model.

[0042] Through the coordinated operation of these three types of sensors, information on all aspects of the braking process, including electronic control, mechanics, and motion, can be comprehensively acquired, providing reliable and multi-dimensional data support for the initial calibration of the simulation model.

[0043] The measured curves were plotted using the time-series data collected during the no-load braking test as a calibration reference, and braking commands were input into the simulation model to drive the simulation model to run.

[0044] By adjusting the basic physical parameters in the model, the simulation output curve of the no-load braking process is made to match the measured curve of the no-load braking process with the preset degree of agreement, thus completing the initial calibration of the model. The basic physical parameters include at least the system's moment of inertia and friction coefficient.

[0045] The no-load braking process curve output by the above-mentioned model simulation refers to a set of physical quantity curves that the simulation model calculates and outputs over time after receiving the braking command. These curves mainly include the main drive shaft speed-time curve, the motor current-time curve, and the brake displacement-time curve, which are exactly the same as the measured curves collected in the no-load braking test.

[0046] The specific calibration process is as follows: First, the time-series curves output by the parameterized simulation model under no-load braking conditions are aligned point by point with the measured curves synchronously collected in the no-load physical test, and the fitting error of each curve, such as the root mean square error, is calculated.

[0047] Subsequently, corresponding basic physical parameters to be calibrated are assigned to different curves. For example, the rotational speed curve is mainly associated with the equivalent rotational inertia and friction coefficient of the system, the displacement curve is associated with the brake action delay and damping characteristics, and the current curve is associated with the electrical response time constant. The corresponding parameters are iteratively adjusted based on the error amount. For example, when the simulation deceleration is too fast, the rotational inertia is appropriately increased or the friction coefficient is decreased.

[0048] Repeat the above simulation, comparison and adjustment process until the errors of all curves converge to the preset calibration tolerance range.

[0049] The calibration tolerance range error can be referenced from the allowable deviation requirements for performance indicators in the relevant safety standards for escalator braking performance.

[0050] Finally, the parameter values ​​are locked at this point, and the parameterized simulation model has now been personalized as a digital twin of the escalator.

[0051] S2. Extract load distribution based on historical operating data of escalators, and design a gradient simulation load scheme covering the load range from no load to the upper limit of safe load.

[0052] After constructing a digital twin of the escalator with initial parameter calibration, the initial calibration only ensures the model's accuracy under no-load conditions, while the escalator's braking performance is highly dependent on the load. To ensure the model maintains high predictive accuracy under full-load conditions, it is necessary to perform advanced calibration through multi-level simulated load tests, enabling it to grasp the mapping relationship between load and performance.

[0053] Based on the above, the multi-level simulated load is not set subjectively or uniformly, but is designed based on the load distribution characteristics actually observed in the historical operation data of escalators. This ensures that the selected simulated load can reflect the real usage scenario, thereby improving the information effectiveness of the calibration data.

[0054] Optionally, the gradient simulation load scheme is designed as follows: extract the load time series from the historical monitoring data of escalator operation.

[0055] It's important to understand that during escalator operation, passengers standing on the steps exert an additional load torque on the drive system due to their own weight. The greater the load, the greater the downward gravitational force the escalator drive system needs to overcome, thus directly affecting braking performance.

[0056] Since the distribution, number, and dwell time of passengers on escalators change dynamically over time, weight sensing units can be installed on the escalator steps to accurately obtain the actual operating load and sense the local load borne by each step in real time.

[0057] By synchronously collecting signals from all sensing units and summing the instantaneous load values ​​on all steps at each moment, the equivalent total load acting on the whole machine at that moment can be obtained.

[0058] The equivalent total load is recorded continuously in chronological order, thus forming a load time series that reflects the actual usage intensity of the escalator.

[0059] Histograms were used to construct a load probability distribution map reflecting the actual operating conditions from the extracted load time series.

[0060] Understandably, histograms clearly show the frequency distribution of load values ​​by dividing the data into several intervals and counting the number of data points in each interval.

[0061] The load probability distribution map is divided into several load intervals through cluster analysis.

[0062] According to the safety regulations for the use of escalators, the upper limit of the safe load for escalators shall be determined.

[0063] It should be noted that the upper limit of the safe load of an escalator refers to the maximum weight that the escalator can safely bear, and this value does not refer to the load when the escalator is fully loaded.

[0064] Between the no-load and the upper limit of the safe load, a number of discrete load points are selected proportionally based on the area of ​​each load interval relative to the overall distribution.

[0065] In the load probability distribution, the width of each interval on the horizontal axis represents the span of its load range, and the height reflects the frequency of occurrence of a unit load segment within that interval. The area formed by the product of the two represents the total frequency of load occurrence within that interval, i.e., its statistical weight in historical operating data.

[0066] Given that larger sections account for a higher proportion in actual operation and are more representative of the daily performance of escalator braking, more load points should be allocated in the gradient simulation load scheme; while smaller sections, although occurring less frequently, may involve safety-critical operating conditions, and appropriate sampling should also be retained to ensure coverage integrity.

[0067] In practice, the total number of discrete load points to be selected is first determined, then the load points are allocated proportionally according to the area ratio of each cluster interval, and the load points are evenly distributed within their corresponding load width range.

[0068] In a specific selection example, assuming a total of 6 sampling points, three load intervals A, B, and C are obtained through clustering, with their area proportions being 50%, 30%, and 20%, respectively: interval A is assigned 3 load points; interval B is assigned 2 load points; and interval C is assigned 1 load point.

[0069] The specific values ​​of each load point are determined by uniform distribution within the load range of its respective interval.

[0070] The selected discrete load points are arranged in ascending order to form a gradient simulation load test scheme covering the entire working condition range.

[0071] The gradient simulation load test scheme determined based on the actual operating load distribution of escalators can ensure that the selected load points take into account both operational representativeness and working condition coverage.

[0072] S3. According to the gradient simulation load scheme, apply each simulated load sequentially on the escalator.

[0073] To determine the gradient simulation load scheme, loads need to be applied to the escalator to obtain measured braking data under different loads before further calibration of the parametric simulation model can be performed.

[0074] As an optional implementation of the above steps, the test mode is automatically activated when the escalator is not in operation.

[0075] Because braking performance testing requires actively triggering emergency braking, the escalator will suddenly decelerate or stop during the process. If conducted during operating hours, this could easily lead to passengers losing their balance, falling, or even causing a stampede. Therefore, testing must be conducted during non-operating hours.

[0076] According to the designed gradient simulation load scheme, gradient loads are applied to the escalator drive system in sequence through a load simulation device that is non-contactly coupled with the escalator drive system.

[0077] The aforementioned load simulation device can accurately output a constant equivalent torque according to a preset gradient. Specifically, it can be a magnetic powder brake or an eddy current brake, installed on the non-load-side extension shaft of the escalator drive unit or on a dedicated testing fixture. The excitation current is adjusted by a controller to accurately simulate different load torques, flexibly reproducing any operating condition from no-load to the upper limit of the safe load. Compared to the traditional method of stacking physical weights on the steps, this non-contact loading method eliminates the need for physical counterweights, fundamentally avoiding safety risks such as heavy object handling and falls from heights.

[0078] It should be noted that although the load simulation device has the ability to simulate full-load conditions, the equivalent torque it applies will actually trigger the mechanical braking process of the escalator. If such physical braking tests are frequently performed under full-load conditions, it will inevitably accelerate the aging of the brake pads and transmission chain, which will not only shorten the service life of the equipment, but may also introduce unnecessary maintenance costs.

[0079] Therefore, this invention selects only a few gradient load points within the upper limit of the safe load range for a limited number of real braking tests on the load simulation device. The obtained multi-condition measured data is used to perform advanced calibration of the parametric simulation model. Based on this, the braking performance under full load conditions no longer depends on physical testing, but is predicted through simulation using the calibrated model.

[0080] This method avoids repeated high-stress impacts on the core braking components of escalators, effectively protecting the health of the equipment. On the other hand, it provides a feasible path for subsequent long-term, high-frequency braking performance monitoring.

[0081] S4. Trigger a braking command after each simulated load is applied, and synchronously collect multi-signal timing data of the entire braking process. The multi-signal timing data includes the three-phase current of the motor, the displacement of the brake action, and the speed signal of the main drive shaft. Key feature points are identified and the braking process stages are decoupled. At the same time, braking performance characteristics are extracted in each braking stage.

[0082] After applying a simulated load and triggering a braking command, the braking process of the escalator is not a single transient behavior, but rather goes through multiple dynamic stages in sequence. Each stage has its own braking performance. In order to fully characterize the braking performance, it is necessary to decouple the entire braking process into stages and extract physically meaningful performance indicators in each sub-stage.

[0083] To achieve accurate stage division, it is necessary to rely on multi-source signals generated during the braking process, such as the three-phase current of the motor, the displacement of the brake action, and the speed signal of the main drive shaft, to identify the key feature points that mark the start and end of each stage.

[0084] As one way to achieve the above scheme, the key feature point identification process is as follows: the moment when the braking command is issued is taken as the start time of the electrical response.

[0085] Understandably, the braking command is an external triggering event for the entire braking process, with a clear timestamp and determinism. It is defined as the start time of the electrical response, providing a synchronous time origin for all subsequent feature points and ensuring the timing alignment of multiple signals.

[0086] The three-phase current signals of the motor acquired synchronously are processed by first-order differentiation to obtain the current change rate signal.

[0087] The first local peak point is captured in the current rate of change signal, and the time corresponding to this peak is defined as the electrical response completion time.

[0088] Understandably, after the braking command is issued, the inverter or contactor actuates, causing a sudden change in the motor's power supply state and triggering a rapid change in current. This change is not instantaneous but involves a brief electronic control response process. By performing first-order differential processing on the current signal, its rate of change can be amplified, highlighting the steep edge at the start of the response. The first local peak point can be identified in the current rate of change signal, corresponding to the moment when the current change is most drastic, i.e., the critical point where the electronic control system completes the switching and braking energy begins to be injected.

[0089] The first derivative of the brake displacement signal is used to obtain the brake speed signal. The mechanical braking start time is determined by detecting the first peak moment in the speed signal.

[0090] Understandably, after receiving an electrical signal, the brake needs to overcome the spring preload, mechanical clearance, and frictional resistance to generate effective braking force, which results in a mechanical delay. By taking the first derivative of the brake displacement signal, the speed is obtained. When the brake begins to move, the speed will rise rapidly from zero. The first speed peak corresponds to the moment when the brake actuator breaks through static friction and begins to effectively press the brake disc, marking the official start of pure mechanical friction braking.

[0091] The main drive shaft speed signal is continuously monitored for its numerical decay process. The moment when the speed first drops to zero and remains at zero without rising again is defined as the moment of complete stop.

[0092] Understandably, under the action of frictional torque, the spindle speed continuously decays to zero. However, due to inertial rebound, sensor noise, or slight vibration, the speed signal may oscillate slightly near zero. When the speed remains at zero and no longer rises, it means that the system's kinetic energy has been completely dissipated, and the braking process has completely ended.

[0093] In a further feasible approach, decoupling the braking process phases using identified key feature points includes the following process: defining the time period from the start of the electrical response to the completion of the electrical response as the electrical response phase.

[0094] The period from the completion of the electrical response to the start of mechanical braking is defined as the mechanical pressure build-up phase.

[0095] The period from the start of mechanical braking to the moment of complete stopping is defined as the pure friction braking phase.

[0096] In a further feasible approach, the braking performance characteristics of each braking phase are extracted as follows: the duration of the electrical response phase is used as an electrical response delay index to reflect the response speed of the control system and drive unit. The longer the duration, the more delayed the braking initiation.

[0097] The ratio of the brake's displacement during the mechanical pressure build-up phase to the phase duration is used as the brake's average pressure build-up rate, which characterizes the brake's mechanical execution efficiency. The lower the average pressure reduction rate, the slower the pressure build-up, resulting in a delay in the establishment of braking force and insufficient deceleration.

[0098] During the pure friction braking phase, the initial velocity and phase duration are extracted from the timing data of the main drive shaft speed signal during this phase to calculate the average deceleration.

[0099] Specifically, since the main drive shaft speed monotonically decreases from its initial value to zero during this stage, the average deceleration can be obtained as the ratio of the initial velocity to the stage duration. The decrease in average deceleration usually means that the coefficient of friction has decreased due to factors such as wear, contamination, or aging, which will lead to a reduction in actual braking capability and thus weaken the safety margin of the braking system.

[0100] The main drive shaft speed signal is integrated over time during this stage to obtain the main shaft rotation angle, which is then multiplied by the step pitch circle radius to convert it into a linear running distance, i.e., pure friction braking distance. This represents the distance the escalator travels during the actual application of effective braking force, eliminating the invalid sliding during the electrical and mechanical delay stages. It is a pure indicator for evaluating the performance of the friction system.

[0101] The total braking time is obtained by summing the durations of the electrical response phase, the mechanical pressure build-up phase, and the pure friction braking phase, which comprehensively reflects the time consumed from issuing the braking command to complete stopping.

[0102] The total braking distance is calculated by integrating the spindle speed signal and multiplying it by the step pitch circle radius during the entire braking process. This is the most critical safety compliance indicator, directly corresponding to the mandatory requirement in national standards that the braking distance must be between the minimum and maximum values. If the braking distance is too short, it will cause excessive deceleration, leading to passenger imbalance; if the braking distance is too long, it will cause the risk of brake failure, resulting in serious safety accidents.

[0103] The extracted braking performance characteristics not only reveal the functional states of each braking stage and achieve a multi-dimensional characterization of the braking process, but also comprehensively reflect the temporal response capability of the overall braking system. These quantitative indicators, combining stage-by-stage and global approaches, provide highly identifiable multi-source constraint data for the advanced calibration of subsequent parametric simulation models.

[0104] S5. Using the braking performance characteristics extracted from multiple gradient load points at each braking stage, the initially calibrated parametric simulation model is further calibrated.

[0105] Given the large amount of braking performance data under multi-gradient loads, performing simulation calibration point by point sequentially would be inefficient. Furthermore, the parameter sensitivity of escalator braking systems varies under different loads. In low-load or high-load regions, the system response tends to be stable, and parameter changes have little impact on the results; however, in medium-load or transitional regions, braking performance is more sensitive to parameter changes, and the load-performance relationship fluctuates significantly. Therefore, prioritizing the identification and focusing on these highly sensitive load ranges as the initial calibration targets can reduce the number of simulations while achieving more efficient parameter correction and model optimization.

[0106] Based on the above considerations, the advanced calibration process is as follows: the braking performance characteristics extracted at each gradient load point are plotted with the load level as the abscissa and the corresponding braking performance as the ordinate, and the load performance variation curve is drawn.

[0107] Numerical differentiation is performed on each load performance curve. By detecting the abrupt change in the sign of the first derivative, the inflection point in the original curve is identified. Thus, the load interval between adjacent inflection points is defined as the fluctuating load interval.

[0108] The intersection of all the fluctuating load ranges corresponding to the braking performance characteristics is calculated to obtain the load range that the braking performance is commonly sensitive to, which is defined as the first-round optimization range.

[0109] Perform a union operation on the fluctuating load intervals corresponding to all braking performance characteristics, and subtract the result from the intersection operation to obtain the secondary sensitive load interval, which is used as the optimization object in the second round.

[0110] The remaining gradient load points not included in the union are then used as the objects for the third round of supplementary optimization.

[0111] Based on the above three-stage optimization sequence, the corresponding gradient load points are sequentially input into the parameterized simulation model that has completed initial calibration, the simulation outputs of various braking performance characteristics are obtained, and compared with the measured values ​​to calculate the error.

[0112] For example, the error can be a relative error. Specifically, for each load point, multiple braking performance errors are calculated, forming an error vector.

[0113] Based on the error, the basic physical parameters in the initially calibrated parameterized simulation model are optimized collaboratively through parameter optimization.

[0114] In the embodiments of the above scheme, the collaborative optimization process is as follows: First, for multiple load points in a round of optimization, the error vectors under each load point are synthesized into a comprehensive error objective function such as root mean square error.

[0115] Then, starting from the initial calibrated basic physical parameters, numerical optimization algorithms such as least squares method are used to automatically search for the parameter combination that minimizes the comprehensive error objective function.

[0116] Specifically, if the simulated braking distance is too long under a certain load, it indicates insufficient friction, and the friction coefficient should be increased.

[0117] If the pressure build-up time is too short, the mechanical response may be too fast. Adjust the brake damping or clearance parameters.

[0118] All parameters are adjusted synchronously to ensure that the overall error of multiple indicators decreases, rather than overfitting a single indicator.

[0119] By co-optimizing, the model is forced to approximate the real behavior under all performance indicators and load conditions, thereby obtaining a physically consistent and generalizable parameter set.

[0120] When the error between the simulated and measured values ​​of various braking performance characteristics under all gradient load points converges to the preset calibration tolerance range after a certain round of optimization, the advanced calibration of the model is determined to be completed, and the current parameter set is locked as the final calibrated model parameters.

[0121] The calibration tolerance range mentioned above still refers to the allowable deviation requirements for braking performance in the relevant safety standards for escalator braking performance.

[0122] It should be noted that although the parameterized simulation model after initial calibration can reproduce the no-load braking behavior well, when different loads are introduced for simulation, there may still be a large deviation from the measured data. Therefore, further calibration is required.

[0123] Since the calibration process is divided into multiple rounds of optimization based on load sensitivity, the first round focuses on the load range that is most sensitive to the parameters. If, after a certain round of optimization, the errors between the simulated and measured values ​​of various braking performance indicators under all gradient load points have met the preset tolerance requirements, then there is no need to continue the optimization in subsequent rounds, so as to improve the calibration efficiency.

[0124] S6. Input the standard full-load conditions into the parameterized simulation model after advanced calibration to perform simulation, obtain the predicted value of full-load braking performance, and make a braking safety judgment based on it.

[0125] The specific implementation is as follows: Input the standard full-load conditions into the parameterized simulation model after advanced calibration to obtain the predicted braking distance under this working condition.

[0126] The above-mentioned standard full-load conditions are based on the most unfavorable braking conditions formed by the rated load, rated speed, and running direction, which are usually downward, as specified in the current national standards.

[0127] The predicted braking distance will be compared with the full-load braking distance limit specified in the current escalator safety standards.

[0128] If the predicted braking distance exceeds the limit for braking distance under full load, the braking performance is deemed substandard.

[0129] It should be added that, under full load conditions, the current safety standards only use braking distance as the sole mandatory compliance indicator for whether braking performance meets the standards. Although other performance parameters are used for process analysis, they do not directly determine the result of whether the performance is qualified. Therefore, in this invention, the braking safety of full load braking simulation is determined only by braking distance.

[0130] Example 2

[0131] See Figure 3 As shown, this invention proposes an intelligent detection system for elevator braking performance, including the following modules: Simulation calibration module: constructs a parameterized simulation model of the escalator braking system, and performs initial parameter calibration of the simulation model based on a single no-load physical braking test data.

[0132] Test planning module: Extract load distribution based on historical operation data of escalators, and design a gradient simulation load scheme covering from no load to the upper limit of safe load.

[0133] The simulation implementation module is connected to the test planning module: based on the gradient simulation load scheme, each simulated load is applied sequentially on the escalator. After each simulated load is applied, a braking command is triggered. Simultaneously, multi-dimensional signal time-series data including the three-phase current of the motor, the displacement of the brake action, and the speed of the main drive shaft are collected throughout the braking process. Key feature points are identified and the braking process stages are decoupled. At the same time, braking performance characteristics are extracted in each braking stage.

[0134] The advanced calibration module is connected to the simulation implementation module and the simulation calibration module respectively: it uses the braking performance characteristics extracted from multiple gradient load points at each braking stage to perform advanced calibration on the initially calibrated parametric simulation model.

[0135] The simulation analysis module is connected to the advanced calibration module: the standard full-load conditions are input into the parameterized simulation model after advanced calibration for simulation, the predicted value of full-load braking performance is obtained, and the braking safety is determined accordingly.

[0136] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0137] Those skilled in the art will recognize that the algorithmic steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0138] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0139] 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.

[0140] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent detection of elevator braking performance, characterized in that, Includes the following steps: S1. Construct a parameterized simulation model of the escalator braking system, and calibrate the initial parameters of the simulation model based on a single no-load physical braking test. Based on the mechanical topology and dynamic characteristics of escalators, a parametric simulation model of the escalator braking system is established. A sensor network is deployed on the output shaft of the escalator's drive motor, the brake actuator, and the transmission nodes. When the escalator is in an unloaded state, a braking command is triggered, and the sensor network synchronously collects time-series data reflecting the braking process, including motor current, brake displacement, and spindle speed. The time-series data collected during the unloaded braking test is used to plot a measured curve as a calibration reference, and a braking command is input to the simulation model to drive its operation. By adjusting the basic physical parameters in the model, the errors between the simulated unloaded braking process curve and the measured unloaded braking process curve are brought within a preset calibration tolerance range, thus completing the initial calibration of the model. The basic physical parameters include at least the system's equivalent moment of inertia and coefficient of friction. S2. Extract load distribution based on historical operating data of escalators, and design a gradient simulation load scheme covering the load from no load to the upper limit of safe load. S3. According to the gradient simulation load scheme, apply each simulated load sequentially on the escalator; S4. Trigger a braking command after each simulated load is applied, synchronously collect multi-signal time data of the entire braking process, identify key feature points and decouple the braking process stages, and extract braking performance characteristics in each braking stage. The multi-signal timing data collected during braking include the three-phase current of the motor, the brake displacement, and the main drive shaft speed signal. The moment the braking command is issued is taken as the start moment of the electrical response. The synchronously collected three-phase current signal of the motor is processed by first-order differentiation to obtain the current change rate signal. The first local peak point in the current change rate signal is captured, and the moment corresponding to the peak point is defined as the completion moment of the electrical response. The brake displacement signal is processed by first-order differentiation to obtain the brake speed signal. The moment of the first peak in the speed signal is determined as the start moment of mechanical braking. The main drive shaft speed signal is continuously monitored for its numerical decay process. The moment when the speed first drops to zero and remains at zero without rising again is defined as the moment of complete stopping. S5. Using the braking performance characteristics extracted from multiple gradient load points at each braking stage, perform advanced calibration on the initially calibrated parametric simulation model. S6. Input the standard full-load conditions into the parameterized simulation model after advanced calibration to perform simulation, obtain the predicted value of full-load braking performance, and make a braking safety judgment based on it.

2. The intelligent detection method for elevator braking performance as described in claim 1, characterized in that: S2 is implemented as follows: Extract load time series from historical monitoring data of escalator operation; Histograms were used to construct a load probability distribution map reflecting the actual operating conditions from the extracted load time series. The load probability distribution map is divided into several load intervals through cluster analysis; According to the safety regulations for the use of escalators, the upper limit of the safe load for escalators shall be determined. Between the no-load and the upper limit of the safe load, a number of discrete load points are selected proportionally based on the area of ​​each load interval relative to the overall distribution. Arrange the selected discrete load points in ascending order to form a gradient simulation load scheme that covers the entire operating condition range.

3. The intelligent detection method for elevator braking performance as described in claim 2, characterized in that: S3 includes the following: The test mode is automatically activated when the escalator is not in operation. According to the designed gradient simulation load scheme, gradient loads are applied to the escalator drive system in sequence through a load simulation device that is non-contactly coupled with the escalator drive system.

4. The intelligent detection method for elevator braking performance as described in claim 3, characterized in that: The decoupling of the braking process phase includes the following processes: The time period from the start time of the electrical response to the completion time of the electrical response is defined as the electrical response phase. The period from the completion of the electrical response to the start of mechanical braking is defined as the mechanical pressure build-up phase. The period from the start of mechanical braking to the moment of complete stopping is defined as the pure friction braking phase.

5. The intelligent detection method for elevator braking performance as described in claim 4, characterized in that: The braking performance characterization extracted during each braking stage includes the following: The duration of the electrical response phase is used as an indicator of electrical response delay. The ratio of the brake's displacement during the mechanical pressure build-up phase to the phase duration is taken as the average pressure build-up rate of the brake. During the pure friction braking phase, the initial velocity and phase duration are extracted from the timing data of the main drive shaft speed signal during this phase to calculate the average deceleration. The main drive shaft speed signal is integrated over time during this stage to obtain the main shaft rotation angle, which is then multiplied by the step pitch circle radius to convert it into a linear running distance, i.e., pure friction braking distance. The total braking time is obtained by summing the durations of the electrical response phase, the mechanical pressure build-up phase, and the pure friction braking phase. The total braking distance is obtained by integrating the spindle speed signal and converting it throughout the braking process.

6. The intelligent detection method for elevator braking performance as described in claim 1, characterized in that: S5 includes the following: The braking performance extracted at each gradient load point is characterized by plotting the load level as the abscissa and the corresponding braking performance as the ordinate, and the load performance variation curve is plotted. Numerical differentiation is performed on each load performance curve. By detecting the abrupt change in the sign of the first derivative, the inflection point in the original curve is identified. Thus, the load interval between adjacent inflection points is defined as the fluctuating load interval. The intersection of all the fluctuating load ranges corresponding to the braking performance characteristics is calculated to obtain the load range that the braking performance is commonly sensitive to, which is defined as the first-round optimization range. Perform a union operation on all the fluctuating load intervals corresponding to the braking performance characteristics, and subtract the result from the intersection operation to obtain the secondary sensitive load interval, which is used as the object of the second round of optimization. The remaining gradient load points not included in the union are then used as the objects for the third round of supplementary optimization. Based on the above three-stage optimization sequence, the corresponding gradient load points are sequentially input into the parameterized simulation model that has completed initial calibration, the simulation outputs of various braking performance characteristics are obtained, and compared with the measured values ​​to calculate the error. Based on the error, the basic physical parameters in the initially calibrated parameterized simulation model are optimized collaboratively through parameter optimization. When the error between the simulated and measured values ​​of various braking performance characteristics under all gradient load points converges to the preset calibration tolerance range after a certain round of optimization, the advanced calibration of the model is determined to be completed, and the current parameter set is locked as the final calibrated model parameters.

7. The intelligent detection method for elevator braking performance as described in claim 1, characterized in that: S6 includes the following: By inputting the standard full-load conditions into the parameterized simulation model after advanced calibration, the predicted braking distance under this working condition can be obtained. The predicted braking distance will be compared with the full-load braking distance limit specified in the current escalator safety standards. If the predicted braking distance exceeds the limit for braking distance under full load, the braking performance is deemed substandard.

8. An intelligent elevator braking performance detection system, used to execute the steps of the intelligent elevator braking performance detection method according to any one of claims 1-7, characterized in that, include: Simulation calibration module: Constructs a parameterized simulation model of the escalator braking system and performs initial parameter calibration of the simulation model based on a single no-load physical braking test. Test planning module: Extract load distribution based on historical operation data of escalators, and design a gradient simulation load scheme covering from no load to the upper limit of safe load. Simulation Implementation Module: Based on the gradient simulation load scheme, simulated loads are applied sequentially on the escalator. After each simulated load is applied, a braking command is triggered. Simultaneously, multi-signal time data including motor three-phase current, brake action displacement, and main drive shaft speed are collected throughout the braking process. Key feature points are identified and braking process stages are decoupled. Braking performance characteristics are extracted in each braking stage. Advanced calibration module: Utilizes braking performance characteristics extracted from multiple gradient load points at each braking stage to perform advanced calibration on the initially calibrated parametric simulation model; Simulation analysis module: Input the standard full-load conditions into the parameterized simulation model after advanced calibration to perform simulation, obtain the predicted value of full-load braking performance, and make a braking safety judgment based on it.

Images