Large-scale industrial door motor operation state fault prediction method and system

Abstract

The invention relates to the technical field of fault prediction and health management, in particular to a large industrial door motor operation state fault prediction method and system. The method comprises the following steps: collecting high-frequency multi-source driving data of a dual-motor driving system and carrying out working condition segmentation preprocessing; performing space-time alignment on the current sequences of the double motors by using a dynamic time warping algorithm, and eliminating phase deviation caused by a mechanical transmission gap; constructing a torque confrontation entropy characteristic index based on the aligned current data in combination with a load level and a local fluctuation characteristic; and constructing a cumulative damage model based on the torque confrontation entropy and the working condition stress weight, evaluating the residual health degree of the equipment, and carrying out fault early warning. According to the method, the problem of phase asynchronous misinformation caused by a mechanical transmission gap is solved, implicit torque confrontation between motors can be accurately identified, and fault prediction and condition-based maintenance in the whole life cycle are realized.

Description

Technical Field

[0001] This invention relates to the field of fault prediction and health management technology. More specifically, this invention relates to a method and system for predicting the operating status faults of a large industrial door motor. Background Technology

[0002] Large industrial doors, such as aircraft hangar doors and shipyard extra-wide overhead doors, typically employ dual-motor or multi-motor coordinated drive systems due to their significant structural characteristics, including large spans and heavy weight. This ensures redundancy in driving force and dynamic balance during operation. Existing drive motor monitoring technologies generally utilize traditional single-motor protection logic. This logic primarily monitors the operating parameters of individual motors, such as current or temperature, and only triggers an alarm mechanism when the monitored values ​​exceed preset rated thresholds, thereby ensuring basic system operational safety.

[0003] This traditional monitoring method has blind spots in practical engineering applications. Due to objective physical factors such as uneven wear of the reducer and differences in track resistance, dual-motor drive systems are prone to torque conflict, i.e., an abnormal state where one motor outputs power while the other is being driven. In this state, the operating current of each motor often does not reach the overcurrent alarm threshold, making it impossible for existing single-motor protection logic to identify such faults. This results in the system consuming unnecessary electrical energy for a long time without alarms and accelerating the fatigue wear of the mechanical structure.

[0004] To compensate for the shortcomings of single-machine monitoring, while methods that directly calculate the current difference between motors can be improved, they still suffer from false alarms in practical applications. Because large industrial doors have long mechanical transmission chains, the physical gaps between gears and couplings cause phase shifts between the two motors during start-up and stop. If the current difference is directly calculated, this transient phase shift will be converted into a false high-difference signal during start-up and stop, leading to misjudgment and alarm triggering by the monitoring system. Therefore, it is impossible to accurately capture hidden torque interference while eliminating phase shift interference, making it difficult to meet the needs of full lifecycle health assessment. Summary of the Invention

[0005] The purpose of this invention is to propose a method and system for predicting the operating status faults of large industrial door motors. By dynamically aligning multi-source drive data and constructing physical countermeasure indicators, accurate prediction of the motor's operating status can be achieved. This invention provides solutions in the following two aspects.

[0006] In a first aspect, the present invention provides a method for predicting the operating status faults of a large industrial door motor, comprising:

[0007] Multi-source drive data of a dual-motor drive system for a large industrial door is acquired. The effective current sequences of motor A and motor B are synchronously collected using current sensors. Based on the speed signal, the continuous time-series data is segmented into analysis samples of single-run events. The current sequence of one motor is selected as a reference sequence. A dynamic time warping algorithm is used to map the current sequence of the other motor onto the time axis of the reference sequence, generating an aligned current sequence. Based on the instantaneous imbalance, total load level, and local fluctuations between the reference sequence and the aligned current sequence, a torque resistance entropy characterizing the degree of internal friction between the motors is constructed. The expression for the torque resistance entropy is: In the formula, For the index variable of the runtime cycle, For the first The torque resistance entropy calculated in this run To analyze the total number of sampling points in the sample, Let be the effective current sequence of motor A at time t. Let be the aligned motor B current at time t. Let be the instantaneous average current of the two motors at time t. As a reference constant, The fluctuation sensitivity coefficient, The standard deviation of the current difference within a local sliding window near time t is given; based on the torque resistance entropy and the working stress weight of a single operating event, a device health degradation model is constructed, the remaining health score of the system is calculated, and a fault warning is given based on the remaining health score.

[0008] Preferably, the acquisition of multi-source drive data of the dual-motor drive system of a large industrial door further includes: acquiring a real-time speed sequence through a rotary encoder and acquiring a door travel position sequence through a displacement sensor; using the speed sequence to slice the data stream, extracting the current sequence within the motor running time period as a complete analysis sample, and performing sliding median filtering on the original current data.

[0009] Preferably, the step of segmenting continuous time series data into analysis samples of single running events based on rotational speed signals includes: monitoring the real-time rotational speed sequence; marking the start of a running event when the rotational speed sequence is detected to be greater than zero for N consecutive periods; marking the end of a running event when the rotational speed sequence returns to zero and remains stable; and extracting the current sequence from the start to the end of the running event as a complete analysis sample.

[0010] Preferably, the step of mapping the current sequence of another motor onto the time axis of a reference sequence using a dynamic time warping algorithm includes: constructing a cumulative distance matrix, where the elements of the matrix represent the Euclidean distance between data points in the reference sequence and data points in the sequence to be aligned; finding a shortest warping path in the cumulative distance matrix that satisfies boundary, continuity, and monotonicity constraints, such that the cumulative distance of all points on the path is minimized; and nonlinearly mapping the data points in the sequence to be aligned onto the time axis of the reference sequence according to the shortest warping path.

[0011] Preferably, the standard deviation within the local sliding window is obtained by setting a length of... Calculate the absolute value sequence of current differences within a time window. The standard deviation is used to characterize the intensity of the interaction between motors.

[0012] Preferably, the expression for the system's remaining health score in the equipment health decline model is: In the formula, Assess the system's remaining health. As the initial health of the system, For the index variable of the runtime cycle, For the first The torque resistance entropy calculated in this run For the first Stress weights under operating conditions for each run As the baseline damage rate, This is the fatigue acceleration index.

[0013] Preferably, the expression for the stress weight under the working condition is: In the formula, For the first Stress weights under operating conditions for each run For the first The total distance the gate moves in each operation. This refers to the duration during which the motor operates at high torque output but with minimal displacement. This is the stall damage conversion coefficient.

[0014] Preferably, the fault warning based on the remaining health score includes: using the isolated forest algorithm model to detect the torque resistance entropy generated in the historical operating cycle and identify anomalies; issuing an early maintenance warning when the remaining health score is lower than a preset warning line; and triggering a shutdown lockout command when the torque resistance entropy is determined to be a serious anomaly and the remaining health score shows a sharp drop with a rapidly increasing rate of decline.

[0015] In the second aspect, a fault prediction system for the operating status of a large industrial door motor includes: The processor; the memory storing computer instructions for predicting the operating status faults of large industrial door motors, which, when executed by the processor, cause the system to perform a method for predicting the operating status faults of large industrial door motors.

[0016] The beneficial effects of this invention are as follows: By introducing a dynamic time warping algorithm, this invention solves the problem of false alarms due to phase asynchrony caused by mechanical backlash; by constructing a torque counter-entropy model that integrates load rate, difference ratio, and volatility, it achieves accurate capture of minor internal friction faults; by establishing a cumulative damage model based on torque counter-entropy and operating stress, it achieves quantitative assessment and scientific early warning of the entire equipment life cycle, thereby improving the safety and reliability of industrial door operation. Attached Figure Description

[0017] Figure 1 The flowchart illustrating the steps of the large industrial door motor operation status fault prediction method in this embodiment is shown in the schematic diagram. Figure 2 This is a schematic diagram comparing the processing results of the existing direct interpolation method and the method of the present invention; Figure 3 It is a device lifecycle health prediction curve based on torque-resistance entropy. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0019] like Figure 1 As shown in this embodiment, a method for predicting the operating status faults of a large industrial door motor includes the following steps: Step S1: Obtain multi-source drive data of the dual-motor drive system of a large industrial door, and segment the continuous time series data into analysis samples of single running events based on the speed signal.

[0020] In this step, we take a large industrial door driven by two motors as an example, and use an industrial fieldbus or analog data acquisition card to acquire data from the system. The motors are designated as motor A and motor B.

[0021] Specifically, the effective current sequences of motor A and motor B are synchronously acquired at a high sampling rate of, for example, 100Hz, using Hall current sensors installed at the output of the frequency converter or the stator of the motor. Simultaneously, a real-time speed sequence is acquired using a rotary encoder installed at the tail of the motor, and a door travel position sequence is acquired using a pull rope displacement sensor installed at the end of the door.

[0022] Subsequently, the continuous data stream is segmented and cleaned based on rotational speed. For example, when the rotational speed sequence has N consecutive periods greater than zero, it is marked as the start of an event; when the rotational speed sequence returns to zero and remains stable, it is marked as the end of an event. The effective current sequences of motors A and B within this time period are extracted as a complete analysis sample. Simultaneously, a sliding median filter is used to preprocess the raw current data, filtering out high-frequency glitches and noise caused by the inverter's switching frequency. In this embodiment, N is set to 5.

[0023] In this way, by using high-frequency synchronous acquisition and motion state-based working condition segmentation, effective data segments of the motor during actual operation can be accurately extracted, invalid data interference during the static waiting period can be eliminated, and the true load change trend can be preserved, providing a high-quality data foundation for subsequent analysis.

[0024] Step S2: Select the current sequence of one motor as the reference sequence, and use the dynamic time warping algorithm to map the current sequence of the other motor onto the time axis of the reference sequence to generate the aligned current sequence.

[0025] Due to the presence of mechanical transmission clearances such as gear clearance and coupling clearance, motor A and motor B have slight asynchrony in physical operation. Directly subtracting the current at the same time point will produce false high differences during the start-up and braking phases.

[0026] To solve the above problem, specifically, the effective value current sequence of motor A is selected as the reference sequence. The effective current sequence of motor B As a sequence to be aligned.

[0027] First, construct a cumulative distance matrix, where the elements of the matrix are... express The Middle Data at each time point and The Middle Euclidean distance between data points at different time points.

[0028] Subsequently, dynamic programming is used to find the shortest regular path from the starting point to the ending point of the matrix. This path satisfies the constraints of boundary, continuity, and monotonicity, minimizing the cumulative distance between all points on the path. In essence, this path finds the optimal matching relationship between the two sequences in terms of peak and trough waveform characteristics.

[0029] Finally, based on the found optimal regularization path, The data points in the nonlinear mapping are mapped to On the time axis, an aligned sequence of motor B currents is generated, denoted as... .

[0030] For example, suppose that at the moment of startup, motor A is in The current reaches its peak value, while motor B lags due to transmission clearance. It only reaches its peak value at that time. Without alignment, in... The system will calculate a huge difference, and using the DTW algorithm, it will adjust motor B accordingly. Data point mapping aligned to motor A This eliminates misalignment on the timeline.

[0031] In this way, through the dynamic time warping algorithm, the feature points of the two motors on the current waveform are forced to align, effectively eliminating the phase difference interference caused by mechanical backlash, avoiding false alarms during the start-stop phase, and ensuring that the differences in subsequent calculations truly reflect the physical-level confrontation.

[0032] Step S3: Based on the instantaneous imbalance between the reference sequence and the aligned current sequence, the total load level, and the local fluctuations, construct the torque resistance entropy that characterizes the degree of internal friction between motors.

[0033] After data alignment is completed, it is necessary to quantify the internal friction between the two motors. This embodiment constructs a composite physical index that integrates load rate, difference ratio, and volatility—torque-resistance entropy.

[0034] The expression for torque versus entropy is: ; In the formula, For the index variable of the runtime cycle, For the first The torque resistance entropy calculated in this run To analyze the total number of sampling points in the sample, Let be the effective current sequence of motor A at time t. Let be the aligned motor B current at time t. Let be the instantaneous average current of the two motors at time t. As a reference constant, The fluctuation sensitivity coefficient, Let be the standard deviation of the current difference within a local sliding window around time t.

[0035] in, Represents instantaneous imbalance; This represents the current total load level of the system and is used as the denominator for load normalization. The method for obtaining it is: set a length of Calculate the sequence within the time window. The standard deviation. In this embodiment, Take 5% of the motor's rated current to prevent the denominator from approaching 0 and causing the calculation to diverge; The value is 1.0. The value is 20.

[0036] Thus, by introducing a torque-against-entropy index, not only is the relative magnitude of the current difference considered, but the volatility of the difference is also amplified through an exponential weighting term. If there are intermittent impacts between motors caused by gear breakage or track deformation, the difference will fluctuate drastically, leading to... As the torque counteraction entropy increases, the torque counteraction entropy value increases significantly, thus enabling sensitive detection of latent mechanical counteraction faults.

[0037] Step S4: Based on the torque resistance entropy and the working stress weight of a single operating event, construct a device health degradation model, calculate the remaining health score of the system, and issue a fault warning based on the remaining health score.

[0038] Failures in industrial door mechanical transmission systems are usually cumulative effects. This step uses the torque-resistance entropy obtained from S3 to construct a health degradation model.

[0039] The expression for the system's remaining health score is: ; In the formula, For the equipment to complete the first The remaining health score after the next run (set to 0 if it is less than 0). As the initial health of the system, For the index variable of the runtime cycle, For the first The torque resistance entropy calculated in this run For the first Stress weights under operating conditions for each run As the baseline damage rate, The fatigue acceleration index (typically greater than 1) is used. In this embodiment, the initial health of the system is set to 100; the fatigue acceleration index is 1.5, which is used to reflect the nonlinear loss of lifespan due to high-intensity resistance.

[0040] For the first The stress weight for the current operating condition is expressed as follows: ; In the formula, For the first The total distance the gate moves in each operation (calculated from the position sequence); This refers to the duration during which the motor is in high torque output but with minimal displacement, used to capture stall conditions. This is the stall damage conversion coefficient.

[0041] The early warning logic is as follows: Isolation forest algorithm model is used to identify torque anti-entropy anomalies. When... When the temperature falls below the preset warning line, the system will display a message indicating "increased mechanical wear"; when the torque-to-entropy ratio is determined to be severely abnormal and... During a sudden drop, a shutdown and lockout command is triggered. In this embodiment, the preset warning threshold is set at 80%.

[0042] Figure 2 This demonstrates a comparison between the results of the existing direct interpolation method and the processing results of this invention. From Figure 2 As can be seen in (a), during the 0-2.5 second startup phase, the existing technology generates a huge spurious peak due to phase lag; from Figure 2 As can be seen from (b) in the present invention, after DTW alignment and TCE calculation, the interference waveform during the start-stop phase is corrected, and only the true mechanical internal friction characteristics during the middle of operation (about 10 seconds) are retained, which proves the ability of the present invention to eliminate interference and accurately lock.

[0043] Figure 3 The system displays a health prediction curve for the entire lifecycle of the equipment. As the operating cycle increases, the torque resistance entropy shown in the bar chart increases non-linearly, while the remaining system health shown in the line chart decreases accordingly. When the health drops below 80%, the system triggers an early warning.

[0044] Thus, by establishing a cumulative damage model that includes operating stress and fatigue acceleration factors, a single operating state can be transformed into a visualized long-term health trend, effectively correcting the error of simply using distance or number of times to assess lifespan, and realizing scientific management and condition-based maintenance of equipment throughout its entire life cycle.

[0045] This invention also provides a fault prediction system for the operating status of a large industrial door motor. The system includes a processor and a memory, the memory storing computer program instructions. When the processor executes the computer program instructions, it implements the fault prediction method for the operating status of a large industrial door motor according to the present invention.

[0046] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and therefore will not be described in detail here.

[0047] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device. Any application or module described in this invention can be implemented by computer-readable / executable instructions stored or otherwise maintained on such a computer-readable medium.

[0048] In the description of this specification, "multiple" means at least two, such as two, three or more, etc., unless otherwise expressly and specifically defined.

[0049] While various embodiments of the invention have been shown and described in this specification, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention.

Claims

1. A method for predicting faults in the operating status of a large industrial door motor, characterized in that, include: The system acquires multi-source drive data of a dual-motor drive system for large industrial doors. It synchronously collects the effective current sequence of motor A and the effective current sequence of motor B through current sensors, and divides the continuous time series data into analysis samples of single running events based on the speed signal. The current sequence of one motor is selected as the reference sequence, and the current sequence of the other motor is mapped onto the time axis of the reference sequence using the dynamic time warping algorithm to generate the aligned current sequence. Based on the instantaneous imbalance, total load level, and local fluctuations between the reference sequence and the aligned current sequence, a torque resistance entropy is constructed to characterize the degree of internal friction between motors. The expression for the torque-resistance entropy is: In the formula, For the index variable of the runtime cycle, For the first The torque resistance entropy calculated in this run To analyze the total number of sampling points in the sample, Let be the effective current sequence of motor A at time t. Let be the aligned motor B current at time t. Let be the instantaneous average current of the two motors at time t. As a reference constant, The fluctuation sensitivity coefficient, This represents the standard deviation of the current difference within a local sliding window around time t; Based on the torque resistance entropy and the working stress weight of a single operating event, a device health degradation model is constructed, the remaining health score of the system is calculated, and a fault warning is given based on the remaining health score.

2. The method for predicting the fault status of a large industrial door motor according to claim 1, characterized in that, The acquisition of multi-source drive data for a dual-motor drive system of a large industrial door also includes: Real-time rotation speed sequence is acquired by a rotary encoder, and door travel position sequence is acquired by a displacement sensor; The data stream is sliced ​​using the speed sequence, and the current sequence within the motor's operating time period is extracted as a complete analysis sample. The original current data is then subjected to sliding median filtering.

3. The method for predicting the fault status of a large industrial door motor according to claim 1, characterized in that, The method of segmenting continuous time series data into single running events based on speed signals includes: monitoring the real-time speed sequence; marking the start of a running event when the speed sequence is detected to be greater than zero for N consecutive periods; marking the end of a running event when the speed sequence returns to zero and remains stable; and extracting the current sequence from the start to the end of the running event as a complete analysis sample.

4. The method for predicting the fault status of a large industrial door motor according to claim 1, characterized in that, The method of mapping the current sequence of another motor onto the time axis of a reference sequence using a dynamic time warping algorithm includes: Construct a cumulative distance matrix, where the elements represent the Euclidean distance between data points in the reference sequence and data points in the sequence to be aligned; Find the shortest regular path in the cumulative distance matrix that satisfies the constraints of boundary, continuity and monotonicity, such that the cumulative distance of all points on the path is minimized; Based on the shortest normalization path, the data points in the sequence to be aligned are nonlinearly mapped onto the time axis of the reference sequence.

5. The method for predicting the fault status of a large industrial door motor according to claim 2, characterized in that, The standard deviation within the local sliding window is obtained as follows: Set a length of Calculate the absolute value sequence of current differences within a time window. The standard deviation is used to characterize the intensity of the interaction between motors.

6. The method for predicting the fault status of a large industrial door motor according to claim 1, characterized in that, The expression for the system's remaining health score in the equipment health decline model is as follows: ； In the formula, Assess the system's remaining health. As the initial health of the system, For the index variable of the runtime cycle, For the first The torque resistance entropy calculated in this run For the first Stress weights under operating conditions for each run As the baseline damage rate, This is the fatigue acceleration index.

7. The method for predicting the fault status of a large industrial door motor according to claim 6, characterized in that, The expression for the stress weight under the working condition is: ； In the formula, For the first Stress weights under operating conditions for each run For the first The total distance the gate moves in each operation. This refers to the duration during which the motor operates at high torque output but with minimal displacement. This is the stall damage conversion coefficient.

8. The method for predicting the fault status of a large industrial door motor according to claim 1, characterized in that, The fault warning based on the remaining health score includes: The isolated forest algorithm model is used to detect the torque anti-entropy generated in historical operating cycles and identify outliers; When the remaining health score falls below the preset warning line, an early maintenance warning will be issued. When the torque resistance entropy is determined to be a serious anomaly and the remaining health score shows a sharp drop with a rapidly increasing rate of decline, a shutdown lockout command is triggered.

9. A fault prediction system for the operating status of a large industrial door motor, characterized in that, include: processor; A memory storing computer instructions for predicting the operating status faults of large industrial door motors, which, when executed by the processor, cause the system to perform the method for predicting the operating status faults of large industrial door motors according to any one of claims 1-8.

Images