A port equipment fault prediction method and system based on edge computing

Abstract

The present application relates to the technical field of port equipment fault monitoring, and particularly relates to a port equipment fault prediction method and system based on edge computing. In the calculation of the load coefficient, the baseline load corresponding to the port equipment is first obtained, then the load deviation of the port equipment is calculated, and finally the load coefficient is calculated according to the load deviation. The calculation process integrates multi-dimensional physical parameters and time series trends, so that the calculation result of the load coefficient can overcome the problems of time variation and strong variability of the port equipment.

Description

Technical Field

[0001] This invention relates to the field of port equipment fault monitoring technology, and in particular to a port equipment fault prediction method and system based on edge computing. Background Technology

[0002] As a core hub of global logistics, ports require 24 / 7 continuous operation of equipment such as quay cranes, yard cranes, AGVs, and conveyors, often in harsh working environments. Currently, port equipment failure prediction mainly relies on centralized analysis, requiring all sensor data to be uploaded to remote servers, which suffers from network latency, high data transmission costs, and significant privacy and security risks. Furthermore, cloud computing resources are limited by network transmission speeds, making it difficult to achieve millisecond-level real-time responses. Edge computing, by deploying computing nodes at the device or in the near field, can complete data preprocessing and preliminary analysis locally, uploading only key features or early warning results to the cloud, offering advantages such as real-time performance, low bandwidth consumption, and high reliability.

[0003] In the existing technology, there is a scheme to predict port equipment failures by using load factors, but the existing methods for determining load factors have the problem of low accuracy. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a port equipment fault prediction method and system based on edge computing, which solves the problems existing in the prior art.

[0005] This invention provides a port equipment fault prediction method based on edge computing, comprising the following steps:

[0006] S1: Collect port equipment operation data through multi-source sensors;

[0007] S2: Perform data preprocessing operations on the port equipment operation data at the edge node;

[0008] S3: Extract edge features from the preprocessed port equipment operation data;

[0009] S4: Compare each edge-side feature with the corresponding fault threshold. If the fault threshold is exceeded, the port equipment is preliminarily determined to be in a fault state; and proceed to the next step.

[0010] The formula for calculating the fault threshold is:

[0011] ;

[0012] In the formula, T dynamicThe dynamic threshold of the edge-side feature at the current time, μ is the rolling mean of the edge-side feature over the past preset time window, σ is the rolling standard deviation of the edge-side feature over the same preset time window, and L is the dynamic threshold of the edge-side feature at the current time. final α represents the load factor of the port equipment, and β are adjustable weighting parameters.

[0013] The method for determining the load factor is as follows:

[0014] Sa: Perform time-series baseline modeling for each port equipment to obtain the baseline load corresponding to that port equipment;

[0015] Sb: Calculate the load deviation based on the baseline load of the port equipment;

[0016] Sc: Calculate the load factor corresponding to the port equipment based on the load deviation;

[0017] S5: The edge node transmits the operating data of the port equipment to the cloud server to realize the fault prediction of the port equipment.

[0018] Preferably, the load factor The calculation formula is:

[0019] ;

[0020] In the formula, The weight coefficient for the i-th edge-side feature of the port equipment. This is the normalized value of the i-th edge-side feature of the port equipment. These are the time series weighting coefficients. This refers to the load deviation value of the port equipment. The fault weights of the port equipment are... The load values ​​associated with historical faults of the port equipment.

[0021] Preferably, Sa specifically involves: acquiring edge-side feature data of the port equipment in the same time period over the past 7 days, constructing a multivariate time series sample, and then inputting the edge-side feature data of the port equipment in the same time period over the past 7 days into a lightweight long short-term memory network model to predict the edge-side feature data of the port equipment for the next day as the baseline load of the port equipment.

[0022] Preferably, the input layer dimension of the lightweight long short-term memory network model is the product of the number of edge-side feature classes and the time step; the hidden layer has 2 layers with 64 and 32 neurons respectively; the output of the output layer is the edge-side feature data of the port equipment for the next day; the loss function is MAE and the optimizer is Adam.

[0023] Preferably, Sb specifically involves: calculating the Mahalanobis distance between the real-time edge-side feature data of the port equipment and the baseline load of the port equipment, as the load deviation of the port equipment.

[0024] Preferably, in step S1, the multi-source sensor includes a vibration sensor, a temperature sensor, a current / voltage sensor, and a displacement sensor.

[0025] Preferably, in step S2, the preprocessing operation includes: noise reduction filtering and missing value compensation.

[0026] Preferably, the denoising filtering specifically involves: using discrete wavelet transform, selecting the db4 wavelet basis, and performing a 5-level decomposition on the port equipment operation data: levels 1-3 are low-frequency components; levels 4-5 are high-frequency components; and then using a soft thresholding method to filter out random noise from the high-frequency coefficients of levels 4-5 before reconstructing the signal.

[0027] Specifically, the missing value compensation is as follows: for single-point missing values, linear interpolation is used to compensate for missing values; for continuous multi-point missing values, the data in that time period is marked as invalid and a local alarm is triggered. Edge nodes will prompt through a buzzer or send sensor abnormal information to the operation and maintenance platform.

[0028] Preferably, in step S3, time-domain and frequency-domain features are extracted from the vibration data in the preprocessed port equipment operation data. The time-domain features include root mean square value, peak factor, and kurtosis. The frequency-domain features include rotation frequency and its harmonic components, sideband energy proportion, and frequency band energy distribution. Temperature features are extracted from the temperature data in the preprocessed port equipment operation data. The temperature features include steady-state temperature value, temperature rise rate, and temperature difference ratio. Current features are extracted from the current data in the preprocessed port equipment operation data. The current data features include current RMS value and current harmonic features. Pose features are extracted from the pose data in the preprocessed port equipment operation data. The pose features include displacement variance and lifting speed deviation.

[0029] According to another aspect of the present invention, a port equipment failure prediction system based on edge computing is provided. The system employs the aforementioned port equipment failure prediction method based on edge computing, and the system includes:

[0030] The data acquisition module is used to collect port equipment operation data through multi-source sensors;

[0031] The data preprocessing module is used to perform data preprocessing operations on the port equipment operation data at the edge node;

[0032] The feature extraction module is used to extract edge features from the preprocessed port equipment operation data;

[0033] The preliminary fault status determination module is used to compare each edge-side feature with the corresponding fault threshold. If the fault threshold is exceeded, the port equipment is preliminarily determined to be in a fault state, and then proceeds to the next step.

[0034] The fault prediction module is used to transmit the operating data of the port equipment to the cloud server to realize fault prediction of the port equipment.

[0035] The embodiments of the present invention have the following technical effects:

[0036] When calculating the load factor, this invention first aligns with the baseline load of the port equipment, then calculates the load deviation of the port equipment, and finally calculates the load factor based on the load deviation. The calculation process integrates multi-dimensional physical parameters and time series trends, enabling the load factor calculation results to overcome the problems of time variability and high variability of port equipment. Attached Figure Description

[0037] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0038] Figure 1 This is a flowchart of a port equipment fault prediction method based on edge computing provided in an embodiment of the present invention;

[0039] Figure 2 This is a flowchart of the method for determining the load factor provided in an embodiment of the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0041] Figure 1 A flowchart of a port equipment failure prediction method based on edge computing is shown, such as... Figure 1 As shown, a port equipment fault prediction method based on edge computing includes the following steps:

[0042] S1: Collect port equipment operation data through multi-source sensors;

[0043] The multi-source sensors include vibration sensors, temperature sensors, current / voltage sensors, and displacement sensors. The vibration sensor is an accelerometer with a sampling rate of 10kHz to 50kHz, used to monitor the vibration amplitude and frequency of rotating components such as bearings and gearboxes of the port equipment. The temperature sensor is a PT100 thermocouple with a sampling rate of 1Hz to 10Hz, used to monitor the temperature of the motor windings, bearing housings, and hydraulic oil of the port equipment. The current sensor has a sampling rate of 1kHz to 10kHz and is used to monitor the stability of the motor drive current of the port equipment. The displacement sensor is used to monitor mechanical position parameters of the port equipment, such as the lifting height of the spreader and the amount of conveyor belt deviation.

[0044] An edge computing gateway is installed in the control cabinet of the port equipment. The edge computing gateway is an industrial-grade gateway equipped with an ARM Cortex-A72 / Intel Xeon D processor, with a computing power of ≥4 TOPS, supporting 5G / Ethernet / Wi-Fi 6 communication, and collecting port equipment operation data in real time through CAN bus, Modbus RTU / TCP or IO-Link protocol.

[0045] S2: Perform data preprocessing operations on the port equipment operation data at the edge node;

[0046] Data preprocessing is a crucial step in edge computing to ensure the accuracy of subsequent fault feature extraction and analysis. The port equipment operation data collected by the multi-source sensors often suffers from noise interference, random fluctuations, inconsistent sampling rates, and occasional missing values. If these problems are not properly addressed, they will directly affect the reliability of port equipment fault prediction.

[0047] The preprocessing operations include: noise reduction filtering and missing value compensation.

[0048] Specifically, the denoising filtering involves: using Discrete Wavelet Transform (DWT), selecting a suitable db4 wavelet basis for non-stationary signals, and performing a 5-level decomposition on the port equipment operation data: Levels 1-3 are low-frequency components, retaining the fundamental frequency information of normal equipment operation; Levels 4-5 are high-frequency components, focusing on analyzing fault characteristics; and then the high-frequency coefficients of Levels 4-5 are filtered out for random noise using a soft thresholding method before reconstructing the signal.

[0049] Specifically, the missing value compensation is as follows: for single-point missing values, linear interpolation is used to compensate for missing values; for continuous multi-point missing values, the data in that time period is marked as invalid and a local alarm is triggered. Edge nodes will prompt through a buzzer or send sensor abnormal information to the operation and maintenance platform.

[0050] S3: Extract edge features from the preprocessed port equipment operation data;

[0051] Specifically, the vibration data in the preprocessed port equipment operation data undergoes time-domain and frequency-domain feature extraction. The time-domain features include root mean square value, peak factor, and kurtosis; the frequency-domain features include rotational frequency and its harmonic components, sideband energy proportion, and frequency band energy distribution. Temperature features are extracted from the temperature data in the preprocessed port equipment operation data, including steady-state temperature value, temperature rise rate, and temperature difference ratio. Current data features are extracted from the current data in the preprocessed port equipment operation data, including RMS current value and current harmonic characteristics. Pose features are extracted from the pose data in the preprocessed port equipment operation data, including displacement variance and hoisting speed deviation.

[0052] S4: Compare each edge-side feature with the corresponding fault threshold. If the fault threshold is exceeded, the port equipment is preliminarily determined to be in a fault state; and proceed to the next step.

[0053] The operating status of port equipment exhibits significant time-varying and variability. For example, quay cranes lifting heavy loads during morning and evening peak hours show significantly higher vibration and current characteristics than during off-peak hours at night. Environmental factors such as high temperature, high humidity, and salt spray can cause the equipment's temperature baseline to rise. Furthermore, the characteristic range of the same equipment's normal operating state may gradually shift with increasing service life. Therefore, fixed thresholds are not accurate for determining the operating status of port equipment. This embodiment dynamically calculates a reasonable threshold range under the current operating conditions based on the equipment's real-time operating status and historical data patterns. Specifically, the formula for calculating the fault threshold is:

[0054] ;

[0055] In the formula, T dynamic The dynamic threshold of the edge-side feature at the current time, μ is the rolling mean of the edge-side feature over the past preset time window, σ is the rolling standard deviation of the edge-side feature over the same preset time window, and L is the dynamic threshold of the edge-side feature at the current time. final α represents the load factor of the port equipment, and β represents adjustable weighting parameters.

[0056] Furthermore, the rolling mean and rolling standard deviation are calculated as follows: for each edge node, a sliding window of length N is maintained, and the mean and standard deviation of all historical data within the window are calculated in real time. For example, for a vibration signal sampled at 10kHz, the RMS value after 1kHz downsampling is calculated every 10 seconds, and this RMS value is stored in the sliding window. Based on 1440 RMS values ​​from the past 24 hours, μ and σ are updated every minute.

[0057] Furthermore, the accurate assignment of the load factor has a significant impact on the fault threshold. Compared with the prior art, this embodiment proposes a method for determining the load factor to ensure that the determination of the load factor can be adapted to the entire life cycle of the equipment and multiple operating conditions.

[0058] Specifically, such as Figure 2 As shown, the method for determining the load factor is as follows:

[0059] Sa: Perform time-series baseline modeling for each port equipment to obtain the baseline load corresponding to that port equipment;

[0060] Port equipment exhibits cyclical patterns such as shift work and seasonal changes, as well as a tendency for increased load due to equipment aging. Therefore, it is necessary to separate the normal fluctuations and abnormal loads of port equipment.

[0061] The Sa process involves: acquiring edge-side feature data of the port equipment in the same time period over the past 7 days, constructing a multivariate time series sample, and then inputting the edge-side feature data of the port equipment in the same time period over the past 7 days into a lightweight long short-term memory network model (LSTM) to predict the edge-side feature data of the port equipment for the next day as the baseline load of the port equipment.

[0062] The lightweight long short-term memory network model has an input layer dimension that is the product of the number of edge-side feature classes and the time step; two hidden layers with 64 and 32 neurons respectively; the output layer outputs the edge-side feature data of the port equipment for the next day; the loss function is MAE, and the optimizer is Adam.

[0063] Sb: Calculate the load deviation based on the baseline load of the port equipment;

[0064] Specifically, the Mahalanobis distance between the real-time edge-side feature data of the port equipment and the baseline load of the port equipment is calculated as the load deviation of the port equipment.

[0065] Sc: Calculate the load factor corresponding to the port equipment based on the load deviation;

[0066] Wherein, the load factor The calculation formula is:

[0067] ;

[0068] In the formula, The weight coefficient for the i-th edge-side feature of the port equipment. This is the normalized value of the i-th edge-side feature of the port equipment. These are the time series weighting coefficients. This refers to the load deviation value of the port equipment. The fault weight of the port equipment is determined based on its real-time operating status. When the port equipment is fault-free, its fault weight is 0; when the port equipment is faulty, its fault weight is 0.1. The load value related to historical faults of the port equipment is 0.2 when the fault rate of the port equipment in the past year is less than a preset fault rate threshold, and 0.4 when the fault rate of the port equipment in the past year is greater than or equal to the preset fault rate threshold.

[0069] In this embodiment, when calculating the load factor, the baseline load of the port equipment is first compared with that of the port equipment, then the load deviation of the port equipment is calculated, and finally the load factor is calculated based on the load deviation. The calculation process integrates multi-dimensional physical parameters and time series trends, so that the calculation result of the load factor can overcome the problem of the time variability and high variability of port equipment.

[0070] S5: The edge node transmits the operating data of the port equipment to the cloud server to realize the fault prediction of the port equipment.

[0071] The cloud server aggregates the operational data of port equipment from all edge nodes and runs a relatively complex fault prediction model to achieve accurate fault prediction of port equipment.

[0072] The more complex fault prediction model is either a hierarchical long short-term memory network model or a multi-head attention long short-term memory network model.

[0073] Example 2: This invention also provides a port equipment fault prediction system based on edge computing. The system employs a port equipment fault prediction method based on edge computing from Example 1. The system includes:

[0074] The data acquisition module is used to collect port equipment operation data through multi-source sensors;

[0075] The data preprocessing module is used to perform data preprocessing operations on the port equipment operation data at the edge node;

[0076] The feature extraction module is used to extract edge features from the preprocessed port equipment operation data;

[0077] The preliminary fault status determination module is used to compare each edge-side feature with the corresponding fault threshold. If the fault threshold is exceeded, the port equipment is preliminarily determined to be in a fault state, and then proceeds to the next step.

[0078] The fault prediction module is used to transmit the operating data of the port equipment to the cloud server to realize fault prediction of the port equipment.

[0079] Example 3: The present invention also provides an electronic device, including one or more processors and a memory.

[0080] A processor can be a central processing unit (CPU) or other form of processing unit with data processing and / or instruction execution capabilities, and can control other components in an electronic device to perform desired functions.

[0081] The memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and a processor may execute the program instructions to implement the edge computing-based port equipment fault prediction method of any embodiment of this application described above, and / or other desired functions. Various contents such as initial extrinsic parameters and thresholds may also be stored in the computer-readable storage medium.

[0082] In addition to the methods and devices described above, embodiments of this application may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to implement the function of the port equipment fault prediction method based on edge computing provided in any embodiment of this application.

[0083] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0084] Furthermore, embodiments of this application may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to implement the port equipment fault prediction method based on edge computing provided in any embodiment of this application.

[0085] The computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0086] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A port equipment failure prediction method based on edge computing, characterized in that, The method comprises the following steps: S1: collecting port equipment operation data through multi-source sensors; S2: performing data preprocessing on the port equipment operation data at an edge node; S3: extracting edge-side features from the preprocessed port equipment operation data; S4: comparing each edge-side feature with a corresponding fault threshold value, and if the fault threshold value is exceeded, it is preliminarily determined that the port equipment is in a fault state; and proceeding to the next step; The formula for calculating the fault threshold value is: ； In the formula, T dynamic is the dynamic threshold value of the edge side feature at the current time, μ is the rolling mean value of the edge side feature in a preset time window in the past, σ is the rolling standard deviation of the edge side feature in the same preset time window, L final is the load coefficient of the port equipment, and α and β are adjustable weight parameters. The determination method of the load coefficient is specifically: Sa: modeling the time sequence baseline of each port equipment to obtain the baseline load of the corresponding port equipment; Sb: calculating the load deviation based on the baseline load of the port equipment; Sc: calculating the load coefficient corresponding to the port equipment based on the load deviation; S5: the edge node transmits the operation data of the port equipment to a cloud server to realize fault prediction of the port equipment.

2. The port equipment fault prediction method based on edge computing according to claim 1, wherein: The load factor The calculation formula is: ； wherein, is a weight coefficient of the i-th edge side feature of the port facility, is a normalized value of the i-th edge side feature of the port facility, is a time series weight coefficient, is a load deviation value of the port facility, is a failure weight of the port facility, is a historical failure related load value of the port facility.

3. The port equipment fault prediction method based on edge computing according to claim 1, wherein: The Sa is specifically: obtaining the edge-side feature data of the port equipment in the same time period in the past 7 days, constructing a multivariate time sequence sample, and then inputting the edge-side feature data of the port equipment in the same time period in the past 7 days into a lightweight long short-term memory network model to predict the edge-side feature data of the port equipment in the future 1 day as the baseline load of the port equipment.

4. The port equipment fault prediction method based on edge computing according to claim 3, wherein: The dimension of the input layer of the lightweight long short-term memory network model is the product of the number of edge-side feature categories and the time step; the hidden layer has 2 layers, and the number of neurons is 64 and 32; the output of the output layer is the edge-side feature data of the port equipment in the future 1 day; the loss function is MAE, and the optimizer is Adam.

5. The port equipment fault prediction method based on edge computing according to claim 3, wherein: The Sb is specifically: calculating the Mahalanobis distance between the real-time edge-side feature data of the port equipment and the baseline load of the port equipment as the load deviation of the port equipment.

6. The port equipment fault prediction method based on edge computing according to claim 1, wherein: In S1, the multi-source sensors include vibration sensors, temperature sensors, current / voltage sensors, and displacement sensors.

7. The port equipment fault prediction method based on edge computing according to claim 1, wherein: In S2, the preprocessing operation includes denoising filtering and missing value compensation.

8. The port equipment fault prediction method based on edge computing according to claim 7, wherein: The denoising filter is specifically: adopting discrete wavelet transform, selecting db4 wavelet base, and performing 5-layer decomposition on the port equipment operation data: the first to third layers are low-frequency components; the fourth to fifth layers are high-frequency components; then, the high-frequency coefficients of the fourth to fifth layers are filtered to remove random noise by using a soft threshold method, and the signal is reconstructed; The missing value compensation is specifically: for single-point missing, linear interpolation method is used for missing value compensation; for continuous multi-point missing, the data in this time period is marked as invalid, and a local alarm is triggered; the edge node prompts through a buzzer or sends sensor abnormal information to an operation and maintenance platform.

9. The port equipment fault prediction method based on edge computing according to claim 1, characterized in that: In S3, time domain features and frequency domain features of vibration data in the preprocessed port equipment operation data are extracted, the time domain features include root mean square value, peak factor and kurtosis; the frequency domain features include rotation frequency and its multiple frequency components, edge band energy ratio and frequency band energy distribution; Temperature features of temperature data in the preprocessed port equipment operation data are extracted, the temperature features include steady-state temperature value, temperature rise rate and temperature difference ratio; current data features of current data in the preprocessed port equipment operation data are extracted; The current data features include current effective value and current harmonic characteristics; Pose features of pose data in the preprocessed port equipment operation data are extracted; the pose features include displacement variance and lifting speed deviation. 10.A port equipment failure prediction system based on edge computing, characterized in that, The system adopts the port equipment fault prediction method based on edge computing according to any one of claims 1-9, and the system comprises: A data acquisition module for acquiring port equipment operation data through multi-source sensors; A data preprocessing module for performing data preprocessing operation on the port equipment operation data at an edge node; A feature extraction module for extracting edge-side features from the preprocessed port equipment operation data; A fault state preliminary determination module for comparing each edge-side feature with a corresponding fault threshold value, and preliminarily determining that the port equipment is in a fault state if the fault threshold value is exceeded; A fault prediction module for transmitting the operation data of the port equipment to a cloud server to realize fault prediction of the port equipment.

Images