Repeater motion attenuation coupling simulation system for elevator shaft

Abstract

The invention discloses an elevator shaft-oriented repeater motion attenuation coupling simulation system, particularly relates to the field of ultra-dense networks, and is used for solving the problems of propagation path dynamic change, isolation coupling fluctuation and door area transient oscillation noise raising black field caused by on-way operation of a lift car in a long and narrow metal boundary space of an elevator shaft. The method comprises the following steps: establishing an on-way position index and a door area window output unified time reference through a motion label generation unit; the coupling margin mapping unit is used for calculating a positional coupling curve and the isolation margin estimation to form a mapping relation; a gain threshold simultaneous unit carries out simultaneous solution to generate a position gain strategy and vibration suppression threshold strategy output parameter table; a simulation error iteration unit plays back a strategy to execute time domain simulation record error iteration optimization; the curing risk output unit stabilizes the curing parameter set and outputs a risk map and an abnormal trigger point; dynamic boundary prediction and closed-loop optimization are realized, and coverage continuity is ensured.

Description

Technical Field

[0001] This invention relates to the field of ultra-dense networks, and more specifically, to a simulation system for motion attenuation coupling of repeaters in elevator shafts. Background Technology

[0002] Elevator shafts are long, narrow, metal-bounded spaces, often containing heterogeneous nodes such as repeaters, distributed antennas, and floor-level clusters, forming an ultra-dense network topology within a single building. Donor antennas and service antennas are located in the shaft and car, respectively. The car's movement introduces strong motion scattering and waveguide effects. Door opening and closing, guide rails, and accessories trigger boundary condition switching, rapidly altering the propagation path and energy distribution with position and orientation. Co-frequency reuse and near-far-end concurrency cause inter-link co-frequency interference and coupling channels to fluctuate with the elevator's travel distance. Isolation margins change abruptly within a short window at the station and door zone, making coverage quality highly sensitive to door zone transients, prone to drops, fallbacks, and blackouts.

[0003] Existing repeater links, under the constraints of ultra-dense networks, lack position-attitude driven prediction and feedforward control, and do not integrate shaft boundary states and gain control in time-domain simulations. They also lack joint modeling of donor-serving antenna coupling chains, co-channel interference maps, and gate window conditions, making it difficult to generate along-path coupling margin mappings and parameter tables for deployment calibration and threshold calculation. Furthermore, the operations and maintenance side lacks precise location and contingency plans for high-risk sections and windows. These gaps make it difficult to maintain coverage continuity and stability throughout the entire process, and oscillations and noise spikes are more easily triggered under conditions of coexisting ultra-dense nodes and co-channel reuse. To address these issues, corresponding technical solutions are proposed. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a motion attenuation coupling simulation system for repeaters in elevator shafts. A motion tag generation unit establishes a path position index and door zone window based on arrival signals and door status events, and outputs motion-driven tags as a unified time reference. A coupling margin mapping unit reads the motion-driven tags, calculates the positional coupling curve and isolation margin estimate, and associates the estimation results with the path position index to form a mapping relationship. A gain threshold combination unit combines the gain control model and the isolation margin estimate to generate a positional gain strategy and a vibration suppression threshold strategy, and outputs a strategy parameter table for simulation use. A simulation error iteration unit performs time-domain simulation using the motion-driven tag playback strategy, records errors based on oscillation and noise verification results, and iteratively optimizes the coupling estimator and strategy parameters. A risk solidification output unit solidifies the strategy into a deployment parameter set when the iteration reaches stability and outputs a path risk map, a door zone-specific strategy, and anomaly trigger points for operation and maintenance, thus solving the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] Motion tag generation unit: Based on the arrival signal and door status event, establish the along-the-path position index and door area window, and output motion drive tags as a unified time reference;

[0007] Coupling margin mapping unit: Reads motion drive tags, calculates positional coupling curves and isolation margin estimates, and associates and stores the estimation results with the along-line position index to form a mapping relationship;

[0008] Gain threshold combined unit: Combines the gain control model with the isolation margin estimate to generate a positional gain strategy and a vibration damping threshold strategy, and outputs a strategy parameter table for simulation use;

[0009] Simulation error iteration unit: The motion-driven tag playback strategy is used to perform time-domain simulation. Based on the verification results of oscillation and noise rise, the error is recorded and the coupled estimator and strategy parameters are iteratively optimized.

[0010] Solidified risk output unit: When the iteration reaches stability, the strategy is solidified into a set of deployment parameters, and the risk map along the process, the gate-specific strategy, and the abnormal trigger points for operation and maintenance are output.

[0011] Furthermore, the motion tag generation unit collects the deceleration threshold of the car speed and the floor encoder position feedback to generate the arrival signal pulse sequence, records the start time and duration, and records the binary sequence of door status events through the door position sensor switch feedback, including the status transition timestamp and transition type.

[0012] Furthermore, the motion tag generation unit uses the Kalman filtering algorithm to smooth the pulse sequence timestamps and align them with the nominal velocity curve time to calculate the cumulative displacement. It then superimposes the gate state sequence onto the corresponding pulse displacement to form a position correction and generates an associative array sorted by displacement values ​​as a position index along the path.

[0013] Furthermore, the coupling margin mapping unit parses the motion-driven label JSON sequence, loads the label item list, verifies the timestamp increment and displacement continuous coverage of the entire shaft length, and supplements the missing displacements through linear interpolation to form a memory buffer list sorted by displacement.

[0014] Furthermore, the coupling margin mapping unit traverses the label sequence list and uses the ray tracing algorithm to simulate the corresponding displacement electromagnetic coupling path. It defines the fixed position of the donor antenna at the top of the shaft and the position of the service antenna in the car. It uses image theory to process the multiple reflections of the metal wall, enumerates the direct and first-order reflection paths, and aggregates the coherent contributions to generate a set of positional coupling curve points.

[0015] Furthermore, the gain threshold simultaneous unit loading mapping relationship file is expanded into an associated array, the displacement value along the path position index is extracted and the composite object is verified to be monotonically increasing and the margin is non-empty, and the gap margin value is supplemented by cubic spline interpolation to form an ordered list that binds the floor identifier and window status.

[0016] Furthermore, the gain threshold simultaneous unit initialization gain control model cascade function is used. Taking the isolation margin estimate of the extended correlation array as the constraint input, the linear programming algorithm is used to minimize the coverage strength minus margin penalty objective function. The gain variables are iteratively adjusted to ensure that the link gain does not exceed the margin boundary, and an intermediate gain boundary sequence is generated.

[0017] Furthermore, the simulation error iteration unit parses the motion-driven label JSON sequence, loads the label item list to verify the timing integrity, infers missing states through forward padding, and imports the policy parameter table CSV file to match displacement values ​​to expand the label item policy fields, forming a simulation buffer association list that integrates label items and parameters.

[0018] Furthermore, the simulation error iteration unit replays the composite data buffer sequence in timestamp order, queries the corresponding displacement strategy parameters to dynamically set the gain base value and the lower limit of the vibration suppression threshold, initializes the link state vector, uses the Runge-Kutta method to generate a time-domain trajectory file using an integral differential equation system, and records the signal amplitude and margin observations.

[0019] Furthermore, the solidified risk output unit extracts the error evolution curve from the iterative archive file, calculates the rate of change of the most recent three rounds of oscillation error and noise rise error, uses first-order differential to aggregate the average slope to determine a stable state when it is lower than the convergence threshold, and generates a stable flag file that accompanies the final error curve and references the motion-driven label timestamp.

[0020] The technical effects and advantages of the present invention regarding the motion attenuation coupling simulation system for repeater stations in elevator shafts are as follows:

[0021] This invention transforms car arrival signals and door opening / closing events into unified motion-driven tags, serving as spatiotemporal coordinate points for the entire simulation process. Based on this, a positional coupling curve is constructed to depict the signal propagation path changes in the shaft, and isolation margins are calculated to assess safety margins, forming a precise displacement mapping relationship. Subsequently, by solving the gain model and margin simultaneously, a gain strategy and vibration suppression threshold parameter table that adjust with position are generated to avoid excessive signal feedback. Finally, tag playback drives time-domain simulation to capture oscillation and noise issues, iteratively optimizing parameters until stability, and outputting a fixed parameter set, a risk visualization diagram, and maintenance trigger points. This chain-like collaboration cleverly transforms dynamic boundary disturbances into predictable control signals, achieving pre-synchronous matching of gain and coupling. Brief fluctuations during door opening and closing no longer cause signal interruptions or noise surges, and the self-excited loop of upward interference is promptly blocked by the threshold mechanism, ensuring smooth signal transmission throughout the elevator's journey. The iterative closed-loop further transforms risks into intuitive charts and contingency plan prompts, allowing maintenance personnel to easily pinpoint potential hazards and make targeted adjustments. This overall mechanism simplifies the deployment calibration process and improves coverage reliability and policy reuse convenience. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the motion attenuation coupling simulation system for repeater stations in elevator shafts according to the present invention. Detailed Implementation

[0023] 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.

[0024] Example 1: Figure 1 This invention presents a motion attenuation coupling simulation system for repeaters in elevator shafts, comprising:

[0025] Motion tag generation unit: Based on the arrival signal and door status event, establish the along-the-path position index and door area window, and output motion-driven tags as a unified time reference.

[0026] Coupling margin mapping unit: Reads motion-driven tags, calculates positional coupling curves and isolation margin estimates, and associates and stores the estimation results with the along-the-path position index to form a mapping relationship.

[0027] Gain threshold combined unit: Combines the gain control model with the isolation margin estimate to generate a positional gain strategy and a vibration suppression threshold strategy, and outputs a strategy parameter table for simulation use.

[0028] Simulation error iteration unit: The motion-driven tag playback strategy is used to perform time-domain simulation. Based on the verification results of oscillation and noise, the error is recorded and the coupling estimator and strategy parameters are iteratively optimized.

[0029] Solidified risk output unit: When the iteration reaches stability, the strategy is solidified into a set of deployment parameters, and the risk map along the process, the gate-specific strategy, and the abnormal trigger points for operation and maintenance are output.

[0030] Within the narrow, elongated metal boundary space of an elevator shaft, the car's movement along its length causes continuous changes in the propagation path and energy distribution. In particular, the dynamic impact of door opening and closing and guide rail accessories on the equivalent boundary causes significant fluctuations in the isolation and coupling of the repeater, making coverage quality especially sensitive at the arrival and door zones, prone to dropout and fallback phenomena. In this environment, repeater link oscillations, noise rise, and blackouts stem not only from transient fluctuations in position and attitude but also from the lack of position- and attitude-driven prediction and feedforward mechanisms. This makes it difficult to generate risk curves and parameter tables along the route, affecting deployment calibration, threshold calculation, and the location and contingency planning of high-risk sections by the maintenance team. Therefore, establishing a unified time reference framework is crucial for repeaters covering all floors. It transforms the dynamic events of car movement into quantifiable positional benchmarks, providing reliable spatiotemporal anchors for subsequent coupling estimation and gain strategies, ensuring stable coverage continuity throughout the entire travel range.

[0031] However, existing repeater link strategies often neglect the real-time correlation between arrival signals and gate status events, failing to transform these events into precise indexes of along-path positions and definitions of gate windows. This results in the lack of motion-driven labels, making it difficult to achieve a unified reference benchmark for time-domain simulation and thus hindering the prediction and optimization of location-based risks. To address this, the motion label generation unit constructs along-path position indices and gate windows through systematic analysis of arrival signals and gate status events, generating motion-driven labels. These labels serve as a unified time reference for subsequent steps, laying the spatiotemporal foundation for the entire simulation process.

[0032] The specific processing logic of the motion tag generation unit:

[0033] The motion tag generation unit breaks down the processing into four sub-steps: first, it collects the arrival signal and gate status events; second, it establishes a position index along the route based on the collected data; third, it defines the gate area window accordingly; and finally, it outputs motion-driven tags.

[0034] Sub-step 1.1: Acquire arrival signals and gate status events.

[0035] The data acquisition process obtains real-time data from the car's sensor interface. Specifically, arrival signals are captured by monitoring the car's deceleration threshold and the floor encoder's position feedback. When the car speed drops below the preset deceleration threshold and the floor encoder matches the current floor identifier, a timestamped pulse sequence is generated. This sequence records the start time and duration of each arrival event; for example, the start time is the moment the car enters the deceleration phase, and the duration is the interval from deceleration to complete stop. Door status events are acquired through the on / off feedback of the door position sensor. When the door moves from the closed position to the open threshold or returns from the open threshold to the closed position, the corresponding binary state sequence is recorded, including the precise timestamp of the state transition and the transition type (open or closed). Both the acquired arrival signals and door status events are stored in an event log buffer with timestamps as keys, forming an ordered list of sequences. Each entry contains a timestamp, event type, and associated parameters (such as floor identifier or state type) to ensure that subsequent sub-steps can directly reference these sequences for position mapping. The buffer's capacity is set to cover the expected number of events for a single full journey to avoid overflow and interruption of data acquisition.

[0036] Sub-step 1.2: Establish a location index along the route based on the collected data.

[0037] This sub-step directly references the sequence list of arrival signals and gate status events collected in sub-step 1.1, and uses the Kalman filtering algorithm to smooth the timestamps in the sequence to eliminate the interference of sensor noise on position accuracy. Specifically, the pulse sequence of the arrival signal is first time-aligned with the car's nominal speed curve (a predefined ideal operating speed function): for the start time of each pulse, the cumulative displacement of the car from the hoistway starting point to that time is calculated using the formula... Where d k This represents the cumulative displacement of the k-th arrival event, where t0 is the shaft start time, and t k Let v(t) be the pulse start time, and v(t) be the nominal velocity function. The integral is approximated using the numerical trapezoidal rule to ensure that the displacement calculation is dimensionally consistent and conforms to kinematic principles. Then, the binary sequence of door status events is superimposed on the displacement of the corresponding arrival pulse to form a position correction: if the door status is open, the displacement offset is set to the positive door zone expansion distance (preset to half the car length), otherwise it is zero. This process generates a position index along the hoistway, an associative array sorted by displacement values, where the key is the event timestamp and the value is the corrected displacement and its floor identifier. This ensures that the index covers the entire hoistway length and seamlessly connects with the sequence list in sub-step 1.1, providing a positional reference for defining the door zone window.

[0038] Sub-step 1.3: Define the door zone window based on the location index along the route.

[0039] Based directly on the associative array of along-the-path position indices generated in sub-step 1.2, the Dynamic Time Warping (DTW) algorithm is used to match the position values ​​with the binary sequences of gate state events to identify the start and end boundaries of the gate window. Specifically, for each arrival event, the corrected displacement d... k Map the timestamp of the open state transition immediately following the gate state sequence to the starting position of the window: set the starting position to d. k -δ, where δ represents the deceleration buffer distance (preset as the average displacement during the car's deceleration phase); the ending position is the displacement corresponding to the closed state transition timestamp plus δ. The DTW algorithm is used to minimize the distorted path cost between the timestamp sequence and the displacement sequence, ensuring that the window boundary captures the transient expansion of door opening and closing. Each door zone window is recorded as a quadruple: starting position, ending position, associated floor identifier, and duration (ending position minus starting position divided by average car speed). These quadruples are appended to the associative array of the travel position index to form an extended index. This extension ensures that the door zone window is accurately embedded in the travel position index, avoiding position drift and providing a dual event-position anchor for the generation of motion-driven labels.

[0040] Sub-step 1.4: Output motion-driven labels.

[0041] The final sub-step references the extended along-the-path position index from sub-step 1.3 to generate motion-driven labels as a unified time reference. Specifically, all timestamps, displacement values, and door zone window quadruples in the extended index are integrated into a label sequence: each label item contains a timestamp, corresponding displacement, floor identifier, and window status (marked as "Door Zone Active" if within the window, otherwise "Stable Operation"). The sequence is arranged in ascending order of timestamps and serialized and stored as a motion-driven label file in JSON format. The time resolution of this file is set to an integer multiple of the acquisition interval to ensure that the label sequence covers the entire path and is replayable, used to drive the time-domain simulation of subsequent steps. After output, the motion-driven label file is immediately available for reading by the coupling margin mapping unit, realizing a closed loop from event acquisition to label generation.

[0042] Through the progressive advancement of the above sub-steps, the arrival signal and door status event are transformed into structured along-path position indexes and door zone windows, and finally condensed into motion-driven labels. This process provides an accurate spatiotemporal reference framework in the dynamic boundary changes of the narrow space of the elevator shaft. It not only suppresses the impact of position fluctuations on the repeater link, but also lays a reliable mapping foundation for the calculation of coupling curves, ensuring that the subsequent isolation margin estimation can be directly related to position changes, thereby improving the continuity and stability of coverage in the arrival and door zones.

[0043] The motion tag generation unit generates motion-driven tags as spatiotemporal anchors by parsing arrival signals and door status events. These tags accurately capture the dynamic correlation between the along-path position index and the door zone window, providing a replayable reference sequence for subsequent calculations. This enables positional analysis and avoids the neglect of transient events in traditional strategies. However, current repeater link simulations are often limited to static isolation evaluation, failing to transform the tag sequence of car motion into position-dependent coupling curves and margin estimates. This results in a lack of targeted gain control and an inability to predict along-path risk curves. To address this deficiency, the coupling margin mapping unit starts by reading motion-driven tags and progressively advances to the construction of coupling curves, the quantification of isolation margins, and finally the formation of a mapping relationship with the along-path position index. This process ensures accurate anchoring of the estimation results and injects position-driven constraints into the simultaneous solution of the gain threshold simultaneous unit, achieving predictive optimization of door zone sensitive coverage.

[0044] The specific processing logic of the coupling margin mapping unit:

[0045] The coupling margin mapping unit breaks down the processing into four sub-steps: first, reading the motion-driven labels; second, calculating the positional coupling curve; third, calculating the isolation margin estimate based on the curve; and finally, associating and storing the estimation results with the along-the-path position index to form a mapping relationship.

[0046] Sub-step 2.1: Read the motion-driven tags.

[0047] The read operation extracts the tag sequence from the motion-driven tag file output by the motion tag generation unit. Specifically, it parses the JSON-formatted file, loads each tag item line by line, including the timestamp, corresponding displacement, floor identifier, and window status (active or stable door zone), and verifies the integrity of the sequence: checking the monotonically increasing timestamps and the continuous coverage of the displacement values ​​across the entire shaft length; if missing items are detected, displacements are supplemented through linear interpolation to ensure the sequence length matches the preset full-stroke resolution. The read tag sequence is temporarily stored in a memory buffer, forming a list sorted by displacement, where each element is bound to its original timestamp and window status. The buffer is designed as an indexable array, facilitating direct traversal of displacement values ​​for coupled simulation in sub-step 2.2, avoiding the overhead of repeatedly parsing the file, thus maintaining the real-time performance of the calculation.

[0048] Sub-step 2.2: Calculate the positional coupling curve.

[0049] This sub-step iterates through the tag sequence list of sub-step 2.1, using a ray tracing algorithm to simulate the electromagnetic coupling path at each corresponding displacement. Specifically, for the displacement corresponding to each tag item in the sequence, the donor antenna is first defined as fixed at the top of the shaft, and the service antenna is positioned in the car. Image theory is used to handle multiple reflections from the metal wall: path enumeration includes direct paths and first-order reflection paths (mirrors of the four shaft walls). The length of each path is calculated by summing geometric vectors, and the total coupling strength is the sum of the coherent contributions of all paths. The coupling curve is adjusted to a multipath form using the Friesian transmission formula to calculate path loss: for single-path loss, the following is used... Where L p d represents the path loss (in decibels) of the p-th path. p Let λ be the path length (in meters), λ be the signal wavelength (in meters), and f be the carrier frequency. The formula ensures that the loss is expressed in decibels and conforms to the free-space propagation law. The total multipath loss is calculated by coherently superimposing the amplitude squared. The result generates a positional coupling curve, a discrete set of points with the corresponding displacement on the horizontal axis and the coupling loss (negative values ​​indicate strength) on the vertical axis, considering window state corrections: if it is a gate activity, an additional 5 dB attenuation from gate reflection is added. This curve point set is output as an array, with the key being the corresponding displacement, for direct reference in the margin calculation of substep 2.3.

[0050] Sub-step 2.3: Calculate the isolation margin estimate.

[0051] Based on the point set array of the location-based coupling curves from sub-step 2.2, the isolation margin for each corresponding displacement is quantified using the margin link budget algorithm. Specifically, for the coupling loss value in each point set, the isolation margin is defined as the difference between the minimum acceptable isolation and the actual coupling loss: first, the minimum isolation threshold (fixed as the upper limit of link gain plus the feedback suppression factor) is extracted from the preset repeater specifications, and then I is calculated point by point. m =I min -L c , where I m I represents the isolation margin estimate for the m-th displacement (unit: decibels, a positive value indicates sufficient margin). min L is the minimum isolation threshold (in decibels). c To correspond to the coupling loss (unit: decibel) of the positional coupling curve, the formula maintains consistent decibel dimensions and reflects the relativity of the margin. If the window state is gate activity, then in L... cA weighted transient factor (empirically 2 dB) is applied to capture additional coupling fluctuations caused by opening and closing. The calculation results form an isolation margin estimation sequence, a list of values ​​running parallel to the point set array, ensuring that each value is bound to its corresponding displacement and the floor identifier of the original label item. The generation of this sequence emphasizes door zone sensitivity, avoids biases in static estimation, and provides a margin anchor for the mapping in substep 2.4.

[0052] Sub-step 2.4: Associate the estimation results with the location index along the route to form a mapping relationship.

[0053] The final sub-step integrates the isolation margin estimation sequence from sub-step 2.3 with the associated array of the along-path position indices from the motion tag generation unit to form a position-margin mapping. Specifically, each value in the estimation sequence is inserted under the corresponding key value of the along-path position index: for a matching displacement, the array entry is expanded with new fields including coupling loss, isolation margin estimation, and associated window status; if a displacement gap exists, spline interpolation is used to fill the margin value, ensuring the mapping covers the entire stroke without blind spots. Storage uses a key-value database format, serializing the expanded associated array into a mapping relationship file, where the key is the displacement value of the along-path position index, and the value is a composite object (containing margin estimation, coupling curve points, and floor identifiers). The timestamp of this file is referenced from the motion-driven tag to ensure traceability, and an access lock is set to prevent concurrent modification. After output, the mapping relationship file is directly called by the gain threshold combination unit, realizing a complete link from tag reading to margin anchoring.

[0054] In the dynamic propagation environment of the metal boundary of the elevator shaft, the coupling margin mapping unit constructs a positional coupling curve and derives an isolation margin estimate by analyzing and simulating the motion drive label layer by layer. Finally, it embeds the mapping relationship of the position index along the travel path. This targeted processing not only quantifies the precise impact of the car motion on the coupling fluctuations of the direct-drive link, but also provides quantifiable boundary constraints for the transient margin of the door zone, ensuring that the subsequent gain coupling can pre-align the risk, thereby suppressing the coverage drop and maintaining stability throughout the entire travel.

[0055] The motion tag generation unit and coupling margin mapping unit have laid the foundation. The former outputs motion-driven tags as spatiotemporal references, while the latter calculates positional coupling curves and isolation margin estimates by reading these tags, forming a mapping relationship with the position index along the travel path. This mapping accurately anchors the distribution of the margin in displacement, providing a quantifiable input boundary for simultaneous solution. However, existing gain control models are often independent of the isolation margin, causing the strategy parameters to be unable to adapt to nonlinear changes in the position along the travel path. The vibration suppression threshold also lacks dynamic adjustment for the door zone window, making it difficult to generate a positional strategy table, thus restricting the accuracy of time-domain simulation and the formulation of operation and maintenance plans. To address this, the gain threshold simultaneous solution unit starts by reading the mapping relationship, advancing simultaneous calculation and strategy generation, and finally outputting a strategy parameter table. This process ensures the alignment of gain and margin positions, enabling predictive response to changes in car attitude, injecting optimization constraints into the simulation playback of the simulation error iteration unit, and improving the continuity of the repeater link in high-risk sections.

[0056] The specific processing logic of the gain threshold simultaneous unit:

[0057] The gain threshold combination unit divides the processing into four sub-steps: first, reading the mapping relationship; second, combining the gain control model with the isolation margin estimate; third, generating the positional gain strategy; fourth, generating the vibration suppression threshold strategy; and finally, outputting the strategy parameter table.

[0058] Sub-step 3.1: Read the mapping relationship.

[0059] The reading process loads an extended associative array from the mapping relationship file output by the coupling margin mapping unit. Specifically, the key-value database format file is opened, and the displacement value and its composite object for each along-the-loop position index are extracted key by key, including the isolation margin estimate, the coupling loss point of the positional coupling curve, and the associated gate window state. Simultaneously, the integrity of the array is verified by checking the monotonically increasing displacement values ​​and the non-emptiness of the margin estimate. If gaps are found, cubic spline interpolation is applied to supplement the missing margin values ​​to cover the entire shaft length. The read extended associative array is temporarily stored in an ordered list, with each list element bound to its floor identifier and window state. This facilitates direct indexing of the margin estimate corresponding to the displacement in the simultaneous process of sub-step 3.2, avoiding redundant data loading operations.

[0060] Sub-step 3.2: Combine the gain control model with the isolation margin estimate.

[0061] This sub-step addresses the extended associative array in sub-step 3.1, employing a linear programming algorithm to solve the gain control model and isolation margin estimate simultaneously. Specifically, the gain control model is first initialized as a cascaded function containing the forward amplifier gain and feedback suppression modules. The output objective of this model is to maximize signal coverage strength while constraining self-oscillation risk. Then, for the isolation margin estimate at each displacement in the array, it serves as the input to the simultaneous constraint: the objective function is minimized using a linear programming algorithm, i.e., coverage strength minus the margin penalty term. The coverage strength is defined by multiplying the preset transmit power by the gain function, and the margin penalty term is the linear attenuation when the isolation margin estimate falls below a preset standard. The solution process iteratively adjusts the gain variable until all constraint equations are satisfied, including the boundary condition that the link gain does not exceed the isolation margin estimate. If the gate window is active, a transient margin correction factor (based on the additional loss of the positional coupling curve) is embedded in the constraints. The combined results generate an intermediate gain boundary sequence, a list of values ​​in parallel with the extended associative array, where each value represents the upper limit of the optimized gain at that shift. This sequence emphasizes the margin-driven dynamic balance and provides a combined anchor point for the location strategy in sub-step 3.3.

[0062] Sub-step 3.3: Generate the location-based gain strategy.

[0063] Based on the intermediate gain boundary sequence from sub-step 3.2, a piecewise linear interpolation algorithm is used to construct a positional gain strategy. Specifically, for each displacement in the extended correlation array, the intermediate gain boundary value is used as an anchor point to calculate the linear transition between adjacent displacements: starting from the nominal gain (preset link equalization value) of the stable operating segment, a slope adjustment is introduced at the door zone window boundary to smooth the impact of margin fluctuations; the interpolation process considers the discreteness of floor identifiers, and if the displacement span between consecutive floors exceeds the preset standard, the segments are subdivided and a second smoothing is applied to suppress jumps. The generated positional gain strategy is a composite array, where the key is the displacement index along the travel path, and the value is the gain strategy function parameter set, including the gain base value, slope coefficient, and margin linkage factor of the segment; this array covers the entire travel path, ensuring the gradual response of the strategy during car deceleration and arrival at the station, avoiding noise rise induced by abrupt gain changes. The output of this strategy array directly constrains the vibration suppression threshold calculation in sub-step 3.4, achieving synergy between gain and oscillation suppression.

[0064] Sub-step 3.4: Generate vibration suppression threshold strategy and output strategy parameter table.

[0065] The final sub-step integrates the gate window states of the positional gain strategy array and the extended association array from sub-step 3.3, and generates a vibration suppression threshold strategy using a threshold adaptive algorithm. Specifically, the array is first traversed. For each displacement's gain strategy parameter set, the vibration suppression threshold is calculated as the inverse ratio of the gain base value multiplied by the margin factor of the isolation margin estimate. The threshold is designed as a dynamic upper limit; when the window state is gate active, the threshold shifts downward to enhance feedback suppression and prevent transient self-excitation. The adaptive process monitors the threshold continuity of adjacent displacements; if fluctuations exceed a preset tolerance, the threshold sequence is smoothed using low-pass filtering. The vibration suppression threshold strategy forms an extended parameter set containing the lower threshold limit, trigger condition, and recovery delay for each displacement. Subsequently, the positional gain strategy array and the vibration suppression threshold strategy parameter set are combined into a strategy parameter table, a tabular file where rows correspond to displacement values, and columns include the gain base value, slope coefficient, vibration suppression threshold, and associated floor identifier. The output file is serialized in CSV format and embedded with timestamps referenced from motion drive tags to ensure traceability for simulation error iteration unit calls. The generation of this table completes the closed loop of the entire joint process.

[0066] To address the dynamic impact of elevator car attitude changes on the isolation margin of the repeater link, a gain threshold simultaneous unit is developed. Through precise solution of the simultaneous gain control model and margin estimation, a positional gain strategy and a vibration suppression threshold strategy are derived and condensed into a strategy parameter table. This positional anchoring process not only pre-aligns gain fluctuations and coupling fluctuations but also strengthens the suppression of oscillations under door transients, ensuring stability during deceleration and arrival zones. This provides a parameter benchmark for subsequent time-domain simulation error iteration and improves the operational response to high-risk windows.

[0067] The motion tag generation unit and the gain threshold simultaneous unit have gradually built a spatiotemporal foundation and strategy framework. The former outputs motion-driven tags as a unified reference, the middle unit forms a mapping relationship between the positional coupling curve and the isolation margin estimate, and the latter generates positional gain strategy and vibration suppression threshold strategy through simultaneous solution and outputs a strategy parameter table. This table accurately anchors the displacement dependence of gain and threshold, providing a callable parameter set for simulation execution. However, current repeater link verification mostly relies on offline static testing, ignoring the playback capability of motion-driven tags and failing to capture the temporal characteristics of oscillation and noise rise in real time. This leads to a disconnect between error recording and parameter iteration, making it difficult to converge the coupled estimator to the dynamic boundary of the shaft. To overcome this limitation, the simulation error iteration unit uses tag playback as a driver to promote temporal simulation execution, verification, and iterative optimization. This closed-loop process ensures the robust adjustment of strategy parameters throughout the car's entire stroke, injects verification constraints into the stable solidification of the solidified risk output unit, and achieves precise suppression of high-risk sections and reliability of the operation and maintenance plan.

[0068] The specific processing logic of the simulation error iteration unit:

[0069] The simulation error iteration unit divides the processing into four sub-steps: first, it reads the motion-driven label and policy parameter table; second, it performs time-domain simulation using the label playback strategy; third, it records the error based on the oscillation and noise verification results; and finally, iteratively optimizes the coupling estimator and policy parameters.

[0070] Sub-step 4.1: Read the motion-driven label and strategy parameter table.

[0071] The read operation extracts core data from the motion-driven label file of the motion label generation unit and the strategy parameter table of the gain threshold unit. Specifically, firstly, the JSON sequence of motion-driven labels is parsed, and a list of label items, including timestamps, corresponding displacements, floor identifiers, and window states, is loaded. The temporal integrity of the sequence is verified: the list is traversed to confirm that the timestamps are increasing and the corresponding displacements cover the entire shaft length. If the gap exceeds the resolution threshold, the missing state is inferred from adjacent label items through a forward padding mechanism to avoid playback interruption. Secondly, the CSV file of the strategy parameter table is opened, and the table row data is imported row by row. Each row is bound to the displacement value and its column value along the position index, including the gain base value, slope coefficient, and lower threshold limit and trigger condition of the positional gain strategy, and the vibration suppression threshold strategy. During the import process, the displacement values ​​are matched to expand the strategy fields of the label items. If they do not match, they are marked as default nominal parameters. The read composite data is temporarily stored in the simulation buffer as an associated list structure, where each element integrates the label item and the corresponding strategy parameter, so that the playback module in sub-step 4.2 can directly query the displacement-driven gain adjustment. The buffer is designed to support parallel access, ensuring low-latency execution of tag playback.

[0072] Sub-step 4.2: Perform time-domain simulation using the tag playback strategy.

[0073] This sub-step utilizes the composite data buffer from sub-step 4.1 to simulate the temporal dynamics of the repeater link using the Runge-Kutta method. Specifically, the sequence is replayed in the order of the motion-driven tag timestamps: for each timestamp, the corresponding displacement strategy parameters are queried, and the current gain base value and vibration suppression threshold lower limit of the gain control model are dynamically set; then, the link state vector is initialized, including the donor antenna transmit power, the serving antenna receive signal, and the feedback loop gain. The Runge-Kutta method is used to apply a set of integral-differential equations, which describe the evolution of the signal amplitude over time, with the derivative term capturing the transient impact of coupling fluctuations on the margin. The simulation step size is set to a subset of the timestamp interval to capture the opening and closing transients in the gate window state: when the window state is gate activity, real-time loss corrections from the positional coupling curve are injected into the equation input to ensure the simulation reflects boundary fluctuations caused by guide rail attachments. The execution process generates a time-domain trajectory file, a time-series dataset, recording the signal amplitude, applied gain value, and margin observation value for each step; this file covers the entire simulation cycle and embeds floor identifiers for segmented analysis. The output of this trajectory file provides the raw waveform data for verification in sub-step 4.3, enabling seamless integration of playback and simulation.

[0074] Sub-step 4.3: Record the error based on the oscillation and noise verification results.

[0075] Based on the time-domain trajectory file from sub-step 4.2, spectral analysis algorithms are used to verify oscillation and noise rise indices. Specifically, the trajectory dataset is loaded, and the entire journey is divided into stable operation segments and gate zone active segments. For each segment, a fast Fourier transform is applied to extract the spectral components of the signal amplitude. The oscillation peak is identified as the harmonic amplitude near the carrier frequency, and the noise rise peak is identified as the instantaneous jump in out-of-band noise power. The verification threshold is set to a predefined upper limit of self-excitation risk. If the peak exceeds the verification threshold, the timestamp is marked as an anomaly, and error indices are calculated: oscillation error is the relative deviation between the peak amplitude and the lower limit of the vibration suppression threshold strategy, and noise rise error is the difference between the noise power and the inverse of the isolation margin estimate. The recording process summarizes the anomalies into an error log, a structured table where rows correspond to timestamps and corresponding displacements, and columns include oscillation error value, noise rise error value, associated floor identifier, and window status. If the anomaly density of the gate zone active segment is higher than average, it is additionally marked as a high-risk sub-segment. The generation of this log emphasizes transient sensitivity, avoids global average deviation, and provides quantitative feedback for the iteration of sub-step 4.4.

[0076] Sub-step 4.4: Iteratively optimize the coupling estimator and policy parameters.

[0077] The final sub-step references the error log from sub-step 4.3 and uses the gradient descent algorithm to iteratively adjust the coupling estimator and policy parameters. Specifically, the coupling estimator is first initialized with the parameter set of the positional coupling curve generation module of the coupling margin mapping unit, including the reflection path weights for ray tracing. Then, for the oscillation error and noise rise error at each outlier point in the log, the gradient direction is calculated: the error gradient is the partial derivative with respect to coupling loss, pointing towards the curve adjustment to reduce margin deviation. In the iterative loop, the estimator parameters are updated by multiplying the step size by the negative gradient, and the time-domain simulation of sub-step 4.2 is rerun to verify convergence. If the total error is below a preset tolerance, the process stops. The policy parameters are simultaneously optimized: for the displacement of high-risk segments, the slope coefficient of the positional gain policy is adjusted to match the updated isolation margin estimate, and the lower threshold of the vibration suppression threshold policy is shifted downwards by the error proportion. After optimization, the policy field in the composite data buffer is updated, and the final version of the error log is serialized to the iteration archive file. This file records the error evolution curve and parameter change history for each loop, ensuring traceability. The output of this file completes the closed loop of the simulation error iteration unit, providing an optimized benchmark for the stable judgment of the solidified risk output unit.

[0078] Under the dynamic boundary disturbance of elevator shaft car operation, the simulation error iteration unit captures the verification error of oscillation and noise rise through the accurate playback of motion-driven tags and the deep execution of time-domain simulation, and drives the iterative convergence of coupled estimators and strategy parameters. This processing of door zone transients not only suppresses the chain reaction of gain overshoot, but also refines the margin boundary under the along-path position index, ensuring the coverage continuity of the repeater link in deceleration and arrival areas, and laying the error calibration foundation for the location of operation and maintenance anomaly trigger points.

[0079] The motion tag generation unit and the simulation error iteration unit have formed a closed-loop foundation. The former establishes motion-driven tags as spatiotemporal references, the middle calculates the mapping relationship between the positional coupling curve and the isolation margin estimate, and the latter simultaneously generates a strategy parameter table for the positional gain strategy and the vibration suppression threshold strategy. Finally, the oscillation and noise rise errors are recorded through time-domain simulation of tag playback, and the coupling estimator and strategy parameters are iteratively optimized. This optimization process refines the error log and iteration archive, ensuring robust convergence of parameters under the dynamic boundary of the shaft. However, existing repeater link deployments often lack stable judgment and solidification mechanisms after iteration, failing to transform the optimized strategy into a deployable parameter set. At the same time, they ignore the output of risk visualization and operation and maintenance triggers, resulting in delayed positioning of high-risk sections and the deviating of contingency plan formulation from positional constraints. To fill this gap, the solidified risk output unit starts with checking the stability of iterations, and promotes strategy solidification, risk map generation and dedicated output. This final link ensures the displacement anchoring of the deployment parameter set and the accurate marking of anomalies, realizes long-term suppression of deceleration zone sway and uplink noise, provides predictive tools for operation and maintenance, and improves stability and reuse efficiency throughout the entire process.

[0080] The specific processing logic of the solidified risk output unit:

[0081] The solidified risk output unit breaks down the processing into four sub-steps: first, it checks if the iteration has reached stability; second, it solidifies the strategy into a deployment parameter set; third, it generates a risk map along the process; and finally, it outputs gate-specific strategies and anomaly trigger points for operation and maintenance.

[0082] Sub-step 5.1: Check that the iteration has reached stability.

[0083] The inspection process extracts the error evolution curve from the iteration archive file of the simulation error iteration unit. Specifically, it loads the sequence data from the archive, calculates the rate of change of the oscillation error value and the noise rise error value in the most recent three iterations, differentiates the curve using the first-order difference method, obtains the slope sequence for each timestamp, and then aggregates the average slope of the entire process. If the absolute value is lower than the preset convergence threshold (defined as a 10% relative limit of error change) and remains so for two consecutive iterations, it is determined to be in a stable state. At the same time, for the sub-segments of the gate window state, the local slope of the high-risk sub-segments is additionally verified to ensure the convergence priority of transient errors. During the inspection, if stability is not reached, it rolls back to the optimization loop entry of the simulation error iteration unit and records the current iteration and slope statistics in the inspection log. When stability is reached, a stability flag file is generated, with a Boolean flag accompanying the final error curve. The timestamp of this file is referenced from the motion-driven label, which is convenient for sub-step 5.2 to directly read to confirm the solidification premise. The creation of this flag file emphasizes consistency throughout the entire process and avoids false positives of local convergence.

[0084] Sub-step 5.2: Solidify the strategy into a deployment parameter set.

[0085] Based on the stability flag file in sub-step 5.1, the composite data buffer optimized by the simulation error iteration unit is integrated to solidify the strategy parameters. Specifically, the process first verifies the flag is true, then iterates through the associated list of the buffer. For each displacement value along the path location index, the final gain base value, slope coefficient, and the lower threshold and trigger condition of the positional gain strategy are extracted. The solidification operation employs a parameter freezing mechanism, converting these values ​​from floating-point optimized form to a fixed-precision integer set (rounded to decibels), and embedding a margin linkage factor as a checksum to prevent deployment drift. For the discrete distribution of floor identifiers, continuous displacement segments are aggregated into parameter blocks, each containing the boundary range of the average gain base value and the lower threshold. Finally, the deployment parameter set is serialized into a binary configuration file, where the key is the displacement range and the value is a frozen parameter vector, which includes the gain base value, slope coefficient, lower threshold, and trigger condition. The output of this configuration file covers the entire shaft length and is set to read-only to ensure immutability for external deployment calls, providing a solidified baseline for the risk calculation in sub-step 5.3.

[0086] Sub-step 5.3: Generate a risk map along the route.

[0087] Using the deployment parameter set configuration file and the error log of the simulation error iteration unit from sub-step 5.2, a heatmap rendering algorithm is employed to generate a risk map along the path. Specifically, the configuration file and log are opened, and the frozen parameters of each displacement are mapped to risk indicators: the risk value is calculated as the reciprocal of the lower limit of the vibration suppression threshold multiplied by the peak proportion of the corresponding noise rise error value, plus the cumulative contribution of the oscillation error, forming a composite risk score; then, these scores are sorted according to the corresponding displacement, and the heatmap algorithm is used to map the scores to a color gradient (blue for low risk, red for high risk), with the horizontal axis representing the displacement of the path position index and the vertical axis representing the discrete level of the floor identifier, embedding the outline of the door zone window state to highlight transient bright areas. The generation process weights the risk scores in the door zone activity sub-segment, and the relative deviation of the lower threshold is used to amplify the visualization comparison; the result is output as a risk map image file, a PNG format raster map, accompanied by a vector SVG backup to support scaling. The resolution of this image matches the timestamp density of the motion-driven labels to ensure smooth rendering of the risk curve. The metadata of the image file is embedded with stability flags and final error statistics, which facilitates the extraction of trigger points in sub-step 5.4.

[0088] Sub-step 5.4: Output gate-specific policies and operation and maintenance-oriented exception trigger points.

[0089] The final sub-step integrates the risk map image files and deployment parameter sets from sub-step 5.3, extracting and outputting dedicated content. Specifically, the process first filters high-risk segments of the door zone window status in the risk map. For the displacement range of each segment, a door zone-specific strategy is customized: the gain base value and slope coefficient are copied from the deployment parameter set, with an added transient recovery delay (based on the extension of the trigger condition, set as a margin reset two seconds after door closure). Then, for abnormal trigger points, abnormal points where oscillation errors or noise rise errors exceed the threshold are located from the error log, generating a trigger point list. This list is an ordered array containing timestamps, corresponding displacements, floor identifiers, risk scores, and contingency plans (e.g., gain reduced to 80% of the lower threshold). The output operation serializes the door zone-specific strategy into a dedicated configuration file, containing the strategy vector and recovery delay for each window. The abnormal trigger point list is exported as an operation and maintenance report table, with row data in CSV format and embedded links to the risk map. All output files are uniformly named and placed in the output directory to ensure that abnormal trigger points for operation and maintenance have a location traceability link. The completeness of this output marks the end of the risk output unit process.

[0090] To address the transient instability risks in the door and deceleration zones during elevator car operation, the solidified risk output unit uses iterative and stable precision checks and strategy solidification to generate visual anchoring of the risk map along the route. It also outputs door zone-specific strategies and maintenance tools for abnormal trigger points. This final processing not only constrains the self-excitation propagation of upward noise but also achieves deployment-level stability with continuous coverage, ensuring rapid location of high-risk sections and reuse of contingency plans. It provides a feasible framework for the full-journey optimization of repeater links.

[0091] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0092] It should be noted that the system of the present invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting a variety of hardware environments and usage requirements.

[0093] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0094] It should be noted that, in this document, the use of relational terms such as "first" and "second" is merely to distinguish one entity or operation from another, and does not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0095] 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. A simulation system for the motion attenuation coupling of a repeater facing an elevator shaft, characterized by, Comprising: a motion tag generation unit that establishes an along-track position index and a door zone window based on a stop signal and a door state event, and outputs a motion drive tag as a unified time reference; a coupling margin mapping unit that reads the motion drive tag, calculates a positionized coupling curve and an isolation margin estimate, and stores the estimate in association with the along-track position index to form a mapping relationship; a gain threshold co-solution unit that co-solves a gain control model and the isolation margin estimate, generates a positionized gain strategy and a vibration suppression threshold strategy, and outputs a strategy parameter table for simulation invocation; a simulation error iteration unit that replays the strategy based on the motion drive tag and performs time-domain simulation, records errors based on verification results of oscillation and noise lifting, and iteratively optimizes the coupling estimator and the strategy parameters; a solidification risk output unit that solidifies the strategy into a deployment parameter set when the iteration reaches stability, and outputs an along-track risk map and a door zone-specific strategy, as well as an exception trigger point for operation and maintenance.

2. The elevator shaft-oriented repeater motion attenuation coupling simulation system according to claim 1, wherein: the motion tag generation unit collects a deceleration threshold of the car speed and a floor encoder position feedback to generate a stop signal pulse sequence, records a start time and a duration, and records a door state event binary sequence through a door body position sensor switch feedback, including a state transition timestamp and a transition type.

3. The elevator shaft-oriented repeater motion attenuation coupling simulation system according to claim 2, wherein: the motion tag generation unit uses a Kalman filter algorithm to smooth the pulse sequence timestamp and align the nominal speed curve to calculate the cumulative displacement, and superimposes the door state sequence to the corresponding pulse displacement to form a position correction, generating an associated array sorted by displacement values as an along-track position index.

4. The elevator shaft-oriented repeater motion attenuation coupling simulation system according to claim 3, wherein: the coupling margin mapping unit parses the motion drive tag JSON sequence to load the tag item list to verify that the timestamp is increasing and the displacement is continuous, and supplements the missing displacement by linear interpolation to form a memory buffer list sorted by displacement.

5. The elevator shaft-oriented repeater motion attenuation coupling simulation system according to claim 4, wherein: the coupling margin mapping unit traverses the tag sequence list to simulate the corresponding displacement electromagnetic coupling path using a ray tracing algorithm, defines the donor antenna at the top of the shaft and the service antenna at the car position, and uses image theory to handle multiple reflections of metal walls to enumerate direct and first-order reflection paths to aggregate coherent contributions and generate a positionized coupling curve point set.

6. The elevator shaft-oriented repeater motion attenuation coupling simulation system according to claim 5, wherein: the gain threshold co-solution unit loads the mapping relationship file to expand the associated array, extracts the along-track position index displacement value and the composite object verification displacement monotonicity, and supplements the gap margin value by cubic spline interpolation to form an ordered list bound to the floor identifier and the window state.

7. The elevator shaft-oriented repeater motion attenuation coupling simulation system according to claim 6, wherein: The gain threshold unit initializes the gain control model cascade function, isolates the margin estimate as a constraint input for the extended associative array, uses a linear programming algorithm to minimize the coverage strength minus margin penalty objective function, iteratively adjusts the gain variable to satisfy the link gain not exceeding the margin boundary, and generates an intermediate gain boundary sequence.

8. The elevator shaft-oriented repeater motion attenuation coupling simulation system according to claim 7, characterized in that: The simulation error iteration unit analyzes the motion drive label JSON sequence, loads the label item list to check the time sequence integrity, infers the missing state through forward filling, and imports the strategy parameter table CSV file to match the displacement value to expand the label item strategy field, forming a simulation buffer area association list that integrates label items and parameters.

9. The elevator shaft-oriented repeater motion attenuation coupling simulation system according to claim 8, characterized in that: The simulation error iteration unit plays back the composite data buffer sequence in timestamp order, queries the corresponding displacement strategy parameter to dynamically set the gain base value and the lower limit of the vibration suppression threshold, initializes the link state vector to use the Runge-Kutta method to integrate the differential equation set to generate a time-domain trajectory file, and records the signal amplitude and margin observation value.

10. The elevator shaft-oriented repeater motion attenuation coupling simulation system according to claim 9, characterized in that: The solidification risk output unit extracts the error evolution curve from the iteration archive file to calculate the latest three rounds of oscillation error and noise lifting error change rate, uses first-order difference to aggregate the average slope below the convergence threshold to determine the stable state, and generates a stable flag file accompanied by a final error curve referencing the motion drive label timestamp.

Images