Remote fault diagnosis method and system for sand absorbing, screening and separating all-in-one machine based on Internet of Things

Abstract

The invention relates to a remote fault diagnosis method and system for a sand absorbing, screening and separating all-in-one machine based on the Internet of Things, and the method comprises the following steps: obtaining the historical data of equipment, and deploying an Internet of Things sensor group based on the key part of the sand absorbing, screening and separating all-in-one machine, the method comprises the steps of collecting vibration data, temperature data, pressure data and displacement data of equipment in real time, preprocessing historical data of the equipment, performing normalization processing on the historical data of the equipment, extracting features from the historical data, fusing the historical data of the equipment, performing fault prediction on time series data through an LSTM network, and obtaining a fault prediction result. The equipment fault mode can be recognized in advance, early warning, edge computing and cloud computing are combined, delay is reduced, more accurate diagnosis can be provided, the deep reinforcement learning is introduced to optimize the equipment self-adaptive repair process, automatic adjustment and intelligent repair are achieved, and through the multi-task learning and big data analysis technology, the equipment fault diagnosis efficiency is improved. And the precision and reliability of fault diagnosis are improved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of remote control of equipment, and particularly to a remote fault diagnosis method and system for a sand suction and screening separator integrated machine based on the Internet of Things. Background Art

[0002] The sand suction and screening separator integrated machine has been widely used in many fields such as sand and gravel processing, mineral mining, and sewage treatment. Long-term operation of the equipment may encounter various faults, such as excessive vibration, overheating, blockage, etc. If these problems are not diagnosed and repaired in time, it may lead to equipment shutdown, reduced production efficiency, and even more serious damage. Traditional fault diagnosis methods rely on manual inspections or regular detections, and cannot achieve real-time monitoring and remote diagnosis. Therefore, the remote fault diagnosis system based on the Internet of Things (IoT) technology has great development potential.

[0003] Analysis of Existing Technical Problems 1 In the existing technology, the fault diagnosis of equipment usually depends on simple threshold judgment or rule-based reasoning after the sensor obtains data. However, the existing technology has the following problems: 2 Sensor accuracy and data noise problems: Existing sensors may be affected by environmental interference, resulting in unstable data and affecting the accuracy of fault diagnosis.

[0004] 3 Multi-source data processing problems: The fusion and real-time processing of various types of sensor data are difficult, and it is difficult to achieve accurate integration and efficient analysis of data.

[0005] 4 Insufficient intelligence of the fault diagnosis model: Most of the existing fault diagnosis models are based on static rules or simple statistical methods, lacking intelligent analysis and unable to identify complex fault patterns in time.

[0006] Communication delay and stability problems: The reliability of remote data transmission is poor in low-bandwidth and unstable network environments, which may affect the effect of real-time diagnosis. Summary of the Invention

[0007] The purpose of the present invention is to provide a remote fault diagnosis method and system for a sand suction and screening separator integrated machine based on the Internet of Things. The solution of the present invention aims to achieve remote fault diagnosis of the sand suction and screening separator integrated machine through the combination of Internet of Things technology and advanced data analysis algorithms.

[0008] To achieve the above purpose, in the first aspect of the embodiments of the present disclosure, a remote fault diagnosis method for a sand suction and screening separator integrated machine based on the Internet of Things is provided, which is applied to the sand suction and screening separator integrated machine. The method includes the following steps: Obtain the historical data of the equipment, deploy an Internet of Things sensor group based on the key parts of the sand suction and screening separation integrated machine, and collect the vibration data, temperature data, pressure data and displacement data of the equipment in real time. Preprocess the historical data of the equipment, normalize the historical data of the equipment and extract features from the historical data, and fuse the historical data of the equipment; Deploy an edge computing unit at the equipment end based on the Internet of Things sensor group, receive the data of the Internet of Things sensor group in real time, preliminarily analyze the data of the Internet of Things sensor group, classify faults based on machine learning algorithms and identify the fault types, upload the historical fault data to the cloud platform based on the communication layer, and issue local alarms; Upload the processed historical fault data to the cloud platform for storage and management based on the edge computing unit. The cloud platform uses deep learning algorithms to train models. In the models, fault type identification, occurrence time and prediction of the remaining life of the equipment are carried out synchronously, and the data of the Internet of Things sensor group is fused using the particle filter algorithm; Build a deep reinforcement learning model to simulate the repair strategies under different faults of the equipment, automatically adjust the equipment operation parameters according to the model output, and form a closed-loop control by feedback-adjusting the repair strategy through real-time monitoring of the repair effect; Respond to the construction of a remote monitoring interface on the cloud platform to display the equipment operation status, fault diagnosis results and maintenance suggestions, and generate maintenance suggestions and optimized operation strategies based on the fault diagnosis results and equipment status through a decision support system to assist in operation and maintenance decision-making.

[0009] In a possible implementation manner, the obtaining of the historical data of the equipment, deploying an Internet of Things sensor group based on the key parts of the sand suction and screening separation integrated machine, collecting the vibration data, temperature data, pressure data and displacement data of the equipment in real time, preprocessing the historical data of the equipment, normalizing the historical data of the equipment and extracting features from the historical data, and fusing the historical data of the equipment specifically includes: Data acquisition: Install an Internet of Things sensor group at the key parts of the sand suction and screening separation integrated machine. The key parts include the screen, vibration motor and feed inlet. The Internet of Things sensor group includes vibration sensors, temperature sensors, pressure sensors and flow sensors, and transmits the data to the edge computing unit through a wireless communication module; The preprocessing of the historical data of the equipment includes denoising, normalization and missing value filling processing; Adopt time domain, frequency domain and time-frequency domain analysis methods to extract features from the historical data of the equipment. The features include mean value, variance and spectral features; Use the Kalman filter algorithm or the particle filter algorithm to fuse the data of the Internet of Things sensor group.

[0010] In a possible implementation manner, an edge computing unit is deployed at the device end based on the Internet of Things sensor group, and data of the Internet of Things sensor group is received in real time, the data of the Internet of Things sensor group is preliminarily analyzed, fault classification is performed based on a machine learning algorithm to identify the fault type, historical fault data is uploaded to the cloud platform based on the communication layer, and a local alarm is issued, specifically including: The edge computing unit performs preliminary data analysis and fault diagnosis through direct connection with the Internet of Things sensor group; The machine learning algorithm includes a support vector machine and a decision tree. The support vector machine responds to the edge computing unit using the support vector machine algorithm for real-time fault diagnosis and identifies the fault type; When the support vector machine responds to the edge computing unit to determine that the state of the device is abnormal, the edge computing unit issues a local alarm and uploads the information to the cloud platform for analysis through the communication layer.

[0011] In a possible implementation manner, the edge computing unit uploads the processed historical fault data to the cloud platform for storage and management. The cloud platform trains a model using a deep learning algorithm. In the model, fault type identification, occurrence time, and prediction of the remaining life of the device are carried out simultaneously. The particle filter algorithm is used to fuse the data of the Internet of Things sensor group, specifically including: Based on the historical fault data, the cloud platform trains a model using a deep learning algorithm. The deep learning algorithm includes a convolutional neural network and a long short-term memory network. Among them, the long short-term memory network is used to process the time series data during the operation of the device, and the time series data includes vibration signals and temperature changes; Multiple fault type identifications, prediction of fault occurrence time, and prediction of the remaining service life of the device are carried out simultaneously in the same model; The particle filter algorithm is used to fuse the data of the Internet of Things sensor group.

[0012] In a possible implementation manner, a deep reinforcement learning model is constructed to simulate the repair strategies under different faults of the device, automatically adjust the device operation parameters according to the model output, and form a closed-loop control by real-time monitoring of the repair effect feedback to adjust the repair strategy, specifically including: A deep reinforcement learning model is constructed to simulate the repair strategies under different faults of the device. Through the big data analysis module of the cloud platform, the fault patterns in the historical data of the device are identified based on the K-Means algorithm clustering analysis; Based on the historical data analysis and clustering results of the device, the cloud platform generates an automated fault repair strategy. The fault repair strategy includes recommended maintenance steps, recommended component replacement cycles, and preventive measures for common device faults; Automatically adjust the operating parameters of the device according to the repair strategy output by the deep reinforcement learning model, where the parameters include vibration frequency and screen tension; And form a closed-loop control by monitoring the repair effect feedback in real time to adjust the repair strategy.

[0013] In a possible implementation manner, in response to building a remote monitoring interface on the cloud platform to display the device operation status, fault diagnosis results, and maintenance suggestions, generate maintenance suggestions and optimized operation strategies based on the fault diagnosis results and device status through a decision support system to assist in operation and maintenance decision-making, specifically including: Build an adaptive control system. Based on deep reinforcement learning, when the system detects minor faults, where the minor faults include excessive vibration and temperature rise, perform adaptive repair on the device through deep reinforcement learning; Remote monitoring interface: Provide a user-friendly remote monitoring interface on the cloud platform to display the device operation status, fault diagnosis results, and maintenance suggestions; Provide maintenance suggestions and optimized operation strategies according to the fault diagnosis results and device status.

[0014] In the second aspect of the embodiments of the present disclosure, a remote fault diagnosis method for a sand suction and screening separation integrated machine based on the Internet of Things further includes the following steps: The Internet of Things sensor group further includes high-frequency acoustic sensors and industrial cameras, and the device historical data preprocessing further includes preprocessing acoustic and visual data, and; Build a digital twin basic model. Based on the CAD design drawings, physical parameters, component connection relationships, kinematic and dynamic equations of the device, construct the geometric model and physical behavior basic model of the device, and initially construct a knowledge graph; Adopt distributed edge nodes to process visual and acoustic data and perform preliminary real-time image and acoustic analysis at the edge; Process multi-modal sensing data in real time at the edge and trigger local alarms; Perform multi-dimensional state diagnosis through a deep learning model enhanced by a knowledge graph in the cloud; Realize the closed-loop feedback between the physical device and the virtual model through digital twin synchronization; Apply deep reinforcement learning to generate and present an interpretable maintenance strategy through AR.

[0015] In the third aspect of the embodiments of the present disclosure, a remote fault diagnosis system for a sand suction and screening separation integrated machine based on the Internet of Things is provided, which is applied to the sand suction and screening separation integrated machine. The system includes: A data acquisition module, configured to obtain device historical data, deploy an Internet of Things sensor group based on key parts of the sand suction and screening separation integrated machine, collect vibration data, temperature data, pressure data, and displacement data of the device in real time, preprocess the device historical data, normalize the device historical data, extract features from the historical data, and fuse the device historical data; An edge computing module, configured to deploy an edge computing unit at the device end based on the Internet of Things sensor group, receive Internet of Things sensor group data in real time, preliminarily analyze the Internet of Things sensor group data, classify faults and identify fault types based on machine learning algorithms, upload historical fault data to the cloud platform based on the communication layer, and issue local alarms; A communication module, configured to upload processed historical fault data to the cloud platform for storage and management based on the edge computing unit. The cloud platform trains a model using deep learning algorithms. Fault type identification, occurrence time, and device remaining life prediction are carried out synchronously in the model, and the particle filter algorithm is used to fuse Internet of Things sensor group data; A cloud platform module, configured to build a deep reinforcement learning model to simulate repair strategies under different device faults, automatically adjust device operation parameters according to the model output, and form a closed-loop control by feedback-adjusting the repair strategy through real-time monitoring of the repair effect; An application module, configured to respond to building a remote monitoring interface on the cloud platform to display the device operation status, fault diagnosis results, and maintenance suggestions, and generate maintenance suggestions and optimized operation strategies based on the fault diagnosis results and device status through a decision support system to assist in operation and maintenance decision-making.

[0016] In a fourth aspect of the embodiments of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the steps of the method described in any one of the first aspects are implemented.

[0017] In a fifth aspect of the embodiments of the present disclosure, a sand suction and screening separation integrated machine is provided, including: A memory, on which a computer program is stored; A processor, configured to execute the computer program in the memory to implement the steps of the method described in any one of the first aspects.

[0018] The present invention provides a remote fault diagnosis method and system for a sand suction and screening separation integrated machine based on the Internet of Things. Compared with the prior art, the following beneficial effects are achieved: 1. The present invention uses an LSTM network to predict faults for time series data, and can identify device fault modes in advance and give early warnings; 2. The combination of edge computing and cloud computing in the present invention not only reduces latency, but also provides more accurate diagnosis; 3. The present invention introduces deep reinforcement learning to optimize the device adaptive repair process, achieving automatic adjustment and intelligent repair; 4. Through multi-task learning and big data analysis technology, the present invention further improves the accuracy and reliability of fault diagnosis; 5. The present invention can significantly enhance the remote fault diagnosis ability of the sand suction and screening separator integrated machine, reduce downtime and maintenance costs, and improve the reliability of the device.

[0019] Other features and advantages of the present disclosure will be described in detail in the following specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings are used to provide a further understanding of the present disclosure, and constitute a part of the specification. Together with the following specific implementation, they are used to explain the present disclosure, but do not constitute a limitation to the present disclosure. In the drawings: Figure 1 is a flowchart of a remote fault diagnosis method for a sand suction and screening separator integrated machine based on the Internet of Things shown in the embodiments of the specification; Figure 2 is a schematic structural diagram of a remote fault diagnosis system for a sand suction and screening separator integrated machine based on the Internet of Things shown in the embodiments of the specification; Figure 3 is a block diagram of another sand suction and screening separator integrated machine shown in the embodiments of the specification. SPECIFIC IMPLEMENTATION

[0021] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0022] The following will describe in detail the specific implementation of the present disclosure in conjunction with the drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the present disclosure, and is not used to limit the present disclosure.

[0023] Embodiment 1 The sand suction and screening separator integrated machine is widely used in fields such as sand and gravel processing and mineral mining. Long-term high-load operation of the equipment is prone to various faults, such as excessive vibration and abnormal temperature. Existing fault diagnosis methods rely on traditional manual detection or simple rule diagnosis, lacking real-time performance and intelligence. Therefore, this solution will combine cutting-edge technologies such as the Internet of Things technology, deep learning, and big data analysis to design a remote fault diagnosis method for a sand suction and screening separator integrated machine based on the Internet of Things; The selected model of the sand suction, screening and separation integrated machine is: SXSF series sand and stone separator or SJF30 type concrete cleaning sand and stone separator.

[0024] Edge intelligent data processing: Combining the data of Internet of Things sensors, real-time preliminary analysis is carried out through edge computing to reduce communication latency.

[0025] Dynamic fault prediction model: Using the long short-term memory network (LSTM) of deep learning to achieve dynamic fault prediction, and optimizing the prediction model through multi-task learning.

[0026] Time series data fusion: Fusing multi-dimensional sensor data such as vibration, temperature, and pressure, and using the most advanced data fusion algorithms to improve diagnostic accuracy.

[0027] Adaptive repair mechanism: Based on deep reinforcement learning (DRL), the adaptive control of the device is realized, the operating state of the device is optimized, and the occurrence of faults is reduced.

[0028] Figure 1 It is a flowchart of a remote fault diagnosis method for a sand suction, screening and separation integrated machine based on the Internet of Things shown according to an embodiment. The method includes: In step S11, acquiring the device historical data, deploying an Internet of Things sensor group based on the key parts of the sand suction, screening and separation integrated machine, collecting the vibration data, temperature data, pressure data and displacement data of the device in real time, preprocessing the device historical data, normalizing the device historical data and extracting features from the historical data, and fusing the device historical data; specifically including: Data acquisition: Install an Internet of Things sensor group at the key parts of the sand suction, screening and separation integrated machine. The key parts include the screen mesh, vibration motor and feed inlet. The Internet of Things sensor group includes vibration sensors, temperature sensors, pressure sensors and flow sensors, and transmits the data to the edge computing unit through a wireless communication module; The preprocessing of the device historical data includes denoising, normalization and missing value filling processing; Adopt time domain, frequency domain and time-frequency domain analysis methods to extract features from the device historical data. The features include mean value, variance and spectrum features; Use the Kalman filter algorithm or particle filter algorithm to fuse the data of the Internet of Things sensor group.

[0029] In the embodiment of the present disclosure, data acquisition installs an Internet of Things sensor group at the key parts of the sand suction, screening and separation integrated machine. Common sensors include: Vibration sensor: Detect the vibration condition of the device.

[0030] Temperature sensor: Monitor the temperature of components such as motors and bearings.

[0031] Pressure sensor: Detects the pressure change of the working fluid.

[0032] Flow sensor: Measures the flow rate of the liquid.

[0033] These sensors will collect various types of data in real time and transmit the data to the edge computing unit through a wireless communication module (such as Wi-Fi, LoRa, etc.).

[0034] Since the sensor data may contain noise or errors, data denoising is performed after collection. The Kalman filtering algorithm is used to process the historical data.

[0035] The Kalman filter can extract signals from noise. The formula is as follows: ; Where: : The estimated value at the current time (the denoised data).

[0036] : The sensor measurement value at the current time.

[0037] : The observation matrix, representing the relationship between the measurement value and the state.

[0038] : The Kalman gain, indicating how to combine the measurement value and the predicted value.

[0039] Data standardization To eliminate the dimensional difference of the sensor data, the Z-Score standardization method is used to normalize the data.

[0040] The standardization formula is: ; Where: : The historical data value.

[0041] : The mean value of the data.

[0042] : The standard deviation of the data.

[0043] : The standardized data.

[0044] Data fusion Fuses multi-dimensional sensor data such as vibration, temperature, pressure, and flow rate, and uses the Kalman filtering algorithm or the particle filtering algorithm to improve the data accuracy and reduce noise.

[0045] The particle filter can be estimated through the following formula: ; Where: : State estimation at the current moment.

[0046] : Particle weight, indicating the contribution of each particle to the estimation result.

[0047] : State of the particle.

[0048] In step S12, the edge computing unit is deployed at the device end based on the Internet of Things sensor group, receives the data of the Internet of Things sensor group in real time, preliminarily analyzes the data of the Internet of Things sensor group, classifies faults based on a machine learning algorithm and identifies the fault type, uploads historical fault data to the cloud platform based on the communication layer, and issues a local alarm, specifically including: The edge computing unit performs preliminary data analysis and fault diagnosis through direct connection with the Internet of Things sensor group; The machine learning algorithm includes a support vector machine and a decision tree. The support vector machine responds to the edge computing unit using the support vector machine algorithm for real-time fault diagnosis and identifies the fault type; When the support vector machine responds to the edge computing unit to determine that the state of the device is abnormal, the edge computing unit issues a local alarm and uploads the information to the cloud platform for analysis through the communication layer.

[0049] Based on the edge computing unit uploading the processed historical fault data to the cloud platform for storage and management, the cloud platform trains a model using a deep learning algorithm. In the model, fault type identification, occurrence time, and prediction of the remaining life of the device are carried out synchronously. The particle filter algorithm is used to fuse the data of the Internet of Things sensor group, specifically including: According to the historical fault data, the cloud platform trains a model using a deep learning algorithm. The deep learning algorithm includes a convolutional neural network and a long short-term memory network. Among them, the long short-term memory network is used to process the time series data during device operation, and the time series data includes vibration signals and temperature changes; Multiple fault type identifications, prediction of fault occurrence time, and prediction of the remaining service life of the device are carried out simultaneously in the same model; The particle filter algorithm is used to fuse the data of the Internet of Things sensor group.

[0050] In the embodiment of the present disclosure, the edge computing unit performs preliminary data analysis and fault diagnosis through direct connection with the sensor.

[0051] The goal of edge computing is to reduce data transmission and processing delays and hand over some real-time fault warning tasks to the device end for processing.

[0052] The Support Vector Machine (SVM) performs real-time fault diagnosis using the SVM algorithm in response to the edge computing unit.

[0053] SVM is a classification model that can classify fault types by maximizing the classification margin.

[0054] The optimization objective of SVM is: ; Subject to: ; Where: : The normal vector of the hyperplane.

[0055] : Bias.

[0056] : The th sample data.

[0057] : The label of the sample (0 or 1, indicating whether a fault has occurred); : Constraint condition.

[0058] When the SVM model determines that the device's state is abnormal, the edge computing unit generates a preliminary fault alarm and sends the relevant data to the cloud platform for further analysis.

[0059] In step S13, the edge computing unit uploads the processed historical fault data to the cloud platform for storage and management. The cloud platform trains a model using a deep learning algorithm. In the model, fault type identification, occurrence time, and prediction of the remaining life of the device are carried out simultaneously. The particle filter algorithm is used to fuse the data of the Internet of Things sensor group, specifically including: Based on the historical fault data, the cloud platform trains a model using a deep learning algorithm. The deep learning algorithm includes a convolutional neural network and a long short-term memory network. Among them, the long short-term memory network is used to process the time-series data during device operation. The time-series data includes vibration signals and temperature changes; Multiple fault type identifications, prediction of fault occurrence time, and prediction of the remaining service life of the device are carried out simultaneously in the same model; The particle filter algorithm is used to fuse the data of the Internet of Things sensor group.

[0060] In the embodiments of the present disclosure, the LSTM network is used to perform fault prediction on the time-series data during device operation (such as vibration signals and temperature changes) using the long short-term memory (LSTM) network. LSTM can capture long-term dependencies and is very effective for trends and patterns in time-series data.

[0061] The update formula of LSTM is as follows: ; ; ; ; ; ; Where: : Hidden state.

[0062] : Cell state.

[0063] : Forget gate, controlling the degree of memory forgetting.

[0064] : Input gate, controlling the storage of new information.

[0065] : Candidate cell state.

[0066] is the weight matrix, is the bias.

[0067] The fault prediction LSTM model is trained based on historical data and outputs predictions of future device states. If the prediction results indicate that the device is about to fail, the system will issue an alarm in advance and prompt the possible types of faults.

[0068] Multi-task learning (MTL) combines with the LSTM network for multi-task learning, training multiple tasks (such as fault type prediction, fault occurrence time prediction, etc.) simultaneously to improve the accuracy of the diagnostic model. Each task shares some parameters to enhance the overall model performance.

[0069] In step S14, the construction of the deep reinforcement learning model simulates the repair strategies under different faults of the device, automatically adjusts the device operation parameters according to the model output, and forms a closed-loop control by feedback-adjusting the repair strategy through real-time monitoring of the repair effect, specifically including: Construct a deep reinforcement learning model to simulate the repair strategies under different faults of the device. Through the big data analysis module of the cloud platform, identify the fault patterns in the historical data of the device based on the K-Means algorithm clustering analysis; According to the analysis of the device historical data and the clustering results, the cloud platform generates an automated fault repair strategy, and the fault repair strategy includes recommended maintenance steps, recommended component replacement cycles, and preventive measures for common device faults; Automatically adjust the operating parameters of the device according to the repair strategy output by the deep reinforcement learning model, where the parameters include vibration frequency and screen tension; And form a closed-loop control by monitoring the repair effect feedback in real time to adjust the repair strategy.

[0070] The specific operation of automatically adjusting the device parameters includes: Fuse device sensor data (vibration amplitude, separation efficiency, energy consumption, etc.) and fault modes (historical fault classification based on K-Means clustering).

[0071] The DRL model (such as DDPG) outputs the optimal operating parameters (vibration frequency, screen tension, etc.), supporting continuous parameter space and multi-parameter collaborative optimization.

[0072] The parameters are sent to the device through industrial protocols (Modbus, OPC UA), and the repair effect (efficiency improvement, fault elimination, etc.) is collected in real time as a reward signal to update the model.

[0073] In the embodiments of the present disclosure, data analysis and fault mode recognition are performed through the big data analysis module of the cloud platform, and clustering analysis (such as the K-Means algorithm) is used to identify the fault modes in the historical data of the device.

[0074] The K-Means clustering algorithm is as follows: ; Where: : Clustering error.

[0075] : Indicates the weight of the th sample belonging to the th class.

[0076] : The th sample data.

[0077] : The mean of the th class.

[0078] Based on historical data analysis and clustering results, the cloud platform will generate an automated fault repair strategy, including: Recommended maintenance steps, recommended component replacement cycles, and preventive measures for common device faults.

[0079] In step S15, in response to building a remote monitoring interface on the cloud platform to display the device operating status, fault diagnosis results, and maintenance suggestions, generate maintenance suggestions and optimized operation strategies based on the fault diagnosis results and device status through the decision support system to assist in operation and maintenance decisions, specifically including: Build an adaptive control system based on deep reinforcement learning. When the system detects minor faults, including excessive vibration and increased temperature, perform adaptive repair of the device through deep reinforcement learning; Remote monitoring interface: Provide a user-friendly remote monitoring interface on the cloud platform to display the operating status of the device, fault diagnosis results, and maintenance suggestions; Provide maintenance suggestions and optimize operation strategies based on the fault diagnosis results and device status.

[0080] In the embodiments of the present disclosure, when the system detects minor faults (such as excessive vibration or increased temperature), deep reinforcement learning (DRL) is used to perform adaptive repair of the device. Reinforcement learning learns how to take optimal actions to maximize the operating efficiency of the device.

[0081] The update formula for deep Q-learning is: ; Where: : The value of taking an action in the current state.

[0082] : Learning rate.

[0083] : Immediate reward.

[0084] : All possible actions.

[0085] : Discount factor.

[0086] : Current state.

[0087] : The action currently taken.

[0088] Remote repair and control If the impact of the fault is minor, the system will adjust the operating parameters of the device, such as speed, cooling system, etc., through remote control to reduce the impact of the fault.

[0089] Specifically, 1. Edge computing technology enables data to be preliminarily processed and fault diagnosed at the device end (edge node) at the first time of collection. Only the data after preliminary screening will be transmitted to the cloud for further analysis and processing.

[0090] By performing partial data processing at the device end, communication latency can be significantly reduced, real-time performance can be improved, and the computing pressure on the cloud can also be alleviated. Edge computing makes fault diagnosis not only more efficient but also more reliable, avoiding the negative impact of network latency.

[0091] 2. LSTM (Long Short-Term Memory) is a sequential model in deep learning, which is particularly suitable for processing time series data. LSTM is used to analyze multi-dimensional sensor data such as vibration, temperature, and pressure of the sand suction and screening separation integrated machine, and conduct dynamic prediction of faults.

[0092] LSTM can capture long-term dependencies in sequential data, which is very important for identifying potential faults in equipment. For example, it may take several hours or days of accumulation for vibration signals and temperature changes to show signs of faults. Through the LSTM model, the system can identify potential faults in the equipment in advance, provide early warnings, and thus achieve predictive maintenance, rather than just diagnosis based on real-time data.

[0093] 3. Multi-task learning (MTL) is a deep learning technique that can handle multiple related tasks simultaneously. Each task shares some parameters in the network to improve the generalization ability of the model. MTL is used to perform multiple tasks simultaneously, such as fault type prediction, fault occurrence time prediction, remaining useful life prediction of equipment, etc.

[0094] Through multi-task learning, the model can improve the accuracy of each task by sharing the learned features, and avoid the overfitting problem that may occur in a single-task model. At the same time, it can meet different types of fault diagnosis needs, enabling the system to not only predict the fault type, but also estimate the time of fault occurrence, providing more comprehensive support for maintenance personnel.

[0095] 4. Deep reinforcement learning (DRL) is a technique that combines deep learning and reinforcement learning. It allows the system to autonomously learn and take optimal actions to maximize the reward. DRL is used to optimize the adaptive repair mechanism of equipment. When the equipment detects a minor fault, the system can automatically adjust the operating parameters of the equipment (such as speed, cooling system, etc.) according to different environmental and fault conditions, thus minimizing the impact of the fault on the equipment.

[0096] Traditional equipment repair often relies on manual operation and fixed rules. Through DRL, the system can autonomously learn the best repair strategy, reduce human intervention, and improve the repair efficiency. The system can not only handle known fault patterns, but also adjust autonomously based on the feedback during operation to optimize the overall operating state of the equipment.

[0097] 5. Particle filter is a recursive filtering technique based on the Monte Carlo method, which is suitable for non-linear and non-Gaussian noise systems. Particle filter is used to fuse multi-dimensional sensor data, unify the analysis of sensor data from different sources (such as vibration, temperature, pressure, etc.), and reduce the errors that may be brought by a single sensor.

[0098] Through the particle filtering method, the device state can be estimated more accurately. Especially in the process of multi-source data fusion, particle filtering can effectively handle data biases and noises that may exist in different sensors, thereby improving the accuracy and reliability of fault diagnosis.

[0099] 6. Through the model-based control method, combining deep reinforcement learning and model predictive control (MPC) technologies, an adaptive repair mechanism was developed. When the device shows abnormalities, the system can automatically adjust the operating parameters of the device (such as rotational speed, cooling system, etc.) according to the current fault mode and historical data to minimize the impact of the fault.

[0100] This model-based adaptive repair method can make intelligent responses when the device has problems. It not only relies on preset rules but also can be optimized and adjusted according to real-time data and historical data, avoiding errors caused by manual intervention and improving the repair efficiency and device stability.

[0101] Summary: Through edge computing, LSTM time series data prediction, multi-task learning, deep reinforcement learning, particle filtering, and model-based adaptive repair mechanism, this embodiment provides an efficient and intelligent fault diagnosis and repair method. The combination of these technologies not only improves the accuracy of fault diagnosis but also optimizes the operation and maintenance efficiency of the device. Through means such as real-time fault prediction, adaptive repair, and remote control, it effectively reduces the device downtime, lowers the maintenance cost, and enhances the reliability and intelligence level of the device.

[0102] Embodiment Two Based on Embodiment One, the present invention also provides a remote fault diagnosis method for a sand suction and screening separation integrated machine based on the Internet of Things, which further includes the following steps: The Internet of Things sensor group further includes high-frequency acoustic sensors and industrial cameras, and the preprocessing of the device historical data further includes preprocessing acoustic and visual data, and; Digital twin basic model construction. Based on the CAD design drawings, physical parameters, component connection relationships, kinematic and dynamic equations of the device, construct the geometric model and physical behavior basic model of the device, and preliminarily construct a knowledge graph; Adopt distributed edge nodes to process visual and acoustic data and perform preliminary real-time image and acoustic analysis at the edge; Process multi-modal sensing data in real time at the edge and trigger local alarms; Perform multi-dimensional state diagnosis through a deep learning model enhanced by a knowledge graph in the cloud; Achieve closed-loop feedback between the physical device and the virtual model through digital twin synchronization; Generating and presenting interpretable maintenance strategies through AR using deep reinforcement learning; In the embodiments of the present disclosure, Step 1: Multi-modal data collection, preprocessing, and digital twin foundation construction: Sensor deployment: Based on the original vibration, temperature, pressure, and displacement sensors, it also includes: High-frequency acoustic sensors (microphone arrays): Capture abnormal sound patterns during device operation (such as bearing abnormal sounds, material impact sounds).

[0103] Industrial cameras (machine vision): Installed at key observation points (such as the surface of the sieve, the discharge port, and connecting components) to capture image and video data and monitor visually visible problems such as surface wear, cracks, blockages, and leaks.

[0104] Ensure that all sensors are time-synchronized to lay the foundation for subsequent data fusion.

[0105] Data preprocessing: It also includes preprocessing acoustic and visual data: Acoustics: Noise reduction, feature extraction (such as Mel-frequency cepstral coefficients MFCC, spectrogram).

[0106] Vision: Image, deblurring, object detection, feature extraction (such as texture features, shape features).

[0107] Perform unified denoising, normalization, and missing value filling on all sensor data.

[0108] Digital Twin (DigitalTwin) basic model construction: Static model: Based on the device's CAD design drawings, physical parameters (material properties, dimensions, mass), component connection relationships, kinematic and dynamic equations, construct the geometric model and physical behavior basic model of the device.

[0109] Initial construction of the Knowledge Graph (KnwledgeGraph): Start constructing the Knowledge Graph, including structured knowledge such as device component information, relationships between components, known failure modes, historical maintenance records, operating procedures, and the association between sensors and components.

[0110] Step 2: Edge intelligent computing and preliminary state perception: Deploy edge computing units with strong performance or use distributed edge nodes to process visual and acoustic data.

[0111] Perform real-time preliminary analysis of images and acoustics at the edge (such as simple anomaly detection, specific pattern recognition).

[0112] Edge preliminary fault / status recognition: Utilize traditional machine learning (SVM, decision tree) and lightweight deep learning models to conduct preliminary fusion and analysis on multi-modal data (vibration, temperature, pressure, displacement, acoustics, visual features) and identify obvious or common fault patterns.

[0113] Edge-side digital twin state update and preliminary XAI explanation: The edge node uses the processed data and preliminary identification results to preliminarily update the digital twin state parameters of the components within its monitoring scope.

[0114] Deploy lightweight XAI methods (such as feature importance analysis) to provide preliminary explanations for the alarms at the edge side (e.g., "High vibration is mainly detected by X sensor, accompanied by specific acoustic patterns").

[0115] Local alarm and data upload: When potential faults or significant state changes are identified, local alarms are issued, and the processed feature data, preliminary diagnosis results, preliminary XAI explanations, and updated local twin state information are uploaded to the cloud platform.

[0116] Step 3: Cloud platform in-depth analysis, digital twin synchronization, and knowledge graph diagnosis: Cloud data fusion and storage: The cloud platform receives multi-modal data and information from each edge node for more comprehensive fusion, storage, and management. Advanced deep learning model training: Utilize the powerful computing resources of the cloud to train more complex multi-modal deep learning models (such as including CNN for image processing, LSTM / Transformer for time series and acoustics, and attention mechanism for fusing multi-source information).

[0117] Integrate knowledge graph (KG): Introduce knowledge graph information (such as graph embedding) in model training to help the model understand component relationships and fault propagation paths and improve diagnostic accuracy.

[0118] Digital Twin synchronization and simulation prediction: The cloud platform integrates all edge information and cloud analysis results, synchronizes and calibrates the global digital twin model in real time to accurately reflect the current state of the physical device.

[0119] Utilize digital twin for simulation analysis: Simulate the future evolution under the current state to conduct more accurate remaining useful life (RUL) prediction.

[0120] Conduct "What-if" analysis to evaluate the impact of different operating parameters or loads on device health.

[0121] Simulate the propagation path and influence scope of potential faults.

[0122] Deep application of Explainable Artificial Intelligence (XAI): Apply more complex XAI technologies (such as SHAP, LIME, Grad-CAM for vision, AttentionMap analysis) in the cloud, elaborate on the diagnostic results of the deep learning model, and illustrate which sensor data, which features, and data from which time periods contribute the most to the diagnostic conclusion.

[0123] Associate the XAI results with the knowledge graph to provide more engineering-logic-compliant explanations (for example, "The model believes there is a bearing fault because high-frequency vibration features are detected, and the acoustic sensor captures an abnormal sound pattern that matches the bearing wear record in ").

[0124] Multi-task learning: On the basis of the original fault classification, time prediction, and RUL prediction, add root cause analysis of faults (RtCauseAnalysis) and fault severity assessment as learning tasks.

[0125] Step 4: Digital twin-driven deep reinforcement learning and adaptive repair: DRL training based on digital twin: Use the digital twin model as the training environment for DRL.

[0126] XAI-guided adaptive repair strategy: The DRL model not only outputs repair strategies (such as adjusting the vibration frequency), but also combines with XAI to explain why this strategy is selected.

[0127] Before applying the strategy to the physical device, first simulate the effect and potential risks of the strategy in the digital twin for verification.

[0128] Knowledge graph-assisted repair: Compare and associate the strategies generated by DRL with the standard operating procedures (SP) in the knowledge graph to ensure the safety and compliance of the strategies, or provide more standardized guidance for operators.

[0129] Real-time feedback and closed-loop optimization: After the physical device executes the repair strategy, the sensor data is fed back in real time to update the status of the physical device and the digital twin.

[0130] Compare the prediction effect of the digital twin with the actual effect, and continuously optimize the DRL model and the digital twin model.

[0131] Step 5: Integrated remote monitoring, decision support, and AR-assisted maintenance: Integrated intelligent monitoring platform: Provide a unified monitoring interface that integrates real-time data, diagnostic results, XAI explanations, knowledge graph information, and digital twin visualization. Users can intuitively view the device status, fault causes, prediction trends, and interact with the digital twin model.

[0132] Decision support system: Combine diagnostic results, XAI explanations, digital twin simulations, and knowledge graph information (such as maintenance history, spare part information, standard procedures) to generate more comprehensive and reliable maintenance suggestions, spare part demand forecasts, and operation optimization plans.

[0133] Augmented reality (AR)-assisted maintenance: Develop an application (runnable on a tablet or glasses). When on-site maintenance personnel use the device: The camera identifies the device components, and the application automatically overlays and displays the real-time operating data, health status, and virtual perspective from the digital twin of the component.

[0134] Display the fault diagnosis results, XAI explanations, and knowledge graph association information.

[0135] Visually guide the maintenance process step by step (e.g., highlighting the bolts that need to be inspected, showing the animation of the disassembly sequence), and these guiding information directly comes from the knowledge graph and the optimized maintenance strategy.

[0136] Continuous learning and model evolution: The system continuously collects new operating data, fault cases, and maintenance feedback for regularly updating and optimizing all models (including deep learning models, digital twin models, DRL models) and the knowledge graph, realizing the self-evolution of the system.

[0137] Embodiment 3 Based on Embodiment 1, the present invention also provides a remote fault diagnosis method for a sand suction and screening separation integrated machine based on the Internet of Things, which further includes the following steps: Step 1: Data acquisition of the Internet of Things sensor group and multi-dimensional feature fusion: Data acquisition is carried out by installing a variety of sensors (vibration sensors, temperature sensors, flow sensors, etc.) on the sand suction and screening separation integrated machine for real-time data acquisition. The data acquisition frequency is high (e.g., 1 - 10 times per second) to ensure real-time monitoring of the device operation status.

[0138] Multi-dimensional data fusion: The Kalman filtering algorithm is used to preprocess the sensor data to remove noise and smooth the data. The Kalman filtering can be expressed by the following formula: ; Wherein: : The estimated value at the current moment (data after denoising).

[0139] : The sensor measurement value at the current moment.

[0140] : The observation matrix, representing the relationship between the measurement value and the state.

[0141] : The Kalman gain, indicating how to combine the measurement value and the predicted value.

[0142] Data normalization: All sensor data are uniformly standardized to eliminate the dimension difference. Z-Score normalization is adopted, and the formula is: ; Wherein, is the data collected by the sensor, is the mean value of the data, is the standard deviation, : The data after standardization.

[0143] Step 2: Preliminary real-time fault diagnosis of the edge computing unit: The edge computing unit: The edge computing device is installed on each sand suction and screening separation integrated machine and can perform preliminary data processing on-site. By real-time analyzing various sensor data, a support vector machine (SVM) is used for preliminary fault classification.

[0144] Support vector machine model: By training the SVM model, using the sensor data of each device as input, fault mode classification is carried out. The goal of SVM is to find a hyperplane that maximizes the classification margin, which can be described by the following formula: ; Satisfied with: ; Wherein, is the weight vector, is the bias, is the input data, is the class label, : The constraint condition.

[0145] Preliminary diagnosis result: If the device state exceeds the normal range, the edge computing unit will immediately send an alarm to the cloud platform and record the relevant data to assist remote analysis.

[0146] Step 3: Deep learning and dynamic fault prediction model: For time series data, a deep learning model (LSTM) uses a long short-term memory network (LSTM) for fault prediction. When processing time series data, LSTM can capture long-term dependencies, greatly improving the prediction accuracy.

[0147] The core update formula of the LSTM network is: ; ; ; ; ; ; Among them, is the input at the current time, is the hidden state at the current time, is the cell state, : forget gate, controlling the degree of forgetting of memory.

[0148] : input gate, controlling the storage of new information.

[0149] : candidate cell state, is the weight matrix, is the bias.

[0150] Fault prediction is carried out by training the LSTM model, based on historical device operation data (such as vibration, temperature, etc.), to predict the device state changes in the next few time steps. If the prediction result indicates that the device may fail, the cloud platform will issue a warning and conduct a detailed fault analysis.

[0151] Multi-task learning (MTL) combines with the LSTM network. Using the multi-task learning (MTL) method, multiple related tasks (such as fault type identification, fault occurrence time prediction, etc.) are trained simultaneously to further improve the generalization ability of the model. Each task improves the overall performance by sharing some parameters.

[0152] Step 4: Data analysis and fault repair suggestions: Data analysis is based on the cloud platform to conduct in-depth analysis of the historical operation data of various devices. Using big data analysis techniques, the most common fault modes of the devices and the key factors affecting the device stability are identified. Common analysis methods include clustering analysis (such as K-Means) and association rule mining.

[0153] The fault repair suggestion generation system generates automated fault repair suggestions according to the analysis results and pushes them to the device operation and maintenance personnel. The repair suggestions include: Recommended equipment maintenance procedures; Recommended component replacement cycles; Best operating methods for preventing failures.

[0154] Step Five: Adaptive Repair and Remote Control: Adaptive control system When the device has minor failures (such as abnormal mild vibration or temperature increase), the system will automatically activate the adaptive control mechanism, such as adjusting the speed, cooling system, or vibration damping system of the device. This process is optimized based on deep reinforcement learning (DRL) to learn how to maximize the stability of device operation by adjusting control parameters.

[0155] Reinforcement learning model Use a reinforcement learning model to make automatic repair decisions. Assume that the repair operation is a state space, and the state transition function and reward mechanism are as follows: The reward mechanism is as follows: The update formula for deep Q-learning is: ; Where: : The value of taking an action in the current state.

[0156] : Learning rate.

[0157] : Immediate reward.

[0158] : All possible actions.

[0159] : Discount factor.

[0160] : Current state.

[0161] : Current action taken.

[0162] Specifically, edge intelligent data processing: Based on real-time analysis of sensor data, the data is intelligently processed and fault identified through a deep learning model, and partial fault diagnosis is performed at the device end to reduce transmission delay.

[0163] Dynamic fault prediction model: Based on long short-term memory network (LSTM) and multi-task learning, combined with the device's historical operation data, to achieve more accurate fault prediction and pattern recognition.

[0164] Temporal data fusion and multi-source sensor data analysis: Deeply fuse multi-dimensional sensor data (vibration, temperature, flow, etc.), and use the latest temporal data processing technology to improve the accuracy of diagnosis.

[0165] Model-based Fault Adaptive Repair Mechanism: Through the fault features learned by the model, combined with edge computing and an adaptive control system, automatic repair or adjustment of the device is achieved.

[0166] Embodiment 4 Figure 2 It is a schematic structural diagram of the remote fault diagnosis system for the sand suction and screening separation integrated machine based on the Internet of Things according to the embodiments of the present application. As Figure 2 shown, the embodiments of the present disclosure also provide a remote fault diagnosis system for the sand suction and screening separation integrated machine based on the Internet of Things, which is applied to the sand suction and screening separation integrated machine. The system includes: A data acquisition module, configured to obtain device historical data, deploy an Internet of Things sensor group based on key parts of the sand suction and screening separation integrated machine, collect vibration data, temperature data, pressure data, and displacement data of the device in real time, preprocess the device historical data, perform normalization processing on the device historical data, extract features from the historical data, and fuse the device historical data; An edge computing module, configured to deploy an edge computing unit at the device end based on the Internet of Things sensor group, receive Internet of Things sensor group data in real time, perform preliminary analysis on the Internet of Things sensor group data, classify faults and identify fault types based on machine learning algorithms, upload historical fault data to the cloud platform based on the communication layer, and issue local alarms; A communication module, configured to upload the processed historical fault data to the cloud platform for storage and management based on the edge computing unit. The cloud platform trains a model using deep learning algorithms. Fault type identification, occurrence time, and prediction of the remaining life of the device are carried out synchronously in the model, and particle filter algorithms are used to fuse Internet of Things sensor group data; A cloud platform module, configured to build a deep reinforcement learning model to simulate repair strategies under different faults of the device, automatically adjust device operation parameters according to the model output, and form a closed-loop control by feedback-adjusting the repair strategy through real-time monitoring of the repair effect; An application module, configured to respond to building a remote monitoring interface on the cloud platform to display the device operation status, fault diagnosis results, and maintenance suggestions, and generate maintenance suggestions and optimized operation strategies based on the fault diagnosis results and device status through a decision support system to assist in operation and maintenance decisions.

[0167] For other details of the implementation of each module in the above-described embodiment device, reference can be made to the description in the remote fault diagnosis method for the sand suction and screening separation integrated machine based on the Internet of Things in Embodiment 1 above, which will not be elaborated here.

[0168] It should be noted that the embodiments in this specification are all described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the embodiments, reference can be made to each other. For the method embodiments, since they are basically similar to the method embodiments, the description is relatively simple. For the relevant parts, reference can be made to the corresponding descriptions in the method embodiments.

[0169] The embodiments of the present disclosure also provide a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps of the method described in any one of the foregoing embodiments are implemented.

[0170] The embodiments of the present disclosure also provide a sand suction and screening separation integrated machine, including: A memory, on which a computer program is stored; A processor, configured to execute the computer program in the memory to implement the steps of the method described in any one of the foregoing embodiments.

[0171] As Figure 3 FIG. is a schematic structural diagram of an electronic device provided by an embodiment of the present disclosure. It shows a schematic structural diagram of an electronic device suitable for implementing the electronic device in the embodiments of the present disclosure. Figure 3 The shown electronic device is only an example and should not impose any limitation on the functions and usage scope of the embodiments of the present disclosure.

[0172] Figure 3 The shown sand suction and screening separation integrated machine 100 includes: a processor 1001 and a memory 1003. Among them, the processor 1001 and the memory 1003 are connected, such as through a bus 1002. Optionally, the sand suction and screening separation integrated machine 100 may further include a communication component 1004, and the communication component 1004 may be used for data interaction between the device 100 and other devices, such as sending and / or receiving data, etc. It should be noted that in actual scheduling, the communication component 1004 is not limited to one, and the structure of the sand suction and screening separation integrated machine 100 does not constitute a limitation to the embodiments of the present application.

[0173] The processor 1001 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute various exemplary logical blocks, modules, and circuits described in connection with the disclosure of this application. The processor 1001 can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0174] The bus 1002 can include a path for transmitting information between the above components. The bus 1002 can be a PCI (Peripheral Component Interconnect) bus, an EISA (Extended Industry Standard Architecture) bus, or the like. The bus 1002 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 only a thick line is used to represent it in the figure, but it does not mean that there is only one bus or one type of bus.

[0175] The memory 1003 can be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or it can also be an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory), or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media, other magnetic storage devices, or any other medium that can be used to carry or store program code and can be read by a computer, which is not limited herein.

[0176] The memory 1003 is used to store program codes for executing the embodiments of the present disclosure, and the execution is controlled by the processor 1001. The processor 1001 is used to execute the program codes stored in the memory 1003 to implement the steps shown in the embodiment of the remote fault diagnosis method for the sand suction, screening and separation integrated machine based on the Internet of Things.

[0177] The disclosed embodiment also provides a computer-readable storage medium having program code stored thereon. When the program code is executed by a processor, the steps and corresponding contents of the aforementioned embodiment of the remote fault diagnosis method for the sand suction, screening and separation integrated machine based on the Internet of Things can be implemented.

[0178] The preferred embodiments of the present disclosure are described in detail above in conjunction with the accompanying drawings; however, the present disclosure is not limited to the specific details in the above embodiments; within the technical concept of the present disclosure, various changes, modifications, substitutions and variations may be made to these embodiments, and these changes, modifications, substitutions and variations all fall within the protection scope of the present disclosure.

[0179] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any appropriate manner without contradiction, and they should also be regarded as the contents disclosed in this disclosure. In order to avoid unnecessary repetition, this disclosure will not further describe various possible combinations. The technical scope of this application is not limited to the contents in the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A remote fault diagnosis method for a sand suction and screening separation integrated machine based on the Internet of Things, characterized in that Applied to the sand suction and screening separation integrated machine, wherein the method comprises the following steps: Obtain the device historical data, deploy an Internet of Things sensor group based on the key parts of the sand suction and screening separation integrated machine, collect the vibration data, temperature data, pressure data and displacement data of the device in real time, preprocess the device historical data, normalize the device historical data and extract features from the historical data, and fuse the device historical data; Deploy an edge computing unit at the device end based on the Internet of Things sensor group, receive the Internet of Things sensor group data in real time, preliminarily analyze the Internet of Things sensor group data, classify and identify the fault types based on machine learning algorithms, upload the historical fault data to the cloud platform based on the communication layer, and issue a local alarm; Upload the processed historical fault data to the cloud platform for storage and management based on the edge computing unit. The cloud platform trains a model using deep learning algorithms. Fault type identification, occurrence time and prediction of the remaining life of the device are carried out synchronously in the model, and the Internet of Things sensor group data is fused using the particle filter algorithm; Construct a deep reinforcement learning model to simulate the repair strategies under different faults of the device, automatically adjust the device operation parameters according to the model output, and form a closed-loop control by feedback-adjusting the repair strategy through real-time monitoring of the repair effect; Respond to the construction of a remote monitoring interface on the cloud platform to display the device operation status, fault diagnosis results and maintenance suggestions, and generate maintenance suggestions and optimized operation strategies based on the fault diagnosis results and device status through a decision support system to assist in operation and maintenance decision-making.

2. The remote fault diagnosis method for the integrated sand suction and screening separator based on the Internet of Things according to claim 1, characterized in that, The obtaining of the device historical data, deploying an Internet of Things sensor group based on the key parts of the sand suction and screening separation integrated machine, collecting the vibration data, temperature data, pressure data and displacement data of the device in real time, preprocessing the device historical data, normalizing the device historical data and extracting features from the historical data, and fusing the device historical data specifically includes: Data acquisition: Install an Internet of Things sensor group at the key parts of the sand suction and screening separation integrated machine. The key parts include the screen, vibration motor and feed inlet. The Internet of Things sensor group includes vibration sensors, temperature sensors, pressure sensors and flow sensors, and transmits the data to the edge computing unit through a wireless communication module; The preprocessing of the device historical data includes denoising, normalization and missing value filling processing; Adopt time domain, frequency domain and time-frequency domain analysis methods to extract features from the device historical data. The features include mean value, variance and spectral features; Use the Kalman filter algorithm or the particle filter algorithm to fuse the Internet of Things sensor group data.

3. The remote fault diagnosis method for the integrated sand suction and screening separator based on the Internet of Things according to claim 1, characterized in that The deploying of an edge computing unit at the device end based on the Internet of Things sensor group, receiving the Internet of Things sensor group data in real time, preliminarily analyzing the Internet of Things sensor group data, classifying and identifying the fault types based on machine learning algorithms, uploading the historical fault data to the cloud platform based on the communication layer, and issuing a local alarm specifically includes: The edge computing unit conducts preliminary data analysis and fault diagnosis through direct connection with the Internet of Things sensor group; The machine learning algorithms include support vector machines and decision trees. The support vector machine responds to the edge computing unit to perform real-time fault diagnosis using the support vector machine algorithm and identify the fault type; When the support vector machine responds to the edge computing unit to determine that the device state is abnormal, the edge computing unit issues a local alarm and uploads the information to the cloud platform through the communication layer for analysis.

4. The remote fault diagnosis method for the integrated sand suction and screening separator based on the Internet of Things according to claim 1, characterized in that, Based on the edge computing unit uploading the processed historical fault data to the cloud platform for storage and management, the cloud platform uses deep learning algorithms to train a model. In the model, fault type identification, occurrence time, and prediction of the remaining life of the device are carried out synchronously. The particle filter algorithm is used to fuse the data of the Internet of Things sensor group, specifically including: According to the historical fault data, the cloud platform uses deep learning algorithms to train a model. The deep learning algorithms include convolutional neural networks and long short-term memory networks. Among them, the long short-term memory network is used to process the time-series data during device operation. The time-series data includes vibration signals and temperature changes; Multiple fault type identifications, prediction of the fault occurrence time, and prediction of the remaining service life of the device are carried out simultaneously in the same model; The particle filter algorithm is used to fuse the data of the Internet of Things sensor group.

5. The remote fault diagnosis method for the integrated sand suction and screening separator based on the Internet of Things according to claim 1, characterized in that, The construction of the deep reinforcement learning model simulates the repair strategies under different device faults, automatically adjusts the device operation parameters according to the model output, and forms a closed-loop control by feedback-adjusting the repair strategy through real-time monitoring of the repair effect, specifically including: Construct a deep reinforcement learning model to simulate the repair strategies under different device faults. Through the big data analysis module of the cloud platform, the fault patterns in the historical data of the device are identified based on K-Means algorithm clustering analysis; According to the device historical data analysis and clustering results, the cloud platform generates an automated fault repair strategy. The fault repair strategy includes recommended maintenance steps, recommended component replacement cycles, and preventive measures for common device faults; According to the repair strategy output by the deep reinforcement learning model, automatically adjust the device operation parameters. The parameters include vibration frequency and screen tension; And form a closed-loop control by feedback-adjusting the repair strategy through real-time monitoring of the repair effect.

6. The remote fault diagnosis method for the integrated sand suction and screening separator based on the Internet of Things according to claim 1, wherein, In response to building a remote monitoring interface on the cloud platform to display the device operation status, fault diagnosis results, and maintenance suggestions, based on the fault diagnosis results and device status, generate maintenance suggestions and optimized operation strategies through a decision support system to assist in operation and maintenance decision-making, specifically including: Construct an adaptive control system. Based on deep reinforcement learning, when the system detects minor faults, the minor faults include excessive vibration and temperature rise, and the device is adaptively repaired through deep reinforcement learning; Remote monitoring interface: Provide a user-friendly remote monitoring interface on the cloud platform to display the device operation status, fault diagnosis results, and maintenance suggestions; Provide maintenance suggestions and optimized operation strategies according to the fault diagnosis results and device status.

7. The remote fault diagnosis method for the integrated sand suction and screening separator based on the Internet of Things according to claim 1, wherein, It also includes the following steps: The Internet of Things sensor group also includes high-frequency acoustic sensors and industrial cameras. The preprocessing of the device historical data also includes preprocessing acoustic and visual data, and; Construct a digital twin basic model. Based on the CAD design drawings, physical parameters, component connection relationships, kinematic and dynamic equations of the device, construct the geometric model and the basic model of the physical behavior of the device, and initially construct a knowledge graph; Adopt distributed edge nodes to process visual and acoustic data, and perform preliminary real-time image and acoustic analysis at the edge; Process multi-modal sensing data in real time at the edge and trigger local alarms; Perform multi-dimensional state diagnosis through a deep learning model enhanced by a knowledge graph in the cloud; Achieve closed-loop feedback between the physical device and the virtual model through digital twin synchronization; Apply deep reinforcement learning to generate and present interpretable maintenance strategies through AR.

8. A remote fault diagnosis system for a sand suction and screening separation integrated machine based on the Internet of Things, characterized in that, Apply to the sand suction and screening separation integrated machine, and the system includes: A data acquisition module, configured to obtain device historical data, deploy an Internet of Things sensor group based on the key parts of the sand suction and screening separation integrated machine, collect the vibration data, temperature data, pressure data and displacement data of the device in real time, preprocess the device historical data, normalize the device historical data, extract features from the historical data, and fuse the device historical data; An edge computing module, configured to deploy an edge computing unit at the device end based on the Internet of Things sensor group, receive the Internet of Things sensor group data in real time, perform preliminary analysis on the Internet of Things sensor group data, classify faults and identify fault types based on machine learning algorithms, upload historical fault data to the cloud platform based on the communication layer, and issue local alarms; A communication module, configured to upload the processed historical fault data to the cloud platform for storage and management based on the edge computing unit. The cloud platform uses a deep learning algorithm to train a model. In the model, fault type identification, occurrence time and prediction of the remaining life of the device are carried out simultaneously, and the data of the Internet of Things sensor group is fused using a particle filter algorithm; A cloud platform module, configured to construct a deep reinforcement learning model to simulate the repair strategies under different faults of the device, automatically adjust the device operation parameters according to the model output, and form a closed-loop control by feedback-adjusting the repair strategy through real-time monitoring of the repair effect; An application module, configured to respond to the construction of a remote monitoring interface on the cloud platform to display the device operation status, fault diagnosis results and maintenance suggestions, and generate maintenance suggestions and optimized operation strategies through a decision support system based on the fault diagnosis results and device status to assist in operation and maintenance decision-making.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the steps of the method described in any one of claims 1-7.

10. A sand suction and screening separation integrated machine, characterized in that, Including: A memory, on which a computer program is stored; A processor, configured to execute the computer program in the memory to implement the steps of the method described in any one of claims 1-7.

Images