Crane hoisting equipment fault pre-diagnosis system and method based on multi-source sensors

Abstract

The application discloses a hoisting and lifting equipment fault pre-diagnosis system and method based on a multi-source sensor, relates to the technical field of hoisting and lifting equipment, and aims to realize comprehensive real-time monitoring of core parameters of hoisting equipment, early potential fault prediction, multi-parameter collaborative analysis, and operation and maintenance response closed-loop management, and proposes the following technical scheme: a multi-source sensing perception layer, a data transmission layer, a fault pre-diagnosis analysis layer, and a remote management and early warning layer are included, the multi-source sensing perception layer collects operation parameters of hoisting and lifting equipment, the data transmission layer transmits the collected data to a cloud analysis platform, the fault pre-diagnosis analysis layer performs fault pre-diagnosis on the data uploaded by the data transmission layer, and the remote management and early warning layer visualizes diagnosis results, pushes early warning information through multiple channels, and manages work orders. The application realizes real-time monitoring of hoisting equipment structural members, braking systems, hoisting mechanisms, control systems and safety protection devices, and improves the accuracy of fault pre-diagnosis.

Description

Technical Field

[0001] This invention relates to the field of lifting and hoisting equipment technology, and specifically to a fault pre-diagnosis system and method for lifting and hoisting equipment based on multi-source sensors. Background Technology

[0002] The fault detection and maintenance management of existing lifting and hoisting equipment mainly rely on manual inspections and single-parameter monitoring. Manual inspections cannot achieve real-time monitoring of core parameters, and faults are often only discovered after they occur, resulting in delayed fault detection. Single-parameter monitoring can only identify obvious faults that have occurred by exceeding thresholds, and cannot predict potential faults based on parameter change trends, thus lacking pre-diagnostic capabilities. The parameter monitoring of multiple systems such as structural components, braking systems, and hoisting mechanisms is fragmented, lacking multi-parameter collaborative analysis, making it difficult to identify complex faults caused by the correlation between parameters. When equipment is distributed widely, manual inspections have incomplete coverage, fault alarms and maintenance work orders are not linked enough, and maintenance response is disconnected.

[0003] Some existing technologies attempt to introduce intelligent monitoring methods. For example, Chinese patent application CN121493812A discloses an online assessment system for the safety status of tower crane jacking based on edge computing. This system uses an edge computing layer to process sensor data locally in real time and assess its safety status. However, it only monitors a limited number of parameters during the tower crane jacking process and does not cover core system parameters such as structural components, braking systems, and hoisting mechanisms of the lifting equipment. Another example is Chinese patent application CN121493745A, which discloses an intelligent inspection and management system for faults in elevator internal components. This system uses a fusion model of convolutional neural networks and long short-term memory networks for fault diagnosis. However, its application in the elevator field is problematic because the sensor types differ significantly from those in lifting equipment, and it does not achieve multi-parameter collaborative analysis and closed-loop operation and maintenance management. Summary of the Invention

[0004] This invention provides a fault pre-diagnosis system and method for lifting and hoisting equipment based on multi-source sensors, so as to realize comprehensive real-time monitoring of core parameters of lifting equipment, early prediction of potential faults, multi-parameter collaborative analysis, and closed-loop management of operation and maintenance response.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] A fault pre-diagnosis system for lifting and hoisting equipment based on multi-source sensors includes: a multi-source sensing layer for collecting core operating parameters of the lifting and hoisting equipment; it includes multiple types of sensors for monitoring structural component parameters, braking system parameters, hoisting mechanism parameters, control system parameters, and safety protection device parameters, wherein the multiple types of sensors include at least laser deflection sensors, ultrasonic flaw detection sensors, displacement sensors, torque sensors, and high-definition vision sensors.

[0007] Laser deflection sensors are deployed at the mid-span of the main beam of the lifting equipment to monitor the vertical displacement of the main beam; ultrasonic flaw detection sensors are deployed at the welds of structural components to detect internal cracks in the welds; displacement sensors are deployed at the brake gaps to monitor the gap between the brake shoes and the brake wheel; torque sensors are deployed at the brake wheel axle to monitor the braking torque; and high-definition vision sensors are deployed along the wire rope's running path to identify the number of broken wires on the wire rope surface.

[0008] The data transmission layer has its input end connected to the signal output end of the multi-source sensing layer and its output end connected to the input end of the fault pre-diagnosis analysis layer. It is used to transmit the collected data to the cloud analysis platform.

[0009] The fault pre-diagnosis analysis layer, deployed on the cloud analysis platform, includes a parameter threshold verification module, a trend prediction and fault prediction module, a multi-parameter correlation analysis module, and a fault level classification module, which are used to perform fault pre-diagnosis on the data uploaded from the data transmission layer.

[0010] The parameter threshold verification module is used to compare the collected data with the preset threshold and trigger a primary alarm when the threshold is exceeded.

[0011] The trend prediction and fault prediction module has a built-in time series prediction model, which integrates historical monitoring data, equipment runtime and environmental parameters to analyze the changing trends of each parameter and predict potential faults.

[0012] The multi-parameter correlation analysis module has built-in parameter correlation rules, which are used to perform collaborative analysis of multi-parameter trend prediction results and identify compound fault causes.

[0013] The fault severity classification module is used to classify faults into multiple levels based on their severity and output fault severity signals.

[0014] The remote control and early warning layer has its input end connected to the output end of the fault level classification module, and is used to visualize diagnostic results, push early warning information through multiple channels, and manage work orders in a closed loop.

[0015] Furthermore, the multi-source sensing layer also includes: an ultrasonic thickness sensor, an infrared ranging sensor, a signal triggering sensor, and a laser wear sensor, the outputs of which are all communicatively connected to the inputs of the edge computing nodes.

[0016] Ultrasonic thickness sensors are deployed at stress concentration points on components to monitor corrosion and thinning; infrared ranging sensors are deployed at the hook to monitor changes in hook opening; signal trigger sensors are installed in the power supply circuit of the brake electromagnet to monitor braking response time; and laser wear sensors are installed on the brake wheel tread to monitor brake wheel wear.

[0017] Furthermore, the multi-source sensing layer also includes: a diameter measurement sensor, an ultrasonic wall thickness sensor, a temperature sensor, a sensitivity detection sensor, a speed regulation accuracy monitoring module, a delay tester, a response time sensor, an accuracy calibration sensor, a drawstring displacement sensor, and a pressure sensor.

[0018] A diameter measurement sensor is installed at the wire rope to monitor changes in wire rope diameter; an ultrasonic wall thickness sensor is installed on the drum wall to monitor changes in drum wall thickness; a temperature sensor is deployed in the bearing housing housing to monitor bearing operating temperature; a sensitivity detection sensor is installed at the limit switch trigger position to detect the limit switch's action sensitivity; a speed regulation accuracy monitoring module is deployed in the frequency converter control cabinet to monitor motor speed regulation accuracy; a delay tester is installed in the signal transmission path to measure signal transmission delay; a response time sensor is integrated into the emergency stop button to record the emergency stop response time; a precision calibration sensor is configured at the output end of the overload limiter weighing sensor to calibrate the overload limiter's accuracy; a rope-type displacement sensor is installed at the buffer compression part to monitor the buffer compression amount; and a pressure sensor is deployed in the windproof rail clamping mechanism to monitor clamping force.

[0019] Furthermore, the parameter threshold verification module has a built-in standard threshold library of core parameters of the lifting equipment, which is used to compare the collected data with the threshold in real time. When the collected data exceeds the threshold, a primary alarm signal is output, and the compared data is transmitted to the trend prediction and fault prediction module.

[0020] The trend prediction and fault prediction module has a built-in CNN-LSTM neural network algorithm, which is used to receive the comparison data output by the parameter threshold verification module, integrate historical monitoring data, equipment running time and environmental parameters, analyze parameter change trends and predict potential faults, and output trend prediction results to the multi-parameter correlation analysis module.

[0021] The multi-parameter correlation analysis module has a built-in parameter correlation model, which is used to receive the trend prediction results output by the trend prediction and fault prediction module, identify the compound fault causes, and output the correlation analysis results to the fault level classification module.

[0022] The fault level classification module has built-in fault level classification rules, which are used to receive the correlation analysis results output by the multi-parameter correlation analysis module, classify faults into three levels: minor, moderate and severe, and output fault level signals to the remote control and early warning layer.

[0023] Furthermore, the CNN-LSTM neural network algorithm sequentially includes an input layer, a CNN feature extraction layer, an LSTM temporal modeling layer, a fully connected layer, and an output layer;

[0024] The CNN feature extraction layer consists of two one-dimensional convolutional layers, followed by a Batch Normalization layer.

[0025] The LSTM temporal modeling layer consists of two LSTM units and employs an Attention mechanism;

[0026] The output layer is a Softmax classification layer, which outputs three prediction results: normal, potential fault, or impending fault.

[0027] Furthermore, the data transmission layer includes edge computing nodes, 5G communication modules, and Ethernet communication modules;

[0028] The input end of the edge computing node is communicatively connected to the signal output end of the multi-source sensing layer to receive data collected by the sensors;

[0029] The edge computing node integrates a data preprocessing module to perform local data caching, outlier removal, data compression, and breakpoint resume functions;

[0030] The input ends of the 5G communication module and the Ethernet communication module are respectively connected to the output ends of the edge computing node, and the output ends of the 5G communication module and the Ethernet communication module are respectively connected to the cloud analysis platform, forming a dual-link redundant transmission architecture;

[0031] The data transmission layer uses the AES encryption algorithm to encrypt the transmitted data.

[0032] Furthermore, the data transmission layer adopts a three-tier network architecture:

[0033] The first level is the sensor field network, which consists of sensors from the multi-source sensing layer connected to the edge computing node via RS485 bus or industrial Ethernet.

[0034] The second level is the edge computing network, which consists of edge computing nodes and their integrated data preprocessing modules.

[0035] The third level is the cloud transmission network, which consists of edge computing nodes connected to the cloud analysis platform via 5G communication modules or wired Ethernet.

[0036] Furthermore, the remote control and early warning layer includes a visualization module, a multi-channel early warning module, and a work order closed-loop management module;

[0037] The input end of the visualization module is connected to the output end of the fault level classification module to receive fault level signals and provide equipment distribution maps, real-time dashboards of core parameters, fault trend curves, and historical data tracing functions.

[0038] The input end of the multi-channel early warning module is connected to the output end of the fault level classification module to receive fault level signals and generate alarm notifications such as pop-ups, SMS, APP push or email according to the fault level.

[0039] The input end of the work order closed-loop management module communicates with the output end of the multi-channel early warning module. It is used to automatically generate maintenance work orders after an alarm is triggered, and to perform full-process management of work order allocation, progress tracking, acceptance confirmation, and closed-loop archiving.

[0040] Furthermore, it also includes a hardware encryption circuit, which is bidirectionally connected to the encryption interface of the edge computing node via an I2C or SPI data bus to perform AES-256 hardware encryption on the data to be transmitted output by the edge computing node.

[0041] A fault pre-diagnosis method for a fault pre-diagnosis system for lifting and hoisting equipment based on the above-mentioned multi-source sensors includes the following steps:

[0042] Step S1: Collect the core operating parameters of the lifting and hoisting equipment through the multi-source sensing layer;

[0043] Step S2: The collected data is transmitted to the fault pre-diagnosis analysis layer of the cloud analysis platform through the data transmission layer. The edge computing node performs local caching, outlier removal, and data compression on the collected data, and transmits it to the cloud analysis platform through 5G or Ethernet dual-link encryption.

[0044] Step S3: Process the data through the fault pre-diagnosis analysis layer: First, the parameter threshold verification module performs threshold comparison, and triggers a primary alarm when the threshold is exceeded; then, the compared data is transmitted to the trend prediction and fault prediction module to analyze the parameter change trend and predict potential faults; then, the trend prediction results are transmitted to the multi-parameter correlation analysis module to identify compound fault causes; finally, the correlation analysis results are transmitted to the fault level classification module to classify faults into three levels: minor, moderate, and severe, and the fault level signal is transmitted to the remote control and early warning layer.

[0045] Step S4: The diagnostic results are visualized based on the fault level through remote control and early warning layer, and early warning information is pushed through multiple channels. Repair work orders are automatically generated for closed-loop management. Alarms are pushed to the corresponding personnel according to the fault level, and closed-loop management of work order allocation, progress tracking, and acceptance confirmation is executed.

[0046] The present invention has the following beneficial effects:

[0047] This invention achieves comprehensive real-time monitoring of lifting equipment by deploying sensors such as laser deflection sensors, ultrasonic flaw detection sensors, inductive displacement sensors, torque sensors, and high-definition vision sensors. Based on this, through the sequential collaboration of four modules—parameter threshold verification, trend prediction (improved CNN-LSTM with added Batch Normalization and Attention mechanisms), multi-parameter correlation analysis, and fault level classification—it realizes intelligent diagnosis throughout the entire process, from anomaly detection to trend prediction, and from complex fault identification to fault level classification, significantly improving the accuracy and comprehensiveness of fault pre-diagnosis.

[0048] This invention achieves a closed-loop process from fault discovery to resolution by providing visualized remote control and early warning layers, multi-channel differentiated early warning, and closed-loop work order management. Simultaneously, the data transmission layer utilizes edge computing nodes for local data caching and breakpoint resumption, combined with a dual-link redundant transmission architecture of 5G and Ethernet and AES-256 hardware encryption, ensuring the reliability, security, and continuity of data transmission. This effectively shortens fault response time, enabling a shift from manual inspection to remote intelligent monitoring, from passive response to proactive early warning, and from decentralized management to closed-loop control. Attached Figure Description

[0049] Figure 1 This is a block diagram of the architecture of the fault pre-diagnosis system for lifting and hoisting equipment based on multi-source sensors of the present invention;

[0050] Figure 2 This is a flowchart of the fault pre-diagnosis method of the present invention. Detailed Implementation

[0051] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0052] Please refer to Figure 1 This embodiment provides a fault pre-diagnosis system for lifting and hoisting equipment based on multi-source sensors. The system includes a multi-source sensing layer, a data transmission layer, a fault pre-diagnosis analysis layer, and a remote control and early warning layer. Each layer is sequentially connected to form a complete closed loop from data acquisition, transmission, analysis to early warning and control.

[0053] The fault pre-diagnosis remote analysis system includes: a multi-source sensing layer for collecting core operating parameters of lifting and hoisting equipment; it includes multiple types of sensors for monitoring structural component parameters, braking system parameters, hoisting mechanism parameters, control system parameters, and safety protection device parameters, including at least laser deflection sensors, ultrasonic flaw detection sensors, displacement sensors, torque sensors, and high-definition vision sensors.

[0054] Laser deflection sensors are deployed at the mid-span of the main beam of the lifting equipment to monitor the vertical displacement of the main beam; ultrasonic flaw detection sensors are deployed at the welds of structural components to detect internal cracks in the welds; inductive displacement sensors are deployed at the brake gaps to monitor the gap between the brake shoes and the brake wheel; torque sensors are deployed at the brake wheel axle to monitor the braking torque; and high-definition vision sensors are deployed along the wire rope's running path to identify the number of broken wires on the wire rope surface.

[0055] The data transmission layer has its input end connected to the signal output end of the multi-source sensing layer and its output end connected to the input end of the fault pre-diagnosis analysis layer. It is used to transmit the collected data to the cloud analysis platform. The fault pre-diagnosis analysis layer is deployed on the cloud analysis platform and includes a parameter threshold verification module, a trend prediction and fault prediction module, a multi-parameter correlation analysis module, and a fault level classification module. It is used to perform fault pre-diagnosis on the data uploaded by the data transmission layer.

[0056] The parameter threshold verification module compares the collected data with preset thresholds and triggers a primary alarm when the threshold is exceeded. The trend prediction and fault prediction module has a built-in time-series prediction model that integrates historical monitoring data, equipment runtime, and environmental parameters to analyze the changing trends of each parameter and predict potential faults. The multi-parameter correlation analysis module has built-in parameter correlation rules that are used to perform collaborative analysis of multi-parameter trend prediction results and identify compound fault causes. The fault level classification module classifies faults into multiple levels according to their severity and outputs fault level signals for remote control and early warning. Its input end is connected to the output end of the fault level classification module for visualizing diagnostic results, pushing early warning information through multiple channels, and managing work orders in a closed loop.

[0057] The multi-source sensing layer is the data acquisition front-end of this system, covering the three core subsystems of the lifting and hoisting equipment: structural components, braking system, and hoisting mechanism. Specifically:

[0058] For structural components, the multi-source sensing layer includes laser deflection sensors and ultrasonic flaw detection sensors. The laser deflection sensor is deployed at the mid-span of the main beam, employing the laser triangulation principle to monitor the vertical displacement of the main beam under load in real time, with a sampling frequency of no less than 1 time / minute and a measurement accuracy of ±0.1mm. The ultrasonic flaw detection sensor is deployed at the weld seams of the structural components, using the pulse reflection principle and operating at a frequency of 5MHz, to detect internal defects such as cracks and porosity in the weld seams.

[0059] For the braking system, the multi-source sensing layer includes an inductive displacement sensor and a torque sensor. The inductive displacement sensor is deployed at the brake gap location, employing an inductive measurement principle with a measurement range of 0-5mm and a measurement accuracy of ±0.01mm, used to monitor changes in the gap between the brake shoe and the brake wheel. The torque sensor is deployed at the brake wheel axle end, employing a strain gauge bridge principle, used to monitor the magnitude of the braking torque.

[0060] For the hoisting mechanism, the multi-source sensing layer includes a high-definition vision sensor. This sensor is deployed along the wire rope's running path, using an industrial camera and a ring-shaped LED light source to identify the number of broken wires on the wire rope surface through machine vision algorithms.

[0061] The input of the data transmission layer is communicatively connected to the signal output of the multi-source sensing layer. In this embodiment, each sensor transmits the acquired analog signals to the data transmission layer via an RS485 bus or industrial Ethernet. The data transmission layer then transmits the acquired data to the fault pre-diagnosis analysis layer of the cloud analysis platform.

[0062] The fault pre-diagnosis analysis layer is deployed on a cloud-based analysis platform, and its input end is communicatively connected to the output end of the data transmission layer. This layer includes a parameter threshold verification module, a trend prediction and fault prediction module, a multi-parameter correlation analysis module, and a fault level classification module, which are used to perform fault pre-diagnosis on the data uploaded by the data transmission layer.

[0063] The parameter threshold verification module has a built-in standard threshold library of core parameters of the lifting equipment. It compares the collected data with the preset threshold in real time. When the collected data exceeds the threshold, it outputs a primary alarm signal and transmits the compared data (including the collected value, the degree of deviation, whether it exceeds the threshold, etc.) to the trend prediction and fault prediction module.

[0064] The trend prediction and fault prediction module incorporates a CNN-LSTM neural network algorithm. It receives the comparison data output by the parameter threshold verification module, integrates historical monitoring data, equipment runtime and environmental parameters, analyzes parameter change trends and predicts potential faults, and outputs trend prediction results to the multi-parameter correlation analysis module.

[0065] The multi-parameter correlation analysis module has a built-in parameter correlation model, which is used to receive the trend prediction results output by the trend prediction and fault prediction module, identify the compound fault causes, and output the correlation analysis results to the fault level classification module.

[0066] The fault level classification module has built-in fault level classification rules. It receives the correlation analysis results output by the multi-parameter correlation analysis module and classifies the fault into three levels: minor, moderate, and severe, based on the degree of parameter deviation, the scope of fault impact, and the level of safety risk. It then outputs the fault level signal to the remote control and early warning layer.

[0067] The remote management and early warning layer is used to visualize diagnostic results, push early warning information through multiple channels, and manage work orders in a closed loop.

[0068] As a preferred embodiment, the following sensors or instruments can also be used to comprehensively cover the core parameters of the lifting equipment control system and safety protection devices, forming a core parameter monitoring system together with the sensors of the aforementioned structural components, braking system, and hoisting mechanism.

[0069] Specifically, the multi-source sensing layer also includes: an ultrasonic thickness sensor, an infrared ranging sensor, a signal triggering sensor, and a laser wear sensor. The ultrasonic thickness sensor is deployed at stress concentration points on the component (such as the connection between the main beam and the end beam, and the root of the lifting lug) to monitor corrosion and thinning of the component; the infrared ranging sensor is deployed at the hook to monitor changes in hook opening; the signal triggering sensor is installed in the brake electromagnet power supply circuit to monitor braking response time; and the laser wear sensor is installed on the brake wheel tread to monitor the amount of brake wheel wear.

[0070] As a further preferred embodiment, the multi-source sensing layer also includes: a diameter measurement sensor, an ultrasonic wall thickness sensor, a temperature sensor, a sensitivity detection sensor, a speed regulation accuracy monitoring module, a delay tester, a response time sensor, a precision calibration sensor, a drawstring displacement sensor, and a pressure sensor. Specifically, the diameter measurement sensor is installed at the wire rope to monitor changes in wire rope diameter; the ultrasonic wall thickness sensor is installed on the drum wall to monitor changes in drum wall thickness; the temperature sensor is deployed on the bearing housing housing to monitor bearing operating temperature; the sensitivity detection sensor is installed at the limit switch trigger position to detect the limit switch's action sensitivity; the speed regulation accuracy monitoring module is deployed inside the frequency converter control cabinet to monitor motor speed regulation accuracy; the delay tester is installed on the signal transmission path to measure signal transmission delay; the response time sensor is integrated into the emergency stop button to record the emergency stop response time; the precision calibration sensor is configured at the output end of the overload limiter weighing sensor to calibrate the overload limiter's measurement accuracy; the drawstring displacement sensor is installed at the buffer compression part to monitor the buffer compression amount; and the pressure sensor is deployed on the windproof rail clamping mechanism to monitor clamping force.

[0071] As a preferred implementation, the parameter threshold verification module incorporates a built-in standard threshold library for core parameters of the lifting equipment. This threshold library, established based on national standards, includes three levels of thresholds: normal range, warning threshold, and alarm threshold for 12 categories of core parameters. For example, the normal range for the static deflection at mid-span of the main beam is ≤L / 1000 (where L is the main beam span), the normal range for the number of broken wires in the wire rope is ≤5% of the total number of wires, the normal range for the braking clearance is 0.5-1.5mm, and the normal range for the bearing temperature is 55-65℃.

[0072] The parameter threshold verification module receives sensor data uploaded from the data transmission layer in real time and compares the measured values ​​with the threshold database. When the measured value is within the normal range, the module outputs a normal status; when the measured value is within the warning threshold range, the module outputs a warning status and triggers a yellow primary alarm signal; when the measured value exceeds the alarm threshold, the module outputs an alarm status and triggers a red primary alarm signal. After the comparison is completed, the module transmits the compared data (including the original collected values, the degree of deviation, whether the threshold is exceeded, etc.) to the trend prediction and fault prediction module.

[0073] The trend prediction and fault prediction module incorporates a CNN-LSTM neural network algorithm. This module receives comparison data from the parameter threshold verification module, and simultaneously reads historical monitoring data (stored in a cloud database), equipment runtime (read from the equipment control system), and environmental parameters (such as ambient temperature, humidity, and wind speed, read from environmental sensors). The module fuses this multi-source information and analyzes the changing trends of each parameter using the CNN-LSTM neural network algorithm to predict the timing and type of potential faults.

[0074] For example, when the wear of the brake wheel increases by more than 0.5% per month, the module calculates the estimated time to reach the wear limit based on the current wear rate and remaining wear, predicting that the wear limit will be reached in 30 days. When the bearing temperature continues to rise and the daily average increase exceeds 2°C, the module predicts a risk of bearing lubrication failure. After completing the trend prediction, the module outputs the trend prediction results (including the predicted fault type, estimated occurrence time, confidence level, etc.) to the multi-parameter correlation analysis module.

[0075] The multi-parameter correlation analysis module incorporates a parameter correlation model. This model is built upon a Bayesian network, where nodes represent parameter variables and edges represent causal relationships between parameters. This module receives trend prediction results from the trend prediction and fault prediction modules and identifies complex fault causes through Bayesian network inference.

[0076] For example, when increased brake clearance and insufficient braking torque occur simultaneously, the module infers that brake shoe wear is the root cause, classifying it as a brake system linkage failure. When abnormal main beam deflection and excessive component corrosion thickness occur simultaneously, the module infers that component corrosion is the root cause, classifying it as a decrease in structural load-bearing capacity. After the correlation analysis is completed, the module outputs the correlation analysis results (including composite fault types, root causes, and a list of correlation parameters) to the fault level classification module.

[0077] The fault classification module has built-in fault classification rules. This module receives the correlation analysis results output by the multi-parameter correlation analysis module and classifies faults into three levels: minor, moderate, and severe, based on three dimensions: parameter deviation, fault impact range, and safety risk level.

[0078] Minor fault: The parameter is close to the warning threshold but not exceeded. The fault is limited to a single component and does not affect the normal operation of the equipment. The safety risk level is low. For example, if the brake clearance is close to the warning threshold but not exceeded, it is judged as a minor fault, and the handling recommendation is to carry out maintenance within a time limit (within 30 days).

[0079] Medium-level fault: Parameters exceed the warning threshold but do not reach the alarm threshold. The fault affects multiple related components, impacting equipment operating efficiency, and the safety risk level is medium. For example, if the brake clearance exceeds the warning threshold and the braking torque decreases, it is classified as a medium-level fault, and the recommended handling is to perform maintenance within a specified period (within 7 days).

[0080] Serious Fault: Parameters exceed alarm thresholds, the fault affects the overall safety of the machine, posing a significant safety hazard, and the safety risk level is high. For example, if the brake clearance exceeds the alarm threshold, the braking torque drops significantly, and the braking response time is significantly prolonged, it is determined to be a serious fault, and the recommended action is to immediately stop the machine for repair.

[0081] After the fault level is classified, the module outputs the fault level signal (including fault level, handling suggestions, and maintenance time limit) to the remote control and early warning layer.

[0082] As a preferred implementation, the CNN-LSTM neural network algorithm sequentially includes an input layer, a CNN feature extraction layer, an LSTM temporal modeling layer, a fully connected layer, and an output layer.

[0083] The input layer receives historical monitoring data from multiple sensors. The data dimension is the number of samples × time step × number of features, where the time step is set to 720, corresponding to 720 minutes, or 12 hours, of historical data; the number of features consists of 12 core parameters; and the number of samples is the number of training samples generated by the sliding window.

[0084] The CNN feature extraction layer consists of two one-dimensional convolutional layers. The first layer has 64 kernels, a kernel size of 3, a stride of 1, and uses ReLU activation. The output dimension is (number of samples × 718 × 64). The second layer has 128 kernels, a kernel size of 3, a stride of 1, and uses ReLU activation. The output dimension is (number of samples × 716 × 128). A BatchNormalization layer is added after each convolutional layer to accelerate training convergence. The CNN feature extraction layer extracts local variation features of parameters through local receptive fields, such as short-term fluctuations in brake wheel wear and periodic fluctuations in bearing temperature.

[0085] The LSTM temporal modeling layer consists of two LSTM layers. The first layer has 128 LSTM units, and the second layer has 64. The LSTM layer models the long-term dependencies of parameters through gating mechanisms (input gate, forget gate, output gate), such as the cumulative trend of brake wheel wear and the continuous increase in bearing temperature. The LSTM layer employs an attention mechanism to enhance the focus on key parameters, enabling the model to automatically identify parameter features that contribute significantly to fault prediction. The output dimension of the LSTM layer is the number of samples × 64.

[0086] A fully connected layer contains a fully connected neural network with 32 neurons, ReLU activation function, and an output dimension of 32 (number of samples).

[0087] The output layer is a Softmax classification layer with an output dimension of sample number × 3, corresponding to three prediction categories: normal, potential fault, and impending fault. The Softmax function calculates the probability of each category, and the category with the highest probability is the prediction result.

[0088] In one preferred embodiment, the data transmission layer includes an edge computing node, a 5G communication module, and an Ethernet communication module.

[0089] Edge computing nodes are deployed within the electrical control cabinet of the lifting equipment. Their input terminals are connected to the signal output terminals of the multi-source sensing layer via an RS485 bus or industrial Ethernet to receive data collected by the sensors. In this embodiment, the edge computing nodes utilize industrial-grade embedded computers.

[0090] The edge computing node integrates a data preprocessing module. This module runs an embedded Linux operating system and performs the following functions: encapsulates raw sensor data in standard JSON format and stores it on a local storage chip, capable of storing 72 hours of historical data; identifies and removes sensor outliers, such as data exceeding the sensor's range or sudden abrupt changes, using statistical analysis methods; and compresses the data using the LZ77 algorithm to reduce the amount of data transmitted. When the network is interrupted, the edge computing node continuously caches data; once the network is restored, it automatically resumes transmission from the point of interruption to avoid data loss.

[0091] The input terminals of the 5G communication module and the Ethernet communication module are respectively connected to the output terminals of the edge computing node. The 5G communication module adopts a 5G industrial gateway, supports SA (Standalone) networking architecture, dual SIM dual standby, and automatic network switching. The Ethernet communication module adopts the Gigabit Ethernet standard. The output terminals of the 5G communication module and the Ethernet communication module are respectively connected to the cloud analysis platform, forming a dual-link redundant transmission architecture: when the primary network (such as wired Ethernet) fails, the system automatically switches to the backup network (5G) to ensure the continuity of data transmission.

[0092] The data transmission layer employs the AES-256 encryption algorithm to encrypt transmitted data. The encryption process is executed at the edge computing nodes, which have a built-in TPM2.0 trusted platform module hardware encryption chip. The key length is 256 bits, and the key update cycle is 24 hours. The encryption algorithm uses CBC mode to ensure the security of data transmission.

[0093] In this embodiment, the data transmission layer adopts a three-level network architecture:

[0094] Level 1: Sensor Field Network. Sensors in the multi-source sensing layer are connected to edge computing nodes via RS485 bus or industrial Ethernet. The field network adopts a star topology, and a single edge computing node can access 32 sensor signals with a transmission distance of ≤100m.

[0095] Level 2: Edge computing network. Edge computing nodes run an embedded Linux operating system and integrate a data preprocessing module. This node is deployed inside the crane equipment's electrical control cabinet and is responsible for local preprocessing, caching, and compression of raw data.

[0096] Level 3: Cloud Transmission Network. Edge computing nodes connect to the cloud analytics platform via 5G modules or wired Ethernet. The cloud transmission network prioritizes wired Ethernet, automatically switching to 5G when the wired network becomes unavailable. Transmission latency is measured by timestamp comparison; the time difference between sensor sampling and cloud analytics platform reception is ≤5ms.

[0097] In this embodiment, the remote control and early warning layer includes a visualization module, a multi-channel early warning module, and a work order closed-loop management module.

[0098] The input of the visualization module is communicatively connected to the output of the fault level classification module, receiving fault level signals. This module provides the following visualization functions:

[0099] Equipment distribution map: Based on the GIS geographic information system, the location of lifting equipment is marked on the electronic map, and the color of the equipment icon indicates the equipment status (green normal, yellow warning, orange medium fault, red serious fault).

[0100] Real-time dashboard for core parameters: It adopts a combination of instrument panel and digital display. Each of the 12 types of parameters corresponds to one instrument panel. The instrument panel pointer points to the current measurement value, and the background color partition indicates the normal range, warning range, and alarm range.

[0101] Fault trend curve: A time series line chart is used, with time on the horizontal axis and parameter values ​​on the vertical axis. The curve color indicates the parameter status and supports the display of multiple parameters.

[0102] Historical data traceability: Supports drill-down analysis by device, parameter type, and time dimension, and allows querying historical data of any device, any parameter, and any time period.

[0103] The input of the multi-channel early warning module communicates with the output of the fault level classification module, receiving fault level signals and generating alarm notifications based on the fault level. This module supports four alarm methods: 1) Pop-up alarm window in a PC browser or app, displaying alarm details and a handling button; 2) Sending alarm SMS to maintenance personnel's mobile phones via SMS gateway, with the SMS content including device name, parameter type, current value, threshold, and fault level; 3) Pushing alarm notifications to maintenance personnel's mobile app via push service, with the push message displayed in the phone's notification bar; 4) Sending alarm emails to maintenance personnel's mailboxes via email server, with the email content including alarm details, fault trend charts, and handling suggestions; 5) Notifying the device administrator for minor faults, the device administrator and maintenance supervisor for medium faults, and the device administrator, maintenance supervisor, and safety manager for serious faults.

[0104] The input end of the work order closed-loop management module communicates with the output end of the multi-channel early warning module. It is used to automatically generate maintenance work orders after an alarm is triggered, and to perform full-process management of work order allocation, progress tracking, acceptance confirmation, and closed-loop archiving.

[0105] The work order content includes: work order number, equipment name, fault location, parameter anomaly details, fault level, predicted fault type, handling suggestions, creation time, and required completion time. Work order allocation supports automatic allocation (based on maintenance personnel skill tags and workload balancing algorithms) and manual allocation (specified by the equipment administrator). Work order status includes six states: pending allocation, allocated, processing, pending acceptance, completed, and archived. Maintenance personnel can update work order progress and upload on-site photos and maintenance records via a mobile app. After maintenance is completed, the equipment administrator conducts acceptance confirmation, and the work order is archived and added to the historical work order database, supporting queries and statistics by equipment, fault type, maintenance personnel, and time range.

[0106] In this embodiment, the system also includes a hardware encryption circuit. The hardware encryption circuit is bidirectionally connected to the edge computing nodes in the data transmission layer via a data bus (such as an I2C bus or an SPI bus) and is used to perform AES-256 hardware encryption on the transmitted data. Specifically, the hardware encryption circuit uses the TPM2.0 trusted platform module chip, which integrates an AES-256 encryption engine and a true random number generator. The encryption process is as follows: the processor of the edge computing node writes the data packet to be transmitted (containing device ID, timestamp, sensor values, etc.) into the input buffer of the encryption chip. The encryption chip calls its internal encryption engine to perform hardware-level encryption operations on the data in CBC mode. After encryption, the encrypted data is returned via the data bus. The edge computing node reads the encrypted data and concatenates it with the data packet header to form the final transmission frame for transmission.

[0107] This embodiment also provides a fault pre-diagnosis method based on the above system, including the following steps:

[0108] Step S1, Data Acquisition

[0109] The core operating parameters of the lifting and hoisting equipment are collected through a multi-source sensing layer. The key components of the lifting and hoisting equipment include structural parts, braking systems, and hoisting mechanisms. Specifically, laser deflection sensors, ultrasonic flaw detection sensors, displacement sensors, torque sensors, and high-definition vision sensors synchronously collect each parameter at a set sampling frequency (≥1 time / minute) and output the collected analog signals to the data transmission layer.

[0110] Step S2, Data Transmission

[0111] The collected data is transmitted to the cloud-based analytics platform via a data transmission layer. Specifically, edge computing nodes receive sensor data, perform local data caching, outlier removal, data compression, and breakpoint resumption processing, and then transmit the data to the fault pre-diagnosis analysis layer of the cloud-based analytics platform via a 5G or Ethernet communication module. The transmission process employs AES-256 hardware encryption to ensure data security.

[0112] Step S3, Fault Pre-diagnosis

[0113] The data is processed through a fault pre-diagnosis analysis layer. The processing flow is as follows:

[0114] First, the parameter threshold verification module performs threshold comparison. This module compares the collected data with the built-in standard threshold library. When the collected data exceeds the threshold, a primary alarm signal is triggered, and the compared data (including the collected value, the degree of deviation, whether the threshold is exceeded, etc.) is transmitted to the trend prediction and fault prediction module.

[0115] Then, the trend prediction and fault prediction module receives the compared data, integrates historical monitoring data, equipment runtime and environmental parameters, analyzes the parameter change trend and predicts potential faults through the CNN-LSTM neural network algorithm, and outputs the trend prediction results (including the predicted fault type, expected occurrence time, confidence level, etc.) to the multi-parameter correlation analysis module.

[0116] Next, the multi-parameter correlation analysis module receives the trend prediction results, identifies the compound fault causes through the parameter correlation model, and outputs the correlation analysis results (including compound fault types, root causes, and a list of correlation parameters) to the fault level classification module.

[0117] Finally, the fault level classification module receives the correlation analysis results and classifies the fault into three levels: minor, moderate, and severe, based on the degree of parameter deviation, the scope of fault impact, and the level of safety risk. The fault level signal (including fault level, handling suggestions, and maintenance time limit) is then transmitted to the remote control and early warning layer.

[0118] Step S4: Remote Control and Early Warning

[0119] The remote control and early warning layer visualizes the diagnostic results based on the fault level, pushes early warning information through multiple channels, automatically generates maintenance work orders for closed-loop management, pushes alarms to the corresponding personnel according to the fault level, and performs closed-loop management of work order allocation, progress tracking, and acceptance confirmation.

[0120] Specifically, the visualization module receives fault level signals, identifies equipment status using different colors on the equipment distribution map, highlights abnormal parameters on the real-time dashboard, and marks the fault occurrence time on the trend curve. The multi-channel early warning module generates corresponding alarm notifications based on the fault level: minor faults are notified to the equipment administrator via APP push or email; medium faults are notified to both the equipment administrator and maintenance manager via SMS and APP push; and severe faults are notified to the equipment administrator, maintenance manager, and safety supervisor simultaneously via telephone, SMS, and APP push. The work order closed-loop management module automatically generates maintenance work orders after an alarm is triggered, assigns them to designated maintenance personnel, and tracks the entire process of work order allocation, processing, acceptance, and archiving until the fault is completely resolved.

[0121] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multi-source sensor-based fault pre-diagnosis system for a hoisting and lifting equipment, characterized in that, include: The multi-source sensing layer is used to collect the core operating parameters of lifting and hoisting equipment; It includes multiple types of sensors for monitoring structural parameters, braking system parameters, hoisting mechanism parameters, control system parameters, and safety protection device parameters. These multiple types of sensors include at least laser deflection sensors, ultrasonic flaw detection sensors, displacement sensors, torque sensors, and high-definition vision sensors. The laser deflection sensor is deployed at the mid-span of the main beam of the lifting equipment to monitor the vertical displacement of the main beam; the ultrasonic flaw detection sensor is deployed at the weld of the structural component to detect internal cracks in the weld; and the displacement sensor is deployed at the brake gap to monitor the gap between the brake shoe and the brake wheel. The torque sensor is deployed at the brake wheel axle end to monitor braking torque; the high-definition vision sensor is deployed along the wire rope running path to identify the number of broken wires on the wire rope surface. The data transmission layer has its input end connected to the signal output end of the multi-source sensing layer and its output end connected to the input end of the fault pre-diagnosis analysis layer, and is used to transmit the collected data to the cloud analysis platform. The fault pre-diagnosis analysis layer, deployed on the cloud analysis platform, includes a parameter threshold verification module, a trend prediction and fault prediction module, a multi-parameter correlation analysis module, and a fault level classification module, which are used to perform fault pre-diagnosis on the data uploaded from the data transmission layer. The parameter threshold verification module is used to compare the collected data with the preset threshold and trigger a primary alarm when the threshold is exceeded. The trend prediction and fault prediction module has a built-in time series prediction model, which is used to integrate historical monitoring data, equipment runtime and environmental parameters, analyze the changing trends of each parameter and predict potential faults. The multi-parameter correlation analysis module has built-in parameter correlation rules, which are used to perform collaborative analysis of multi-parameter trend prediction results and identify compound fault causes. The fault level classification module is used to classify faults into multiple levels according to their severity and output fault level signals. The remote control and early warning layer has its input end connected to the output end of the fault level classification module, and is used to visualize the diagnostic results, push early warning information through multiple channels, and manage work orders in a closed loop.

2. The fault pre-diagnosis system for lifting and hoisting equipment based on multi-source sensors according to claim 1, characterized in that, The multi-source sensing layer also includes an ultrasonic thickness sensor, an infrared ranging sensor, a signal triggering sensor, and a laser wear sensor, the outputs of which are all communicatively connected to the input of the edge computing node. The ultrasonic thickness sensor is deployed at the stress concentration point of the component to monitor the corrosion and thinning of the component; the infrared ranging sensor is deployed at the hook to monitor the change in the hook opening; the signal triggering sensor is installed in the power supply circuit of the brake electromagnet to monitor the braking response time; and the laser wear sensor is installed on the brake wheel tread to monitor the amount of brake wheel wear.

3. The fault pre-diagnosis system for lifting and hoisting equipment based on multi-source sensors according to claim 1, characterized in that, The multi-source sensing layer also includes: a diameter measurement sensor, an ultrasonic wall thickness sensor, a temperature sensor, a sensitivity detection sensor, a speed regulation accuracy monitoring module, a delay tester, a response time sensor, an accuracy calibration sensor, a drawstring displacement sensor, and a pressure sensor. The diameter measurement sensor is installed at the wire rope to monitor changes in wire rope diameter; the ultrasonic wall thickness sensor is installed on the drum wall to monitor changes in drum wall thickness; the temperature sensor is deployed on the bearing housing housing to monitor bearing operating temperature; the sensitivity detection sensor is installed at the limit switch trigger position to detect the limit switch's action sensitivity; the speed regulation accuracy monitoring module is deployed in the frequency converter control cabinet to monitor motor speed regulation accuracy; the delay tester is installed on the signal transmission path to measure signal transmission delay; the response time sensor is integrated into the emergency stop button to record the emergency stop response time; the accuracy calibration sensor is configured at the overload limiter weighing sensor output end to calibrate the overload limiter accuracy; the rope-type displacement sensor is installed at the buffer compression part to monitor the buffer compression amount; and the pressure sensor is deployed on the windproof rail clamping mechanism to monitor clamping force.

4. The fault pre-diagnosis system for lifting and hoisting equipment based on multi-source sensors according to claim 1, characterized in that: The parameter threshold verification module has a built-in standard threshold library of core parameters of lifting equipment, which is used to compare the collected data with the threshold in real time. When the collected data exceeds the threshold, it outputs a primary alarm signal and transmits the compared data to the trend prediction and fault prediction module. The trend prediction and fault prediction module has a built-in CNN-LSTM neural network algorithm, which is used to receive the comparison data output by the parameter threshold verification module, integrate historical monitoring data, equipment running time and environmental parameters, analyze parameter change trends and predict potential faults, and output trend prediction results to the multi-parameter correlation analysis module. The multi-parameter correlation analysis module has a built-in parameter correlation model, which is used to receive the trend prediction results output by the trend prediction and fault prediction module, identify the compound fault causes, and output the correlation analysis results to the fault level classification module. The fault level classification module has built-in fault level classification rules, which are used to receive the correlation analysis results output by the multi-parameter correlation analysis module, classify faults into three levels: minor, moderate and severe, and output fault level signals to the remote control and early warning layer.

5. The fault pre-diagnosis system for lifting and hoisting equipment based on multi-source sensors according to claim 4, characterized in that, The CNN-LSTM neural network algorithm comprises, in sequence, an input layer, a CNN feature extraction layer, an LSTM temporal modeling layer, a fully connected layer, and an output layer; The CNN feature extraction layer consists of two one-dimensional convolutional layers, followed by a Batch Normalization layer. The LSTM temporal modeling layer contains two LSTM units and employs an Attention mechanism; The output layer is a Softmax classification layer, which outputs three types of prediction results: normal, potential fault, or impending fault.

6. The fault pre-diagnosis system for lifting and hoisting equipment based on multi-source sensors according to claim 1, characterized in that, The data transmission layer includes edge computing nodes, 5G communication modules, and Ethernet communication modules; the data transmission layer also includes a link switching control module, whose input is connected to the status output of the 5G communication module and the Ethernet communication module respectively, and whose output is connected to the control terminal of both, for automatically switching to the backup link when the primary link fails; The input end of the edge computing node is communicatively connected to the signal output end of the multi-source sensing layer, and is used to receive data collected by the sensor; The edge computing node integrates a data preprocessing module for performing local data caching, outlier removal, data compression, and breakpoint resume functions. The input ends of the 5G communication module and the Ethernet communication module are respectively connected to the output end of the edge computing node, and the output ends of the 5G communication module and the Ethernet communication module are respectively connected to the cloud analysis platform, forming a dual-link redundant transmission architecture. The data transmission layer uses the AES encryption algorithm to encrypt the transmitted data.

7. The fault pre-diagnosis system for lifting and hoisting equipment based on multi-source sensors according to claim 6, characterized in that, The data transmission layer adopts a three-level network architecture: The first level is the sensor field network, which consists of sensors from the multi-source sensing layer connected to the edge computing node via RS485 bus or industrial Ethernet. The second level is the edge computing network, which consists of the edge computing nodes and their integrated data preprocessing modules. The third level is the cloud transmission network, which consists of the edge computing nodes connected to the cloud analysis platform via the 5G communication module or wired Ethernet.

8. The fault pre-diagnosis system for lifting and hoisting equipment based on multi-source sensors according to claim 1, characterized in that, The remote control and early warning layer includes a visualization module, a multi-channel early warning module, and a work order closed-loop management module; The input end of the visualization module is communicatively connected to the output end of the fault level classification module, and is used to receive fault level signals and provide equipment distribution maps, real-time dashboards of core parameters, fault trend curves, and historical data tracing functions. The input end of the multi-channel early warning module is communicatively connected to the output end of the fault level classification module, and is used to receive fault level signals and generate alarm notifications such as pop-ups, SMS, APP push or email according to the fault level. The input end of the work order closed-loop management module is communicatively connected to the output end of the multi-channel early warning module. It is used to automatically generate maintenance work orders after an alarm is triggered, and to perform full-process management of work order allocation, progress tracking, acceptance confirmation, and closed-loop archiving.

9. The fault pre-diagnosis system for lifting and hoisting equipment based on multi-source sensors according to claim 1, characterized in that, It also includes a hardware encryption circuit, which is bidirectionally connected to the encryption interface of the edge computing node via an I2C or SPI data bus, and is used to perform AES-256 hardware encryption on the data to be transmitted output by the edge computing node.

10. A fault pre-diagnosis method based on the multi-source sensor-based fault pre-diagnosis system for lifting and hoisting equipment as described in claim 4, characterized in that, Includes the following steps: Step S1: Collect the core operating parameters of the lifting and hoisting equipment through the multi-source sensing layer; Step S2: The collected data is transmitted to the fault pre-diagnosis analysis layer of the cloud analysis platform through the data transmission layer. The edge computing node performs local caching, outlier removal, and data compression on the collected data, and transmits it to the cloud analysis platform through 5G or Ethernet dual-link encryption. Step S3: The data is processed through the fault pre-diagnosis analysis layer: First, the parameter threshold verification module performs threshold comparison, and triggers a primary alarm when the threshold is exceeded; then, the compared data is transmitted to the trend prediction and fault prediction module to analyze the parameter change trend and predict potential faults; then, the trend prediction results are transmitted to the multi-parameter correlation analysis module to identify compound fault causes; finally, the correlation analysis results are transmitted to the fault level classification module to classify faults into three levels: minor, moderate, and severe, and the fault level signal is transmitted to the remote control and early warning layer. Step S4: The remote control and early warning layer visualizes the diagnostic results according to the fault level, pushes early warning information through multiple channels, automatically generates maintenance work orders for closed-loop management, pushes alarms to the corresponding personnel according to the fault level, and performs closed-loop management of work order allocation, progress tracking, and acceptance confirmation.

Images