Multi-device timing coordination control method for hoisting power master

Abstract

The application relates to the technical field of industrial control, in particular to a multi-device timing coordination control method for hoisting power master controllers, which comprises the following steps: obtaining a target device power-on interval duration to determine a thermal pretreatment state flag, combining a load rate and a power grid voltage deviation rate to divide a working condition partition grid, updating a response time delay mean value and obtaining a residual standard deviation and an aging trend slope; calculating an adaptive upper bound based on the mean value, the residual standard deviation and the aging trend slope; sampling within a window to construct a state sequence, if not ready, extracting a tail subsection to calculate a first-order and second-order difference sequence, identifying an aging cold start working condition slow response or a real device fault; and executing differential timing treatment and refreshing the mean value based on the identification conclusion. Through the application, the cold and hot state aliasing is eliminated to deviate from the baseline, the aging cold start slow response is effectively avoided from being misjudged as a fault, and the operation continuity is improved.

Description

Technical Field

[0001] This application relates to the field of industrial control technology, specifically to a multi-device timing collaborative control method for a hoisting power supply main controller. Background Technology

[0002] In industrial hoisting operations, lifting equipment is typically equipped with multiple independently controlled electrical units, including hoisting motor frequency converters, horizontal travel drives, electromagnetic brakes, and auxiliary lighting. These devices must be initialized sequentially according to a strict power-on sequence each time an operation begins to ensure logical coordination between mechanical movement and braking protection. The hoisting power supply main controller, as the unified scheduling core of these multiple devices, directly determines the safety, reliability, and operational continuity of the entire machine through the accuracy of its timing control.

[0003] To address the issue of surge current aggregation caused by multiple devices being powered on simultaneously, existing main controllers generally employ a timing control scheme with fixed interval delays and set fixed timeout thresholds. To overcome the limitations of fixed thresholds, the improved scheme introduces a working condition partitioning baseline based on historical response delay data, and calculates the adaptive waiting upper bound using the mean and standard deviation of the data within the window.

[0004] However, the baseline scheme for operating condition zoning assigns equal weight to all historical samples within the sliding window when calculating the mean and standard deviation, implicitly assuming that all historical data within the same zone follows the same steady-state distribution. Because the existing scheme treats new and old data equally in its sliding window, a large number of fast-response historical samples in the early stages of aging will continuously suppress the window mean, causing the current mean to be systematically lower than the actual expected latency under the actual aging state of the equipment. When the equipment enters the mid-to-late stages of aging and is subjected to a long-term shutdown cold start condition, the actual response latency will significantly exceed the upper bound of the waiting time calculated centered on the lower mean. However, at this time, the equipment itself has no functional faults and is exhibiting aging-driven slow response. The main controller cannot distinguish between this aging cold start slow response and a real fault, inevitably triggering an erroneous emergency shutdown, and this misjudgment continues to worsen with the increase in the equipment's service life.

[0005] To address the existing technical problem that the baseline delay of the hoisting power supply main controller, which is based on the equal weighted averaging of historical samples and cannot separate the superimposed effects of equipment aging drift trend and hot pre-processing state, leads to slow response and continuous misjudgment of actual faults under aging cold start conditions, this application provides a multi-device timing collaborative control method for the hoisting power supply main controller. Summary of the Invention

[0006] To address the existing technical problem that the baseline delay of the hoisting power supply main controller, which is based on the equal weighted averaging of historical samples and cannot separate the superimposed effects of equipment aging drift trend and hot pre-processing state, leads to slow response and continuous misjudgment of actual faults under aging cold start conditions, this application provides a multi-device timing collaborative control method for the hoisting power supply main controller.

[0007] This application provides a multi-device timing collaborative control method for a hoisting power supply master controller, including: obtaining the interval between the current power-on and the previous normal power-off of the target device; determining a thermal pre-processing status flag based on the interval; dividing the operating condition into zones based on the load rate and the grid voltage deviation rate; updating the mean of the response delay within the operating condition zone based on an exponentially weighted moving average algorithm; and simultaneously obtaining the residual standard deviation and aging trend slope; when issuing an enable command to the target device, calculating the adaptive waiting upper bound of the current timing step based on the mean, residual standard deviation, and aging trend slope; and in the adaptive... The system continuously samples the ready feedback register of the target device within the window corresponding to the upper bound to construct a state sequence. If the target device has not completed ready confirmation when the window expires, the last segment of the state sequence is extracted to calculate the first-order difference sequence and the second-order difference sequence. Based on the first-order difference sequence and the second-order difference sequence, the system identifies whether the target device is in an aging cold start condition with slow response or a real equipment failure. Based on the conclusion that it is an aging cold start condition with slow response or a real equipment failure, the system performs differentiated timing processing on the target device and refreshes the mean value in the condition partition cell using a differentiated baseline write-back method.

[0008] By introducing the thermal pretreatment state into the operating condition coordinate system and strictly isolating the cold and hot state sequences within the partition grid, the systematic bias of the mean center caused by the aliasing of cold and hot states is eliminated, so that the baseline mean of each operating condition partition grid can accurately reflect the actual expected time delay under the corresponding hot state conditions.

[0009] Preferably, determining the thermal pretreatment status indicator based on the interval duration includes: collecting the average start-up response delay of the same model of equipment under different shutdown intervals, plotting the response curve of the average start-up response delay changing with the shutdown interval; calculating the absolute value of the slope between adjacent gradients of the response curve, and taking the shutdown interval corresponding to the first drop of the absolute value of the slope below the preset proportion of the peak value as the cold-hot boundary point; comparing the interval duration with the cold-hot boundary point, and determining the thermal pretreatment status indicator based on the comparison result.

[0010] Preferably, updating the mean of the response delay within the operating condition grid based on the exponentially weighted moving average algorithm includes: calculating the first-order difference mean of the residual sequence within the operating condition grid, and normalizing the first-order difference mean using the residual standard deviation to obtain a dimensionless drift intensity index; determining the forgetting factor through linear interpolation based on the percentile distribution of the dimensionless drift intensity index in the historical sequence; and using the forgetting factor to perform a weighted summation of the measured response delay and the mean of the previous time step to obtain the updated mean of the response delay.

[0011] By replacing the equal-weighted moving average with an exponentially weighted moving average and introducing a forgetting factor that is adaptively adjusted based on the measured drift intensity, the mean can continuously approach the current equipment state, effectively tracking the monotonic drift trend caused by equipment aging and avoiding the locking of the baseline by early historical data.

[0012] Preferably, the adaptive waiting upper bound for this time series step is calculated based on the mean, residual standard deviation, and aging trend slope, including: calculating the aging forward compensation term based on the aging trend slope and the number of forward compensation steps; calculating the dispersion compensation term based on the residual standard deviation and the benchmark multiple; and summing the mean, the aging forward compensation term, and the dispersion compensation term to obtain the adaptive waiting upper bound.

[0013] By superimposing an aging trend forward compensation term on the mean, the upper bound of the waiting period is transformed from passively lagging behind the aging facts to actively predicting the aging trend. This effectively fills the margin gap left by the tracking lag and reduces the probability of triggering erroneous shutdowns due to the systematically low upper bound of the waiting period.

[0014] Preferably, the ready feedback register of the target device is continuously sampled within the window corresponding to the upper bound of the adaptive waiting period to construct a state sequence, including: obtaining the minimum physical change time constant of the ready signal of the target device; determining a fixed sampling period based on the minimum physical change time constant; continuously sampling the ready feedback register of the target device within the window corresponding to the upper bound of the adaptive waiting period with the fixed sampling period as the time interval; and arranging the progress ratio values ​​obtained by continuous sampling in chronological order to construct a state sequence.

[0015] Preferably, the method for identifying whether a target device is in an aging cold start condition with slow response or a real device fault based on a first-order difference sequence and a second-order difference sequence includes: calculating the percentage of terms with values ​​greater than or equal to zero in the first-order difference sequence and calculating the mean of the second-order difference sequence; determining whether the percentage of terms is not lower than the continuous progress percentage threshold, whether the mean is less than zero, and whether the final value of the last segment is not lower than the final value threshold; if all three conditions are met, the target device is identified as being in an aging cold start condition with slow response; if any one of the three conditions is not met, the target device is identified as being in a real device fault.

[0016] By jointly judging the end segment of the ready signal trajectory, the normal physical state of the capacitor naturally decelerating at the end of charging is distinguished from the non-monotonic termination state caused by the fault, thus cutting off the path of aging cold start slow response being misjudged as a real fault.

[0017] Preferably, based on the conclusion that the response is a slow response due to aging cold start or a real equipment failure, differentiated timing processing is performed on the target equipment, including: in response to the conclusion that the response is a slow response due to aging cold start, suspending the timing commands of all subsequent unstarted equipment during the extended waiting phase; calculating the amount by which the measured response delay exceeds the upper bound of the adaptive waiting phase, and determining the reduction weight based on the ratio of the excess amount to the residual standard deviation; updating and writing back the mean using the reduction weight, and resuming timing progress after the target equipment completes the readiness confirmation, thus completing the differentiated timing processing.

[0018] Preferably, based on the conclusion that the fault is due to aging cold start or a real equipment failure, differentiated timing processing is performed on the target equipment, including: in response to the conclusion that the fault is a real equipment failure, the enabled devices that have been completed and ready are deactivated in a preset safety sequence to perform a reverse power-down sequence; the fault device number, the fault determination time, and the feature summaries of the first-order difference sequence and the second-order difference sequence are extracted and packaged into a diagnostic data packet; the diagnostic data packet is pushed to the host computer, and the delay data corresponding to this fault is prohibited from being written to the operating condition partition grid, thus completing the differentiated timing processing.

[0019] Preferably, the process of simultaneously acquiring the residual standard deviation and the aging trend slope includes: calculating the difference between the measured response delay and the mean to obtain the residual; storing the residual in a fixed-length residual window and calculating the standard deviation of all residuals within the fixed-length residual window to obtain the residual standard deviation; performing linear regression on the measured response delay within the fixed-length residual window and extracting the linear regression slope as the aging trend slope.

[0020] Preferably, during the extended waiting phase, the timing commands for all subsequent unstarted devices are suspended, including: during the extended waiting phase, continuously monitoring the status value of the ready feedback register of the target device; if the status value decreases for a consecutive preset number of sampling points, or if no ready confirmation signal is received when the upper bound of the extended duration is exhausted, then the identification conclusion is forcibly changed to a real device failure; the extended waiting phase is terminated, and the corresponding reverse power-down timing is triggered.

[0021] This application eliminates the systematic bias of the mean center caused by cold start historical data and hot start historical data in the partitioned grid, so that the baseline mean of each working condition partition can accurately reflect the actual expected time delay under the corresponding hot condition. At the same time, it replaces the equal weighted moving average with the exponentially weighted moving average and introduces the forward compensation term of the aging trend slope, so that the waiting upper bound can actively track the time delay drift trend caused by equipment aging.

[0022] Furthermore, this application distinguishes between the natural deceleration at the end of charging and the non-monotonic termination caused by a fault at the morphological level, thus achieving reliable identification of slow response under old operating conditions and real faults. By anchoring the values ​​of all key parameters to the statistical characteristics of factory calibration data or historical field samples, the adaptability failure of fixed parameters under different equipment and operating conditions is avoided, thereby enhancing the continuous effectiveness of the solution throughout its entire life cycle. Attached Figure Description

[0023] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent upon reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of the invention are illustrated by way of example and not limitation, and like or corresponding reference numerals denote like or corresponding parts, wherein: Figure 1 This is a flowchart illustrating a multi-device timing collaborative control method for a hoisting power supply main controller according to the present invention.

[0024] Figure 2 This is a comparison chart of response delay aging drift tracking and waiting upper bound margin in an embodiment of the present invention. Detailed Implementation

[0025] 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, not all, of the embodiments of the present invention. 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.

[0026] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0027] This invention discloses a multi-device timing collaborative control method for a hoisting power supply main controller, referring to... Figure 1 This includes steps S1-S4: S1. Divide the operating conditions into zones and update the average response delay.

[0028] In an optional embodiment, before each job starts, the main controller reads the interval between the current power-on and the previous normal power-off of the target device from its internal runtime accumulation register. This is used as an evaluation index for the state of heat pretreatment.

[0029] Hot and cold dividing point The determination method is as follows: During the factory calibration phase of equipment of the same model, the average start-up response delay is collected at different downtime intervals. ,draw Follow For the changing response curve, calculate the absolute value of the slope between adjacent gradients, and the value corresponding to the first drop of this absolute slope below a predetermined proportion of its peak value. Value as For example, the preset ratio can be 10%, and this inflection point is the physical boundary where response latency is no longer significantly sensitive to downtime, determined by the measured curve. When the current startup is determined to be a cold start, the thermal pretreatment status flag for this startup is determined; the thermal pretreatment status flag is recorded. ;when When this occurs, it is determined to be a hot start, and the heat pretreatment status flag is recorded. .

[0030] Furthermore, in the operating condition coordinate system, the overall load rate Deviation rate between the three levels and the mains voltage Two levels, and heat pretreatment status. The system is divided into two levels, forming 12 operating condition sub-divisions. Each sub-division independently maintains hot and cold state sequences, ensuring strict isolation between the two types of historical data in statistical calculations. Because the pre-charge thermal boundary conditions corresponding to different downtime durations are fundamentally different, this difference can lead to a situation where, if hot and cold state data are mixed into the same sequence for mean calculation, the longer delay samples in the cold state systematically inflate the expected estimate of the hot state, while conversely, the shorter delay samples in the hot state continuously depress the baseline center of the cold state. Therefore, by forcibly isolating the two types of historical data within the sub-divisions, each maintaining its own independent distribution center, this source of bias is eliminated.

[0031] Next, within each partition, to capture the monotonic drift trend caused by device aging, this application tracks response latency by directly replacing the equal-weighted sliding window mean with an exponentially weighted moving average (EWMA). The EWMA is updated as follows:

[0032] in For the first The measured response delay of this operation. This represents the current EWMA mean. Because the equal-weighted sliding window treats both new and old samples equally, the large number of fast-response samples accumulated in the early stages of aging will suppress the window mean for a long period, causing the mean update speed to lag far behind the actual aging process of the equipment. EWMA uses a forgetting factor... Give more weight to recent samples so that the mean can continue to converge toward the current device state, rather than being locked in by earlier historical data.

[0033] Then, the forgetting factor The value is derived based on the measured drift intensity of the current partition grid: taking the most recent First-order difference mean of the secondary residual sequence Characterizing the drift rate, expressed as the residual standard deviation Normalizing it yields a dimensionless drift intensity index. A higher value indicates more significant aging drift. A larger value is used to accelerate tracking; a smaller value indicates a smoother aging trend. Take the smaller value to keep the baseline stable. Specifically, let and These are the history of this partition. The 10th percentile and 90th percentile of the sequence, when Time to take Its lower limit value ,when Time to take Its upper limit Linear interpolation is used in the middle. and The specific values ​​are determined during the factory calibration phase by applying a known drift sequence to each partition and calibrating to minimize the tracking error, ensuring... The tracking speed of aging trends is matched with the actual drift intensity.

[0034] Maintain a capacity of [value] synchronously within each partition. A fixed-length residual window, where the residual is defined as The standard deviation of the residual window This serves as an estimate of the dispersion of the current partition. Since directly calculating the standard deviation from the original time delay data can artificially inflate the overall dispersion of the time delay sequence due to the aging drift trend, resulting in a systematically larger standard deviation, a method is employed. This involves first subtracting the EWMA mean from the original time delay to obtain the residuals, and then calculating the standard deviation from these residuals. This process removes the artificial contribution of the aging trend to the dispersion, thus ensuring a more accurate and stable standard deviation. This more accurately reflects the actual jitter amplitude of the response delay after detrending under the same operating conditions. Simultaneously, linear regression is performed on the measured response delay within the fixed-length residual window, and the linear regression slope is extracted as the aging trend slope. The unit is ms / time. It quantifies the average step size of the increase in response delay due to aging accumulation in each job cycle, thereby achieving simultaneous acquisition of residual standard deviation and aging trend slope.

[0035] Determined by statistical reliability requirements: During the factory calibration phase, calculate the intra-day coefficient of variation of response delay data within each operating condition zone of the same model of equipment. Based on engineering precision requirements The estimated relative error does not exceed the target accuracy. To constrain this, the statistical relationship between sample size and estimation accuracy is used.

[0036] The minimum sample size required is calculated by reverse calculation, and the integer part is used as the partition size. Different working conditions require different zones. Different results can be obtained It is determined by measured data. When the number of historical data points in a certain partition is insufficient... At that time, the initial value and residual window of EWMA are filled with the reference time delay data collected under the same working conditions and thermal state during the factory calibration stage, so as to ensure that a usable baseline is available in the early stage of field operation.

[0037] Thus, by introducing the thermal pretreatment state into the operating condition coordinate system and strictly isolating the hot and cold state sequences within the partitioned grid, the double bias of the baseline center caused by the mixing of hot and cold states and the backlog of aging history samples is eliminated, providing a reliable statistical basis for the accurate calculation of the upper bound in the subsequent waiting period.

[0038] S2. Calculate the adaptive waiting upper bound for this timing step.

[0039] In an optional embodiment, the master controller reads the current EWMA average value from the corresponding operating condition partition cell while issuing an enable command to the target device. residual standard deviation and the slope of the aging trend Calculate the adaptive waiting upper bound for this timing step. .

[0040] The original adaptive scheme centers on the historical mean and uses a fixed multiple of the standard deviation as boundaries, without considering the systematic mean drift caused by the aging trend. Since the EWMA mean inherently lags in tracking aging drift, this discrepancy means that when the aging rate is fast, the EWMA mean will always be systematically lower than the actual expected latency for the next device startup. Therefore, an aging forward compensation term is introduced into the original adaptive scheme. The specific calculation logic is as follows: based on the aging trend slope... With forward compensation steps Calculate the aging forward compensation term Based on residual standard deviation Multiples of the benchmark Calculate the dispersion compensation term ; take the mean The aging forward compensation term and the dispersion compensation term are summed to obtain the adaptive waiting upper bound. The relationship is as follows:

[0041] in The forward compensation steps represent the amount of aging drift that the main controller reserves for several future operating cycles based on the current EWMA average. The larger the value, the greater the forward range of the aging expectation at the upper bound; however, an excessively large value... This can lead to an invalid expansion of the upper bound of the wait, delaying timely responses to real faults. The upper bound is determined by the process-allowed single-step wait expansion constraint: Let For the current partition If it is an integer multiple of , then satisfy ,Right now When actually used, Take the largest integer value within this upper limit range, based on actual measurements. and Inverse reasoning; when At this time, the forward compensation term is automatically reset to zero, without introducing additional waiting margin, thus preventing invalid expansion of the baseline.

[0042] Furthermore, The baseline value is calibrated by the false negative rate of historical samples in the current partition: within a sliding window... Using the proportion of normally ready samples as a constraint, the minimum threshold that causes the false negative rate to first fall below the acceptable level specified in the product safety specifications is selected. The value is used as the current benchmark. When there is a voltage deviation in the power grid ( When this occurs, it indicates that power grid fluctuations may affect the time delay dispersion; in this case, the processor will... For actual measurement The proportion of historical influence is expanded proportionally. Instead of linear interpolation to a preset upper limit, it makes The expansion magnitude is consistent with the actual impact of grid deviation on time delay dispersion. In response to It temporarily suspends timing stepping and outputs power grid anomaly alarms to avoid waiting for the upper bound based on a distorted baseline under severe power grid anomaly conditions.

[0043] In this way, the margin gap left by the EWMA tracking lag is effectively filled, and the probability of erroneous shutdowns triggered by slow response to normal cold starts due to the systematically low upper bound during the mid-to-late stages of aging is significantly reduced.

[0044] S3. Construct a state sequence and identify slow response or real device failure.

[0045] In an optional embodiment, in Within the window, obtain the minimum physical change time constant of the target device's ready signal, and determine the fixed sampling period based on the minimum physical change time constant. With a fixed sampling period The master controller continuously samples the ready feedback register of the target device at time intervals, and arranges the progress ratio values ​​obtained from the continuous sampling in chronological order, constructing a length of [missing information]. state sequence ,in 0 indicates that the device is not started, 1 indicates that the device is ready and confirmed to be completed, and the intermediate value represents the progress of the device's internal power-on self-test. Minimum physical change time constant of the target equipment readiness signal determined during the factory calibration phase Confirm, take for The preset fraction, to meet the sampling margin requirements, is determined by the measured physical characteristics, for example, Can be taken .

[0046] In response to The device still hasn't returned by the time the window expires. (i.e., the target device has not completed readiness confirmation), the master controller enters the trajectory morphology identification stage. The length of the extracted state sequence at the end is... sub-segments Calculate the first-order difference sequence based on this segment. With second-order difference sequence In this sub-segment Corresponding observation duration The typical duration of the deceleration propulsion physical phase at the end of charging, determined by the factory calibration stage, ensures that the sub-segment covers the complete curvature change process at the end of charging.

[0047] Because the two types of typical abnormal operating conditions have fundamental differences in trajectory morphology, the aging cold start slow response The curve shows a decelerating upward trend, that is... and Real faults The curve then shows a plateau followed by a decline or oscillation pattern, i.e., it appears... continuous rollback or The persistently large jitter characteristics may affect the judgment of the equipment status. Therefore, this application identifies the problem based on the contrasting characteristics of the first-order difference sign distribution and the second-order difference mean sign.

[0048] Next, the identification and judgment rules need to be comprehensively and continuously improved to meet the proportion threshold. and final value threshold Continue to promote the percentage threshold From the last segment of each case in the historical normal cold start sample library The distribution of proportions is determined, and the 5th percentile of this distribution is taken as... Final value threshold Similarly, the final value at the window time is taken from a historical normal cold start case. The 5th percentile value of the distribution is identified by the following logic: 1) Calculate the percentage of terms with values ​​greater than or equal to zero in the first-order difference sequence, and then calculate the mean of the second-order difference sequence; finally, determine whether this percentage is not lower than the threshold for continued advancement. Is the mean less than zero (i.e.) ), and the final value of the last sub-field. Is it not lower than the final value threshold? If all three conditions are met, the target device is identified as being in a slow response condition due to aging and cold start, and the processor will enter the extended waiting phase. If any one of the three conditions is not met, the target device is identified as being in a real equipment failure, the waiting phase ends, and the process enters the S4 differentiated handling phase.

[0049] Extended duration upper bound Determined by the actual delay distribution from windowing to final readiness in historical normal cold start cases: the high quantile value of this delay distribution is taken as... If the historical slow-start samples of this partition are insufficient to support distribution estimation, then the maximum out-of-window delay observed during the factory calibration phase under the same operating conditions will be used as the initial value. The value is dynamically updated as samples accumulate. This is in response to the continuous [refresh rate / response] during the expansion waiting phase. Each sampling point appeared Decline, or Exhausted and still not received It is forcibly identified as a real fault and enters S4. Depend on The ratio of the minimum operating time of the inverter protection relay is determined by: Let To protect the minimum operating time specified by the factory for the relay, then To ensure continuity The drop in point does indeed correspond to the relay protection action rather than the measurement noise, and is calculated from the physical action time.

[0050] Thus, by jointly judging the proportion of the first-order differential, the sign of the second-order differential mean, and the final value of the three indicators at the end of the ready signal trajectory, the main controller can distinguish between the normal physical state of the capacitor naturally decelerating at the end of charging and the non-monotonic termination state caused by the fault, effectively cutting off the path of aging cold start slow response being misjudged as a real fault and triggering erroneous shutdown.

[0051] S4. Perform differentiated time-series processing and refresh the mean.

[0052] In an optional embodiment, based on the completed trajectory morphology identification conclusion, the main controller processes the current timing step into three types of execution paths, and refreshes the mean value in the working condition partition cell using a differentiated baseline write-back method for each type of conclusion, so as to maintain the long-term validity of the S1 partition baseline.

[0053] In response to If the device is ready normally within the window, the main controller immediately proceeds to the enable sequence of the next device and records the measured response delay. Write back the corresponding cold or hot sequence of S1 according to the current EWMA update formula, triggering... , and the slope of the aging trend Synchronous scrolling refresh. The criteria for determining accelerated aging rate are as follows: based on the current partition's historical data. The standard deviation of the sequence itself Based on this, when this The updated value is more than the previous increment. When the preset multiple threshold is reached, the main controller writes an aging rate acceleration prompt to the operation and maintenance log, so that maintenance personnel can assess the remaining lifespan of the equipment. For example, the preset multiple can be 2.

[0054] Furthermore, in response to the identification conclusion being a slow response due to aging cold start, the main controller suspends the timing commands of all subsequent unstarted devices during the extended waiting phase. During this period, the status value of the target device's ready feedback register is continuously monitored; if the status value decreases for a consecutive preset number of sampling points, or if no ready confirmation signal is received before the upper limit of the extended duration is exhausted, the identification conclusion is forcibly changed to a real device fault, the extended waiting phase is terminated, and the corresponding reverse power-down sequence is triggered.

[0055] If the above forced change conditions are not triggered, calculate the amount by which the measured response delay exceeds the adaptive waiting upper bound. Based on the ratio of excess to the standard deviation of the residuals Determine the reduction weight This value quantifies the extreme degree of the current delay. A larger value indicates a greater deviation from the normal distribution of delay, thus requiring a smaller actual update weight for the EWMA; conversely, a smaller value indicates a closer approximation of the normal distribution of delay, thus requiring a larger actual update weight for the EWMA. Specifically, The relation is:

[0056] in The attenuation slope coefficient is determined during the factory calibration phase by constraining the EWMA contamination level of historical slow-start samples in each partition. For example, It can be 1. When hour , The larger, The smaller the value, the weaker its contribution to the bias of EWMA; When approaching zero, Approaching 1, it degenerates into normal full-weighted write-back. The mean is updated and written back using reduced weights, i.e., the update is performed with... Replace the original Perform EWMA writeback and update the residual window with the original weights to ensure It can accurately reflect the degree of time delay dispersion under cold conditions. Timing execution resumes after the target equipment completes readiness confirmation, completing differentiated timing processing. Furthermore, the main controller will extend the waiting trigger conditions (thermal pretreatment status flag, ...) for this operation. , , Add it to the aging status tracking log to provide a basis for subsequent judgment on whether the equipment has reached the preventive maintenance node.

[0057] In response to the identification conclusion of a genuine equipment failure, the main controller sequentially deactivates the enabled devices that have completed the readiness process according to a preset safety sequence to execute a reverse power-down sequence: the electromagnetic brake, as the last device in the power-up sequence, is applied first in the reverse sequence to ensure that the mechanical side enters the braking and locking state in the shortest possible time. Simultaneously, the main controller extracts the faulty device number, the fault determination time, and feature summaries of the first-order and second-order difference sequences, encapsulates them into a diagnostic data package, and includes the current baseline parameters used to calculate the adaptive waiting upper bound. , , and the status indicators of the heat pretreatment at that time Push diagnostic data packets to the host computer to allow maintenance personnel to reconstruct the complete operating conditions and equipment aging status at the time of the fault, assisting in root cause analysis. Simultaneously, prohibit writing the latency data corresponding to this fault into the operating condition partition grid to avoid abnormal fault data contaminating the normal baseline distribution and to complete differentiated timing handling.

[0058] like Figure 2 As shown in the figure, this is a comparison diagram of response delay aging drift tracking and waiting upper bound margin in an embodiment of the present invention. It can be seen that the existing method has a larger aging drift trend slope, and the waiting upper bound margin band is generally wider and the center is lower. However, the method of this application, through the introduction of adaptive forgetting factor and aging forward compensation term, makes the trend slope controlled and the waiting upper bound margin band more compact and in line with the actual aging process, effectively reducing the risk of erroneous timeout judgment caused by systematic low margin.

[0059] In this way, by adopting differentiated handling paths for the three types of judgment conclusions, namely, immediate advancement, reduction and write-back with suspension command, and reverse power-down with diagnostic reporting, the main controller can ensure that the partition baseline is not polluted by abnormal data in the long term, while continuously accumulating aging status information into the operation and maintenance log, so that the full life cycle time-series margin management and preventive maintenance judgment of the equipment form a self-consistent data loop.

Claims

1. A multi-device timing collaborative control method for a hoisting power supply main controller, characterized in that, include: The interval between the current power-on and the previous normal power-off of the target device is obtained. Based on the interval, the thermal pre-processing status flag is determined. The operating condition is divided into grids by combining the load rate and the grid voltage deviation rate. Within the operating condition grids, the mean of the response delay is updated based on the exponential weighted moving average algorithm. The residual standard deviation and aging trend slope are obtained simultaneously. When issuing an enable command to the target device, the adaptive waiting upper bound of the current timing step is calculated based on the mean, the residual standard deviation, and the aging trend slope. Within the window corresponding to the upper bound of the adaptive waiting, the ready feedback register of the target device is continuously sampled to construct a state sequence. If the target device has not completed the ready confirmation when the window expires, the last segment of the state sequence is extracted to calculate the first-order difference sequence and the second-order difference sequence. Based on the first-order difference sequence and the second-order difference sequence, it is determined whether the target device is in an aging cold start condition with slow response or a real device failure. Based on the conclusion that the slow response is due to aging cold start or a real equipment failure, differentiated timing processing is performed on the target equipment, and the average value in the operating condition partition is refreshed using a differentiated baseline write-back method.

2. The multi-device timing collaborative control method for a hoisting power supply main controller according to claim 1, characterized in that, The determination of the heat pretreatment status flag based on the interval duration includes: Collect the average startup response delay of the same model of equipment under different downtime intervals, and plot the response curve of the average startup response delay as a function of the downtime interval; Calculate the absolute value of the slope between adjacent gradients of the response curve, and take the shutdown interval corresponding to the first drop of the absolute value of the slope below the peak value as the hot and cold dividing point; The interval duration is compared with the hot-cold boundary point, and the heat pretreatment status indicator is determined based on the comparison result.

3. The multi-device timing collaborative control method for a hoisting power supply main controller according to claim 1, characterized in that, The step of updating the mean response delay based on the exponentially weighted moving average algorithm within the operating condition grid includes: Calculate the first-order difference mean of the residual sequence within the working condition partition grid, and normalize the first-order difference mean using the residual standard deviation to obtain a dimensionless drift intensity index. Based on the percentile distribution of the dimensionless drift intensity index in the historical sequence, the forgetting factor is determined by linear interpolation. The mean of the measured response delay is obtained by weighting and summing the mean of the measured response delay with the mean of the previous time step using the forgetting factor.

4. The multi-device timing collaborative control method for a hoisting power supply main controller according to claim 1, characterized in that, The calculation of the adaptive waiting upper bound for this time series step based on the mean, the residual standard deviation, and the aging trend slope includes: The aging forward compensation term is calculated based on the aging trend slope and the number of forward compensation steps; The dispersion compensation term is calculated based on the residual standard deviation and the benchmark multiple. The mean, the aging forward compensation term, and the dispersion compensation term are summed to obtain the adaptive waiting upper bound.

5. The multi-device timing collaborative control method for a hoisting power supply main controller according to claim 1, characterized in that, The step of continuously sampling the ready feedback register of the target device within the window corresponding to the upper bound of the adaptive waiting to construct a state sequence includes: Obtain the minimum physical change time constant of the target device ready signal, and determine a fixed sampling period based on the minimum physical change time constant; Using the fixed sampling period as the time interval, the ready feedback register of the target device is continuously sampled within the window corresponding to the upper bound of the adaptive waiting; The progress ratio values ​​obtained from continuous sampling are arranged in chronological order to construct the state sequence.

6. The multi-device timing collaborative control method for a hoisting power supply main controller according to claim 1, characterized in that, The method of identifying whether the target device is in an aging cold start condition with slow response or a real device fault based on the first-order difference sequence and the second-order difference sequence includes: The percentage of terms with values ​​greater than or equal to zero in the first-order difference sequence is counted, and the mean of the second-order difference sequence is calculated. Determine whether the percentage of the number of items is not lower than the threshold for the percentage of continuous progress, whether the mean is less than zero, and whether the last value of the last sub-segment is not lower than the threshold for the last value. If all three conditions above are met, the target device is identified as being in the aging cold start condition with slow response. If any one of the above three conditions is not met, then the target device is identified as being in the form of a real device malfunction.

7. The multi-device timing collaborative control method for a hoisting power supply main controller according to claim 1, characterized in that, Based on the conclusion that the slow response is due to aging cold start or a genuine equipment failure, differentiated timing processing is performed on the target equipment, including: In response to the identification conclusion that the aging cold start condition is a slow response, the timing command of all subsequent unstarted devices is suspended during the extended waiting phase. Calculate the amount by which the measured response delay exceeds the adaptive waiting upper bound, and determine the reduction weight based on the ratio of the excess amount to the residual standard deviation; The mean is updated and written back using the reduction weight. Once the target device is confirmed to be ready, the timing process resumes, and the differentiated timing processing is completed.

8. The multi-device timing collaborative control method for a hoisting power supply main controller according to claim 1, characterized in that, Based on the conclusion that the slow response is due to aging cold start or a genuine equipment failure, differentiated timing processing is performed on the target equipment, including: In response to the identification conclusion that the actual equipment is faulty, the enabled devices that have been completed and ready are deactivated in sequence according to a preset safety sequence to execute the reverse power-down sequence; Extract the faulty equipment number, the fault determination time, and the feature summaries of the first-order difference sequence and the second-order difference sequence, and encapsulate them into a diagnostic data package; The diagnostic data packet is pushed to the host computer, and the writing of the delay data corresponding to this fault into the operating condition partition is prohibited, thus completing the differentiated timing processing.

9. A multi-device timing collaborative control method for a hoisting power supply main controller according to claim 1, characterized in that, The simultaneous acquisition of residual standard deviation and aging trend slope includes: Calculate the difference between the measured response delay and the mean to obtain the residual; The residuals are stored in a fixed-length residual window, and the standard deviation of all residuals in the fixed-length residual window is calculated to obtain the standard deviation of the residuals. The measured response delay within the fixed-length residual window is subjected to linear regression processing, and the linear regression slope is extracted as the aging trend slope.

10. A multi-device timing collaborative control method for a hoisting power supply main controller according to claim 7, characterized in that, The timing command to suspend all subsequent unstarted devices during the extended waiting phase includes: During the extended waiting phase, the status value of the ready feedback register of the target device is continuously monitored; If the status value of a preset number of consecutive sampling points decreases, or if no ready confirmation signal is received when the upper limit of the extended duration is exhausted, the identification conclusion will be forcibly changed to the actual equipment failure. The extended waiting phase is terminated, and the corresponding reverse power-down sequence is triggered.

Images