CN112799382A - Robot micro-electro-mechanical system fault diagnosis method and system - Google Patents
Robot micro-electro-mechanical system fault diagnosis method and system Download PDFInfo
- Publication number
- CN112799382A CN112799382A CN202110352970.8A CN202110352970A CN112799382A CN 112799382 A CN112799382 A CN 112799382A CN 202110352970 A CN202110352970 A CN 202110352970A CN 112799382 A CN112799382 A CN 112799382A
- Authority
- CN
- China
- Prior art keywords
- fault diagnosis
- micro
- electro
- mechanical system
- feedback data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 190
- 238000000034 method Methods 0.000 title claims abstract description 107
- 238000012549 training Methods 0.000 claims abstract description 59
- 238000001914 filtration Methods 0.000 claims abstract description 21
- 238000005094 computer simulation Methods 0.000 claims abstract description 10
- 238000005070 sampling Methods 0.000 claims description 21
- 238000012545 processing Methods 0.000 claims description 15
- 238000013527 convolutional neural network Methods 0.000 claims description 11
- 230000000694 effects Effects 0.000 claims description 7
- 230000004927 fusion Effects 0.000 claims description 6
- 238000004088 simulation Methods 0.000 claims description 6
- 238000002405 diagnostic procedure Methods 0.000 claims 1
- 238000004422 calculation algorithm Methods 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 12
- 230000006870 function Effects 0.000 description 11
- 238000004590 computer program Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 238000003860 storage Methods 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 6
- 230000007547 defect Effects 0.000 description 6
- 230000004044 response Effects 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 230000009471 action Effects 0.000 description 3
- 230000006399 behavior Effects 0.000 description 3
- 125000004432 carbon atom Chemical group C* 0.000 description 3
- 230000000875 corresponding effect Effects 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 239000000306 component Substances 0.000 description 2
- 230000001276 controlling effect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000000926 separation method Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000001010 compromised effect Effects 0.000 description 1
- 239000008358 core component Substances 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000007274 generation of a signal involved in cell-cell signaling Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0259—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the response to fault detection
- G05B23/0275—Fault isolation and identification, e.g. classify fault; estimate cause or root of failure
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Manipulator (AREA)
Abstract
The invention discloses a robot micro-electro-mechanical system fault diagnosis method and a system, wherein the method comprises the following steps: acquiring feedback data of the MEMS under a normal working condition, and filtering the data; generating feedback data of the MEMS under various faults by utilizing computer simulation; fusing the computer simulated data and the MEMS feedback data and performing data expansion on the fused data to construct a training sample data set; training the fault diagnosis model by adopting a training sample data set; acquiring feedback data of the MEMS to be diagnosed in real time; and inputting the real-time feedback data of the MEMS into the trained fault diagnosis model to realize fault diagnosis and fault classification. The invention ensures that the problem of MEMS fault diagnosis becomes more reliable, improves the MEMS fault recognition rate, increases the stability of a robot system and promotes the application of an advanced intelligent algorithm in the field of MEMS fault diagnosis.
Description
Technical Field
The invention relates to the technical field of robot fault diagnosis based on big data, in particular to a robot micro-electro-mechanical system fault diagnosis method and system.
Background
With the development and maturity of intelligent robot technology, robots and robot technology are not limited to the manufacturing field any more, but show superior applicability in the fields of resource exploration, disaster relief, medical treatment, military, aerospace and the like. The structural design requirements of the application robot on the robot under the complex environment are strict, and the robot gradually develops from an early single structural model to a modularized and complicated direction. The core component of the robot system is various micro-electro-mechanical systems (MEMS) sensors, and as various sensors such as electronics and machinery can be integrated in a narrow physical space, and the application environment of the robot is complex and various, the MEMS sensors are easily affected by factors such as noise, electromagnetic interference, temperature and vibration, and the like, and accordingly the sensors have high failure rate. Once a robot has a certain sensor failure, not only is the current task interrupted, but economic losses are incurred on a light basis, and the life safety of personnel may be compromised on a heavy basis. Therefore, it is very important to accurately identify and predict the sensor fault that has occurred or is about to occur in the robot.
Currently, in order to implement diagnosis and identification of a robot fault, existing research methods are mainly classified into three categories, namely a model-based fault diagnosis method, a knowledge-based fault diagnosis method and a data-driven robot fault diagnosis method.
Model-based approaches typically use analytical redundancy to detect and diagnose faults. The fault diagnosis of the robot system is carried out qualitatively by analyzing and modeling the correct behaviors of all components in the robot system and comparing expected output with observed output. The model-based robot fault diagnosis system includes two parts, residual signal generation, which is the difference between a measured process variable and an estimate of the variable made based on a process model; residual evaluation and decision making. Due to the problems of difficulty in modeling the robot with a complex structure, inaccuracy in modeling and the like, the application of the fault diagnosis method of the robot based on the model is limited.
Knowledge-based robotic fault diagnosis methods typically associate identified behaviors with predefined known faults and diagnoses. The fault diagnosis based on knowledge is based on a qualitative model of prior knowledge of a monitoring process, and fault diagnosis is realized by operating a mature search algorithm. The core of the system is an expert control system consisting of a knowledge base (knowledge base), a database (inference engine) and an interpretation component. The disadvantage is that it relies on expert knowledge and is not conducive to unknown types of faults occurring in the robotic system.
The robot system fault diagnosis method based on data driving is model-free, and potential faults and normal behaviors are distinguished by comprehensively analyzing data. The intelligent diagnosis method based on data driving still faces the defects of low fault diagnosis precision, long model training time and conflict between precision and diagnosis time, and the diagnosis method still needs to be improved, so that the fault recognition rate of the robot MEMS sensor is improved.
Disclosure of Invention
The invention provides a fault diagnosis method and a fault diagnosis system for a robot micro-electro-mechanical system, which aim to solve the technical problems that a fault diagnosis method for a robot based on a model depends on modeling precision, a fault diagnosis method for a robot based on knowledge depends on expert knowledge, and unknown fault categories cannot be effectively identified.
In order to solve the technical problems, the invention provides the following technical scheme:
in one aspect, the invention provides a robot micro-electro-mechanical system fault diagnosis method, which comprises a model training stage and a fault diagnosis stage; wherein,
in the model training stage, the fault diagnosis method for the robot micro-electro-mechanical system comprises the following steps:
collecting feedback data of the robot micro-electro-mechanical system under normal working conditions;
filtering the acquired feedback data of the micro-electro-mechanical system under the normal working condition;
generating feedback data of the robot micro-electro-mechanical system under the preset type fault by utilizing computer simulation;
performing data fusion on feedback data of a computer-simulated micro-electro-mechanical system under a preset type fault and feedback data of the filtered micro-electro-mechanical system under a normal working condition to obtain fused feedback data, and performing data expansion on the fused feedback data by adopting a sliding sampling method to construct a training sample data set;
training a preset fault diagnosis model based on the training sample data set; wherein the preset fault diagnosis model is a deep convolutional neural network model;
in the fault diagnosis stage, the fault diagnosis method of the robot micro-electro-mechanical system comprises the following steps:
acquiring feedback data of a micro electro mechanical system to be diagnosed in real time;
filtering the acquired real-time feedback data of the micro electro mechanical system to be diagnosed;
and inputting the filtered real-time feedback data of the micro-electro-mechanical system into a trained fault diagnosis model, and performing fault diagnosis and fault classification on the current operating condition of the micro-electro-mechanical system by using the trained fault diagnosis model.
Further, the micro-electro-mechanical system includes a robot joint position sensor and an actuator.
Further, the preset type fault comprises one or more of a sensor deviation fault, a sensor linear gain fault, a sensor damage fault, an actuator deviation fault, an actuator linear gain fault and an actuator stuck-at fault.
Further, the filtering processing of the acquired feedback data of the micro-electromechanical system under the normal working condition includes:
filtering the acquired feedback data of the micro electro mechanical system under the normal working condition by using a digital low-pass filter; wherein the upper limit cut-off frequency of the digital low-pass filter is adjustable.
Further, the robot micro-electro-mechanical system fault diagnosis method further comprises the following steps:
comparing the fault type of the micro-electro-mechanical system to be diagnosed with the known fault type, and adjusting the parameters of the fault diagnosis model according to the comparison result; wherein the parameters of the fault diagnosis model comprise convolution number and convolution depth; the convolution quantity and the convolution depth are adjusted according to the fault diagnosis effect; the convolution depth is set to unequal values on different convolution modules.
In another aspect, the present invention further provides a fault diagnosis system for a micro-electromechanical system of a robot, where the fault diagnosis system includes: the system comprises a micro electro mechanical system feedback data acquisition module, a micro electro mechanical system feedback data simulation module, a data processing module, a model training module, a fault diagnosis module and a working mode selection switch for controlling working modes, wherein the working modes comprise a model training mode and a fault diagnosis mode;
in the training mode of the model, the model is,
the feedback data acquisition module of the micro-electro-mechanical system is used for acquiring feedback data of the micro-electro-mechanical system of the robot under normal working conditions;
the feedback data simulation module of the micro-electro-mechanical system is used for generating feedback data of the micro-electro-mechanical system of the robot under the preset type fault by utilizing computer simulation;
the data processing module is used for filtering the acquired feedback data of the micro-electro-mechanical system under the normal working condition; performing data fusion on feedback data of the computer-simulated micro-electro-mechanical system under the preset type fault and the filtered feedback data of the micro-electro-mechanical system under the normal working condition, and performing data expansion on the fused feedback data by adopting a sliding sampling method to construct a training sample data set;
the model training module is used for training a preset fault diagnosis model based on the training sample data set; wherein the preset fault diagnosis model is a deep convolutional neural network model;
in the case of the failure diagnosis mode, the operation mode is,
the micro-electro-mechanical system feedback data acquisition module is used for acquiring the feedback data of the micro-electro-mechanical system to be diagnosed in real time;
the data processing module is used for filtering the acquired real-time feedback data of the micro electro mechanical system to be diagnosed;
the fault diagnosis module is used for inputting the filtered real-time feedback data of the micro-electromechanical system into a trained fault diagnosis model, and performing fault diagnosis and fault classification on the current operating condition of the micro-electromechanical system by using the trained fault diagnosis model.
In yet another aspect, the present invention also provides an electronic device comprising a processor and a memory; wherein the memory has stored therein at least one instruction that is loaded and executed by the processor to implement the above-described method.
In yet another aspect, the present invention also provides a computer-readable storage medium having at least one instruction stored therein, the instruction being loaded and executed by a processor to implement the above method.
The technical scheme provided by the invention has the beneficial effects that at least:
the fault diagnosis method of the robot micro-electro-mechanical system comprises a model training stage and a fault identification stage; in the model training stage, collecting normal operation data of the MEMS sensor, and processing the data through a digital low-pass filter; outputting data of the MEMS sensor under various fault conditions through computer simulation; performing a data set expansion method; training the fault diagnosis model to obtain a fault diagnosis model with better identification degree; and acquiring robot operation data containing MEMS sensor fault information in real time, and inputting the robot operation data into a trained fault diagnosis model to realize sensor fault diagnosis.
The invention is an end-to-end fault diagnosis method, and a neural network fault diagnosis model is used, so that the problem of precision loss caused by the fault diagnosis method based on the model can be effectively avoided, and the defect that the fault diagnosis method of the knowledge-based robot excessively depends on expert knowledge is also reduced. In addition, the invention realizes that the fault diagnosis of the robot joint position sensor and the actuator sensor is based on sensor feedback data, and is a data-driven fault diagnosis method. In the early model training stage, the fault diagnosis model is trained by using data containing various normal and abnormal working conditions, so that a better fault identification effect can be obtained. The invention provides a new method for the fault diagnosis of the industrial robot, improves the identification rate of the fault diagnosis of the robot, ensures the stability of the operation of the robot, promotes the application of an advanced intelligent algorithm in the field of the fault diagnosis of the robot, and has certain scientific significance.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a flowchart of a method for diagnosing a fault in a MEMS robot according to a first embodiment of the present invention;
FIG. 2 is a basic flowchart of a method for diagnosing a fault in a MEMS robot according to a second embodiment of the present invention;
FIG. 3 is a flowchart illustrating a method for diagnosing faults of a MEMS robot according to a second embodiment of the present invention;
FIG. 4 is a schematic diagram of a frequency response curve of a low-pass filter according to a second embodiment of the present invention;
FIG. 5 is a schematic diagram of a sliding sampling method according to a second embodiment of the present invention;
FIG. 6 is a schematic diagram of Softmax logistic regression according to a second embodiment of the present invention
Fig. 7 is a block diagram of an electronic device according to a fourth embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
First embodiment
The embodiment refers to a fault diagnosis method based on data driving, and designs a robot MEMS fault diagnosis method aiming at the MEMS sensor fault of a complex robot system, wherein the method can be realized by electronic equipment which can be a terminal or a server. The method comprises a model training stage and a fault diagnosis stage; specifically, the execution flow of the robot mems fault diagnosis method is shown in fig. 1, and includes the following steps:
in the model training stage, the fault diagnosis method for the robot micro-electro-mechanical system comprises the following steps:
s101, collecting feedback data of the robot MEMS under normal working conditions;
s102, filtering the acquired feedback data of the MEMS under the normal working condition;
s103, simulating feedback data of the robot MEMS under the preset type fault by using a computer;
s104, performing data fusion on feedback data of the MEMS simulated by the computer under the preset type fault and feedback data of the filtered MEMS under the normal working condition to obtain fused data, and performing data expansion on the fused data by adopting a sliding sampling method to construct a training sample data set;
wherein, the sampling step length of the sliding sampling can be adjusted according to the situation to obtain enough training fault samples;
s105, training a preset fault diagnosis model based on the training sample data set; wherein the preset fault diagnosis model is a deep convolutional neural network model; and inputting the sample data set into a fault diagnosis model, outputting a corresponding fault type by the fault diagnosis model, and realizing accurate identification and classification of the robot sensor fault by the fault diagnosis model according to the MEMS feedback data through training.
In the fault diagnosis stage, the fault diagnosis method of the robot micro-electro-mechanical system comprises the following steps:
s106, acquiring feedback data of the MEMS to be diagnosed in real time;
s107, filtering the acquired real-time feedback data of the MEMS to be diagnosed;
and S108, inputting the filtered real-time feedback data of the MEMS into the trained fault diagnosis model, and performing fault diagnosis and fault classification on the current MEMS operating condition by using the trained fault diagnosis model.
In the fault diagnosis stage, the data input into the neural network fault diagnosis model does not contain the fault data output by the computer simulation.
Specifically, in the present embodiment, the MEMS includes a robot joint position sensor and an actuator.
Correspondingly, the preset type faults comprise sensor faults and actuator faults;
the sensor faults include a sensor deviation fault, a sensor linear gain fault, and a sensor damage fault; the actuator faults comprise actuator deviation faults, actuator linear gain faults and actuator stuck faults; the various faults also include a combination of sensor faults and actuator faults;
through computer simulation, the fault feedback data of the sensor and the actuator under the fault state can be obtained; the fault feedback data contains fault information of one or more robot MEMS sensors.
The step of filtering the feedback data of the MEMS comprises the following steps:
filtering the feedback data of the MEMS by using a digital low-pass filter; wherein the upper cut-off frequency of the digital low-pass filter can be adjusted online or offline.
In addition, when the method of this embodiment performs fault diagnosis in real time, the method further includes: and comparing the fault type with the known fault type, wherein the comparison result can be used as the basis for adjusting the fault diagnosis model parameters. Wherein the model parameters comprise convolution number and convolution depth; the number of convolutions and the depth of the convolutions can be changed offline as appropriate; the adjustment of the convolution quantity and the convolution depth can be determined according to the fault diagnosis effect respectively, and the implementation modes are off-line modes; the convolution depth may be set to be unequal across different convolution modules.
Further, the method of this embodiment further includes: setting a mode selection switch; when the trained fault diagnosis model is used for carrying out fault diagnosis on the diagnosed object, the mode selection switch can ensure that data input into the fault diagnosis model does not contain fault data output by computer simulation.
In summary, in the embodiment, based on feedback data of the robot MEMS sensor under various working conditions, a training data set is obtained by a sliding sampling method; inputting the training data set into a robot fault diagnosis model to obtain a trained fault diagnosis model; and sampling real-time operation data of the MEMS sensor of the robot system to be diagnosed, and inputting the real-time operation data into the trained fault diagnosis model to realize the real-time diagnosis of the fault of the robot sensor. The invention is an end-to-end fault diagnosis method, and the fault diagnosis model of the deep convolutional neural network is used, so that the problem of precision loss caused by the fault diagnosis method based on the model can be effectively avoided, and the defect that the fault diagnosis method of the robot based on knowledge excessively depends on expert knowledge is also reduced. In addition, the invention realizes that the fault diagnosis of the robot MEMS sensor is based on sensor feedback data, and is a data-driven fault diagnosis method. In the early model training stage, the fault diagnosis model is trained by using data containing various normal and abnormal working conditions, so that a better fault identification effect can be obtained. The invention provides a new method for the fault diagnosis of the MEMS sensor of the industrial robot, improves the identification rate of the fault diagnosis of the MEMS sensor of the robot, ensures the stability of the operation of a robot system, promotes the application of an advanced intelligent algorithm in the field of the fault diagnosis of the robot, and has certain scientific significance.
Second embodiment
The method aims at the defects of the existing robot MEMS fault diagnosis method, wherein the difficulty of the robot fault diagnosis method based on the model is that the robot with a complex structure is difficult to model and inaccurate in modeling, so that the problem that the robot fault diagnosis method based on the model is limited in application is caused. The fault diagnosis method of the robot based on knowledge is too dependent on expert knowledge, and has no defect of being incapable of solving the fault of unknown type occurring in the robot system. By taking a fault diagnosis method based on data driving as a reference, the present embodiment designs a fault diagnosis method for a robot micro electro mechanical system for a MEMS sensor fault of a complex robot system, where a basic flow of the method is shown in fig. 2, and a specific implementation flow is shown in fig. 3. By combining the basic flowchart and the specific flowchart, the implementation flow steps of the micro-electromechanical system sensor fault diagnosis method of the embodiment mainly include:
step 3, expanding a fault sample data set by using a sliding sampling method in a training mode; the fault sample data set corresponds to the fault state of the MEMS sensor when the data are acquired one by one, and a diagnosis data set is obtained;
inputting the diagnosis data set into a fault diagnosis model for training to obtain a trained fault diagnosis model; wherein, the fault diagnosis model adopts a deep convolutional neural network method. Outputting a result consistent with the fault state of the MEMS sensor of the diagnosis data set by the trained fault diagnosis model with a great probability;
step 5, under a diagnosis mode, using a trained robot sensor fault diagnosis model to diagnose the operation state of the robot MEMS sensor; wherein the diagnosis mode uses a trained fault diagnosis model; the input of the fault diagnosis model is real-time feedback data of an MEMS sensor of the robot to be diagnosed; wherein the feedback data is processed by a filter.
Specifically, in step 1 of this embodiment, a filter is used to process the data of the MEMS sensor; the filter used is a digital low-pass filter, the general frequency response curve of which is shown in fig. 4. The digital low-pass filter is obtained by performing bilinear transformation on an analog low-pass filter, and the transfer function of the digital low-pass filter is as follows:
in the above formula, the first and second carbon atoms are,in order to be a function of the transfer function,representing complex frequency domain variable, wherein a coefficient k is direct current gain of a filter, and weak signals in feedback data of the MEMS sensor can be amplified through adjustment of the parameter; the coefficient T is the response time constant of the filter and also determines the upper cut-off frequency of the low-pass filter. As shown in fig. 4, curveAndrepresenting two frequency response curves of the filter for different T-actions.
In step 1 of this embodiment, when the robot is in different application environments, the parameters k and T may be selected offline to have better values, so as to meet the noise filtering requirement in the complex electromagnetic environment.
In the above formula, the first and second carbon atoms are,a fault output of the MEMS sensor is indicated,to gain for faultsThe value is normally 1,in order to be the normal output of the sensor,represents a sensor bias constant whose magnitude is positively correlated to the severity of the bias fault occurrence. The fault types are shown in table 1, where C is an arbitrary constant.
TABLE 1 MEMS sensor Fault types
And executing a corresponding fault expression by the computer simulation program according to the fault types in the table 1 to obtain the sensor fault data in a period of time.
In step 3 of this embodiment, a sliding sampling method is used to perform data set expansion on the fault data; the sliding sampling method can improve the utilization rate of data, as shown in fig. 5. And obtaining data output by the robot MEMS sensor after sampling for a certain time, wherein the total data sampling point is N. And (3) carrying out one-time training on the robot fault diagnosis model, wherein the length of a required data segment is N1. Therefore, using conventional data separation methods, the above-mentioned N data points can only result in N/N1 data segments. The basic principle of the sliding sampling method is as follows: the start position of the next data segment is h data points backwards from the end of the previous data segment, wherein h is called a sliding step. After the method is used, the number of data segments which can be obtained by the data with the original length of NThe calculation formulas and the calculation formulas of the difference between the data segments obtained by the conventional data separation method are respectively shown in (3) to (4).
Formula (4) shows that the number of samples obtained by using the sliding sampling method is not less than that of the conventional sampling method, and particularly, when h is large or the number of sample points N is large, the method can greatly expand the utilization rate of fault data.
Step 4 of the present embodiment trains the fault diagnosis model using the augmented fault data set, as shown in fig. 2. The fault diagnosis model is based on a deep convolutional neural network method, and the basic principle is as follows: the training data set is subjected to convolution operation through a plurality of convolution blocks to extract the implicit characteristics of the data. In order to effectively extract features, in the diagnostic model training stage, the number of volume blocks of the deep convolutional neural network can be adjusted in real time according to the diagnostic effect, and the plurality of volume blocks are called volume layer stacking. However, this adjustment of the volume blocks is not on-line, and the fault diagnosis model needs to be retrained each time the number of volume blocks is changed.
In order to further improve the fault diagnosis accuracy of the robot MEMS sensor by the diagnosis model, the embodiment introduces a method for changing the convolution depth. For the feedback data of the MEMS sensor, the implicit failure information is weak, such as the sensor is in the early failure initial stage. In order to improve the sensor fault identification degree under the environment, a parameter adjusting mechanism is introduced into the convolution depth from the convolution block 1 to the convolution block n. If the output of the current fault diagnosis module is not matched with the fault type of the real system sensor, and the probability of error identification is high, the depth of the rolling blocks can be increased, and the capability of extracting the features is improved.
Step 5 of this example uses the reject and full link layers to initially classify the MEMS sensor failure. The full link layer functions as a classifier which corresponds the data features extracted by the neural network convolution module to the fault information corresponding to the fault data set used by the training network one to one. The final classification was performed on the final output using Softmax logistic regression, as shown in FIG. 6.
Wherein, Softmax is a function, and the expression thereof is shown as formula (5). It can be made such that any possible output of the neural network satisfies the probability distribution shown in equation (6).
In the above formula, the first and second carbon atoms are,is arbitrary;representing a neural networkA probability of an output;representing a predicted output of the neural network;in order to sum the symbols, the symbols are summed,is the total number of failure categories studied. According to the fault diagnosis method for the MEMS sensor of the robot, the fault types of the researched MEMS sensor are divided into two directions, namely faults of the MEMS sensor and faults of an actuator, and the faults in each direction can be subdivided into bias faults, linear gain faults and failures. Thus, the Softmax classification function associates each data feature with the six categories described above.
And outputting the fault information contained in the current data at the maximum possible probability in the last step of the embodiment.
In summary, in the embodiment, based on feedback data of the robot MEMS sensor under various working conditions, a training data set is obtained by a sliding sampling method; inputting the training data set into a robot fault diagnosis model to obtain a trained fault diagnosis model; and sampling real-time operation data of the MEMS sensor of the robot system to be diagnosed, and inputting the real-time operation data into the trained fault diagnosis model to realize the real-time diagnosis of the fault of the robot sensor. The invention is an end-to-end fault diagnosis method, and the fault diagnosis model of the deep convolutional neural network is used, so that the problem of precision loss caused by the fault diagnosis method based on the model can be effectively avoided, and the defect that the fault diagnosis method of the robot based on knowledge excessively depends on expert knowledge is also reduced. In addition, the invention realizes that the fault diagnosis of the robot MEMS sensor is based on sensor feedback data, and is a data-driven fault diagnosis method. In the early model training stage, the fault diagnosis model is trained by using data containing various normal and abnormal working conditions, so that a better fault identification effect can be obtained. The invention provides a new method for the fault diagnosis of the MEMS sensor of the industrial robot, improves the identification rate of the fault diagnosis of the MEMS sensor of the robot, ensures the stability of the operation of a robot system, promotes the application of an advanced intelligent algorithm in the field of the fault diagnosis of the robot, and has certain scientific significance.
Third embodiment
The embodiment provides a fault diagnosis system for a micro-electromechanical system of a robot, which comprises: the system comprises a micro electro mechanical system feedback data acquisition module, a micro electro mechanical system feedback data simulation module, a data processing module, a model training module, a fault diagnosis module and a working mode selection switch for controlling working modes, wherein the working modes comprise a model training mode and a fault diagnosis mode;
in the training mode of the model, the model is,
the feedback data acquisition module of the micro-electro-mechanical system is used for acquiring feedback data of the micro-electro-mechanical system of the robot under normal working conditions;
the feedback data simulation module of the micro-electro-mechanical system is used for generating feedback data of the micro-electro-mechanical system of the robot under the preset type fault by utilizing computer simulation;
the data processing module is used for filtering the acquired feedback data of the micro-electro-mechanical system under the normal working condition; performing data fusion on feedback data of the computer-simulated micro-electro-mechanical system under the preset type fault and the filtered feedback data of the micro-electro-mechanical system under the normal working condition, and performing data expansion on the fused feedback data by adopting a sliding sampling method to construct a training sample data set;
the model training module is used for training a preset fault diagnosis model based on the training sample data set; wherein the preset fault diagnosis model is a deep convolutional neural network model;
in the case of the failure diagnosis mode, the operation mode is,
the micro-electro-mechanical system feedback data acquisition module is used for acquiring the feedback data of the micro-electro-mechanical system to be diagnosed in real time;
the data processing module is used for filtering the acquired real-time feedback data of the micro electro mechanical system to be diagnosed;
the fault diagnosis module is used for inputting the filtered real-time feedback data of the micro-electromechanical system into a trained fault diagnosis model, and performing fault diagnosis and fault classification on the current operating condition of the micro-electromechanical system by using the trained fault diagnosis model.
The robot micro-electromechanical system fault diagnosis system of the embodiment corresponds to the robot micro-electromechanical system fault diagnosis method of the first embodiment; the functions realized by the functional modules in the fault diagnosis system of the robot micro-electromechanical system of the embodiment correspond to the flow steps in the fault diagnosis method of the robot micro-electromechanical system of the first embodiment one by one; therefore, it is not described herein.
Fourth embodiment
The present embodiment provides an electronic device, as shown in fig. 7, including a controller and a storage device; one or more computer instructions are stored in the storage device, and the instructions are loaded and executed by the controller to implement the robot MEMS fault diagnosis method of the above embodiment.
The electronic devices may have large differences due to different configurations or performances, and may include one or more controllers and one or more storage devices, which may be Random Access Memory (RAM) or non-volatile memory (non-volatile memory).
In addition, the electronic device shown in fig. 7 may further include a data interface and a display interface. The data interface is used for data transmission of the robot MEMS sensor, and the display interface is used for outputting fault information.
Fifth embodiment
The present embodiment provides a computer-readable storage medium, which stores at least one instruction, and the instruction is loaded and executed by a processor to implement the method of the above embodiment. The computer readable storage medium may be, among others, ROM, random access memory, CD-ROM, magnetic tape, floppy disk, optical data storage device, and the like. The instructions stored therein may be loaded by a processor in the terminal and perform the above-described method.
Furthermore, it should be noted that the present invention may be provided as a method, apparatus or computer program product. Accordingly, embodiments of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present invention may take the form of a computer program product embodied on one or more computer-usable storage media having computer-usable program code embodied in the medium.
Embodiments of the present invention are described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, embedded processor, or other programmable data processing terminal to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks. These computer program instructions may also be loaded onto a computer or other programmable data processing terminal to cause a series of operational steps to be performed on the computer or other programmable terminal to produce a computer implemented process such that the instructions which execute on the computer or other programmable terminal provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It is further noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. The terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or terminal that comprises the element.
Finally, it should be noted that while the above describes a preferred embodiment of the invention, it will be appreciated by those skilled in the art that, once the basic inventive concepts have been learned, numerous changes and modifications may be made without departing from the principles of the invention, which shall be deemed to be within the scope of the invention. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the embodiments of the invention.
Claims (6)
1. The robot micro-electro-mechanical system fault diagnosis method is characterized by comprising a model training stage and a fault diagnosis stage; wherein,
in the model training stage, the fault diagnosis method for the robot micro-electro-mechanical system comprises the following steps:
collecting feedback data of the robot micro-electro-mechanical system under normal working conditions;
filtering the acquired feedback data of the micro-electro-mechanical system under the normal working condition;
generating feedback data of the robot micro-electro-mechanical system under the preset type fault by utilizing computer simulation;
performing data fusion on feedback data of a computer-simulated micro-electro-mechanical system under a preset type fault and feedback data of the filtered micro-electro-mechanical system under a normal working condition to obtain fused feedback data, and performing data expansion on the fused feedback data by adopting a sliding sampling method to construct a training sample data set;
training a preset fault diagnosis model based on the training sample data set; wherein the preset fault diagnosis model is a deep convolutional neural network model;
in the fault diagnosis stage, the fault diagnosis method of the robot micro-electro-mechanical system comprises the following steps:
acquiring feedback data of a micro electro mechanical system to be diagnosed in real time;
filtering the acquired real-time feedback data of the micro electro mechanical system to be diagnosed;
and inputting the filtered real-time feedback data of the micro-electro-mechanical system into a trained fault diagnosis model, and performing fault diagnosis and fault classification on the current operating condition of the micro-electro-mechanical system by using the trained fault diagnosis model.
2. The robotic mems fault diagnostic method of claim 1, wherein the mems includes a robot joint position sensor and an actuator.
3. The robotic mems fault diagnosis method of claim 2 wherein the predetermined types of faults include a combination of one or more of sensor bias faults, sensor linear gain faults, sensor damage faults, actuator bias faults, actuator linear gain faults, and actuator stuck-at faults.
4. The method for fault diagnosis of a robot micro electro mechanical system according to claim 1, wherein the step of filtering the acquired feedback data of the micro electro mechanical system under normal working conditions comprises:
filtering the acquired feedback data of the micro electro mechanical system under the normal working condition by using a digital low-pass filter; wherein the upper limit cut-off frequency of the digital low-pass filter is adjustable.
5. The robot mems fault diagnosis method according to claim 1, wherein the robot mems fault diagnosis method further comprises:
comparing the fault type of the micro-electro-mechanical system to be diagnosed with the known fault type, and adjusting the parameters of the fault diagnosis model according to the comparison result; wherein the parameters of the fault diagnosis model comprise convolution number and convolution depth; the convolution quantity and the convolution depth are adjusted according to the fault diagnosis effect; the convolution depth is set to unequal values on different convolution modules.
6. A robotic MEMS fault diagnosis system, comprising: the system comprises a micro electro mechanical system feedback data acquisition module, a micro electro mechanical system feedback data simulation module, a data processing module, a model training module, a fault diagnosis module and a working mode selection switch for controlling working modes, wherein the working modes comprise a model training mode and a fault diagnosis mode;
in the training mode of the model, the model is,
the feedback data acquisition module of the micro-electro-mechanical system is used for acquiring feedback data of the micro-electro-mechanical system of the robot under normal working conditions;
the feedback data simulation module of the micro-electro-mechanical system is used for generating feedback data of the micro-electro-mechanical system of the robot under the preset type fault by utilizing computer simulation;
the data processing module is used for filtering the acquired feedback data of the micro-electro-mechanical system under the normal working condition; performing data fusion on feedback data of the computer-simulated micro-electro-mechanical system under the preset type fault and the filtered feedback data of the micro-electro-mechanical system under the normal working condition, and performing data expansion on the fused feedback data by adopting a sliding sampling method to construct a training sample data set;
the model training module is used for training a preset fault diagnosis model based on the training sample data set; wherein the preset fault diagnosis model is a deep convolutional neural network model;
in the case of the failure diagnosis mode, the operation mode is,
the micro-electro-mechanical system feedback data acquisition module is used for acquiring the feedback data of the micro-electro-mechanical system to be diagnosed in real time;
the data processing module is used for filtering the acquired real-time feedback data of the micro electro mechanical system to be diagnosed;
the fault diagnosis module is used for inputting the filtered real-time feedback data of the micro-electromechanical system into a trained fault diagnosis model, and performing fault diagnosis and fault classification on the current operating condition of the micro-electromechanical system by using the trained fault diagnosis model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110352970.8A CN112799382A (en) | 2021-04-01 | 2021-04-01 | Robot micro-electro-mechanical system fault diagnosis method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110352970.8A CN112799382A (en) | 2021-04-01 | 2021-04-01 | Robot micro-electro-mechanical system fault diagnosis method and system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112799382A true CN112799382A (en) | 2021-05-14 |
Family
ID=75816059
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110352970.8A Pending CN112799382A (en) | 2021-04-01 | 2021-04-01 | Robot micro-electro-mechanical system fault diagnosis method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112799382A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113670645A (en) * | 2021-09-01 | 2021-11-19 | 上海节卡机器人科技有限公司 | Fault detection method and device, electronic equipment and storage medium |
WO2023125059A1 (en) * | 2021-12-29 | 2023-07-06 | 中兴通讯股份有限公司 | Failure data acquisition method, failure handling method, electronic device, and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110100126A1 (en) * | 2009-10-29 | 2011-05-05 | Hitachi Automotive Systems, Ltd. | Capacitance Sensor |
CN107909585A (en) * | 2017-11-14 | 2018-04-13 | 华南理工大学 | Inner membrance dividing method in a kind of blood vessel of intravascular ultrasound image |
CN110514924A (en) * | 2019-08-12 | 2019-11-29 | 武汉大学 | The Winding in Power Transformer Fault Locating Method of depth convolutional neural networks fusion visual recognition |
CN112171721A (en) * | 2020-11-30 | 2021-01-05 | 北京科技大学 | Robot joint sensor and actuator fault diagnosis method and system |
-
2021
- 2021-04-01 CN CN202110352970.8A patent/CN112799382A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110100126A1 (en) * | 2009-10-29 | 2011-05-05 | Hitachi Automotive Systems, Ltd. | Capacitance Sensor |
CN107909585A (en) * | 2017-11-14 | 2018-04-13 | 华南理工大学 | Inner membrance dividing method in a kind of blood vessel of intravascular ultrasound image |
CN110514924A (en) * | 2019-08-12 | 2019-11-29 | 武汉大学 | The Winding in Power Transformer Fault Locating Method of depth convolutional neural networks fusion visual recognition |
CN112171721A (en) * | 2020-11-30 | 2021-01-05 | 北京科技大学 | Robot joint sensor and actuator fault diagnosis method and system |
Non-Patent Citations (1)
Title |
---|
彭开香 等: ""复杂工业过程质量相关的故障检测与诊断技术综述"", 《自动化学报》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113670645A (en) * | 2021-09-01 | 2021-11-19 | 上海节卡机器人科技有限公司 | Fault detection method and device, electronic equipment and storage medium |
WO2023125059A1 (en) * | 2021-12-29 | 2023-07-06 | 中兴通讯股份有限公司 | Failure data acquisition method, failure handling method, electronic device, and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP3809220B1 (en) | Method and system for semi-supervised deep anomaly detection for large-scale industrial monitoring systems based on time-series data utilizing digital twin simulation data | |
D’Angelo et al. | A new fault classification approach applied to Tennessee Eastman benchmark process | |
EP3171239B1 (en) | Predictive monitoring and diagnostics system and method | |
JP5538597B2 (en) | Anomaly detection method and anomaly detection system | |
CN113128561A (en) | Machine tool bearing fault diagnosis method | |
EP1982301B1 (en) | Method of condition monitoring | |
CN111385140B (en) | System and method for generating data for monitoring network physical system to determine abnormality early | |
US20020013664A1 (en) | Rotating equipment diagnostic system and adaptive controller | |
JP2010191556A (en) | Abnormality detecting method and abnormality detecting system | |
CN112799382A (en) | Robot micro-electro-mechanical system fault diagnosis method and system | |
CN108257365B (en) | Industrial alarm design method based on global uncertainty evidence dynamic fusion | |
Jalali et al. | Predicting time-to-failure of plasma etching equipment using machine learning | |
EP3674946B1 (en) | System and method for detecting anomalies in cyber-physical system with determined characteristics | |
CN114625009A (en) | Fault detection method based on system identification and optimal filtering | |
CN114861522A (en) | Precision manufacturing quality monitoring method and device based on artificial intelligence meta-learning technology | |
Gamanayake et al. | A hybrid deep learning model-based remaining useful life estimation for reed relay with degradation pattern clustering | |
CN116360384A (en) | System and method for diagnosing and monitoring anomalies in information physical systems | |
Arellano–Espitia et al. | Anomaly detection in electromechanical systems by means of deep-autoencoder | |
Meyes et al. | Transparent and interpretable failure prediction of sensor time series data with convolutional neural networks | |
EP3674828A1 (en) | System and method of generating data for monitoring of a cyber-physical system for early determination of anomalies | |
CN116235121A (en) | Apparatus and method for identifying anomalies in industrial facilities for performing a production process | |
Alfaro-Nango et al. | Predictive maintenance algorithm based on machine learning for industrial asset | |
Ahmed | Training of Neural Networks Using the Smooth Variable Structure Filter with Application to Fault Detection | |
Saddem et al. | Benefits of using Digital Twin for online fault diagnosis of a manufacturing system | |
Li et al. | An intelligent predictive engine for milling machine prognostic monitoring |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210514 |
|
RJ01 | Rejection of invention patent application after publication |