Digital intelligent PHM platform for driving mechanism of coal unloading and storing mechanical equipment of coal wharf

Abstract

The invention discloses a digital intelligent PHM platform for a driving mechanism of coal unloading and storage mechanical equipment of a coal wharf, which comprises five core parts, namely a data acquisition layer, a data transmission layer, a data storage and processing layer, a digital modeling layer and a visualization and decision support layer, and is characterized in that the data acquisition layer comprises a multi-modal sensor arranged at a key part of the equipment; the data storage and processing layer comprises a framework combining edge computing and cloud computing, performs cleaning, noise reduction and feature extraction on original data, and constructs a driving mechanism operation state portrait through a multi-source information fusion technology; the digital modeling layer comprises fusion based on physical modeling and a data driving algorithm, and equipment digital twin bodies capable of being updated in real time are formed; the visualization and decision support layer provides a visual interface and operation and maintenance strategy recommendation; according to the invention, by realizing unified data acquisition and centralized management, the problems of data dispersion and incompatibility in a traditional monitoring system are solved, and a data foundation is laid for full-life-cycle management of equipment.

Description

Technical Field

[0001] This invention belongs to the field of engineering machinery condition monitoring technology, specifically relating to a digital intelligent PHM platform for the drive mechanism of coal unloading and storage machinery in coal terminals. Background Technology

[0002] As a crucial hub in the port logistics system, coal terminals primarily handle the large-scale loading, unloading, transshipment, and storage of coal. With the continuous expansion of coal logistics, the automation and scale of terminal equipment are constantly increasing. Among these, ship unloaders and belt conveyors, as key coal unloading and storage machinery, bear the most important power transmission and control functions in the entire coal unloading and storage process. However, due to the high dust concentration, heavy equipment workload, and long operating hours in the coal terminal operating environment, coupled with the complex equipment structure and numerous drive components, the drive system is prone to fatigue damage, wear, and failures, such as reducer wear, drum mechanism failure, and motor overload. Once such failures occur, they often cause the entire operation chain to shut down, resulting in economic losses and even safety risks.

[0003] At present, the operation and maintenance of ship unloaders and belt conveyor drive systems at coal terminals mainly rely on manual inspections and regular maintenance, lacking real-time and systematic health status monitoring and fault early warning capabilities, and the following technical bottlenecks exist: (1) The operating status of key components of the drive mechanism is difficult to be grasped online in real time, and the investigation of potential faults is lagging behind; (2) Equipment health data is scattered in different control systems and management platforms, resulting in information silos and a lack of effective data fusion and intelligent diagnostic methods; (3) Existing maintenance strategies are mostly preventive and experience-driven, making it difficult to achieve intelligent operation and maintenance optimization based on the entire life cycle of equipment.

[0004] In recent years, the intelligent operation and maintenance concept based on condition monitoring and predictive maintenance (PHM, Prognostics and Health Management) has gradually become an important trend in port equipment management. PHM technology, through real-time acquisition, modeling analysis, and health assessment of multi-source heterogeneous sensor data, can realize the transformation of equipment maintenance from reactive repair to predictive maintenance. However, at present, the research and application of digital intelligent PHM platforms for ship unloaders and belt conveyor drive mechanisms under the complex operating conditions of coal terminals are still in their initial stage, with problems such as an imperfect data acquisition system, difficulty in extracting multi-variable operating status characteristics of drive systems, and insufficient generalization ability of intelligent diagnostic models.

[0005] Therefore, developing a digital intelligent PHM platform for the drive mechanism of coal unloading and storage machinery at coal terminals, and constructing an integrated solution covering multi-source data acquisition, drive system health diagnosis, fault prediction and maintenance decision support, has significant engineering application value and industrial promotion prospects for improving the reliability of port coal unloading and storage equipment, reducing operation and maintenance costs, and ensuring the safe and efficient operation of coal logistics. Summary of the Invention

[0006] To achieve the above objectives, the technical solution of the present invention is as follows: The present invention proposes a digital intelligent PHM (Prognostics and Health Management) platform and its implementation method for the drive mechanism of coal unloading and storage machinery in coal terminals. Addressing the problem that real-time health monitoring and predictive maintenance are difficult to achieve for the drive mechanisms of key operating equipment such as ship unloaders and belt conveyors under high dust, high load, and highly corrosive conditions, the present invention achieves full lifecycle health management and intelligent operation and maintenance decision support for the drive mechanism by constructing a digital model of the equipment, a multi-source sensor data perception network, artificial intelligence diagnostic algorithms, and a digital twin system.

[0007] The platform of this invention comprises five core components: a data acquisition layer, a data transmission layer, a data storage and processing layer, a digital modeling layer, and a visualization and decision support layer.

[0008] By arranging high-precision multimodal sensors for temperature, vibration, and strain in key parts of the equipment, including motors, gearboxes, hoisting mechanisms, and drive drums, comprehensive and real-time perception of the equipment's operating status can be achieved.

[0009] By utilizing an architecture that combines edge computing and cloud computing, the raw data is cleaned, denoised, and feature extracted. A profile of the driving mechanism's operating status is then constructed using multi-source information fusion technology.

[0010] Data cleaning: The sliding window method is used to remove the mean from the signal within a length N.

[0011] (1),

[0012] A bandpass filter is designed based on the equipment's rotation frequency and mechanical characteristic frequency, with the retained frequency range typically between 0.5 and 10 times the rotation frequency, to filter out power frequency interference and high-frequency noise. Then, Savitzky-Golay filtering is used to smooth the temperature or slowly varying signals to eliminate low-frequency drift. Finally, the z-score normalization formula z=(x−μ) / σ is used to obtain the preprocessed clean signal, which is then used for subsequent feature extraction and modeling.

[0013] Time-domain and frequency-domain features are extracted from the preprocessed signal. The time-domain features include root mean square value, peak value, and kurtosis. The frequency-domain features are extracted by fast Fourier transform to extract the frequency amplitude of the bearing and gear fault characteristics.

[0014] Based on the integration of physical modeling and data-driven algorithms, a digital twin of the equipment that can be updated in real time is formed, enabling operating condition simulation, health status prediction, and abnormal operating condition tracing.

[0015] ① Using real-time data collected by sensors, the data-driven engine continuously adjusts the posture of each component of the coal unloading and storage equipment in the digital twin model, ensuring that the motion state of the virtual model is highly synchronized with the actual equipment, thus showcasing the real operating conditions. This involves collecting real-time data from sensors, such as position and acceleration information, and processing it efficiently to provide accurate data input to the drive engine, ensuring precise feedback on the dynamic posture of the coal unloading and storage equipment.

[0016] ② When a potential fault or anomaly is detected, the fault information is synchronized to the digital twin model as soon as possible, and the fault status is mapped and presented in real time in the virtual environment;

[0017] By introducing machine learning and deep learning algorithms, intelligent diagnosis and remaining useful life (RUL) prediction are performed on abnormal vibrations, overheating risks, and aging trends of key components.

[0018] ① Data Sample Construction: Based on historical equipment data and failure time, the RUL tag is defined as the time elapsed since failure, using a sliding window. (Points 256-512) Generate sequence samples. ② Model architecture: Employ a two-layer Long Short-Term Memory (LSTM) network or Temporal Convolutional Network (TCN) structure. The input is a multi-dimensional feature sequence, and the output is the RUL predicted value. ③ Loss function: The mean squared error (MSE) or negative log-likelihood form is used.

[0019] (6),

[0020] This allows us to simultaneously learn the expected value and variance of RUL, thus enabling prediction confidence estimation.

[0021] ④ Model Training: The Adam optimizer is used to train the model until the validation set converges. Samples are split according to device to prevent data leakage.

[0022] ⑤ Uncertainty estimation: In the inference phase, Monte Carlo Dropout is used for repeated forward propagation to calculate the mean and standard deviation, and output the RUL confidence interval. ;

[0023] ⑥ Decision-making mechanism: The system triggers multi-level early warnings based on the RUL predicted value and the confidence interval threshold. When the lower limit of the prediction is lower than the threshold and the confidence level is high, a maintenance suggestion is triggered.

[0024] It provides a visual interface and operation and maintenance strategy recommendations, and supports equipment hierarchical early warning, predictive maintenance, and resource optimization scheduling.

[0025] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0026] 1) This invention establishes a three-dimensional digital model of the ship unloader and belt conveyor, integrates the parameters of the entire system such as the drive system and steel structure, and realizes a precise mapping from equipment structure to operating status. The platform can be interconnected with various control systems such as SCADA, DCS, and PLC to achieve unified data collection and centralized management, solve the problem of scattered and incompatible data in traditional monitoring systems, and lay a data foundation for the full life cycle management of equipment.

[0027] 2) The platform integrates algorithms such as Support Vector Machine (SVM) and Backpropagation Neural Network (BPNN) to automatically identify abnormal patterns from complex multi-source monitoring data. It is applicable to various fault types, including motor failure, gearbox failure, roller bearing failure, and structural safety issues. Through trend analysis of historical operating data, it enables equipment health assessment and life prediction, shifting maintenance strategies from reactive repair to predictive maintenance, effectively reducing unplanned downtime.

[0028] 3) The platform forms a multi-dimensional state perception network by deploying various types of sensors, including vibration, temperature, strain, and insulation monitoring sensors. Combined with a digital twin model, it can automatically associate sensor locations with equipment structure, accurately locate fault locations, and enhance diagnostic robustness and reliability under complex operating conditions.

[0029] 4) This invention not only provides real-time alarms, but also combines historical data analysis and model inference to automatically generate operation and maintenance suggestions and maintenance priority lists, providing scientific decision support for operation and maintenance personnel, shortening troubleshooting time, reducing labor costs, and improving the operating efficiency of production systems;

[0030] 5) The platform adopts a modular and cloud architecture design, which can be flexibly expanded to multiple ship unloaders and multiple belt conveyor production lines and conveying systems of different sizes, realizing centralized remote monitoring and cross-regional equipment group management. The system can be deeply integrated with the enterprise information platform to provide full life cycle asset management capabilities for large-scale continuous production systems.

[0031] 6) By identifying potential hazards in advance, the risk of prolonged downtime and accidents caused by coal unloading and storage equipment failures can be reduced, the lifespan of drive cabinets and key components can be extended, spare parts inventory and maintenance costs can be reduced, providing significant economic benefits to enterprises, while improving the safety and intelligence level of equipment operation. Attached Figure Description

[0032] Figure 1 This is a flowchart of the algorithm of the present invention. Detailed Implementation

[0033] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are for illustrative purposes only and are not intended to limit the scope of the invention.

[0034] Example 1: Platform system architecture and sensor deployment.

[0035] The PHM platform of this invention comprises five levels:

[0036] Data acquisition layer: Multimodal sensors are deployed in key parts of the ship unloader and belt conveyor drive mechanism, including: triaxial acceleration sensors for detecting vibration spectrum and dynamic impact characteristics; temperature sensors for real-time monitoring of the thermal characteristics of the motor and gearbox; strain sensors for assessing the safety of the steel structure; and electrical parameter acquisition modules for detecting the insulation of the motor. The sensor sampling frequency can reach the millisecond level, supporting the capture of transient impacts and minute abnormal signals.

[0037] Data transmission layer: The platform adopts an edge computing node-server architecture: Edge computing nodes are responsible for data preprocessing and initial anomaly detection; Data cleaning: The sliding window method is used to remove the mean from signals within a length N.

[0038] (1),

[0039] A bandpass filter is designed based on the equipment's rotation frequency and mechanical characteristic frequency, with the retained frequency range typically between 0.5 and 10 times the rotation frequency, to filter out power frequency interference and high-frequency noise. Then, Savitzky-Golay filtering is used to smooth the temperature or slowly varying signals to eliminate low-frequency drift. Finally, the z-score normalization formula z=(x−μ) / σ is used to obtain the preprocessed clean signal, which is then used for subsequent feature extraction and modeling.

[0040] Extracting temporal features (root mean square value) from the preprocessed signal. Peak , cliff ) and frequency domain features (extracting the frequency amplitude of bearing and gear fault features through Fast Fourier Transform (FFT);

[0041] The cloud platform is responsible for the fusion, modeling, and analysis of multi-source data to ensure a comprehensive health assessment;

[0042] 1) Establish the dynamic model of the transmission system:

[0043] (2),

[0044] And thermal models:

[0045] (3),

[0046] The parameters are determined by the equipment's design values ​​or experimental calibration.

[0047] 2) Calculate the model's predicted values ​​and generate residuals during runtime:

[0048] (4),

[0049] The RMS, kurtosis, and energy spectrum of the residuals serve as new diagnostic features;

[0050] 3) The fusion strategy adopts a residual enhancement approach: residual statistical features are incorporated into the input of the data-driven model; simultaneously, the system supports output-level fusion.

[0051] (5),

[0052] The data transmission protocol supports 5G / Industrial Ethernet, ensuring real-time performance in large-scale equipment deployments at the dock.

[0053] Data storage layer: Employs a highly scalable time-series database to store historical operation records, real-time status data, and maintenance logs, providing a reliable data source for training machine learning models.

[0054] Digital Model Layer: Constructing a digital twin model of the equipment: Physical modeling is based on the dynamics, thermal and transmission characteristics of the drive system; Data-driven modeling identifies fault modes through deep learning and time series algorithms; Dynamic response prediction of the equipment under different operating conditions is achieved through a simulation platform.

[0055] Visualization and Decision Support Layer: Displays real-time equipment status, abnormal alarm information, and trend prediction results through a 3D digital twin interface; generates maintenance strategy suggestions based on the prediction results to achieve predictive maintenance and optimized spare parts resource scheduling.

[0056] Example 2: Design of an artificial intelligence diagnostic and prediction model.

[0057] Abnormal vibration detection: Vibration signal features are extracted using FFT (Fast Fourier Transform) and wavelet packet decomposition, and intelligent classification of fault types such as bearing failure and gear meshing abnormality is achieved by combining support vector machine (SVM).

[0058] Overheat protection strategy: Temperature sensors monitor the temperature of the drive motor and gearbox in real time, and a deep learning model predicts future temperature trends based on historical heat load data, triggering an alarm mechanism in advance.

[0059] Component aging assessment and remaining life prediction:

[0060] The model combines multi-dimensional parameters such as runtime, vibration characteristics, and lubrication conditions, and uses a long short-term memory network (LSTM) to predict the remaining life (RUL) of key components.

[0061] Digital twin integration with AI models: When an abnormal state is detected, the platform synchronously updates the digital twin model and reproduces the source of the fault in a virtual environment; a visual interface assists maintenance personnel in making quick decisions and improves maintenance efficiency.

[0062] Example 3: Typical application scenario of coal terminal.

[0063] Take the ship unloader and belt conveyor system of a large coal terminal as an example:

[0064] The platform deploys multi-dimensional sensor nodes to achieve full system monitoring of the unloader's lifting and closing mechanism, pitching mechanism, steel structure safety, and conveyor belt unit.

[0065] After the implementation of predictive maintenance strategies, equipment maintenance cycles were extended by 15%, and annual maintenance costs were reduced by approximately 20%.

[0066] It should be noted that the above content merely illustrates the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. For those skilled in the art, various improvements and modifications can be made without departing from the principle of the present invention, and all such improvements and modifications fall within the scope of protection of the claims of the present invention.

Claims

1. A digital intelligent PHM platform for the drive mechanism of coal unloading and storage machinery at a coal terminal, characterized in that: The platform comprises five core components: data acquisition layer, data transmission layer, data storage and processing layer, digital modeling layer, and visualization and decision support layer. The data acquisition layer includes deploying multimodal sensors in key parts of the device; The data storage and processing layer includes an architecture that combines edge computing and cloud computing to clean, reduce noise, and extract features from raw data, and to build a profile of the operating status of the driving mechanism through multi-source information fusion technology. The digital modeling layer integrates physical modeling and data-driven algorithms to form a digital twin of the device that can be updated in real time; The visualization and decision support layer provides a visual interface and operation and maintenance strategy recommendations.

2. The digital intelligent PHM platform for the drive mechanism of coal unloading and storage machinery in a coal terminal as described in claim 1, characterized in that, Data cleaning includes using a sliding window method to remove the mean from signals within a length N: （1）， A bandpass filter is designed based on the equipment's rotation frequency and mechanical characteristic frequency, retaining a frequency range typically between 0.5 and 10 times the rotation frequency. Then, Savitzky-Golay filtering is used to smooth the temperature or slowly varying signals. Finally, the z-score normalization formula z=(x−μ) / σ is used to obtain a preprocessed clean signal, which is then used for subsequent feature extraction and modeling.

3. The digital intelligent PHM platform for the drive mechanism of coal unloading and storage machinery at a coal terminal as described in claim 1, characterized in that, Data preprocessing includes: extracting time-domain and frequency-domain features from the preprocessed signal; The cloud platform is responsible for the fusion, modeling, and analysis of multi-source data to ensure a comprehensive health assessment; 1) Establish the dynamic model of the transmission system: （2）， And thermal models: （3）， The parameters are determined by the equipment's design values ​​or experimental calibration. 2) Calculate the model's predicted values ​​and generate residuals during runtime: （4）， The RMS, kurtosis, and energy spectrum of the residuals serve as new diagnostic features; 3) The fusion strategy adopts a residual enhancement approach: residual statistical features are incorporated into the input of the data-driven model; simultaneously, the system supports output-level fusion. （5）。 4. The digital intelligent PHM platform for the drive mechanism of coal unloading and storage machinery in a coal terminal as described in claim 1, characterized in that, The digital modeling layer includes: With the help of real-time data collected by sensors, the data-driven engine continuously adjusts the posture of each component of the coal unloading and storage equipment in the digital twin model, so that the motion state of the virtual model is highly synchronized with the actual equipment, showing the real operation. The real-time data collected by the sensors is efficiently processed to provide accurate data input to the drive engine, ensuring accurate feedback of the dynamic posture of the coal unloading and storage equipment. When a potential fault or anomaly is detected, the fault information is synchronized to the digital twin model immediately, and the fault status is mapped and presented in real time in the virtual environment.

5. The digital intelligent PHM platform for the drive mechanism of coal unloading and storage machinery in a coal terminal as described in claim 1, characterized in that, The platform also incorporates machine learning and deep learning algorithms to intelligently diagnose abnormal vibrations, overheating risks, and aging trends of key components, and predict their remaining lifespan; including: Data sample construction: Based on historical equipment data and failure time, RUL tags are defined as the time elapsed since failure, using a sliding window. Generate sequence samples; Model architecture: It adopts a two-layer long short-term memory network or a temporal convolutional network structure. The input is a multi-dimensional feature sequence, and the output is the RUL prediction value. ; Loss function: Takes the form of mean squared error or negative log-likelihood. （6）， This allows us to simultaneously learn the expected value and variance of RUL, thus enabling prediction confidence estimation. Model training: The Adam optimizer is used to train until the validation set converges. The sample is split according to the device to prevent data leakage. Uncertainty estimation: In the inference phase, Monte Carlo Dropout is used for repeated forward propagation to calculate the mean and standard deviation, outputting the RUL confidence interval. ; Decision-making mechanism: The system triggers multi-level early warnings based on the RUL predicted value and the confidence interval threshold. When the lower limit of the prediction is lower than the threshold and the confidence level is high, maintenance suggestions are triggered.

6. The digital intelligent PHM platform for the drive mechanism of coal unloading and storage machinery in a coal terminal as described in claim 1, characterized in that, Multimodal sensors are deployed in key parts of the ship unloader and belt conveyor drive mechanism. Triaxial acceleration sensors are used to detect vibration spectrum and dynamic impact characteristics, strain sensors are used to assess the safety of steel structure, and electrical parameter acquisition modules are used to detect motor insulation.

Images