Robot fault diagnosis system based on Mworks platform

Abstract

The invention relates to the technical field of robot fault diagnosis, and discloses a robot fault diagnosis system based on an Mworks platform. According to the system, multiple sensors are used for collecting key data of robot operation in real time, signal processing and noise reduction are carried out through advanced technologies such as wavelet transform and FFT spectrum analysis, it is ensured that extracted features have higher reliability, and multi-dimensional indexes such as RMS values, temperature rise gradients and current harmonic distortion rates calculated by a feature extraction module are more reliable. Comprehensive data support is provided for fault diagnosis, the fault diagnosis module is combined with parameter estimation and an SVM classifier, sources of multiple faults in a complex industrial environment can be effectively recognized, misjudgment and missed judgment in a traditional method are avoided, in addition, the intelligent linkage function of the decision support module ensures that maintenance suggestions and energy efficiency optimization strategies are generated in time, and the fault diagnosis efficiency is improved. And the reliability and the operation efficiency of the robot system are further improved, so that a powerful guarantee is provided for continuous and stable industrial production.

Description

Technical Field

[0001] The present invention relates to the technical field of robot fault diagnosis, and particularly to a robot fault diagnosis system based on the Mworks platform. Background Art

[0002] The Mworks platform is an integrated robot control and management system that provides rich tools and interfaces, aiming to improve the application efficiency of robots in multiple fields such as industry, service, and scientific research. This platform supports the access and data processing of multiple sensors, providing users with functions such as real-time monitoring, data analysis, and fault diagnosis. With the help of the Mworks platform, users can easily achieve complex robot operations and management, improve production efficiency, and reduce operating costs.

[0003] In modern manufacturing, robots are widely used in automated production lines to perform various tasks such as assembly, welding, and handling. These robots can not only improve work efficiency but also ensure product consistency and quality. However, with the increase in usage frequency, robots may encounter various faults during operation, such as abnormal vibration, overheating, and current fluctuations. If these faults are not diagnosed and processed in a timely manner, they may lead to production stagnation, equipment damage, and even safety hazards. Therefore, it is particularly important to establish an efficient fault diagnosis system.

[0004] Although existing fault diagnosis methods can handle some common faults, in a complex industrial environment, robots may face multiple simultaneous faults, making it difficult for traditional diagnosis methods to accurately identify the source of the faults. Summary of the Invention

[0005] (1) Technical Problems to be Solved

[0006] Aiming at the deficiencies of the existing technology, the present invention provides a robot fault diagnosis system based on the Mworks platform, which uses multiple sensors to collect key data of robot operation in real time, and performs signal processing and noise reduction through advanced technologies such as wavelet transform and FFT spectrum analysis to ensure that the extracted features have higher reliability. The multi-dimensional indicators calculated by the feature extraction module, such as RMS value, temperature rise gradient, and current harmonic distortion rate, provide comprehensive data support for fault diagnosis. The fault diagnosis module combines parameter estimation and SVM classifier to effectively identify the sources of multiple faults in a complex industrial environment, avoiding misjudgment and missed judgment in traditional methods. The intelligent linkage function of the decision support module ensures the timely generation of maintenance suggestions and energy efficiency optimization strategies, further improving the reliability and operation efficiency of the robot system, thus providing a strong guarantee for the continuous and stable industrial production.

[0007] (2) Technical Solutions

[0008] To achieve the above object, the present invention provides the following technical solution: A robot fault diagnosis system based on the Mworks platform, including a data acquisition module, a signal preprocessing module, a feature extraction module, a fault diagnosis module, and a decision support module;

[0009] The data acquisition module collects the time-domain vibration signal, temperature data, electrical signal, and motion parameters of the robot through vibration sensors, temperature sensors, current and voltage sensors, and encoders;

[0010] The signal preprocessing module performs wavelet transform denoising, synchronous sampling, and data alignment on the time-domain vibration signal of the robot, and outputs the denoised time-domain signal and FFT spectrogram;

[0011] The feature extraction module calculates the RMS value, kurtosis coefficient, harmonic energy ratio of the vibration signal, the temperature rise gradient of the temperature data, the current harmonic distortion rate of the electrical signal, and the trajectory tracking error of the robot motion parameters according to the denoised time-domain signal and FFT spectrogram;

[0012] The fault diagnosis module generates observer residuals using the parameter estimation method based on the values calculated by the feature extraction module, and outputs the robot fault code and confidence score in combination with the SVM classifier;

[0013] The decision support module links the spare parts inventory system according to the fault diagnosis result, and generates robot repair suggestions and energy efficiency optimization strategies.

[0014] Preferably, the formula for wavelet transform denoising is as follows:

[0015]

[0016] In the formula, s denoised (t) represents the signal output result of wavelet transform denoising, λ represents the denoising threshold, Thresh(*) represents the soft threshold function, ψ j,k (t) represents the wavelet basis function, and W j,k represents the wavelet coefficient at scale j and position k.

[0017] Preferably, the formula for data alignment is as follows:

[0018]

[0019] In the formula, D(i,j) represents the cumulative alignment distance matrix, x i and y j represent two sets of time series data points to be aligned, and ||*|| represents the Euclidean distance.

[0020] Preferably, the formula for calculating the RMS value of the vibration signal is as follows:

[0021]

[0022] In the formula, x n Represents the discrete vibration signal sampling value, N represents the number of sampling points, and n represents the index subscript.

[0023] Preferably, the formula of the kurtosis coefficient is as follows:

[0024]

[0025] In the formula, K represents the kurtosis coefficient, x n represents the discrete vibration signal sampling value, Represents the sampling mean of discrete vibration signal, N represents the number of sampling points, and n represents the index subscript.

[0026] Preferably, the calculation formula of the harmonic energy ratio is as follows:

[0027]

[0028] In the formula, E h armonic represents the harmonic energy ratio, X k represents the amplitude of the kth frequency point in the FFT spectrum, H represents the set of harmonic frequency indices, k represents the frequency index in frequency domain analysis, and N represents the total number of sampling points used for FFT analysis.

[0029] Preferably, the calculation formula for the temperature rise gradient of the temperature data is as follows:

[0030]

[0031] In the formula, G T Indicates the temperature rise gradient, T max Indicates the highest temperature during the monitoring period, T initial represents the initial temperature, and Δt represents the time interval.

[0032] Preferably, the calculation formula for the current harmonic distortion rate of the electrical signal is as follows:

[0033]

[0034] In the formula, THD represents the current harmonic distortion rate of the electrical signal, I h It represents the effective value of the hth harmonic current, where H represents the highest harmonic order.

[0035] Preferably, the calculation formula of the trajectory tracking error of the robot motion parameters is as follows:

[0036]

[0037] In the formula, e(t) represents the trajectory tracking error of the robot's motion parameters, and x ref (t), y ref (t) represent the reference trajectory coordinates, and x actual (t), y actual (t) represent the actual execution trajectory coordinates.

[0038] Preferably, the observer residual is as follows:

[0039]

[0040] In the formula, r θ represents the observer residual, τ measured represents the measured motor torque, and J, B, and K represent the moment of inertia, damping coefficient, and stiffness coefficient respectively, respectively represent the position, velocity, and acceleration of the robot.

[0041] Compared with the prior art, the present invention provides a robot fault diagnosis system based on the Mworks platform, which has the following beneficial effects:

[0042] The present invention uses a variety of sensors to collect key data of the robot's operation in real time, and performs signal processing and noise reduction through advanced technologies such as wavelet transform and FFT spectrum analysis to ensure that the extracted features have higher reliability. The multi-dimensional indicators calculated by the feature extraction module, such as RMS value, temperature rise gradient, and current harmonic distortion rate, provide comprehensive data support for fault diagnosis. The fault diagnosis module combines parameter estimation and SVM classifier to effectively identify the sources of multiple faults in complex industrial environments, avoiding misjudgment and missed judgment in traditional methods. The intelligent linkage function of the decision support module ensures the timely generation of maintenance suggestions and energy efficiency optimization strategies, further improving the reliability and operation efficiency of the robot system, thus providing a strong guarantee for the continuous and stable industrial production. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a schematic diagram of the system flow of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0044] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0045] Aiming at the problem that in a complex industrial environment, a robot may face multiple simultaneous faults, which makes it difficult for traditional diagnostic methods to accurately identify the source of the faults, a robot fault diagnosis system based on the Mworks platform is proposed. Please refer to Figure 1 , the system includes a data acquisition module, a signal preprocessing module, a feature extraction module, a fault diagnosis module and a decision support module;

[0046] The data acquisition module collects the running state data of the robot in real time through a multi-source heterogeneous sensor network, specifically including: using an ICP type piezoelectric vibration sensor (frequency response range 0.5Hz - 10kHz, sensitivity 100mV / g) to synchronously collect the three-axis vibration time-domain signals of the robotic arm joints at a sampling rate of 50kHz, monitoring the temperature rise curve of the reducer lubricating grease through a PT100 platinum resistance temperature sensor (accuracy ±0.5°C), with a sampling period of 1 second, using a Hall effect current sensor (range ±20A, linearity 0.1% FS) and an isolated voltage probe (bandwidth 100MHz) to collect the three-phase current and DC bus voltage waveforms of the servo drive, cooperating with a 17-bit absolute photoelectric encoder (resolution 0.001°) to real-time feedback the angular positions of each joint, and transmitting them to the main control system through the EtherCAT bus (period 1ms). After all sensor data is subjected to hardware anti-aliasing filtering (cut-off frequency is 80% of the Nyquist frequency), a 24-bit high-precision AD conversion is performed by a PXIe-1071 data acquisition card, and finally a multi-channel synchronous data set with strictly aligned timestamps is formed;

[0047] The signal preprocessing module realizes effective noise reduction and data processing for the time-domain vibration signals of the robot through a series of advanced technical means to ensure the accuracy and reliability of subsequent analysis. First, for the collected time-domain vibration signals, wavelet transform is used for noise reduction processing. The noise reduction method of wavelet transform uses the soft threshold method, and its formula is:

[0048]

[0049] In this formula, W j,k represents the wavelet coefficients at different scales j and positions k. Thresh is the function for threshold processing, which can effectively suppress high-frequency noise while retaining the important features in the signal, especially the impact signals caused by faults. This method performs excellently in maintaining the instantaneous characteristics of the signal and is especially suitable for detecting early faults;

[0050] Secondly, synchronous sampling and data alignment are performed to ensure that the data collected by different sensors can be accurately aligned in time for subsequent analysis. The dynamic time warping (DTW) algorithm is used, and its core formula is:

[0051]

[0052] This algorithm dynamically adjusts the sampling interval by calculating the minimum distance between two sets of time-series data, so as to merge the speeds and timestamps of different sensor data. This can eliminate the errors caused by inconsistent sampling and ensure the accuracy of subsequent data processing;

[0053] After being processed through the above steps, the output denoised time-domain signal can not only eliminate the influence of background noise and significantly improve the signal-to-noise ratio of the signal, but also provide the FFT spectrogram of the vibration signal for frequency-domain analysis. FFT analysis can reveal the frequency characteristics of the signal and help engineers further identify whether there are fault modes at specific frequencies (such as gear meshing frequency or motor speed frequency), so as to timely identify potential faults or performance degradation during the operation of the robot;

[0054] Through this systematic signal preprocessing process, the module can provide high-quality and highly reliable input signals for the fault diagnosis module, which is crucial for ensuring the safe and efficient operation of the robot;

[0055] The core task of the feature extraction module is to extract key features from the denoised time-domain signal and FFT spectrogram. These features can effectively characterize the health status of the robot's operating state. First, calculate the root mean square value (RMS) of the vibration signal from the denoised time-domain vibration signal. The formula is as follows:

[0056]

[0057] In this formula, x n represents the sampling value of the discrete time-domain signal, N is the number of sampling points, and the RMS value reflects the effective energy of the signal, which helps to evaluate the vibration intensity of the robot. Excessive vibration may indicate the existence of mechanical faults such as imbalance or wear;

[0058] Secondly, calculate the kurtosis coefficient to characterize the sharpness of the signal with the following formula:

[0059]

[0060] Here, is the average value of the signal. A kurtosis coefficient greater than 3 usually indicates the existence of impacts or abnormal fluctuations in the signal, which is of great significance for judging whether machine components have suffered sudden faults;

[0061] Then, calculate the harmonic energy ratio using the following formula:

[0062]

[0063] Here, H represents the maximum harmonic frequency sampled, X kis the amplitude of each frequency point in the frequency-domain FFT analysis. The higher the harmonic energy ratio, the greater the likelihood of electrical wear, electrical equipment failure, or non-linear distortion phenomenon;

[0064] In addition, the temperature rise gradient is calculated from the temperature data, and its calculation formula is:

[0065]

[0066] where T max is the highest temperature during the monitoring period, T initial is the initial temperature, and Δt is the time interval. This feature can evaluate whether the robot overheats during operation and timely identify potential failures caused by friction or insufficient lubrication;

[0067] When analyzing the current signal, the total harmonic distortion (THD) of the current is calculated, and its calculation method is:

[0068]

[0069] Here, I1 is the effective value of the fundamental wave current, and I h is the effective value of the hth harmonic current. The increase in the total harmonic distortion of the current often indicates an overloaded motor or electrical equipment failure;

[0070] Finally, the feature extraction module calculates the trajectory tracking error of the robot to evaluate the accuracy of the movement, and its calculation method is:

[0071]

[0072] Here, x ref and y ref represent the reference trajectory coordinates, while x actual and y actual are the actually executed trajectory coordinates. The smaller the trajectory tracking error, the higher the accuracy of the robot when performing tasks, and it can effectively avoid potential damages caused by displacement deviation;

[0073] Through the above calculations, the feature extraction module can integrate a variety of representative diagnostic features, which provide a solid foundation for subsequent fault diagnosis, help quickly identify the fault type, evaluate the fault severity, and intuitively display the health status of the machine, thus providing a scientific basis for the robot operation and maintenance decision-making;

[0074] The core function of the fault diagnosis module is to perform real-time monitoring and fault identification based on the values calculated by the feature extraction module through parameter estimation techniques. This module uses the observer design method in modern control theory to monitor the health status of the robot system. By constructing a state observer, the subsequent generated observer residuals can be characterized by the following formula:

[0075]

[0076] where r θ represents the observer residual, τ measured represents the measured motor torque, and J, B, and K represent the moment of inertia, damping coefficient, and stiffness coefficient respectively, represent the robot position, velocity, and acceleration respectively. The advantage of this formula is that the residual r θ reflects the deviation between the actual measurement value and the predicted value. If the system state is normal, the residual should be maintained within a low amplitude range (e.g., by setting a control threshold, such as ±3σ±3σ). Once the residual exceeds this threshold, it can be determined that there is a risk of system failure. This method significantly enhances the system's ability to detect small faults and helps improve the accuracy of fault location;

[0077] Meanwhile, the fault diagnosis module combines a support vector machine (SVM) classifier to process the generated residual and other eigenvalues. After training, the SVM can output the corresponding fault code and the corresponding confidence score, representing the fault probability in the robot task. This integrated fault identification method not only improves the accuracy of diagnosis but also provides effective decision support for the operator to help them quickly take countermeasures. Therefore, by combining the ability of the observer to generate residuals with the efficient classification ability of the SVM classifier, precise monitoring and fault analysis of complex systems are achieved, providing a strong guarantee for ensuring the reliable operation of the robot and improving the overall safety of the system;

[0078] The decision support module realizes efficient maintenance management and energy efficiency optimization by analyzing the fault diagnosis results and linking with the spare parts inventory system. This module uses the fault code and confidence score generated by the fault diagnosis module, combines historical maintenance records and system operating status, and automatically evaluates the severity and urgency of the current fault. In this process, machine learning and data-driven algorithms are used to predict the possible impact of the fault on the production schedule based on real-time data analysis. For example, the system can optimize the corresponding maintenance strategy by analyzing the response time and repair duration of past similar faults, thus achieving the timeliness of fault handling;

[0079] Then, based on the fault assessment results, the module connects to the spare parts inventory management system in real time and automatically queries the inventory status of the required spare parts. This link uses an intelligent information system that can automatically generate a spare parts procurement list according to the fault type and maintenance suggestions, and through the set safety inventory threshold, ensures the timely replenishment of necessary parts and reduces the downtime caused by lack of parts. If the inventory of a specific part is insufficient, the module can also quickly submit a procurement application through the interface with the supply chain management system to ensure the seamless connection of maintenance work;

[0080] In addition, the decision support module not only focuses on fault repair but also provides energy efficiency optimization strategies based on data analysis. By comprehensively analyzing the operation data and maintenance records of the robot, this module can identify operating conditions and processes with insufficient energy efficiency and then recommend adjustment measures, such as optimizing the motion path, improving the load configuration during operation, or adjusting the working speed. The system can also help operators dynamically adjust the robot parameters during actual operation through a real-time monitoring and feedback mechanism to achieve more efficient energy management and resource utilization;

[0081] In summary, through intelligent fault analysis and resource linkage, this system not only improves the efficiency and accuracy of fault response but also provides practical guidance for the operation and maintenance decisions of enterprises, ultimately promoting the development of intelligent manufacturing towards a more efficient and reliable direction.

[0082] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A robot fault diagnosis system based on the Mworks platform, characterized in that: It includes data acquisition module, signal preprocessing module, feature extraction module, fault diagnosis module and decision support module; The data acquisition module collects the robot's time domain vibration signal, temperature data, electrical signal and motion parameters through vibration sensors, temperature sensors, current and voltage sensors and encoders; The signal preprocessing module performs wavelet transform noise reduction, synchronous sampling and data alignment on the robot's time domain vibration signal, and outputs the denoised time domain signal and FFT spectrum; The feature extraction module calculates the RMS value, kurtosis coefficient, harmonic energy ratio of the vibration signal, temperature rise gradient of the temperature data, current harmonic distortion rate of the electrical signal, and trajectory tracking error of the robot motion parameters based on the denoised time domain signal and FFT spectrum; The fault diagnosis module generates observer residuals based on the values calculated by the feature extraction module using parameter estimation method and outputs robot fault codes and confidence scores in combination with the SVM classifier; The decision support module links the spare parts inventory system according to the fault diagnosis results to generate robot maintenance suggestions and energy efficiency optimization strategies.

2. The robot fault diagnosis system based on the Mworks platform according to claim 1, wherein: The wavelet transform denoising formula is as follows: In the formula, s denoised (t) represents the signal output result of wavelet transform noise reduction, λ represents the denoising threshold, Thresh(*) represents the soft threshold function, and ψ j,k (t) represents the wavelet basis function, and W j,k represents the wavelet coefficient at scale j and position k.

3. The robot fault diagnosis system based on the Mworks platform according to claim 2, characterized in that: The formula for data alignment is as follows: In the formula, D(i,j) represents the cumulative alignment distance matrix, and x i , y j represent two groups of time series data points to be aligned, and ||*|| represents the Euclidean distance.

4. The robot fault diagnosis system based on the Mworks platform according to claim 3, characterized in that: The formula for calculating the RMS value of the vibration signal is as follows: In the formula, x n represents the sampling value of the discrete vibration signal, N represents the number of sampling points, and n represents the index subscript.

5. The robot fault diagnosis system based on the Mworks platform according to claim 4, characterized in that: The formula for the kurtosis coefficient is as follows: In the formula, K represents the kurtosis coefficient, and x n represents the sampling value of the discrete vibration signal, represents the sampling mean of the discrete vibration signal, N represents the number of sampling points, and n represents the index subscript.

6. The robot fault diagnosis system based on the Mworks platform according to claim 5, wherein: The calculation formula of the harmonic energy ratio is as follows: In the formula, E harmonic represents the harmonic energy ratio, X k represents the amplitude of the k-th frequency point of the FFT spectrum, H represents the set of harmonic frequency indices, k represents the frequency index in the frequency domain analysis, and N represents the total number of sampling points used for FFT analysis.

7. A robot fault diagnosis system based on the Mworks platform according to claim 6, characterized in that: The calculation formula for the temperature rise gradient of the temperature data is as follows: In the formula, G T represents the temperature rise gradient, T max represents the highest temperature during the monitoring period, T initial represents the initial temperature, and Δt represents the time interval.

8. The robot fault diagnosis system based on the Mworks platform according to claim 7, characterized in that: The calculation formula of the current harmonic distortion rate of the electrical signal is as follows: In the formula, THD represents the current harmonic distortion rate of the electrical signal, and I h represents the effective value of the h-th harmonic current, and H represents the highest harmonic order.

9. The robot fault diagnosis system based on the Mworks platform according to claim 8, characterized in that: The calculation formula of the trajectory tracking error of the robot motion parameters is as follows: In the formula, e(t) represents the trajectory tracking error of the robot's motion parameters, x ref (t), y ref (t) represent the reference trajectory coordinates, and x actual (t), y actual (t) represent the actual execution trajectory coordinates.

10. The robot fault diagnosis system based on the Mworks platform according to claim 9, characterized in that: The observer residual is as follows: In the formula, r θ represents the observer residual, τ measured represents the measured motor torque, J, B, and K respectively represent the moment of inertia, damping coefficient, and stiffness coefficient, and θ, respectively represent the position, velocity, and acceleration of the robot.

Images