Hoisting machinery wheel fault monitoring system and detection method

Abstract

The invention discloses a hoisting machinery wheel fault monitoring system and a detection method. The monitoring system comprises a multi-axis vibration sensing module, an anti-interference transmission module, an edge calculation module and a data management module. The multi-axis vibration sensing module comprises a plurality of three-axis acceleration sensors, and the three-axis acceleration sensors are installed on wheel hubs of hoisting machinery wheels, bearing pedestals and track contact faces respectively; the anti-interference transmission module comprises a main channel and a standby channel, and adopts an industrial Ethernet to communicate with 5G dual modes; an FPGA chip is arranged in the edge calculation module; the data management module comprises a data storage layer, a data service layer and a data interface layer. Through multi-axis sensor fusion, an anti-interference transmission architecture and an edge computing technology, a set of high-precision and full-time-domain wheel rolling fault vibration signal acquisition system is constructed, so that the detection precision and efficiency are improved, the anti-interference and reliability are enhanced, the operation and maintenance cost and risk are reduced, and a reliable guarantee is provided for safe operation of hoisting machinery.

Description

Technical Field

[0001] The present invention relates to the technical field of hoisting machinery detection, and particularly relates to a wheel fault monitoring system and detection method for hoisting machinery. Background Art

[0002] As a core equipment in industrial production, hoisting machinery needs to regularly detect the running states of large and small wheels. Existing wheel fault detection devices for hoisting machinery generally have problems such as single signal dimension, poor real-time performance, and weak anti-interference ability, which are difficult to meet the requirements of dynamic monitoring and risk warning, resulting in accidents such as derailment and overturning caused by faults such as flange wear, bearing jamming, and track deformation.

[0003] The existing detection technologies have the following problems in practical applications: First, the detection means are lagging behind and cannot meet the timeliness requirements of regulations. Traditional methods rely on manual inspections or regular disassembly inspections, with low efficiency and easy omission of instantaneous faults; for example, hidden dangers such as microcracks generated during frequent start-stop of large vehicle running wheels and bearing lubrication failure are difficult to capture in time by the naked eye or contact sensors, resulting in long-term accumulation of safety hazards. Second, the signal acquisition dimension is single and the data reliability is insufficient, unable to comprehensively reflect the wheel rolling state; in actual operation, wheel faults are often accompanied by abnormal radial, tangential, and composite vibration signals, and single-point detection is prone to misjudgment due to signal interference or installation deviation; the false alarm rate of faults due to unreasonable sensor layout is high. Third, the data transmission and processing capabilities are insufficient and it is difficult to give early warnings in time. Hoisting machinery needs to have real-time status monitoring functions, but existing devices mostly use wired transmission or low-sampling-rate wireless modules, with poor signal stability in a strong electromagnetic interference environment, and data processing relies on offline analysis, unable to meet the real-time requirements. Fourth, the maintenance records are disconnected from the monitoring data and the compliance is in doubt. In the existing technology, sensors, data storage modules, and management systems are independent units, resulting in difficult data traceability. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art. For this purpose, the present invention provides a wheel fault monitoring system and method for hoisting machinery, and constructs a high-precision, full-time-domain wheel rolling fault vibration signal acquisition system through multi-axis sensor fusion, anti-interference transmission architecture, and edge computing technology, providing reliable guarantee for the safe operation of hoisting machinery.

[0005] On the one hand, an embodiment of the present invention provides a wheel fault monitoring system for hoisting machinery, including: A multi-axis vibration sensing module, the multi-axis vibration sensing module includes a plurality of three-axis acceleration sensors, and the plurality of three-axis acceleration sensors are respectively installed on the hub, bearing seat, and track contact surface of the hoisting machinery wheel for real-time acquisition of the vibration signals of the wheel; An anti-interference transmission module, which includes a main channel and a standby channel, and uses industrial Ethernet and 5G dual-mode communication to ensure stable signal transmission of the system in a strong interference environment; An edge computing module, which has an FPGA chip built in, and is used to perform Fourier transform and wavelet denoising preprocessing on the original vibration signal in real time, and extract fault feature values such as kurtosis coefficient, envelope spectrum peak value and root mean square value; A data management module, which includes a data storage layer, a data service layer and a data interface layer, and is used for the integrated management of vibration data, maintenance records and fault codes, visually analyze the vibration data, and display vibration waveform diagrams, spectrograms and fault history curves.

[0006] According to some embodiments of the present invention, the multi-axis vibration sensing module is provided with a differential amplifier circuit and an adaptive band-stop filter to suppress the starting and stopping impact of the crane and electromagnetic braking interference.

[0007] According to some embodiments of the present invention, the model of the triaxial acceleration sensor is IEPE 356A32, and the triaxial acceleration sensor adopts a magnetic adsorption and bolt composite fixing method to adapt to wheels with different diameters.

[0008] According to some embodiments of the present invention, the main channel uses industrial Ethernet, and the standby channel is 5G NR. The automatic switching time ≤ 50 ms, and adaptive filtering is performed based on the LMS algorithm to dynamically suppress pulse noise to ensure that the signal-to-noise ratio ≥ 40 dB.

[0009] According to some embodiments of the present invention, the FPGA chip is provided with multiple fault feature libraries, and the fault feature libraries include wheel flange wear fault features, bearing spalling fault features and track deformation fault features.

[0010] According to some embodiments of the present invention, the data management module encrypts the fault feature values and uploads them to the cloud or a local server, and stores the hash value of the vibration data on the blockchain to achieve blockchain evidence storage of vibration data and maintenance records.

[0011] On the other hand, an embodiment of the present invention provides a method for detecting faults in crane wheels, including: Install multiple triaxial acceleration sensors on the hub, bearing seat and track contact surface of the crane wheel respectively, and use a laser calibrator to adjust the probe angle of the triaxial acceleration sensor to ensure perpendicularity to the measurement surface; Amplify the analog signal collected by the triaxial acceleration sensor through a differential amplifier circuit, and filter out the starting and stopping impact of the crane and electromagnetic interference through an adaptive band-stop filter; Perform signal preprocessing, divide the original vibration signal into multiple data segments according to a time window; perform fast Fourier transform on the data segments to generate a spectrogram and identify the main frequency components; perform multi-level decomposition using the db8 wavelet basis, filter out high-frequency noise through the soft threshold method, and retain the effective frequency band of 1 Hz to 5 kHz; Calculate the kurtosis coefficient, extract the peak value of the envelope spectrum, and calculate the root mean square value, so as to extract the feature vector; Input the feature vector, call the pre-stored fault feature library, calculate the similarity using the RBF kernel function, match the fault model in real time through the SVM classifier, judge the warning level according to the similarity, and use indicator lights of different colors for prompting; Perform visual analysis on the vibration data, display the vibration waveform diagram, spectrogram and fault history curve diagram, and store the vibration data on the blockchain for evidence.

[0012] According to some embodiments of the present invention, the calculating the kurtosis coefficient, extracting the peak value of the envelope spectrum, and calculating the root mean square value includes: Calculate the kurtosis coefficient based on the kurtosis calculation formula, and the kurtosis calculation formula is: where K is the kurtosis coefficient, N is the sampling length, σ is the standard deviation, x i is the signal value at each moment, and μ is the signal mean value, which is used to quantify the impact fault characteristics; The extraction of the peak value of the envelope spectrum is to perform Hilbert transform on the denoised signal, calculate the envelope spectrum, and extract the peak values of the first 3 harmonics; Calculate the root mean square value based on the root mean square calculation formula, and the root mean square calculation formula is: where RMS is the root mean square value, N is the sampling length, σ is the standard deviation, x i is the signal value at each moment.

[0013] According to some embodiments of the present invention, the step of judging the warning level according to the similarity includes: If the similarity < 60%, the warning level is determined to be normal, and the green indicator light is turned on; If 60% ≤ similarity < 80%, the warning level is determined to be attention, the yellow indicator light is turned on, and maintenance inspection is prompted; If 80% ≤ similarity < 95%, the warning level is determined to be a warning, the orange indicator light is turned on, and a speed reduction command is triggered; If the similarity ≥ 95%, the warning level is determined to be dangerous, the red indicator light is turned on, and emergency braking is started; If the similarity continuously exceeds 80% but no known fault is matched, new model training is automatically triggered and the feature library is updated.

[0014] According to some embodiments of the present invention, the step of storing vibration data on the blockchain includes: Generate a 256-bit hash value for the timestamp, device ID, and feature vector in the vibration data using the national cryptographic algorithm SM3; Digitally sign the hash value through a CA certificate to ensure the credibility of the data source; Write the signed hash value into the Hyperledger Fabric consortium blockchain for on-chain storage. The nodes include regulatory agencies, maintenance units, and equipment manufacturers. Write the signed hash value into the Hyperledger Fabric consortium blockchain for on-chain storage. The nodes include regulatory agencies, maintenance units, and equipment manufacturers The embodiments of the present invention at least have the following beneficial effects: Through the multi-axis sensor layout, the fault recognition rate is greatly improved compared with traditional single-point detection, reducing the missed detection rate; through the edge computing module, the data analysis time consumption ≤ 1 second, which is shortened from the previous 30 minutes of offline mode processing to real-time processing, meeting the mandatory requirements of the lifting machinery regular inspection rules for real-time monitoring; adopting anti-interference transmission design to reduce the bit error rate in complex scenarios; ensuring the integrity of locally stored data in case of network disconnection through the dual-mode communication redundancy mechanism; through early fault warning, the downtime loss caused by wheel failures can be reduced, and the maintenance cost can be lowered; adopting blockchain storage to avoid maintenance responsibility disputes and meet the legal requirements of data authenticity; thus improving the detection accuracy and efficiency, enhancing anti-interference and reliability, and reducing operation and maintenance costs and risks.

[0015] The additional aspects and advantages of the present invention will be partially given in the following description, partially become obvious from the following description, or be understood through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The above and / or additional aspects and advantages of the present invention will become obvious and easy to understand from the description of the embodiments in conjunction with the following drawings, where: Figure 1 is a module diagram of the lifting machinery wheel fault monitoring system according to an embodiment of the present invention; Figure 2 is a schematic structural diagram of the multi-axis vibration sensing module of the lifting machinery wheel fault monitoring system according to an embodiment of the present invention; Figure 3 is a functional block diagram of the data management module of the lifting machinery wheel fault monitoring system according to an embodiment of the present invention; Figure 4 is a flowchart of the lifting machinery wheel fault detection method according to an embodiment of the present invention; Figure 5 is a flowchart of storing vibration data on the blockchain for the lifting machinery wheel fault detection method according to an embodiment of the present invention; Figure 6 Flow chart of FPGA resource allocation and interaction processing for the lifting machinery wheel fault detection method according to an embodiment of the present invention; Figure 7 Flow chart of vibration signal processing and fault diagnosis for the lifting machinery wheel fault detection method according to an embodiment of the present invention.

[0017] Reference numerals: Multi-axis vibration sensing module 100, inner hub wall measurement point 110, bearing housing lateral measurement point 120, track contact surface measurement point 130, anti-interference transmission module 200, edge computing module 300, data management module 400, lifting machinery 500, wheel 510, track 520. Specific embodiments

[0018] The embodiments of the present invention will be described in detail below. The examples of the embodiments are shown in the drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the drawings are exemplary only for explaining the present invention and should not be construed as limiting the present invention.

[0019] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by terms such as up, down, front, back, left, right, etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as limiting the present invention.

[0020] In the description of the present invention, the meaning of "several" is one or more, the meaning of "multiple" is two or more, and understandings such as greater than, less than, exceeding, etc. do not include the present number, and understandings such as "above", "below", "within", etc. include the present number. If there is a description of "first", "second", etc., it is only for the purpose of distinguishing technical features and should not be construed as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features or implicitly indicating the sequence relationship of the indicated technical features.

[0021] In the description of the present invention, unless otherwise clearly defined, terms such as "set", "installed", "connected", and "coupled" should be understood in a broad sense, and those skilled in the art can reasonably determine the specific meanings of the above terms in the present invention in combination with the specific content of the technical solution.

[0022] The technical solution of the present invention will be described in detail below with reference to the drawings and specific embodiments.

[0023] Please refer to Figures 1 to 3, this embodiment discloses a fault monitoring system for a hoisting machinery wheel, which includes a multi-axis vibration sensing module 100, an anti-interference transmission module 200, an edge computing module 300, and a data management module 400. The multi-axis vibration sensing module 100 includes a plurality of triaxial acceleration sensors, and the plurality of triaxial acceleration sensors are respectively installed on the hub, bearing seat of the wheel 510 of the hoisting machinery 500, and the contact surface with the track 520, and are used to collect the vibration signals of the wheel 510 in real time; the anti-interference transmission module 200 includes a main channel and a standby channel, adopts industrial Ethernet and 5G dual-mode communication, and combines electromagnetic shielding cables and adaptive filtering technology to ensure stable signal transmission in a strong interference environment; the edge computing module 300 is built with an FPGA chip, and is used to perform Fourier transform and wavelet denoising preprocessing on the original vibration signals in real time, and extract fault characteristic values such as kurtosis coefficient, envelope spectrum peak value, and root mean square value; the data management module 400 includes a data storage layer, a data service layer, and a data interface layer, and is used for the integrated management of vibration data, maintenance records, and fault codes, visually analyzes the vibration data, and displays vibration waveform diagrams, spectrograms, and fault history curve diagrams. Through the layout of multi-axis sensors, the fault recognition rate is greatly improved compared with traditional single-point detection, and the missed detection rate is reduced; through the edge computing module, the data analysis time consumption is ≤1 second, which is shortened from 30 minutes of processing in the previous offline mode to real-time processing, meeting the mandatory requirements of the hoisting machinery regular inspection rules for real-time monitoring; adopting anti-interference transmission design to reduce the bit error rate in complex scenarios; ensuring the integrity of locally stored data in case of network disconnection through the dual-mode communication redundancy mechanism; through early fault warning, the downtime loss caused by wheel faults can be reduced, and the maintenance cost can be reduced; adopting blockchain evidence storage to avoid maintenance responsibility disputes and meet the legal requirements of data authenticity; thereby improving the detection accuracy and efficiency, enhancing the anti-interference and reliability, and reducing the operation and maintenance costs and risks.

[0024] Please refer to Figure 1 and Figure 2, the multi-axis vibration sensing module 100 is provided with a differential amplifier circuit and an adaptive band-stop filter to suppress the start-stop impact of the crane 500 and the electromagnetic braking interference. The differential amplifier circuit uses an AD8221 instrumentation amplifier with a gain of 60 dB and a common-mode rejection ratio of ≥120 dB to amplify the mV-level analog signal output by the sensor; the adaptive band-stop filter dynamically adjusts the stopband frequency based on the LMS algorithm to suppress the high-frequency interference of the crane frequency converter and electromagnetic brake; the sensor is powered by a DC-DC isolated power supply to isolate the ground loop noise. The cable selected is a double-shielded twisted pair with the model of BELDEN 9841, the inner layer is aluminum foil shielding, and the outer layer is tinned copper mesh braided shielding, and the shielding effectiveness is ≥90 dB; the two ends of the cable use aviation plugs with the model of XS12-J6, the pins are gold-plated, and the contact resistance is ≤0.1 Ω; the wiring terminal and the sensor interface are fixed by spring pressing, and the anti-vibration level reaches the IEC 60068-2-6 standard, that is, 10 Hz - 2000 Hz, and the acceleration is 20 g.

[0025] Please refer to Figure 1 and Figure 2 , the model of the triaxial acceleration sensor is IEPE 356A32. The triaxial acceleration sensor adopts a magnetic adsorption and bolt composite fixing method, and is provided with a magnetic adsorption base and bolts to adapt to wheels 510 with different diameters. The magnetic adsorption-bolt composite structure is used for fixing to ensure that the sensor is suitable for wheels 510 with a diameter of 200 - 1500 mm, and the installation error is controlled within ≤0.1 mm. During installation, first clean the installation surface of the wheel 510 to remove oil stains and rust; then adsorb the magnetic adsorption base at the measuring point position, tighten the bolt to the preset torque, such as 5 N·m; and use a laser calibrator to calibrate the angle of the sensor probe. When the deviation exceeds the limit, fine-tune it through a universal joint to ensure perpendicularity to the measuring surface and avoid signal distortion. Regularly calibrate the sensor angle and the connection status of the shielded cable to ensure the normal operation of the system and the timeliness of fault handling.

[0026] Please refer to Figure 1 , the main channel uses industrial Ethernet, and the standby channel is 5G NR. The automatic switching time is ≤50 ms. Adaptive filtering is performed based on the LMS algorithm to dynamically suppress pulse noise to ensure that the signal-to-noise ratio is ≥40 dB. A dual-mode communication design is adopted. The main channel uses industrial Ethernet, and the standby channel is 5G NR in the Sub-6 GHz band, supporting automatic switching with a switching time of ≤50 ms; the outer layer of the cable is covered with a double-layer copper mesh shielding layer, and the shielding effectiveness is ≥90 dB. The interface uses an aviation plug with an IP67 protection level; the adaptive filtering dynamically suppresses pulse noise based on the LMS algorithm to ensure that the signal-to-noise ratio is ≥40 dB.

[0027] Please refer to Figure 6 , the FPGA chip is provided with a variety of fault feature libraries, and the fault feature libraries include fault features of wheel flange wear, bearing spalling, and track deformation.

[0028] Please refer to Figure 3 , the data management module 400 encrypts the fault characteristic values and uploads them to the cloud or local server, and stores the vibration data hash values on the chain to achieve the blockchain evidence storage of vibration data and maintenance records.

[0029] The data management module 400 is docked with the external safety monitoring and management system of lifting machinery through the RESTful API interface, supporting the integrated storage of vibration data, maintenance records and fault codes; the vibration data hash values are written into the consortium chain through the Hyperledger Fabric framework, and the national cryptographic algorithm SM3 is used to generate hash values; the evidence storage information includes time stamps, device IDs, and electronic signatures of inspectors; the visualization interface provides tools such as vibration 52A8 waveforms, spectrograms, and fault diagnosis history curves, supporting the one-click generation of PDF inspection reports containing blockchain evidence storage numbers. Ensure the reliability and traceability of data, facilitating maintenance personnel for monitoring and analysis.

[0030] Please refer to Figure 3 , the data management module 400 is provided with multiple visualization interfaces. Among them, the real-time vibration waveform dynamically displays the radial, axial, and tangential vibration signals of the wheel 510 in a line chart, with a sampling rate of 51.2 kHz, supporting time axis zooming and peak annotation; the spectrum analysis chart uses a waterfall chart to display the FFT spectrum, with a frequency range of 0 - 5 kHz, annotating the main frequency component and the harmonic energy ratio; the fault history curve plots the kurtosis coefficient along the time axis, including the trend of characteristic parameters such as the peak value, and correlates with the change of the warning level.

[0031] Please refer to Figure 4 , this embodiment also provides a method for detecting using the above-mentioned lifting machinery wheel fault monitoring system, mainly including steps S101~S106: S101. Install multiple triaxial acceleration sensors on the hub, bearing seat, and track contact surface of the lifting machinery wheel 510 respectively, and use a laser calibrator to adjust the probe angle of the triaxial acceleration sensor to ensure perpendicularity to the measurement surface.

[0032] S102. Amplify the analog signals collected by the triaxial acceleration sensor through a differential amplifier circuit, and filter out the start-stop impact and electromagnetic interference of the lifting machinery 500 through an adaptive band-stop filter.

[0033] S103. Perform signal preprocessing, divide the original vibration signal into multiple data segments according to a time window; perform a fast Fourier transform on the data segments to generate a spectrogram and identify the main frequency component; perform multi-layer decomposition using the db8 wavelet basis, and filter out high-frequency noise through the soft threshold method, retaining the effective frequency band of 1 Hz ~ 5 kHz.

[0034] S104. Calculate the kurtosis coefficient, extract the peak value of the envelope spectrum, and calculate the root mean square value, so as to extract the feature vector.

[0035] S105. Input the feature vector, call the pre-stored fault feature library, calculate the similarity using the RBF kernel function, match the fault model in real time through the SVM classifier, judge the warning level according to the similarity, and use indicator lights of different colors for prompt.

[0036] S106. Conduct visual analysis on the vibration data, display the vibration waveform diagram, frequency spectrum diagram, and fault history curve diagram, and store the vibration data on the blockchain for evidence.

[0037] It should be noted that the steps of calculating the kurtosis coefficient, extracting the peak value of the envelope spectrum, and calculating the root mean square value in the above S104 step include the following detailed steps: Calculate the kurtosis coefficient based on the kurtosis calculation formula, and the kurtosis calculation formula is: In the formula, K is the kurtosis coefficient, N is the sampling length, σ is the standard deviation, x i is the signal value at each moment, and μ is the signal mean value, which is used to quantify the impact fault feature; Extracting the peak value of the envelope spectrum is to perform Hilbert transform on the denoised signal, calculate the envelope spectrum, and extract the peak values of the first 3 harmonics; Calculate the root mean square value based on the root mean square calculation formula, and the root mean square calculation formula is: In the formula, RMS is the root mean square value, N is the sampling length, σ is the standard deviation, x i is the signal value at each moment.

[0038] Please refer to Figure 6 , the steps of judging the warning level according to the similarity in the above S105 step include: If the similarity < 60%, judge the warning level as normal and turn on the green indicator light; If 60% ≤ similarity < 80%, judge the warning level as attention, turn on the yellow indicator light, and prompt for maintenance inspection; If 80% ≤ similarity < 95%, judge the warning level as warning, turn on the orange indicator light, and trigger the speed reduction instruction; If the similarity ≥ 95%, judge the warning level as dangerous, turn on the red indicator light, and start the emergency brake; If the similarity continuously remains higher than 80% but no known fault is matched, automatically trigger the training of a new model and update the feature library.

[0039] Please refer to Figure 5 , the steps of storing the vibration data on the blockchain for evidence in the above S106 step include: S201. Generate a 256-bit hash value for the timestamp, device ID, and feature vector in the vibration data using the national cryptographic algorithm SM3; S202. Digitally sign the hash value through a CA certificate to ensure the credibility of the data source; S203. Write the signed hash value to the Hyperledger Fabric consortium blockchain for on-chain evidence storage. The nodes include regulatory agencies, maintenance units, and equipment manufacturers.

[0040] Please refer to Figure 6 , the FPGA resource allocation and interaction processing are as follows: S301. FPGA resource allocation Adopt the Xilinx Kintex-7 XC7K325T chip, which has 326,080 logic cells and 16.3 Mb of Block RAM. The FPGA resource allocation is as follows: ① Preprocessing unit: Occupy 20% of the logic resources to implement FFT and wavelet noise reduction; ② Feature extraction unit: Occupy 35% of the logic resources to calculate features such as kurtosis and envelope spectrum; ③ Diagnosis unit: Occupy 30% of the logic resources to run the SVM classifier; ④ Fault feature library: Store it in an external Flash. For example, the Flash capacity is 8 GB and supports online update.

[0041] S302. Interaction processing of the fault feature library ① Feature library structure Each type of fault model contains 100 groups of historical data feature vector machine threshold parameters; the data is stored in JSON format, and the key values include fault type, feature weight, and warning level threshold.

[0042] ② Real-time matching process The FPGA reads the feature library data from the Flash and transmits it to the diagnosis unit through the DMA channel; the SVM classifier compares the real-time features with the feature library and outputs the similarity score; the diagnosis result is transmitted to the communication module through the SPI interface.

[0043] ③ Dynamic update mechanism The newly added fault model is uploaded to the edge computing module 300 through the data management platform, encrypted by AES-256, and written to the Flash. And it supports incremental learning. When the similarity continuously exceeds 80% but no known fault is matched, it automatically triggers new model training and updates the feature library.

[0044] Please refer to Figure 7 , the detailed steps of vibration signal processing and fault diagnosis are as follows: S401. Preprocessing stage ① Signal segmentation: The original vibration signal (sampling rate 51.2 kHz) is segmented into multiple data segments according to a time window (window length 1024 points, overlap rate 50%); ② Fourier transform: Perform a fast Fourier transform (FFT) on each signal segment to generate a spectrogram (frequency resolution 0.5 Hz) and identify the dominant frequency components; ③ Wavelet denoising: Use the db8 wavelet basis for 5-layer decomposition, filter out high-frequency noise through the soft threshold method, and retain the effective frequency band (1 Hz - 5 kHz).

[0045] S402. Feature extraction stage ① Kurtosis coefficient calculation ② Envelope spectrum peak extraction Perform a Hilbert transform on the denoised signal, calculate the envelope spectrum, and extract the first 3 harmonic peaks (frequency error ≤ 2 Hz); ③ Root mean square value RMS calculation Perform a root mean square value RMS calculation to evaluate the signal energy level.

[0046] S403. Fault diagnosis and handling stage (1) SVM classifier matching ① Input multi-dimensional feature vectors, including kurtosis, envelope spectrum peaks, RMS, etc.; ② Call the pre-stored fault feature library, including multiple fault models such as wheel flange wear, bearing spalling, etc., and calculate the similarity through the kernel function RBF.

[0047] (2) Warning level and handling Handle according to the warning level as follows: If the warning level is normal, turn on the green indicator light and there is no warning; If the warning level is attention, turn on the yellow indicator light, push a work order to the maintenance personnel's handheld terminal, and prompt maintenance inspection; If the warning level is warning, turn on the orange indicator light and trigger a speed reduction command to reduce the speed of the crane 500; If the warning level is dangerous, turn on the red indicator light, start emergency braking, stop the braking system, and avoid the expansion of the accident.

[0048] The wheel fault monitoring system and detection method provided by this embodiment first install the signal acquisition sensors of the multi-axis vibration sensing module 100. A plurality of triaxial acceleration sensors are respectively installed at the inner wall measurement points 110 of the wheel hubs of the wheels 510 of the hoisting machinery 500, the lateral measurement points 120 of the bearing seats, and the track contact surface measurement points 130. The mV-level analog signals collected by the triaxial acceleration sensors are amplified by a differential amplifier circuit with a gain of 60 dB, and the start-stop impact and electromagnetic interference of the crane are filtered out through an adaptive band-stop filter. The filtered and conditioned signals are transmitted to the anti-interference transmission module 200 through a double-layer shielded cable. The system adopts a dual-mode communication link switching mechanism to ensure the stability of signal transmission in complex scenarios such as ports and steel mills, with a bit error rate ≤ 0.01%, ensuring the stable transmission of data in complex scenarios. Then, signal preprocessing is performed in the edge computing module 300. The original signal is segmented by a time window with an overlap rate of 50%, a spectrogram is generated through FFT, and noise reduction processing is performed using the db8 wavelet basis. Multidimensional feature vectors such as kurtosis coefficient, envelope spectrum peak value, and root mean square value RMS are calculated for feature extraction. The fault classification and warning SVM classifier calls the pre-stored fault feature library, calculates the similarity using the RBF kernel function, realizes diagnosis and outputs the warning level using an FPGA chip, and gives prompts through indicator lights of different colors. The vibration feature data is uploaded to the blockchain through the Hyperledger Fabric framework, a hash value is generated using the SM3 algorithm, and a CA electronic signature is attached to ensure the reliability and traceability of the data, and it is convenient for maintenance personnel to monitor and analyze. The data management module 400 automatically generates an electronic maintenance report, including equipment information, detection conclusions, and blockchain numbers, and supports scanning the code to verify the data integrity. The visual monitoring interface displays the vibration waveform, spectrogram, and historical curve in real time, and supports fault location and trend analysis. Fault handling is carried out according to the warning level, a work order is pushed to the handheld terminal of the maintenance personnel, or the crane speed is reduced, or the braking system is linked to stop, avoiding the expansion of accidents. Thus, the efficient acquisition, stable transmission, intelligent diagnosis, reliable evidence storage, and visual monitoring of the rolling fault signals of the wheels 510 of the hoisting machinery 500 are realized, and collaborative work and emergency response are carried out to ensure the safe operation of the hoisting machinery 500.

[0049] This embodiment has the following beneficial effects: 1. Improve the detection accuracy and efficiency: Through the layout of multi-axis sensors, the fault recognition rate is greatly improved compared with traditional single-point detection, reducing the missed detection rate; through the edge computing module, the data analysis time consumption ≤ 1 second, which is shortened from the previous 30 minutes of offline mode processing to real-time processing, meeting the mandatory requirements of the hoisting machinery regular inspection rules for real-time monitoring.

[0050] 2. Enhance anti-interference and reliability: Adopt anti-interference transmission design, with bit error rate ≤ 0.01% in complex scenarios such as steel mills and ports, which is significantly optimized compared with traditional solutions; Ensure the integrity of locally stored data in case of network disconnection through a dual-mode communication redundancy mechanism.

[0051] 3. Reduce operation and maintenance costs and risks: Through early fault warning, the downtime loss caused by wheel failures can be reduced, and the maintenance cost can be lowered; Adopt blockchain evidence storage to avoid disputes over maintenance responsibilities and meet the legal requirements for data authenticity.

[0052] 4. Promote industry standardization: Adopt modular design to support rapid deployment, applicable to various types of cranes such as bridge cranes, gantry cranes, and tower cranes; Meet the relevant national and industry standards for lifting machinery and provide technical support for the intelligent supervision of lifting machinery.

[0053] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited to the above embodiments, and various changes can be made without departing from the spirit of the present invention within the scope of knowledge possessed by ordinary technical personnel in the technical field to which it belongs.

Claims

1. A fault monitoring system for a hoisting machinery wheel, characterized in that, Including: A multi-axis vibration sensing module, which includes a plurality of triaxial acceleration sensors. The plurality of triaxial acceleration sensors are respectively installed on the hub, bearing seat and track contact surface of the hoisting machinery wheel, and are used to collect the vibration signals of the wheel in real time; An anti-interference transmission module, which includes a main channel and a standby channel, and uses industrial Ethernet and 5G dual-mode communication to ensure stable signal transmission in a strong interference environment; An edge computing module, which is built with an FPGA chip and is used to perform Fourier transform and wavelet noise reduction preprocessing on the original vibration signal in real time, and extract fault characteristic values such as kurtosis coefficient, envelope spectrum peak value and root mean square value; A data management module, which includes a data storage layer, a data service layer and a data interface layer, and is used for the integrated management of vibration data, maintenance records and fault codes, visually analyze the vibration data, and display vibration waveform diagrams, spectrograms and fault history curve diagrams.

2. The lifting machinery wheel fault monitoring system according to claim 1, wherein, The multi-axis vibration sensing module is provided with a differential amplifier circuit and an adaptive band-stop filter to suppress the start-stop impact of the hoisting machinery and electromagnetic braking interference.

3. The lifting machinery wheel fault monitoring system according to claim 2, characterized in that, The model of the triaxial acceleration sensor is IEPE 356A32, and the triaxial acceleration sensor adopts a magnetic adsorption and bolt composite fixing method to adapt to wheels with different diameters.

4. The hoisting machinery wheel fault monitoring system according to claim 1, wherein The main channel uses industrial Ethernet, and the standby channel is 5GNR. The automatic switching time ≤ 50ms, and adaptive filtering is performed based on the LMS algorithm to dynamically suppress pulse noise to ensure that the signal-to-noise ratio ≥ 40dB.

5. The lifting machinery wheel fault monitoring system according to claim 1, characterized in that, The FPGA chip is provided with a variety of fault feature libraries, and the fault feature libraries include fault features of wheel flange wear, bearing spalling and track deformation.

6. The lifting machinery wheel fault monitoring system according to claim 1, characterized in that, The data management module encrypts the fault characteristic values and uploads them to the cloud or local server, and stores the hash value of the vibration data on the chain to realize the blockchain evidence storage of vibration data and maintenance records.

7. A method for detecting faults in the wheels of a hoisting machine, characterized in that, Based on the hoisting machinery wheel fault monitoring system according to any one of claims 1 to 6, including: Respectively install a plurality of triaxial acceleration sensors on the hub, bearing seat and track contact surface of the hoisting machinery wheel, and use a laser calibrator to adjust the probe angle of the triaxial acceleration sensor to ensure perpendicularity to the measurement surface; Amplify the analog signal collected by the triaxial acceleration sensor through a differential amplifier circuit, and filter out the start-stop impact and electromagnetic interference of the hoisting machinery through an adaptive band-stop filter; Perform signal preprocessing, divide the original vibration signal into multiple data segments according to a time window; perform a fast Fourier transform on the data segments to generate a spectrogram and identify the main frequency components; perform multi-layer decomposition using the db8 wavelet basis, filter out high-frequency noise through the soft threshold method, and retain the effective frequency band of 1Hz to 5kHz; Calculate the kurtosis coefficient, extract the envelope spectrum peak value and calculate the root mean square value to extract the feature vector; Input the feature vector, call the pre-stored fault feature library, calculate the similarity using the RBF kernel function, match the fault model in real time through the SVM classifier, judge the warning level according to the similarity, and use indicator lights of different colors for prompt; Perform visual analysis on the vibration data, display the vibration waveform diagram, frequency spectrum diagram and fault history curve diagram, and store the vibration data on the blockchain for evidence.

8. The method for detecting faults of a hoisting machinery wheel according to claim 7, characterized in that, The calculation of the kurtosis coefficient, extraction of the envelope spectrum peak value and calculation of the root mean square value include: Calculating the kurtosis coefficient based on the kurtosis calculation formula, and the kurtosis calculation formula is: where K is the kurtosis coefficient, N is the sampling length, σ is the standard deviation, x i is the signal value at each moment, and μ is the signal mean, which is used to quantify the characteristics of impact faults; The extraction of the envelope spectrum peak value is to perform Hilbert transform on the denoised signal, calculate the envelope spectrum, and extract the peak values of the first 3 harmonics; Calculating the root mean square value based on the root mean square calculation formula, and the root mean square calculation formula is: Wherein, RMS is the root mean square value, N is the sampling length, σ is the standard deviation, and x i is the signal value at each moment.

9. The method for detecting faults of the lifting machinery wheels according to claim 7, characterized in that, The steps of judging the warning level according to the similarity include: If the similarity < 60%, the warning level is judged to be normal, and the green indicator light is turned on; If 60% ≤ similarity < 80%, the warning level is judged to be attention, the yellow indicator light is turned on, and maintenance inspection is prompted; If 80% ≤ similarity < 95%, the warning level is judged to be warning, the orange indicator light is turned on, and the speed reduction instruction is triggered; If the similarity ≥ 95%, the warning level is judged to be dangerous, the red indicator light is turned on, and the emergency brake is started; If the similarity continuously remains higher than 80% but no known fault is matched, the new model training is automatically triggered and the feature library is updated.

10. The method for detecting faults of a lifting machinery wheel according to claim 7, characterized in that, The steps of storing the vibration data on the blockchain for evidence include: Using the national secret algorithm SM3 to generate a 256-bit hash value for the timestamp, device ID, and feature vector in the vibration data; Performing digital signature on the hash value through the CA certificate to ensure the credibility of the data source; Writing the signed hash value into the Hyperledger Fabric consortium blockchain for on-chain evidence storage, and the nodes include regulatory agencies, maintenance units and equipment manufacturers.

Images