CN112486136B - Fault early warning system and method - Google Patents
Fault early warning system and method Download PDFInfo
- Publication number
- CN112486136B CN112486136B CN201910860038.9A CN201910860038A CN112486136B CN 112486136 B CN112486136 B CN 112486136B CN 201910860038 A CN201910860038 A CN 201910860038A CN 112486136 B CN112486136 B CN 112486136B
- Authority
- CN
- China
- Prior art keywords
- fault
- predicted
- model
- fault prediction
- field 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.)
- Active
Links
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
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0224—Process history based detection method, e.g. whereby history implies the availability of large amounts of data
-
- 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
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/24—Pc safety
- G05B2219/24065—Real time diagnostics
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The embodiment of the invention provides a fault early warning system, which comprises fault early warning equipment and cloud computing equipment, wherein the fault early warning equipment comprises: the field data acquisition module is used for acquiring field data of an object to be predicted; and the edge computing module is used for receiving the field data and a fault prediction model from cloud computing equipment, and inputting the field data into the fault prediction model so as to judge whether the object to be predicted has a fault or is about to have a fault. The invention can ensure the accuracy of the fault prediction model, and can ensure that the fault early warning equipment can always use the accurate fault prediction model issued by the cloud computing equipment to carry out fault prediction, thereby ensuring the accuracy of fault prediction and early warning of the object to be predicted.
Description
Technical Field
The invention relates to the field of fault early warning, in particular to a fault early warning system and a fault early warning method.
Background
The concrete piston is a wear-out component of concrete pumping equipment, and when a fault occurs in the construction process of the concrete piston, the pumping equipment is shut down unplanned, construction of supporting equipment is influenced, construction period is delayed, and additional economic loss is brought. The method has the advantages that the fault early warning is carried out on the concrete piston, the predictive maintenance is implemented, the equipment efficiency can be improved, the maintenance cost is reduced, meanwhile, the service engineer and the spare part scheduling are optimized, and the passive service is changed into the active service.
At present, the concrete piston fault is judged and processed mainly by two methods: one is that the field is artificially judged according to the abnormal performance of the functional performance of the pumping equipment, such as the pumping is weak, slurry leakage and the like, and then the pumping equipment is replaced after a fault occurs. In the other method, firstly, experimental tests are carried out under comparatively simple working conditions such as theoretical simulation calculation, an experiment table and the like, the design service life is determined, and the concrete piston is replaced with a fixed service life in the application process.
However, the existing method for judging and processing the concrete piston fault is to replace the concrete piston according to the designed service life or the field artificial judgment, the concrete piston fault cannot be accurately predicted in advance, the predictive maintenance cannot be realized, and the field shutdown of pumping equipment is easily caused.
Disclosure of Invention
The embodiment of the invention aims to solve the problems that the concrete piston fault cannot be accurately pre-judged in advance, predictive maintenance cannot be realized, the field shutdown of pumping equipment is easily caused, the real-time working condition parameter information is not fully utilized, and early warning is difficult to be timely sent out when the concrete piston works abnormally or is close to the fault.
In order to achieve the above object, according to a first aspect of the present invention, there is provided a fault early warning apparatus including: the field data acquisition module is used for acquiring field data of an object to be predicted; and the edge computing module is used for receiving the field data and a fault prediction model from cloud computing equipment, and inputting the field data into the fault prediction model so as to judge whether the object to be predicted has a fault or is about to have a fault.
Optionally, the field data collecting module includes: a condition parameter acquisition module for acquiring one or more of the following parameters relating to the object to be predicted: the system comprises a working duration parameter, a working amount parameter, an engine rotating speed parameter, an oil pump rotating speed parameter, a pumping pressure parameter and a hydraulic oil temperature parameter.
Optionally, the field data acquisition module includes one or more of the following: the position information acquisition module is used for acquiring the position information of the object to be predicted; and the replacement switch is used for acquiring a replacement record of the object to be predicted.
Optionally, the edge calculation module includes: the system comprises an edge end signal processing module and an edge end real-time fault early warning module; the edge terminal signal processing module is used for receiving the field data, processing the field data and sending the processed field data to the edge terminal real-time fault early warning module and the cloud computing equipment; and the edge end real-time fault early warning module is used for substituting the processed field data into the fault prediction model to judge whether the object to be predicted has a fault or is about to have a fault, and outputting a fault early warning signal under the condition of judging that the object to be predicted has the fault or is about to have the fault.
Optionally, the apparatus further comprises: and the alarm device is used for giving an alarm under the condition that the object to be predicted has or is about to have a fault.
Optionally, the object to be predicted is a concrete piston.
According to a second aspect of the present invention, the present invention provides a cloud computing device for fault pre-warning, the device comprising: the fault prediction model training module is used for training to obtain a fault prediction model based on historical field data of an object to be predicted, which contains a fault label, and storing the fault prediction model into a fault prediction model library; and the fault prediction model base is used for acquiring and storing the fault prediction model from the fault prediction model training module and sending the stored fault prediction model to the edge calculation module.
Optionally, the apparatus further comprises: and the fault prediction module is used for receiving field data of the object to be predicted, calculating a fault prediction result of the object to be predicted according to the fault prediction model stored in the fault prediction model base, and sending the fault prediction result of the object to be predicted to the service application module.
Optionally, the service application module includes one or more of the following: the fault early warning pushing module is used for sending out early warning according to the fault prediction result of the object to be predicted; the service engineer scheduling module is used for sending out service engineer scheduling information according to the fault prediction result of the object to be predicted; and the accessory scheduling module is used for sending accessory scheduling information according to the fault prediction result of the object to be predicted.
Optionally, the fault prediction model includes: the reliability model is used for outputting a first fault prediction result based on the field data of the object to be predicted; the machine learning model is used for outputting a second fault prediction result based on the field data of the object to be predicted; and a second ensemble learning model for generating a final fault prediction result based on the first and second fault prediction results, wherein the reliability model is trained based on fault and maintenance data of the object to be predicted and/or experimental data of the object to be predicted, and the machine learning model is trained based on historical field data of the object to be predicted.
Optionally, the reliability model includes one or more of: a bathtub curve model trained based on the historical subject field data to be predicted and the subject fault and maintenance data to be predicted; and the degradation orbit model is trained on the basis of experimental data of the object to be predicted.
Optionally, the machine learning model includes: the random forest model, the gradient lifting tree model and the deep learning model are used for outputting corresponding fault prediction results according to the field data of the object to be predicted respectively; and the first ensemble learning model is used for generating a final fault prediction result as a second fault prediction result of the machine learning model according to the corresponding fault prediction result.
Optionally, the fault prediction model further includes: and the feature extraction algorithm model is used for extracting shape features, frequency spectrum features and/or statistical features of the working condition time sequence based on the to-be-predicted object field data and/or the historical to-be-predicted object field data, and inputting the extracted shape features, frequency spectrum features and/or statistical features into the random forest model and the gradient lifting tree model.
Optionally, the object to be predicted is a concrete piston.
According to a third aspect of the present invention, there is provided a fault warning system comprising: the above-mentioned failure early warning device; and the cloud computing device.
According to a fourth aspect of the present invention, there is provided a fault warning method, comprising: receiving field data of an object to be predicted; receiving a fault prediction model from a cloud computing device; and inputting the field data into the fault prediction model to judge whether the object to be predicted has a fault or is about to have a fault.
Optionally, the live data comprises one or more of the following parameters relating to the object to be predicted: the system comprises a working duration parameter, a working amount parameter, an engine rotating speed parameter, an oil pump rotating speed parameter, a pumping pressure parameter and a hydraulic oil temperature parameter.
Optionally, the field data includes one or more of: position information of the object to be predicted; and a replacement record of the object to be predicted.
Optionally, the method further includes: and sending the field data to a fault prediction module of the cloud computing equipment.
Optionally, the method further includes: and alarming under the condition that the object to be predicted has or is about to have a fault.
Optionally, the object to be predicted is a concrete piston.
According to a fifth aspect of the present invention, there is provided a method for fault pre-warning performed in a cloud computing device, the method comprising: training to obtain a fault prediction model based on historical field data of the object to be predicted, which contains a fault label; and sending the fault prediction model to fault early warning equipment which is positioned near the object to be predicted.
Optionally, the method further includes: receiving field data of an object to be predicted; and inputting the field data of the object to be predicted into the fault prediction model to obtain a fault prediction result of the object to be predicted, and sending the fault prediction result of the object to be predicted to a service application module.
Optionally, the service application module includes one or more of the following: the fault early warning pushing module is used for sending out early warning according to the fault prediction result of the object to be predicted; the service engineer scheduling module is used for sending out service engineer scheduling information according to the fault prediction result of the object to be predicted; and the accessory scheduling module is used for sending accessory scheduling information according to the fault prediction result of the object to be predicted.
Optionally, the fault prediction model includes: the reliability model is used for outputting a first fault prediction result based on the field data of the object to be predicted; the machine learning model is used for outputting a second fault prediction result based on the field data of the object to be predicted; and a second ensemble learning model for generating a final fault prediction result based on the first and second fault prediction results, wherein the reliability model is trained based on fault and maintenance data of the object to be predicted and/or experimental data of the object to be predicted, and the machine learning model is trained based on historical field data of the object to be predicted.
Optionally, the reliability model includes one or more of: a bathtub curve model trained based on the historical subject field data to be predicted and the subject fault and maintenance data to be predicted; and the degradation orbit model is trained on the basis of experimental data of the object to be predicted.
Optionally, the machine learning model includes: the random forest model, the gradient lifting tree model and the deep learning model are used for outputting corresponding fault prediction results according to the field data of the object to be predicted respectively; and the first ensemble learning model is used for generating a final fault prediction result as a second fault prediction result of the machine learning model according to the corresponding fault prediction result.
Optionally, the method further includes: and extracting shape features, spectrum features and/or statistical features of the working condition time sequence based on the to-be-predicted object field data and/or the historical to-be-predicted object field data, and inputting the extracted shape features, spectrum features and/or statistical features to the random forest model and the gradient lifting tree model.
Optionally, the object to be predicted is a concrete piston.
According to the scheme of the invention, the cloud computing equipment can continuously train the fault prediction model according to the historical field data of the object to be predicted and send the fault prediction model to the fault early warning equipment, so that the fault early warning equipment can predict according to the fault prediction model and by combining the field data of the object to be predicted. According to the scheme, the accuracy of the fault prediction model can be ensured, and the fault early warning equipment can always perform fault prediction by using the accurate fault prediction model issued by the cloud computing equipment, so that the accuracy of fault prediction and early warning is ensured.
It should be noted that, although the following portions all describe the technical solution of the present invention by taking "object to be predicted" as a "concrete piston", the "object to be predicted" is not limited to a "concrete piston", and may include various types of objects requiring fault prediction, such as other components, software and hardware systems, and the like in the field of engineering machinery or in any other field, but when the "object to be predicted" is not a "concrete piston", the operating condition parameters acquired by the operating condition parameter acquisition module may need to be appropriately adjusted according to a specific "object to be predicted".
Drawings
The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the embodiments of the invention without limiting the embodiments of the invention. In the drawings:
fig. 1 is a schematic structural diagram of a concrete piston fault early warning system according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a concrete piston fault early warning system according to another embodiment of the present invention;
FIG. 3 is a schematic structural diagram of a fault prediction model provided by an embodiment of the invention;
FIG. 4 is a schematic diagram of a training process of a fault prediction model provided by an embodiment of the invention;
FIG. 5 is a schematic diagram of a fault prediction process of a fault prediction model provided by an embodiment of the invention;
FIG. 6 is a schematic structural diagram of a concrete piston failure early warning system according to yet another embodiment of the present invention;
and
fig. 7 is a schematic view of a control panel of a construction machine having a concrete piston.
Detailed Description
The following detailed description of embodiments of the invention refers to the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating the present invention, are given by way of illustration and explanation only, not limitation.
In the embodiments of the present invention, unless otherwise specified, the use of directional terms such as "upper, lower, top, and bottom" is generally used with respect to the orientation shown in the drawings or the positional relationship of the components with respect to each other in the vertical, or gravitational direction.
Fig. 1 is a schematic structural diagram of a concrete piston fault early warning system according to an embodiment of the present invention. As shown in fig. 1, a concrete piston fault early warning system provided in an embodiment of the present invention includes a concrete piston fault early warning device 100 and a cloud computing device 200 for early warning of a concrete piston fault. The concrete piston failure early warning apparatus 100 includes: the concrete piston field data acquisition module 110 is used for acquiring concrete piston field data of the concrete piston; and an edge calculation module 120, configured to receive the concrete piston field data and a fault prediction model from a cloud computing device, and input the concrete piston field data into the fault prediction model to determine whether a fault occurs or is about to occur in the concrete piston. The cloud computing device 200 includes: the fault prediction model training module 210 is used for training to obtain a fault prediction model based on historical concrete piston field data containing fault labels, and storing the fault prediction model into the fault prediction model library 220; and a failure prediction model library 220, configured to obtain and store the failure prediction model from the failure prediction model training module, and send the stored failure prediction model to the edge calculation module 120. Therefore, the cloud computing equipment 200 can continuously train a fault prediction model according to historical concrete piston field data and issue the fault prediction model to the concrete piston fault early warning equipment, so that the concrete piston fault early warning equipment can predict according to the fault prediction model and by combining the concrete piston field data. According to the scheme, the accuracy of the fault prediction model can be ensured, the concrete piston fault early warning equipment can always perform fault prediction by using the accurate fault prediction model issued by the cloud computing equipment, and the accuracy of concrete piston fault prediction and early warning is further ensured.
It should be noted that, the edge computing module may store the failure prediction model, and perform failure prediction using the failure prediction model within a period of time, where the failure prediction model may be updated by a failure prediction model issued by the cloud computing device 200; the edge computing module may not store the failure prediction model, and may perform failure prediction by using the failure prediction model issued by the cloud computing device 200 in real time.
The historical concrete piston field data containing the fault label can be provided by the concrete piston field data acquisition module 110, and can also be provided by other equipment. The concrete piston field data collection module 110 may include an operating condition parameter collection module for collecting one or more of the following parameters relating to the concrete piston: working condition parameters related to the service life of the concrete piston, such as working duration parameters, working volume parameters, engine rotating speed parameters, oil pump rotating speed parameters, pumping pressure parameters, hydraulic oil temperature parameters and the like. The operating condition parameter may be an operating condition parameter required by the fault prediction model to perform the prediction.
The concrete piston field data acquisition module can also comprise a position information acquisition module which is used for acquiring the position information of the concrete piston; and the concrete piston replacing switch is used for collecting the replacing record of the concrete piston. For the concrete piston replacing switch, it needs to be explained that, in the traditional method, the concrete piston actual replacing record of the pump equipment is generally recorded manually, the timeliness is difficult to guarantee, and the concrete piston actual replacing record is difficult to be matched with the working condition data strictly, but the concrete piston replacing switch can be arranged on the control panel, and is arranged on the control panel together with other control switches of the pump equipment, and through system logic control and business rules, an operator is required to turn on the concrete piston replacing switch before replacing the piston, so that the piston replacing record can be recorded automatically and uploaded to the cloud. The design of the concrete piston replacing switch on the control panel is shown in figure 7. Therefore, each time the concrete piston is in fault and needs to be replaced, the concrete piston replacement switch can be operated by an operator, so that a fault label can be generated, namely all working condition parameters before the fault label can be used as historical concrete piston field data, and a fault prediction model on the cloud computing equipment can be trained by using the historical concrete piston field data. For example, the concrete piston field data collecting module 110 collects and provides the working condition parameters to the cloud computing device in real time, and the cloud computing device may use the working condition parameters before the "fault label" in time as historical concrete piston field data after receiving the "fault label"
The concrete piston field data acquisition module may need to be further processed for inputting to a fault prediction model for fault prediction or training the fault prediction model as historical concrete piston field data. For this purpose, the edge calculation module may include an edge end signal processing module, which receives the concrete piston field data, and performs processing such as filtering, sampling, and the like on the concrete piston field data, for example, performs signal processing such as filtering, sampling, and the like on real-time working condition time series data with a high sampling frequency, so as to obtain a cleaner and properly sampled signal. On one hand, the obtained signals can be input into an edge end real-time fault early warning module in the edge computing module, the processed concrete piston field data is substituted into the fault prediction model by the edge end real-time fault early warning module to judge whether the concrete piston has a fault or is about to have a fault, and a concrete piston fault early warning signal is output under the condition of judging whether the concrete piston has the fault or is about to have the fault so as to control an alarm device to give an alarm, for example, an acousto-optic reporting device is used for reminding a user of timely replacement in the concrete piston using field in the forms of acousto-optic and the like; on the other hand, the fault prediction model can be uploaded to cloud computing equipment through the Internet of things or other types of networks suitable for information transmission, and the cloud computing equipment trains the fault prediction model as historical concrete piston field data.
The cloud computing device may train a corresponding fault prediction model for one or more of: (1) the cloud computing equipment can adopt historical concrete piston field data with the identifier to train a fault prediction model aiming at the concrete piston, and sends the fault prediction model of the concrete piston to the edge computing module, so that the edge computing module carries out fault prediction on the concrete piston by utilizing the fault prediction model, and therefore accurate fault model training aiming at a specific concrete piston and accurate fault prediction by utilizing the fault model are realized; (2) a certain model and/or batch of concrete pistons produced by a certain manufacturer, wherein for the concrete pistons, the training of the fault prediction model can be trained by using historical concrete piston field data from one or more concrete pistons meeting the model and/or batch, and the fault prediction model can be used for carrying out fault prediction on one or more concrete pistons meeting the model and/or batch; (3) for example, the operation of the concrete piston (for example, mechanical wear, difference of other mechanisms in the engineering machine cooperating with the concrete piston) by different engineering machines may have differences, and these differences may affect the service life of the concrete piston, and for this reason, historical concrete piston field data of one or more concrete pistons of a certain engineering machine may be used to train a fault prediction model for the engineering machine, and the fault prediction model is used to perform fault prediction on one or more concrete pistons of the engineering machine; (4) for this reason, historical concrete piston field data of one or more concrete pistons at a certain geographical position can be used for training a fault prediction model for the certain geographical position, and the fault prediction model is used for carrying out fault prediction on the one or more concrete pistons at the certain geographical position. For the above fault training model for a specific object (e.g., a specific concrete piston, one or more concrete pistons meeting the model and/or batch, a concrete piston of a specific engineering machine, a concrete piston at a specific geographic location), the field data collected by the concrete piston field data collection module may each contain a corresponding identifier to identify the specific object, e.g., the model and/or batch, the specific engineering machine, the specific geographic location, etc., so that the cloud computing device may train the corresponding fault prediction model according to the identifier and issue the trained fault prediction model to the concrete piston fault early warning device 100 with the corresponding identifier.
Fig. 2 is a schematic structural diagram of a concrete piston fault early warning system according to another embodiment of the present invention. Compared with the concrete piston fault early warning system shown in fig. 1, in the concrete piston fault early warning system shown in fig. 2, the cloud computing device 200 further includes a fault prediction module 230, configured to receive concrete piston field data (for example, piston field data directly from the concrete piston field data acquisition module 110, or data obtained after piston field data from the concrete piston field data acquisition module 110 is processed by an edge end signal processing module), calculate a concrete piston fault prediction result according to the fault prediction model stored in the fault prediction model library, and send the concrete piston fault prediction result to the service application module 300.
The business application module 300 includes one or more of the following: the failure early warning pushing module is used for sending out early warning according to the failure prediction result of the concrete piston; the service engineer scheduling module is used for sending out service engineer scheduling information according to the concrete piston fault prediction result; and the accessory scheduling module is used for sending accessory scheduling information according to the concrete piston fault prediction result. Therefore, based on the concrete piston fault prediction result, fault early warning can be pushed through a client APP, service engineer scheduling can be carried out through a CRM work order, and/or accessory scheduling can be carried out through an ECD system.
Fig. 3 is a schematic structural diagram of a fault prediction model provided in an embodiment of the present invention. As shown in fig. 3, the fault prediction model may include: the reliability model is used for outputting a first fault prediction result based on the concrete piston field data; the machine learning model is used for outputting a second fault prediction result based on the concrete piston field data; and a second ensemble learning model for generating a final failure prediction result based on the first and second failure prediction results, wherein the reliability model is trained based on concrete piston failure and maintenance data and/or concrete piston experimental data, and the machine learning model is trained based on the historical concrete piston field data. The failure prediction results can be presented in the form of failure probability, but of course, other forms can also be adopted.
The reliability model can comprise a concrete piston bathtub curve model and a concrete piston degradation track model, and the training process comprises the following steps: (1) synthesizing concrete piston replacement records obtained from fault and maintenance data and concrete piston replacement records judged by historical concrete piston field data to obtain accurate historical actual service life data of the concrete piston, and training a bathtub curve model (for example, related to the working time of the concrete piston, the pumping volume and the like); and (2) establishing a concrete piston degradation track model taking the stroke to position rate and the pumping square amount as degradation amount based on historical concrete piston experiment data. The concrete piston bathtub curve model and the concrete piston degradation track model can respectively and independently give the concrete piston fault probability. The process of using the trained reliability model for fault prediction is as follows: (1) inputting the accumulated working capacity such as the working time of the concrete piston, the pumping volume and the like into a bathtub curve model, and outputting the fault probability of the concrete piston; (2) and inputting the stroke arrival rate and the pumping square quantity of the concrete piston into the degraded track model, and outputting the fault probability of the concrete piston.
The machine learning model may be a stacked model with a first layer comprising a Random deep forest (Random roots) model, a gradient lifting tree (GBDT) model, a deep learning model, and a second layer being a first Boosting ensemble learning model. The training process of the machine learning model is as follows: (1) extracting characteristics of the whole-life working condition data of each piston which is changed historically according to the characteristic extraction algorithm, segmenting the characteristic data, forming a training sample by each segment of the characteristic data and the fault condition of the piston in a specific time in the future, respectively training a random deep forest model and a gradient lifting tree model based on the training sample, and searching for the tree depth, the tree number of the tree and other hyperparameters by adopting a multiple cross validation and heuristic search method; (2) segmenting the service life working condition data of each piston which is changed historically, forming a training sample by the working condition data of each segment and the fault condition of the piston in specific time in the future, training an improved deep learning model, wherein the model is based on CNN, a convolution kernel is rectangular, the width is the number of types of the working condition data, the length is the length of the working condition data segmentation, and Pooling (Pooling) operation is also performed by a rectangular window; (3) and inputting the characteristic data into a trained random deep forest model and a trained gradient lifting tree model respectively, outputting corresponding fault probabilities respectively, inputting the working condition data section into a trained deep learning model simultaneously, outputting corresponding fault probabilities, forming a training sample of the first boosting ensemble learning model by the three fault probabilities and corresponding real fault conditions, and training the first boosting ensemble learning model. The process of fault prediction using the trained machine learning model is as follows: (1) inputting the working condition data of the concrete piston field data into a characteristic extraction algorithm model, outputting the characteristics, inputting the characteristics into a random deep forest model and a gradient lifting tree model, and respectively outputting corresponding fault probabilities; (2) inputting the working condition data of the concrete piston field data into a deep learning model, and outputting corresponding fault probability; (3) and inputting the fault probabilities output by the random deep forest model, the gradient lifting tree model and the deep learning model into a first Boosting ensemble learning model, wherein the first Boosting ensemble learning model can output a fault probability as the total fault prediction result of the machine learning model.
The second ensemble learning model may generate a final failure prediction result based on the failure prediction results of the machine learning model and the reliability model, respectively. The second ensemble learning model training process is as follows: and constructing a training sample based on the concrete piston fault probabilities respectively and independently output by the machine learning model, the bathtub curve model and the concrete piston degradation track model and the corresponding actual concrete piston fault conditions, and obtaining the second ensemble learning model by adopting Boosting ensemble learning. The process of using the trained second ensemble learning model for fault prediction is as follows: the method can receive the concrete piston fault probabilities respectively and independently output by the machine learning model, the bathtub curve model and the concrete piston degradation track model, and output the final concrete piston fault probability as a final fault prediction result.
For the random forest model and the gradient lifting tree model, the concrete piston field data may not be directly processed, and therefore a feature extraction algorithm model is also provided, the feature extraction algorithm model can extract shape features, frequency spectrum features and/or statistical features of a working condition time sequence based on the concrete piston field data and/or the historical concrete piston field data, and the extracted shape features, frequency spectrum features and/or statistical features are input into the random forest model and the gradient lifting tree model. The characteristic extraction algorithm model can extract the characteristics of shape characteristics, frequency spectrum characteristics, statistical characteristics and the like of the working condition time sequence based on the concrete piston related working condition data. Specifically, the Dynamic Time Warping (DTW) algorithm and the wavelet transform can be used to extract the shape characteristics of the working condition time sequence, and the Fast Fourier Transform (FFT) is used in combination with the domain knowledge to extract the spectrum characteristics, and the statistical characteristics mainly include the maximum value, the minimum value, the average value, the variance, the skewness, the kurtosis and other statistical values. The feature extraction algorithm model does not need to be trained, and is mainly used for feature extraction of historical concrete piston field data in the fault prediction model training process and feature extraction of real-time concrete piston field data in the fault prediction process, so that the feature extraction algorithm model can be used by the random forest model and the gradient lifting tree model.
Fig. 4 is a schematic diagram of a training process of a fault prediction model according to an embodiment of the present invention. The training process and the fault prediction process of the fault prediction model provided by the embodiment of the invention have been described above with reference to fig. 3. The input data used in the training process of the fault prediction model will now be described in connection with fig. 4. For the bathtub curve model, the working condition data and the fault and maintenance data of the historical concrete piston field data are adopted; for the concrete piston degradation track model, concrete piston experimental data are adopted; for the machine learning model, the working condition data of historical concrete piston field data is adopted.
Fig. 5 is a schematic diagram of a failure prediction process of the failure prediction model provided by the embodiment of the invention. As shown in fig. 5, in the process of performing fault prediction by using the trained fault prediction model, all the real-time operating condition data of the concrete piston field data are adopted by the model, and the real-time operating condition data can be directly obtained from the piston field data acquired by the concrete piston field data acquisition module 110 in real time or obtained from the piston field data of the concrete piston field data acquisition module 110 after being processed by the edge end signal processing module.
Although fig. 3 to 5 show a specific failure prediction model structure, the present invention is not limited to this, and other failure prediction model structures capable of predicting a concrete piston failure may be applied.
Fig. 6 is a schematic structural diagram of a concrete piston fault early warning system according to still another embodiment of the present invention. As shown in fig. 6, most of the components or features shown therein have been described above in the description of fig. 1-5. The main modules are further explained below.
The concrete piston field data acquisition module can acquire working condition parameters related to the service life of a concrete piston, such as working duration, working volume, engine rotating speed, oil pump rotating speed, pumping pressure, hydraulic oil temperature and the like through a pumping equipment end bus, acquire the position of equipment through a position sensor, and automatically acquire a piston replacing record through a quick-change piston switch. The concrete piston field data acquisition module inputs the acquired data into the edge calculation module.
The edge end signal processing module in the edge computing module can receive the data, and carries out signal processing such as filtering and sampling on real-time working condition time sequence data with high sampling frequency, so that the obtained clean and appropriate signals are input into the edge end real-time fault early warning module on one hand, and are uploaded to a cloud (namely, cloud computing equipment) through an internet of things on the other hand. The real-time fault early warning module at the edge end outputs a concrete piston fault early warning signal based on a feature extraction algorithm and a fault prediction model issued by the cloud, and reminds a user of timely replacement in the form of sound and light on site.
The service system data acquisition module and the experiment table data acquisition module can respectively acquire data related to concrete piston faults and maintenance, working condition parameters generated in the experiment process of the pumping system, the service life of the concrete piston and the like, and store the data in the cloud data warehouse. Compared with pumping equipment on the market actually, the pumping system experiment table is provided with more comprehensive sensors, has higher acquisition frequency, can collect the full-life data of the concrete piston more comprehensively and accurately and is used for establishing a concrete piston degradation track model.
The cloud data warehouse can receive and store real-time working condition data transmitted through the Internet of things after signal processing such as filtering and the like, business data such as fault and maintenance information and the like acquired by the CRM system data acquisition module, and experimental data acquired by the pumping system experiment table data acquisition module.
The fault prediction model training module can be used for training a concrete piston fault prediction model integrating a tree machine learning model, a deep learning model and a reliability model by integrating working condition data, fault and maintenance data and laboratory bench data. And storing the trained model into a cloud fault prediction model library.
The fault prediction model library can store the trained fault prediction models, extract relevant models from the fault prediction models according to needs and send the relevant models to the edge end, and can also send the models to the cloud fault prediction module.
The cloud fault prediction module can be configured according to business requirements, and when the pumping equipment uploads working condition data normally, the cloud fault prediction module gives a concrete piston fault prediction result in real time based on a trained model and sends the fault prediction result to the business application module.
The service application module comprises a fault early warning pushing logic module, a service engineer dispatching logic module and an accessory dispatching logic module, sets related service rules based on a concrete piston fault prediction result, and is respectively used for pushing fault early warning through a client APP, dispatching a service engineer through a CRM (customer relationship management) and dispatching corresponding accessories through an ECC (error correction code).
According to the technical scheme, firstly, the concrete piston fault prediction and early warning system is additionally provided with sensors such as a working condition collection module, a position sensor, a concrete piston quick-change piston switch and the like, can automatically collect information such as a working position, a working condition, the actual service life of the concrete piston, fault records and the like, and provides a big data base for realizing the concrete piston fault prediction and early warning. Secondly, the method fully utilizes the Internet of things and the like to collect big data related to the service life of the concrete piston, such as working condition data, actual service life data, geographic environment data and the like of the concrete piston in all actual works, carries out modeling and service life prediction based on the big data of the concrete piston, fully considers the actual situation and has higher prediction accuracy. And thirdly, the concrete piston fault prediction model combining machine learning and reliability models can accurately predict the concrete piston fault. The edge end of the invention is also provided with a fault real-time early warning module which can send out early warning when a fault is close to or hidden abnormal conditions occur by utilizing real-time working condition information. The invention has novel application mode, based on the concrete piston fault early warning result, the fault early warning is pushed by the client APP, the service engineer scheduling is carried out by the CRM work order, and the accessory scheduling is carried out by the storage system, so that the condition of the concrete piston can be timely adopted by various countermeasures through the service application module, and the concrete piston can be always in a normal working state.
While the embodiments of the present invention have been described in detail with reference to the accompanying drawings, the embodiments of the present invention are not limited to the details of the above embodiments, and various simple modifications can be made to the technical solution of the embodiments of the present invention within the technical idea of the embodiments of the present invention, and the simple modifications are within the scope of the embodiments of the present invention.
It should be noted that the various features described in the above embodiments may be combined in any suitable manner without departing from the scope of the invention. In order to avoid unnecessary repetition, the embodiments of the present invention will not be described separately for the various possible combinations.
Those skilled in the art will appreciate that all or part of the steps in the method for implementing the above embodiments may be implemented by a program, which is stored in a storage medium and includes several instructions to enable a single chip, a chip, or a processor (processor) to execute all or part of the steps in the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In addition, any combination of the various embodiments of the present invention is also possible, and the same should be considered as disclosed in the embodiments of the present invention as long as it does not depart from the spirit of the embodiments of the present invention.
Claims (19)
1. A cloud computing device for fault early warning, the device comprising:
the fault prediction model training module is used for training to obtain a fault prediction model based on historical field data of an object to be predicted, which contains a fault label, and storing the fault prediction model into a fault prediction model library; and
the fault prediction model base is used for acquiring and storing the fault prediction model from the fault prediction model training module and sending the stored fault prediction model to the edge calculation module;
the fault prediction model includes:
the reliability model is used for outputting a first fault prediction result based on the field data of the object to be predicted;
the machine learning model is used for outputting a second fault prediction result based on the field data of the object to be predicted; and
a second ensemble learning model for generating a final fault prediction result based on the first and second fault prediction results,
wherein the reliability model is trained based on fault and maintenance data of the object to be predicted and/or experimental data of the object to be predicted, and the machine learning model is trained based on field data of the historical object to be predicted.
2. The cloud computing device of claim 1, wherein the device further comprises:
and the fault prediction module is used for receiving field data of the object to be predicted, calculating a fault prediction result of the object to be predicted according to the fault prediction model stored in the fault prediction model base, and sending the fault prediction result of the object to be predicted to the service application module.
3. The cloud computing device of claim 2, wherein the business application module comprises one or more of:
the fault early warning pushing module is used for sending out early warning according to the fault prediction result of the object to be predicted;
the service engineer scheduling module is used for sending out service engineer scheduling information according to the fault prediction result of the object to be predicted;
and the accessory scheduling module is used for sending accessory scheduling information according to the fault prediction result of the object to be predicted.
4. The cloud computing device of claim 1, wherein the reliability model comprises one or more of:
a bathtub curve model trained based on the historical subject field data to be predicted and the subject fault and maintenance data to be predicted; and
a degenerate orbit model trained based on experimental data of the subject to be predicted.
5. The cloud computing device of claim 1, wherein the machine learning model comprises:
the random forest model, the gradient lifting tree model and the deep learning model are used for outputting corresponding fault prediction results according to the field data of the object to be predicted respectively; and
and the first ensemble learning model is used for generating a final fault prediction result as a second fault prediction result of the machine learning model according to the corresponding fault prediction result.
6. The cloud computing device of claim 5, wherein the fault prediction model further comprises:
and the feature extraction algorithm model is used for extracting shape features, frequency spectrum features and/or statistical features of the working condition time sequence based on the to-be-predicted object field data and/or the historical to-be-predicted object field data, and inputting the extracted shape features, frequency spectrum features and/or statistical features into the random forest model and the gradient lifting tree model.
7. The cloud computing device of claim 1, wherein the object to be predicted is a concrete piston.
8. A fault warning system, comprising:
fault early warning device, comprising: the field data acquisition module is used for acquiring field data of an object to be predicted; the edge computing module is used for receiving the field data and a fault prediction model from cloud computing equipment, and inputting the field data into the fault prediction model so as to judge whether the object to be predicted has a fault or is about to have a fault; and
the cloud computing device of any of claims 1-7.
9. The fault warning system of claim 8, wherein the field data acquisition module comprises:
a condition parameter acquisition module for acquiring one or more of the following parameters relating to the object to be predicted: the system comprises a working duration parameter, a working amount parameter, an engine rotating speed parameter, an oil pump rotating speed parameter, a pumping pressure parameter and a hydraulic oil temperature parameter.
10. The fault early warning system of claim 8 or 9, wherein the field data acquisition module comprises one or more of:
the position information acquisition module is used for acquiring the position information of the object to be predicted; and
and the replacement switch is used for acquiring a replacement record of the object to be predicted.
11. The fault early warning system of claim 8, wherein the edge calculation module comprises: the system comprises an edge end signal processing module and an edge end real-time fault early warning module;
the edge terminal signal processing module is used for receiving the field data, processing the field data and sending the processed field data to the edge terminal real-time fault early warning module and the cloud computing equipment;
and the edge end real-time fault early warning module is used for substituting the processed field data into the fault prediction model to judge whether the object to be predicted has a fault or is about to have a fault, and outputting a fault early warning signal under the condition of judging that the object to be predicted has the fault or is about to have the fault.
12. The fault early warning system according to claim 8, wherein the fault early warning apparatus further comprises:
and the alarm device is used for giving an alarm under the condition that the object to be predicted has or is about to have a fault.
13. A method for fault early warning performed within a cloud computing device, the method comprising:
training to obtain a fault prediction model based on historical field data of the object to be predicted, which contains a fault label; and
sending the fault prediction model to fault early warning equipment which is positioned near the object to be predicted;
the fault prediction model includes:
the reliability model is used for outputting a first fault prediction result based on the field data of the object to be predicted;
the machine learning model is used for outputting a second fault prediction result based on the field data of the object to be predicted; and
a second ensemble learning model for generating a final fault prediction result based on the first and second fault prediction results,
wherein the reliability model is trained based on fault and maintenance data of the object to be predicted and/or experimental data of the object to be predicted, and the machine learning model is trained based on field data of the historical object to be predicted.
14. The method of claim 13, further comprising:
receiving field data of an object to be predicted; and
and inputting the field data of the object to be predicted into the fault prediction model to obtain a fault prediction result of the object to be predicted, and sending the fault prediction result of the object to be predicted to a service application module.
15. The method of claim 14, wherein the business application module comprises one or more of:
the fault early warning pushing module is used for sending out early warning according to the fault prediction result of the object to be predicted;
the service engineer scheduling module is used for sending out service engineer scheduling information according to the fault prediction result of the object to be predicted;
and the accessory scheduling module is used for sending accessory scheduling information according to the fault prediction result of the object to be predicted.
16. The method of claim 13, wherein the reliability model comprises one or more of:
a bathtub curve model trained based on the historical subject field data to be predicted and the subject fault and maintenance data to be predicted; and
a degenerate orbit model trained based on experimental data of the subject to be predicted.
17. The method of claim 13, wherein the machine learning model comprises:
the random forest model, the gradient lifting tree model and the deep learning model are used for outputting corresponding fault prediction results according to the field data of the object to be predicted respectively; and
and the first ensemble learning model is used for generating a final fault prediction result as a second fault prediction result of the machine learning model according to the corresponding fault prediction result.
18. The method of claim 17, further comprising:
and extracting shape features, spectrum features and/or statistical features of the working condition time sequence based on the to-be-predicted object field data and/or the historical to-be-predicted object field data, and inputting the extracted shape features, spectrum features and/or statistical features to the random forest model and the gradient lifting tree model.
19. The method according to claim 13, wherein the object to be predicted is a concrete piston.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910860038.9A CN112486136B (en) | 2019-09-11 | 2019-09-11 | Fault early warning system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910860038.9A CN112486136B (en) | 2019-09-11 | 2019-09-11 | Fault early warning system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112486136A CN112486136A (en) | 2021-03-12 |
| CN112486136B true CN112486136B (en) | 2022-03-15 |
Family
ID=74920553
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910860038.9A Active CN112486136B (en) | 2019-09-11 | 2019-09-11 | Fault early warning system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112486136B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114987773B (en) * | 2022-05-20 | 2024-06-14 | 成都飞机工业（集团）有限责任公司 | Method, device, equipment and medium for identifying performance abnormality of airplane hydraulic system |
| CN115001943B (en) * | 2022-05-27 | 2024-03-22 | 深圳小湃科技有限公司 | Equipment fault identification method, equipment and storage medium based on big data |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009003561A (en) * | 2007-06-19 | 2009-01-08 | Fuji Xerox Co Ltd | Fault prediction diagnostic system and fault prediction diagnostic system using the same |
| CN107038453A (en) * | 2017-03-06 | 2017-08-11 | 辽宁大唐国际新能源有限公司 | The fault early warning system and determination methods of a kind of Wind turbines |
| CN108092794A (en) * | 2017-11-08 | 2018-05-29 | 北京百悟科技有限公司 | Network failure processing method and device |
| CN108460144A (en) * | 2018-03-14 | 2018-08-28 | 西安华光信息技术有限责任公司 | A kind of coal equipment fault early-warning system and method based on machine learning |
| CN108591104A (en) * | 2018-04-18 | 2018-09-28 | 广东寰球智能科技有限公司 | A kind of Research on Fan Fault Forecasting based on cloud platform and health management system arranged, method |
| CN108829933A (en) * | 2018-05-22 | 2018-11-16 | 北京天泽智云科技有限公司 | A kind of method of the predictive maintenance and health control of semiconductor manufacturing facility |
| CN109543210A (en) * | 2018-09-28 | 2019-03-29 | 国电电力宁夏新能源开发有限公司 | A kind of Wind turbines failure prediction system based on machine learning algorithm platform |
| CN109933004A (en) * | 2019-03-27 | 2019-06-25 | 苏芯物联技术(南京)有限公司 | The machine failure diagnosis and prediction method and system cooperateed with based on edge calculations and cloud |
| CN109947088A (en) * | 2019-04-17 | 2019-06-28 | 北京天泽智云科技有限公司 | Equipment fault early-warning system based on model lifecycle management |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107403480A (en) * | 2017-06-15 | 2017-11-28 | 北汽福田汽车股份有限公司 | A kind of vehicle trouble method for early warning, system and vehicle |
-
2019
- 2019-09-11 CN CN201910860038.9A patent/CN112486136B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009003561A (en) * | 2007-06-19 | 2009-01-08 | Fuji Xerox Co Ltd | Fault prediction diagnostic system and fault prediction diagnostic system using the same |
| CN107038453A (en) * | 2017-03-06 | 2017-08-11 | 辽宁大唐国际新能源有限公司 | The fault early warning system and determination methods of a kind of Wind turbines |
| CN108092794A (en) * | 2017-11-08 | 2018-05-29 | 北京百悟科技有限公司 | Network failure processing method and device |
| CN108460144A (en) * | 2018-03-14 | 2018-08-28 | 西安华光信息技术有限责任公司 | A kind of coal equipment fault early-warning system and method based on machine learning |
| CN108591104A (en) * | 2018-04-18 | 2018-09-28 | 广东寰球智能科技有限公司 | A kind of Research on Fan Fault Forecasting based on cloud platform and health management system arranged, method |
| CN108829933A (en) * | 2018-05-22 | 2018-11-16 | 北京天泽智云科技有限公司 | A kind of method of the predictive maintenance and health control of semiconductor manufacturing facility |
| CN109543210A (en) * | 2018-09-28 | 2019-03-29 | 国电电力宁夏新能源开发有限公司 | A kind of Wind turbines failure prediction system based on machine learning algorithm platform |
| CN109933004A (en) * | 2019-03-27 | 2019-06-25 | 苏芯物联技术(南京)有限公司 | The machine failure diagnosis and prediction method and system cooperateed with based on edge calculations and cloud |
| CN109947088A (en) * | 2019-04-17 | 2019-06-28 | 北京天泽智云科技有限公司 | Equipment fault early-warning system based on model lifecycle management |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112486136A (en) | 2021-03-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115578015B (en) | Sewage treatment whole process supervision method, system and storage medium based on Internet of things | |
| CN112393931B (en) | Detection method, detection device, electronic equipment and computer readable medium | |
| CN112486136B (en) | Fault early warning system and method | |
| CN105593864B (en) | Analytical device degradation for maintenance device | |
| CN109374631B (en) | Tunnel state evaluation method | |
| CN111046582A (en) | Nuclear power station diesel generating set coil temperature early warning method and system | |
| CN111817880A (en) | Oil and gas field production equipment health management system and implementation method | |
| CN109263271A (en) | A kind of printing equipment determination method based on big data | |
| CN112396250B (en) | Diesel engine fault prediction method, device, equipment and storage medium | |
| CN112862233A (en) | Fault relevance analysis system and method based on Internet of vehicles data | |
| CN113847950A (en) | Intelligent ship equipment state monitoring system based on cloud computing and information interaction method | |
| CN115035475A (en) | Water supply pipeline real-time monitoring method and device, electronic equipment and medium | |
| CN114326593B (en) | Cutter life prediction system and method | |
| CN106951360B (en) | Data statistical integrity calculation method and system | |
| CN116822115A (en) | Environment management method and system for intelligent park based on digital twin technology | |
| CN111176226A (en) | Automatic analysis method for alarm threshold of equipment characteristic parameter based on operation condition | |
| CN114462820A (en) | Bearing state monitoring and health management system performance testing and optimizing method and system | |
| KR101884908B1 (en) | Big Data Analytics Based Reliability Prediction Apparatus | |
| CN108613820A (en) | A kind of online allophone monitoring algorithm for GIS bulk mechanicals defect diagonsis and positioning | |
| CN112524077A (en) | Method, device and system for detecting fan fault | |
| CN117114420A (en) | Image recognition-based industrial and trade safety accident risk management and control system and method | |
| CN110878692B (en) | Fault alarm method and device | |
| CN110991779A (en) | Anomaly detection method and device for oil pumping well | |
| CN111058944B (en) | Method and system for detecting temperature abnormity of cylinder of diesel generating set | |
| CN112884170A (en) | Predictive intelligent operation and maintenance system and method for comprehensive pipe gallery |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |