Fault diagnosis method and system for mechanical equipment and storage medium

Abstract

The invention discloses a fault diagnosis method and system for mechanical equipment and a storage medium, and the method comprises the steps: obtaining data related to the mechanical equipment, determining that the data related to the mechanical equipment belongs to a known working condition through an anomaly detection model, determining the working condition related to the data related to the mechanical equipment through a classification model, a diagnostic model is selected for diagnosis based on a working condition of the data about the mechanical equipment. According to the fault diagnosis method of the mechanical equipment, an accurate fault diagnosis result can be provided even if the mechanical equipment is in different working conditions.

Description

Technical Field

[0001] The present disclosure relates to a fault diagnosis method, a fault diagnosis system and a storage medium for mechanical equipment. Background Art

[0002] Various mechanical equipment will inevitably have some usage failures during operation. At present, in order to diagnose mechanical equipment failures, maintenance engineers can evaluate mechanical equipment failures by measuring relevant data when the mechanical equipment is running / stopped. However, depending on the qualifications of the maintenance engineers themselves, there is a great deal of uncertainty in the evaluation of mechanical equipment by maintenance engineers. In addition, since mechanical equipment may be in different working conditions (i.e., operating modes), it may be very challenging to accurately diagnose mechanical equipment failures. Therefore, there is a need for a method that can diagnose mechanical equipment failures according to the working conditions of the mechanical equipment. Summary of the invention

[0003] According to an embodiment of the present disclosure, a fault diagnosis method for mechanical equipment is provided, including: obtaining data about the mechanical equipment, determining that the data about the mechanical equipment belongs to a known operating condition through an anomaly detection model, determining the operating condition of the data about the mechanical equipment through a classification model, and selecting a diagnostic model for diagnosis based on the operating condition of the data about the mechanical equipment.

[0004] According to the fault diagnosis method of the embodiment of the present disclosure, the anomaly detection model includes one or more of the following: a statistical model and an anomaly detection model using machine learning, wherein the anomaly detection model using machine learning includes: one or more of a distribution-based machine learning model, a distance-based machine learning model, a density-based machine learning model, a clustering-based machine learning model, a tree-based machine learning model, a dimensionality reduction-based machine learning model, a classification-based machine learning model, and a prediction-based machine learning model.

[0005] According to the fault diagnosis method of an embodiment of the present disclosure, the classification model includes one or more of the following: a data association model and a classification model using machine learning, wherein the classification model using machine learning is based on one or more of decision trees, random forests, logistic regression, and naive Bayes.

[0006] According to the fault diagnosis method of the embodiment of the present disclosure, the diagnosis model, the abnormality detection model, and the classification model are trained by using previously obtained data about the mechanical equipment.

[0007] According to the fault diagnosis method of an embodiment of the present disclosure, obtaining data about mechanical equipment also includes obtaining characteristics of data about mechanical equipment, and wherein the characteristics of data about mechanical equipment are obtained through one or more of the following: root mean square (RMS), peak to peak (P2P), mean, standard deviation, envelope 3 (Envolop3), time-frequency transform, and numerical operation.

[0008] According to the fault diagnosis method of the embodiment of the present disclosure, it also includes: determining that the data about the mechanical equipment does not belong to a known working condition, and reporting the data about the mechanical equipment.

[0009] According to the fault diagnosis method of an embodiment of the present disclosure, reporting data about mechanical equipment also includes: determining whether the data about the mechanical equipment is a new operating condition, and if it is determined that the data about the mechanical equipment is a new operating condition, adding the new operating condition to the known operating condition; if it is determined that the data about the mechanical equipment is not a new operating condition, discarding the data about the mechanical equipment.

[0010] According to the fault diagnosis method of the embodiment of the present disclosure, it also includes: developing a new diagnostic model based on data about mechanical equipment belonging to new working conditions, and updating the abnormality detection model and classification model based on data about mechanical equipment belonging to known working conditions including the new working conditions.

[0011] According to the fault diagnosis method of an embodiment of the present disclosure, the operating condition is associated with one or more operating parameters of the mechanical equipment.

[0012] According to an embodiment of the present disclosure, a fault diagnosis system for mechanical equipment is provided, comprising: a sensor unit, comprising one or more sensors, configured to collect data about the mechanical equipment; an anomaly detection model, configured to determine whether the data about the mechanical equipment belongs to a known operating condition; a classification model, configured to select a diagnostic model based on the operating condition of the data about the mechanical equipment; a diagnostic model library, comprising one or more diagnostic models configured to diagnose the mechanical equipment; and a processor, configured to: obtain data about the mechanical equipment from the sensor unit, determine whether the data about the mechanical equipment belongs to a known operating condition using the anomaly detection model, determine the operating condition of the data about the mechanical equipment using the classification model, and select a diagnostic model from the diagnostic model library for diagnosis based on the operating condition of the data about the mechanical equipment.

[0013] According to an embodiment of the present disclosure, there is provided a non-transitory storage medium on which instructions are stored. When executed by a processor, the instructions enable the processor to execute the fault diagnosis method of the mechanical equipment as described above.

[0014] According to the fault diagnosis method of mechanical equipment disclosed in the present invention, accurate fault diagnosis results can be provided even when the mechanical equipment is in different working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The above and other aspects, features and advantages of certain embodiments of the present disclosure will become more apparent from the following description in conjunction with the accompanying drawings, in which:

[0016] Figure 1a is a flow chart of a diagnostic method according to an embodiment of the present disclosure.

[0017] Figure 1b is a flow chart of another diagnostic method according to an embodiment of the present disclosure.

[0018] Figure 1c The flowchart of another diagnostic method according to an embodiment of the present disclosure is shown.

[0019] Figure 2a is a diagram of operating conditions and a diagnostic model according to an embodiment of the present disclosure.

[0020] Figure 2b is another diagram of an operating condition and a diagnostic model according to an embodiment of the present disclosure.

[0021] Figure 2c is another diagram of the operating condition and the diagnosis model according to an embodiment of the present disclosure.

[0022] Figure 3 is a system schematic diagram according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0023] Before proceeding to the following detailed description, it may be advantageous to set forth definitions of certain words and phrases used throughout this patent document. The terms "include" and "comprising" and their derivatives are meant to include, but are not limited to. The term "controller" or "control unit" refers to any device, system, or part thereof that controls at least one operation. Such a controller may be implemented with hardware, or a combination of hardware and software and / or firmware. The functionality associated with any particular controller may be centralized or distributed, whether local or remote. The phrase "at least one," when used with a list of items, means that different combinations of one or more of the listed items may be used, and only one item in the list may be required. For example, "at least one of A, B, C" includes any one of the following combinations: A, B, C, A and B, A and C, B and C, A and B and C.

[0024] Definitions for other specific words and phrases are provided throughout this patent document. Those of ordinary skill in the art should understand that in many, if not most instances, such definitions apply to prior and future uses of such defined words and phrases.

[0025] The various embodiments of the principles of the present disclosure in this patent document described below in conjunction with the drawings are only for illustration and should not be interpreted as limiting the scope of the present disclosure in any way. It will be appreciated by those skilled in the art that the principles of the present disclosure can be implemented in any appropriately arranged system or device. In some cases, the actions described in the specification can be performed in a different order and still achieve the desired result. In addition, the process depicted in the drawings does not necessarily require the specific order shown or the sequential order to achieve the desired result. In a particular embodiment, multitasking and parallel processing may be advantageous.

[0026] The mechanical equipment used throughout the present disclosure may include various general and special mechanical equipment, including but not limited to machine tools (e.g., lathes, drilling machines, boring machines, grinders, gear processing machines, thread processing machines, milling machines, planers, broaching machines, sawing machines and other machine tools, etc.), processing and manufacturing equipment (e.g., food processing equipment, textile equipment, chemical equipment, assembly equipment, etc.), vehicles (e.g., motor vehicles, ships, aircraft, public transportation vehicles, construction equipment, agricultural equipment, intelligent entities such as robots, etc.) and their components (e.g., engines, motors, reducers, bearings, vehicle chassis, robot bodies, etc.), construction and transportation mechanical equipment (e.g., pipelines, tracks, contact networks (rails), etc.), etc.

[0027] Figure 1a is another flow chart of a diagnostic method according to an embodiment of the present disclosure.

[0028] In S102, data about the mechanical device may be obtained. The data about the mechanical device may be collected by a sensor unit and may include data such as vibration, sound, temperature, chemical composition, image, pressure, current, voltage, speed, etc.

[0029] At S104, it may be determined through an anomaly detection model that the data about the mechanical equipment belongs to a known operating condition. The known operating condition may be a number of operating conditions determined based on previously collected data about the mechanical equipment. For example, the operating condition may be related to an operating parameter of the mechanical equipment.

[0030] In S106, the operating condition of the data about the mechanical equipment can be determined by the classification model, and a diagnostic model can be selected based on the operating condition for diagnosis. For example, when it is determined that the data about the mechanical equipment belongs to a certain operating condition (e.g., a certain operating speed), a diagnostic model that also belongs to the operating condition (e.g., the operating speed) can be selected for diagnosis.

[0031] Figure 1b is a flow chart of another diagnostic method according to an embodiment of the present disclosure.

[0032] and Figure 1a Similar to the S102, Figure 1bS102 can obtain data about mechanical equipment.

[0033] In S104, it can be determined by an anomaly detection model whether the data about the mechanical equipment belongs to a known working condition. The known working condition can be several working conditions determined based on the data about the mechanical equipment collected previously. For example, the working condition can be related to the operating parameters of the mechanical equipment. The known working condition can be defined as the same class using an anomaly detection model. The anomaly detection model can include a statistical model, an anomaly detection model using machine learning, and various other models. It can be determined whether the data about the mechanical equipment belongs to the class by an anomaly detection model including a statistical model, an anomaly detection model using machine learning, and various other models to determine whether the data about the mechanical equipment belongs to a known working condition. The anomaly detection model can be trained using previously obtained data about the mechanical equipment belonging to the existing working condition (for example, the features of the data about the mechanical equipment belonging to the existing working condition). Among them, the features of the data about the mechanical equipment belonging to the existing working condition can be extracted by root mean square (RMS), peak to peak (P2P), mean value, standard deviation, Envolop3, time-frequency transform (such as FFT transform), numerical operation, etc. For example, the known operating conditions may be ten operating conditions, and the anomaly detection model using machine learning may be trained using data belonging to the ten operating conditions, but the disclosure is not limited thereto.

[0034] In one embodiment, the statistical model may be a relatively lightweight model based on statistical principles. The statistical model may evaluate the probability of certain characteristic values ​​in the data obtained about the mechanical equipment, and determine whether the data about the mechanical equipment belongs to a known working condition by comparing the probability of the characteristic value with a threshold probability. In some cases, the statistical model may occupy relatively less memory space, require lower computing power, and run faster.

[0035] In one embodiment, the anomaly detection model using machine learning may include a distribution-based machine learning model (such as 3sigma, Z-score, boxplot, etc.), a distance-based machine learning model (such as K nearest neighbor (KNN)), a density-based machine learning model (such as Local Outlier Factor (LOF), Connectivity-Based Outlier Factor (COF), Stochastic Outlier Selection (SOS), etc.), a clustering-based machine learning model (such as Density-Based Spatial Clustering of Applications with Noise (DBSCAN)), etc.), a tree-based machine learning model (such as Isolation Forest (iForest)), a dimensionality reduction-based machine learning model (such as Principal Component Analysis (PCA), AutoEncoder, etc.), a classification-based machine learning model (such as One-Class Support Vector Machine (SVM), etc.), etc. SVM), etc.), prediction-based machine learning models (such as Moving Average, Autoregressive Integrated Moving Average model (ARIMA), etc.). Although the example embodiments show some anomaly detection models using machine learning, those skilled in the art should understand that the above description is merely exemplary and not exhaustive, and other existing or future developed models for anomaly detection can be used to determine whether data about mechanical equipment belongs to known working conditions, all of which are within the contemplation of the present disclosure. In some cases, the types of anomaly detection models using machine learning are relatively richer, and the anomaly detection results obtained are more accurate.

[0036] In one embodiment, the statistical model and the anomaly detection model using machine learning can be run serially or in parallel. In another embodiment, when the confidence of the results obtained by the statistical model is insufficient, the anomaly detection model using machine learning can be used.

[0037] If the data about the mechanical equipment belongs to a known operating condition, the process proceeds to S106. In S106, the operating condition of the data about the mechanical equipment can be determined by the classification model, and a diagnostic model can be selected for diagnosis based on the operating condition. For example, when it is determined that the data about the mechanical equipment belongs to a certain operating condition (for example, a certain operating speed), a diagnostic model that also belongs to the operating condition (for example, the operating speed) can be selected for diagnosis. The diagnostic model can correspond to the existing operating conditions respectively. For example, in an embodiment where the known operating conditions are ten operating conditions, the diagnostic models can be ten, and correspond to ten existing operating conditions respectively, but the present disclosure is not limited to this. For example, a diagnostic model belonging to a certain operating condition can be trained by using previously obtained data about the mechanical equipment that also belongs to the operating condition (for example, the features of the data about the mechanical equipment that belongs to the operating condition). Since the data about the mechanical equipment under different operating conditions may have large differences, selecting a diagnostic model belonging to the same operating condition for diagnosis can improve the accuracy of the diagnostic results.

[0038] In one embodiment, the selection of a diagnostic model based on the working condition can be performed through a classification model. For example, the classification model may include a classification model using machine learning, a data correlation model, or any other model. For example, in an embodiment where the known working conditions are ten working conditions, the classification model can be trained by previously obtained data belonging to the ten working conditions (for example, features of data about mechanical equipment belonging to the ten working conditions), but the disclosure is not limited to this. Among them, the features of the data about mechanical equipment belonging to the ten working conditions can be extracted by root mean square (RMS), peak to peak (P2P), mean value, standard deviation, Envolop3, time-frequency transform (such as FFT transform), numerical operation, etc.

[0039] In one embodiment, the data association model may correspond to a data / signal preprocessing method. For example, the data association model may be used to obtain the correlation between data about mechanical equipment and data about known working conditions, thereby selecting a diagnostic model corresponding to a known working condition with a high correlation based on the correlation sorting. In some cases, the data association model may occupy relatively less memory space, require lower computing power, and run faster.

[0040] In one embodiment, the classification model using machine learning can be based on decision tree, random forest, logistic regression, naive Bayes, etc. Although the example embodiments show some classification models using machine learning, those skilled in the art should understand that the above description is merely exemplary and not exhaustive, and other existing or future developed models for classification can be used to determine the known working conditions to which the data about the mechanical equipment belongs, all of which are within the scope of the present disclosure. In some cases, the types of classification models using machine learning are relatively richer, and the obtained abnormality detection results are more accurate.

[0041] In one embodiment, the data association model and the classification model using machine learning can be run serially or in parallel. In another embodiment, when the confidence of the result obtained by the data association model is insufficient, the classification model using machine learning can be used.

[0042] If the data about the mechanical equipment does not belong to a known working condition, the process proceeds to S108. In S108, the data about the mechanical equipment may be reported. For example, the data about the mechanical equipment may be reported to a service plant including human operators for further processing.

[0043] Figure 1c The flowchart of another diagnostic method according to an embodiment of the present disclosure is shown.

[0044] For the sake of brevity and to avoid unnecessary confusion, Figure 1c Zhongyu Figure 1b The same parts will not be described repeatedly.

[0045] At S110, it can be determined whether the data about the mechanical equipment is a new operating condition. For example, it can be determined that the data about the mechanical equipment is a new operating condition based on the data about the mechanical equipment (e.g., its characteristics). In this case, at S112, the new operating condition can be added to the known operating condition. In this way, a new diagnostic model can be developed based on the data about the mechanical equipment belonging to the new operating condition. In addition, the anomaly detection model and the classification model can be updated based on the data about the mechanical equipment belonging to the known operating condition including the new operating condition.

[0046] If it is determined that the data on the mechanical equipment is not a new operating condition, the process proceeds to S114 and the data on the mechanical equipment may be discarded. For example, if it is determined that the data on the mechanical equipment is noise, the data may be discarded without further analysis.

[0047] Through such an iterative approach, known operating conditions and diagnostic models can be continuously expanded to provide a diagnostic model that better matches the data about the mechanical equipment, making the diagnostic results more accurate.

[0048] Figure 2a is a diagram of operating conditions and a diagnostic model according to an embodiment of the present disclosure.

[0049] like Figure 2a As shown, in an embodiment where the example mechanical device has a group of operating parameters, the operating condition can be associated with the operating parameters of the group. In an embodiment where the mechanical device is a machine tool, the first group of operating parameters can be the rotational speed of the machine tool, wherein operating parameter 1.1 can be rotational speed 1, operating parameter 1.2 can be rotational speed 2 which is lower than rotational speed 1, and operating parameter 1.3 can be rotational speed 3 which is lower than rotational speed 2. In other embodiments, a group of operating parameters can include more or fewer operating parameters. Operating conditions 1-operating conditions 3 can correspond to operating parameters 1.1-operating parameters 1.3, respectively. The diagnostic models can correspond to the operating conditions, for example, diagnostic model 1-diagnostic model 3 correspond to operating conditions 1-operating conditions 3, respectively.

[0050] In other embodiments, the operating parameters of the mechanical device may be determined according to the type of the mechanical device. For example, in a vehicle embodiment, the operating parameters may include vehicle speed, road slope, vehicle load, etc. In a robot embodiment, the operating parameters may include the number of bodies operated by the robot and the type of action performed by the robot. However, the present disclosure is not limited thereto.

[0051] In one embodiment, the speed of the machine tool is speed 2, so as to refer to Figure 1a-Figure 1c As described, the data about the mechanical equipment can be determined to belong to the existing working condition through the anomaly detection model, and the data about the mechanical equipment can be determined to belong to working condition 2 through the classification model, so as to select diagnostic model 2 for diagnosis to obtain a diagnostic result. Diagnostic model 2 can be trained using data belonging to working condition 2.

[0052] Figure 2b is another diagram of an operating condition and a diagnostic model according to an embodiment of the present disclosure.

[0053] like Figure 2bAs shown, in an embodiment where the example mechanical device has more than one group of operating parameters, each operating condition can be associated with more than one group of operating parameters. For example, the example mechanical device can have two groups of operating parameters including a first group of operating parameters and a second group of operating parameters. In other embodiments, the example mechanical device can have more groups of operating parameters. In an embodiment where the mechanical device is a machine tool, the first group of operating parameters can be the rotational speed of the machine tool, wherein operating parameter 1.1 can be rotational speed 1, operating parameter 1.2 can be rotational speed 2 which is lower than rotational speed 1, and operating parameter 1.3 can be rotational speed 3 which is lower than rotational speed 2; the second group of operating parameters can be the type of material processed by the machine tool, wherein operating parameter 2.1 can be metal steel, operating parameter 2.2 can be metal copper, and operating parameter 2.3 can be metal aluminum. In other embodiments, each group of operating parameters can include more or fewer operating parameters. Refer to Figure 2b , operating condition 1 may correspond to operating parameters 1.1 and 2.1, and operating condition 2 may correspond to operating parameters 1.1 and 2.2. For the sake of brevity, the corresponding relationship between operating conditions 3-9 and the operating parameters is not described here. The diagnostic models may correspond to the operating conditions respectively, for example, diagnostic model 1-diagnostic model 9 correspond to operating conditions 1-9 respectively.

[0054] In one embodiment, the speed of the machine tool is speed 2 and the material processed by the machine tool is aluminum, so as shown in FIG. Figure 1a-Figure 1c As described, it is possible to determine that the data about the mechanical equipment belongs to the existing working condition through the anomaly detection model, and to determine that the data about the mechanical equipment belongs to working condition 6 through the classification model, thereby selecting the diagnostic model 6 for diagnosis to obtain a diagnostic result. The diagnostic model 6 can be trained using the data belonging to working condition 6. Since the diagnostic model 6 is trained using data that satisfies both operating parameters 1.2 and operating parameters 2.3, the diagnostic result obtained by the diagnostic model 6 can be more accurate than the diagnostic result obtained by the general diagnostic model, which is more significant when there are more operating parameter groups.

[0055] Figure 2c is another diagram of the operating condition and the diagnosis model according to an embodiment of the present disclosure.

[0056] like Figure 2cAs shown, in an embodiment where the example mechanical device has more than one group of operating parameters, each operating condition can be associated with one group of operating parameters. For example, the example mechanical device can have two groups of operating parameters including a first group of operating parameters and a second group of operating parameters. In other embodiments, the example mechanical device can have more groups of operating parameters. In an embodiment where the mechanical device is a machine tool, the first group of operating parameters can be the rotational speed of the machine tool, wherein operating parameter 1.1 can be rotational speed 1, operating parameter 1.2 can be rotational speed 2 which is lower than rotational speed 1, and operating parameter 1.3 can be rotational speed 3 which is lower than rotational speed 2; the second group of operating parameters can be the type of material processed by the machine tool, wherein operating parameter 2.1 can be metal steel, operating parameter 2.2 can be metal copper, and operating parameter 2.3 can be metal aluminum. In other embodiments, each group of operating parameters can include more or fewer operating parameters. Refer to Figure 2c , operating condition 1 may correspond to operating parameter 1.1, operating condition 2 may correspond to operating parameter 1.2, and for the sake of brevity, the corresponding relationship between operating conditions 3-6 and operating parameters will not be described further here. The diagnostic models may correspond to the operating conditions respectively, for example, diagnostic model 1-diagnostic model 6 correspond to operating conditions 1-6 respectively.

[0057] In one embodiment, the speed of the machine tool is speed 2 and the material processed by the machine tool is aluminum, so as shown in FIG. Figure 1a-Figure 1c As described, it is possible to determine that the data about the mechanical equipment belongs to the existing working condition through an abnormal detection method, and to determine that the data about the mechanical equipment belongs to working condition 2 and working condition 6 through a classification method, thereby selecting diagnostic model 2 and diagnostic model 6 for diagnosis to obtain two diagnostic results. Diagnostic model 2 can be trained using data belonging to working condition 2. Diagnostic model 6 can be trained using data belonging to working condition 6. The two diagnostic results can be combined to obtain a final diagnostic result. For example, the combination may include assigning weights to the two diagnostic results. Since each working condition is associated with only one group of operating parameters, the number of diagnostic models can be reduced, thereby saving development costs and storage space, which is more significant when there are more groups of operating parameters. By calling multiple diagnostic models that meet the working conditions and combining the diagnostic results, a relatively accurate diagnostic result can be obtained.

[0058] Figure 3 is a system schematic diagram according to an embodiment of the present disclosure.

[0059] although Figure 3 The storage device is not shown, but those skilled in the art will understand that the processor may include one or more storage devices on which instructions and / or data are stored. Figure 3The processor 320 and the sensor unit 310 are functionally shown as being within a single box, but one of ordinary skill in the art will appreciate that the processor 320 and the sensor unit 310 may actually include multiple processors 320 and multiple sensor units 310, which may or may not be stored in the same physical housing. Therefore, references to the processor 320 and the sensor unit 310 will be understood to include references to a collection of processors 320 and a collection of sensor units 310 operating in parallel or not in parallel.

[0060] Processor 320 may be any conventional processing unit, such as a commercial CPU. Alternatively, processor 320 may be a dedicated device, such as an ASIC or other hardware-based control unit.

[0061] The sensor unit 310 may include Figure 3 The various sensors shown in FIG. 1 may be used in conjunction with a greater or lesser variety of sensors.

[0062] In one embodiment, the vibration sensor may include but is not limited to a mechanical vibration sensor, an optical vibration sensor, and an electrical vibration sensor (such as an inductive vibration sensor, an eddy current vibration sensor, a capacitive vibration sensor, a resistive strain vibration sensor, and a piezoelectric vibration sensor).

[0063] In one embodiment, the sound sensor may include but is not limited to a piezoelectric ceramic acoustic sensor, a capacitive acoustic sensor, a magnetoelectric acoustic sensor, and the like.

[0064] In one embodiment, the chemical composition sensor may include but is not limited to viscosity sensors (for example, various viscosity sensors based on capillary method, falling ball method, vibration method, rotation method, ultrasound, optical technology, electromagnetic principle, and electromagnetic tomography technology), dielectric constant sensors (for example, capacitive type, quartz crystal microbalance type), particle number sensors (for example, fluid mining monitor (FCM), quantitative wear monitor (QDM), metal particle detector (MPD), tracer wear measuring instrument, iron spectrometer, acoustic emission wear detection technology, ultrasonic wear monitoring method, optical wear sensor), etc.

[0065] In one embodiment, the image sensor may include, but is not limited to, a charge coupled device (CCD) image sensor, a complementary metal oxide semiconductor (CMOS) image sensor, and the like.

[0066] In one embodiment, the mechanical sensor may include but is not limited to a multi-component force sensor (e.g., a two-component, three-component, four-component, six-component force sensor), a torque sensor (e.g., a dynamic torque sensor and a static torque sensor), and an acceleration sensor (e.g., a single-axis acceleration sensor, a three-axis acceleration sensor).

[0067] In one embodiment, the voltage sensor may include but is not limited to a voltage transformer, a Hall voltage sensor, a fiber optic voltage sensor, and the like.

[0068] In one embodiment, the current sensor may include, but is not limited to, a shunt, an electromagnetic current transformer, an electronic current transformer, and the like.

[0069] In one embodiment, the temperature sensor may include but is not limited to contact temperature sensors (such as bimetallic thermometers, glass liquid thermometers, pressure thermometers, resistance thermometers, thermistors and thermocouples, etc.) and non-contact temperature sensors (such as various non-contact temperature sensors based on radiation thermometry including brightness method, radiation method and colorimetry).

[0070] In one embodiment, the speed sensor may include but is not limited to a photoelectric speed sensor, a magnetoelectric speed sensor, a Hall speed sensor, and the like.

[0071] The connection between the processor 320 and the sensor unit 310 may be any connection capable of transmitting at least the data about the mechanical device output by the sensor to the processor 320. In one embodiment, the connection includes one or both of a cable connection and a wireless connection.

[0072] The connection part in the form of a cable connection may include a cable that transmits analog signals (e.g., voltage, 4-20mA current) or digital signals (pulse, CAN, RS485, etc.). The connection part in the form of a cable is more suitable for applications that require high-performance acquisition and high reliability.

[0073] Wireless connections can include a variety of configurations and protocols, including Bluetooth TM ,Bluetooth TMLE's short-range communication protocol, sub GHz, WirelessHART, infrared link, ZigBee, radio frequency identification (RFID), WiFi, Internet, World Wide Web, Intranet, Virtual Private Network, Wide Area Network, Local Area Network, Private network using a communication protocol that is proprietary to one or more companies, Ethernet and HTTP, various cellular communication technologies, such as GSM, CDMA, UMTS, EV-DO, WiMAX, LTE or 5th generation "5G" cellular technology and other cellular technologies developed in the future, and various combinations of the foregoing. The connection part in the form of wireless connection is more suitable for requirements such as easy installation and small size.

[0074] As reference Figure 1a and Figure 1b As described, the processor 320 may use an anomaly detection model to determine whether the data about the mechanical equipment belongs to a known working condition. The anomaly detection model may include a statistical model, an anomaly detection model using machine learning, and various other models. For the sake of brevity, the same parts will not be repeated.

[0075] As reference Figure 1a and Figure 1b As described, when the processor 320 determines that the data about the mechanical equipment belongs to a known working condition, the processor 320 can use the classification model to determine the working condition of the data about the mechanical equipment, and based on the working condition, select a diagnostic model from the diagnostic model library 350 for diagnosis. For example, when it is determined that the data about the mechanical equipment belongs to a certain working condition (e.g., a certain operating speed), a diagnostic model that also belongs to the working condition (e.g., the operating speed) can be selected for diagnosis. The diagnostic model can correspond to the existing working conditions respectively. For the sake of brevity, the same part will not be repeated.

[0076] As reference Figure 1a and Figure 1b As described, when the processor 320 determines that the data about the mechanical equipment does not belong to a known working condition, the processor 320 can report the data about the mechanical equipment. For example, the data of the mechanical equipment can be reported to a service factory 360 including a human operator for further processing. The service factory 360 can include a human operator or a processing device with higher computing power. The service factory 360 can share anomaly detection models, classification models, and diagnostic model libraries between different mechanical equipment in real time (if necessary).

[0077] For example, the service factory 360 can share data from the same type of sensors of the same type of mechanical equipment under the same working conditions in real time, so that different service factories can use richer data to develop and train the anomaly detection model 330, the classification model 340, and the diagnostic models in the diagnostic model library 350.

[0078] In one embodiment, the service factory 360 can determine whether the data about the mechanical equipment is a new operating condition. For example, the service factory 360 can determine that the data about the mechanical equipment is a new operating condition based on the data about the mechanical equipment (for example, its characteristics). In this case, the service factory 360 can add the new operating condition to the known operating condition. In this way, a new diagnostic model can be developed based on the new operating condition to update the diagnostic model library 350. In addition, the service factory 360 can update the anomaly detection model 330 and the classification model 340 based on the known operating conditions including the new operating condition. In the case of determining that the data about the mechanical equipment is not a new operating condition, the service factory 360 can discard the data about the mechanical equipment. For example, if it is determined that the data about the mechanical equipment is noise, the service factory 360 can discard the data without further analysis.

[0079] Figure 3 One or more of the processor 320, anomaly detection model 330, classification model 340, and diagnostic model library 350 may be cloud-based, which may be advantageous when local computing power and storage space are insufficient or communication is smooth. Figure 3 One or more of the processor 320, anomaly detection model 330, classification model 340, and diagnostic model library 350 may be edge-based, which may be advantageous when local computing power and storage space are sufficient or communications are blocked.

[0080] According to an embodiment of the present disclosure, a computer-readable medium is provided, on which computer instructions are stored. When the computer instructions are executed by a processor, the processor executes a fault diagnosis method for mechanical equipment.

[0081] The storage medium embodiments for providing the program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, the program code can be downloaded from a server computer by a communication network.

[0082] According to the fault diagnosis solution for mechanical equipment disclosed in the present invention, accurate fault diagnosis results can be provided even when the mechanical equipment is in different working conditions. The method can also detect the early wear of mechanical equipment and provide maintenance suggestions for customers to maintain maintenance in the early wear stage, thereby avoiding unplanned failures. By combining different working conditions, the accuracy of fault detection can be improved and the false alarm rate can be reduced.

[0083] The text and the accompanying drawings are provided as examples only to help understand the present disclosure. They should not be interpreted as limiting the scope of the present disclosure in any way. Although certain embodiments and examples have been provided, based on what is disclosed herein, it is clear to those skilled in the art that the embodiments and examples shown may be changed without departing from the scope of the present disclosure.

[0084] Although the present disclosure has been described with exemplary embodiments, various changes and modifications may be suggested to one skilled in the art. The present disclosure is intended to encompass such changes and modifications as fall within the scope of the appended claims.

[0085] Any description in the present invention should not be construed as implying that any particular element, step, or function is an essential element that must be included in the claims scope. The scope of the patented subject matter is limited solely by the claims.

Claims

1. A method for diagnosing a fault of a mechanical device, comprising: Get data about machinery and equipment, The anomaly detection model determines that the data about the mechanical equipment belongs to a known working condition. The operating condition of the data on the mechanical equipment is determined by the classification model, and a diagnosis model is selected for diagnosis based on the operating condition of the data on the mechanical equipment.

2. The method according to claim 1, wherein: The anomaly detection model includes one or more of the following: Statistical models and anomaly detection models using machine learning, Among them, the anomaly detection model using machine learning includes: one or more of a distribution-based machine learning model, a distance-based machine learning model, a density-based machine learning model, a clustering-based machine learning model, a tree-based machine learning model, a dimensionality reduction-based machine learning model, a classification-based machine learning model, and a prediction-based machine learning model.

3. The method according to claim 1, wherein: The classification model includes one or more of the following: Data association models and classification models using machine learning, The classification model using machine learning is based on one or more of decision tree, random forest, logistic regression, and naive Bayes.

4. The method according to claim 1, wherein: The diagnosis model, anomaly detection model, and classification model are trained by using previously acquired data about the mechanical equipment.

5. The method according to claim 1, wherein: Obtaining data about the mechanical device also includes obtaining characteristics of the data about the mechanical device, and, The characteristics of the data about the mechanical equipment are obtained by one or more of the following: Root mean square (RMS), peak to peak (P2P), average, standard deviation, envelope 3 (Envolop3), time-frequency transform, numerical operation.

6. The method according to claim 4, further comprising: Determine that the data on the mechanical equipment does not belong to a known operating condition, Reports data on machinery and equipment.

7. The method according to claim 6, wherein: The report also includes data on machinery and equipment: Determine whether the data about the mechanical equipment is a new condition, When the data about the mechanical equipment is determined to be a new working condition, the new working condition is added to the known working condition. In a case where it is determined that the data on the mechanical equipment is not a new operating condition, the data on the mechanical equipment is discarded.

8. The method according to claim 7, further comprising: Develop new diagnostic models based on data about the machine pertaining to the new operating conditions, and The anomaly detection model and the classification model are updated based on the data about the mechanical equipment belonging to the known operating conditions including the new operating conditions.

9. The method according to claim 1, wherein: The operating condition is associated with one or more operating parameters of the mechanical equipment.

10. A fault diagnosis system for mechanical equipment, comprising: a sensor unit, including one or more sensors, configured to collect data about the mechanical device; an anomaly detection model configured to determine that data about the mechanical equipment belongs to a known operating condition; a classification model configured to select a diagnostic model based on an operating condition of data regarding the mechanical equipment; A diagnostic model library, including one or more diagnostic models configured to diagnose mechanical equipment; as well as The processor is configured as: Obtain data about the mechanical equipment from the sensor unit, Use anomaly detection models to determine if data about mechanical equipment belongs to known operating conditions. Using the classification model, the operating condition of the data on the mechanical equipment is determined, and based on the operating condition of the data on the mechanical equipment, a diagnostic model is selected from a diagnostic model library for diagnosis.

Images