Mechanical equipment image generation method
Through the multi-source data fusion of mechanical equipment and neural network model processing, detailed mechanical equipment portraits are generated, which solves the problem of insufficient research on mechanical equipment portraits in the oil and gas industry, and realizes accurate monitoring of equipment status and fault detection.
Patent Information
- Application Number
- CN202210132467.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-02-14
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2042-02-14
AI Technical Summary
In the prior art, there is little research on mechanical equipment portraits, especially in large-scale mechanical service equipment in the oil and gas industry. It is difficult to create a complete mechanical equipment portrait to understand the dynamics and operating conditions of the equipment.
By obtaining multi-source data information of mechanical equipment, data fusion and normalization processing are carried out, target attribute data and target state data are generated, attribute tags and state tags are constructed using neural network models, and device portraits are generated by combining clustering algorithms, and visualization is performed by visualization Method presentation.
It realizes comprehensive feature extraction and status monitoring of mechanical equipment, helps to timely and accurately grasp the equipment operation status, and improves fault detection capabilities and production process management.
Smart Images

Figure CN114595943B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of intelligent management of mechanical equipment, and more specifically, to a method for generating a mechanical equipment portrait. Background Art
[0002] Portrait technology is widely used in many industries such as the Internet, mainly using portraits, which can be used to mine user characteristics, grasp user preferences, and thoroughly understand and master the portrayed object. Similarly, large-scale mechanical equipment can also be portrayed like people, and can also be extracted and fully described through portrait technology. The factory, production and maintenance processes of large-scale mechanical equipment generate a large amount of monitoring and monitoring data. These data can directly reflect the significant characteristics of the equipment, but it is impossible to intuitively and deeply mine its characteristics through a large amount of data. Therefore, constructing a portrait of mechanical equipment can accurately mine and extract various aspects of mechanical equipment and present them comprehensively, so that we can timely understand and grasp the status of the equipment and the current status of operation. Summary of the invention
[0003] Technical issues to be solved
[0004] So far, there are relatively few studies on mechanical equipment portraits, and even fewer on large mechanical service equipment in the oil and gas industry. Therefore, creating a complete mechanical equipment portrait is an urgent problem to be solved. With the continuous accumulation and widespread application of equipment data, various large equipment portraits will be created to help timely and accurately grasp the dynamics and operating conditions of equipment during production operations.
[0005] Technical Solution
[0006] In order to achieve the above-mentioned objectives, the present application provides a method for generating a mechanical equipment portrait, which method includes: acquiring all data information related to the mechanical equipment; performing data fusion on the data information related to the mechanical equipment to obtain a multi-source data information set; acquiring target attribute data and target state data of the mechanical equipment based on the multi-source data information set; generating an attribute label of the mechanical equipment based on the acquired target attribute data; inputting the current target state data of the mechanical equipment into a trained neural network model to obtain a current state label of the mechanical equipment, wherein the current state label includes a state label value of the mechanical equipment; generating a device portrait of the mechanical equipment through the attribute label value of the acquired attribute label and the state label value of the acquired current state label.
[0007] Furthermore, the method for generating a mechanical equipment portrait also includes: initializing a neural network model based on the target state data of the mechanical equipment and the number of target state label values of the target state data; generating multiple statistical indicators based on the target state data; and inputting a data matrix including multiple statistical indicators and state classification label data of the target state data into the initialized neural network model to train the neural network model.
[0008] Furthermore, target attribute data and target state data of mechanical equipment are obtained based on the multi-source data information set, including: obtaining initial attribute data and initial state data based on the multi-source data information set; obtaining target attribute data based on the attention and / or importance of the initial attribute data; and normalizing the initial state data to obtain target state data.
[0009] Furthermore, an equipment portrait of the mechanical equipment is generated through the obtained attribute tag values of the attribute tags and the obtained state tag values of the current state tags, including: constructing a set of all tag values and a set of equipment based on the attribute tag values and the current state tag values of the mechanical equipment; constructing a co-occurrence matrix based on the tag value set and the equipment set; obtaining label clustering clusters of the attribute tags and the current state tags through a clustering algorithm based on the co-occurrence matrix; and obtaining the current key label values of the equipment based on the label clustering clusters to generate an equipment portrait of the mechanical equipment.
[0010] Furthermore, based on the label clustering cluster, the current key label value of the equipment is obtained to generate an equipment portrait of the mechanical equipment, including: comparing all label values of each device with the label values contained in the obtained label clustering cluster; and obtaining the first label clustering cluster with the largest number of label value categories in the label clustering cluster covered by the label values contained in each device, and setting the corresponding label value in the first label clustering cluster as the current key label value of the equipment.
[0011] Furthermore, the method for generating a mechanical equipment portrait also includes: visualizing the generated equipment portrait into a portrait word cloud diagram.
[0012] Furthermore, the method for generating a mechanical equipment portrait also includes: comparing the state label value of the current target state data of the mechanical equipment with the set value to obtain the current state label of the mechanical equipment.
[0013] Furthermore, the state tag value of the current target state data of the mechanical equipment is compared with the set value to obtain the current state tag of the mechanical equipment.
[0014] Furthermore, data fusion is performed on the data information related to the mechanical equipment to obtain a multi-source data information set, including: setting different unique identifiers for each device; obtaining a first library table with a unique identifier for each device and a second library table associated with the first library table; and associating the first library table and the second library table with the unique identifier.
[0015] Furthermore, the neural network model includes a back-propagation neural network model.
[0016] Furthermore, the statistical indicators include: mean, standard deviation, root amplitude, effective value, peak value, skewness, kurtosis, maximum value, margin value, waveform factor, and pulse index.
[0017] Furthermore, the status tag values include: normal, faulty, high, medium, and low.
[0018] According to another aspect of the present application, there is provided a computer device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and the processor implements the steps of the above-mentioned mechanical equipment portrait generation method when executing the computer program.
[0019] According to another aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned method for generating a mechanical device portrait are implemented.
[0020] Beneficial Effects
[0021] According to this application, by applying customer portrait technology to plunger pump equipment, a relatively complete plunger pump equipment portrait labeling system is constructed.
[0022] Moreover, the labeling system constructed in the present application includes not only static attributes and instantaneous parameters, but also parameter status evaluation and maintenance status information, and provides a method for constructing current status labels of equipment based on neural networks, so as to facilitate obtaining labels such as the current health status of the equipment, and finally obtain a portrait of the equipment, which is helpful to realize plunger pump fault detection, and can timely and accurately grasp the current operating status of the equipment, so as to facilitate control of the entire production process. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] The drawings constituting a part of the present application are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:
[0024] Figure 1 A flowchart of a method for generating a mechanical device image according to a preferred embodiment of the present application is shown;
[0025] Figure 2A schematic diagram of fusion processing of multi-source data related to mechanical equipment according to an embodiment of the present application is shown;
[0026] Figure 3 Schematically illustrates an example labeling system of sample data constructed according to the present application;
[0027] Figure 4 A schematic diagram showing a method of visualizing a portrait of a mechanical device as a portrait word cloud diagram according to a preferred embodiment of the present application is shown;
[0028] Figure 5 A schematic flow chart of a method for generating a mechanical equipment image according to a specific embodiment of the present application is shown. DETAILED DESCRIPTION
[0029] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0030] According to an embodiment of the present application, Figure 1 As shown, a method for generating a mechanical equipment portrait is provided, the method comprising: acquiring all data information related to the mechanical equipment (S101); performing data fusion on the data information related to the mechanical equipment to obtain a multi-source data information set (S102); acquiring target attribute data and target state data of the mechanical equipment based on the multi-source data information set (S103); generating an attribute label of the mechanical equipment based on the acquired target attribute data (S104); inputting the current target state data of the mechanical equipment into a trained neural network model to obtain a current state label of the mechanical equipment, wherein the current state label includes a state label value of the mechanical equipment (S105); generating a device portrait of the mechanical equipment through the acquired attribute label value of the attribute label and the acquired state label value of the current state label (S106).
[0031] According to the present application, a relatively complete image labeling system can be constructed for mechanical equipment, such as plunger pump equipment, which helps to timely and accurately grasp the dynamics and operating conditions of the equipment during the production process.
[0032] The specific process of the mechanical equipment image generation method is explained below through a detailed description.
[0033] S101: Obtain all data information related to mechanical equipment.
[0034] Mechanical equipment is widely used. For example, the plunger pump in the oil and gas equipment service industry is one of the large-scale mechanical equipment mainly used in the oil and gas industry. It can be used for high-intensity operations such as cementing, acidizing, and fracturing. Mechanical equipment generates a large amount of production data and operation and maintenance data throughout its life cycle, from production to field operation to maintenance and scrapping.
[0035] In addition, the data involved during the operation of mechanical equipment is diverse and scattered. Different types of data are scattered in various equipment management systems and need to be sorted out, mainly including (but not limited to the following data): equipment archive attribute data, equipment production operation data, equipment after-sales maintenance data, and various sensor data of the equipment real-time monitoring system. By sorting out the underlying database tables and library table fields of the equipment-related business system containing the above data, and using this as the data source for mechanical equipment portraits, all data information related to each mechanical equipment can be obtained.
[0036] Next, a fracturing truck-mounted plunger pump at a fracturing well site is used as an example to illustrate how to obtain data information related to the equipment.
[0037] The multi-source data related to the plunger pump and the data source systems involved are sorted out to obtain the following database table:
[0038]
[0039] The above four types of data belong to the main data range involved in the plunger pump. Therefore, the construction of the plunger pump portrait index system needs to be based on the above data sources.
[0040] S102: Perform data fusion on data information related to mechanical equipment to obtain a multi-source data information set.
[0041] The data of mechanical equipment-related business systems are scattered in various business systems from multiple sources. The data are relatively isolated and need to be fused to realize the serial and parallel operation of multi-source data information.
[0042] According to a preferred embodiment of the present application, the above-obtained data information is fused by the following method.
[0043] Preferably, a unique identifier is determined as a unique identity symbol for each device. The unique identifier is different for different devices and can be (but is not limited to) the "device number" of the device when it leaves the factory.
[0044] Sort out all the data tables related to the device in the data information obtained as described above. The data information here should be all the database tables in the underlying structured database of each business system. According to the library table structure (that is, the fields contained in the library table), the first library table with the unique identification of the device and the second library table that can be associated with the first library table through other fields are summarized.
[0045] Specifically, these two types of database tables are sorted out, for example, how many first database tables with unique device identifiers there are for each data source information and how many second database tables can be associated with the first database tables through other fields, and what each database table is and what fields it contains, etc.
[0046] The first library table has a field for uniquely identifying the device, and the second library table does not have a field for uniquely identifying the device. However, other fields in the second library table can be associated with the first library table.
[0047] The first library table and the second library table mentioned above can be provided in multiple numbers, and the "first library table" and the "second library table" are used to distinguish whether the library table has a unique device identifier.
[0048] Open access permissions between the business system databases involved above (such as the first library table and the second library table, etc.), use the device unique identifier as the association field, and create association relationships for the system database tables sorted out above through database statements, such as structured query language (SQL) statements, so as to achieve effective fusion of multi-source device data indexed by the device unique identifier.
[0049] Take the fracturing truck-mounted plunger pump at the fracturing well site as an example to illustrate the process of data fusion of mechanical equipment related data information. Figure 2 A schematic diagram of fusion processing of multi-source data related to mechanical equipment according to an embodiment of the present application is shown.
[0050] like Figure 2 As shown in the figure, the "plunger pump equipment number" is used as the unique identifier. Examples of systems and library tables related to the "plunger pump equipment number" are: fracturing truck operation monitoring system, plunger pump intelligent detection system, fracturing truck after-sales maintenance management system, and equipment production archive management library. For example, the equipment operation parameter table in the fracturing truck after-sales maintenance management system has a total of 23 fields; the equipment vibration signal record table in the plunger pump intelligent detection system has a total of 9 fields; the equipment maintenance record table in the fracturing truck after-sales maintenance management system has a total of 12 fields, and the equipment archive information table in the equipment production archive management library has a total of 25 fields. Based on the unique identifier "plunger pump equipment number", the above four tables are associated, and the duplicate fields are removed. After data fusion, the result table contains a total of 61 fields.
[0051] Of course, you can filter the required fields according to different label requirements and put them into the associated result table, and you can delete and query the fields contained in the above data fusion table according to the label calculation needs.
[0052] Through the operations described above, a fused device multi-source data information set indexed by the device unique identifier is obtained.
[0053] In this application, device tags are divided into two categories, namely, device status hot tags and device attribute cold tags. Among them, device attribute cold tags refer to attribute tags whose tag values are inherent attributes of the device or generally do not change once generated; device status hot tags refer to tag classes whose tag values change periodically or randomly with the device status. All tags of mechanical equipment can be classified into the above two categories.
[0054] For example, Figure 3 As shown, it schematically shows an example labeling system of sample data constructed according to the present application.
[0055] For example, the equipment attribute cold label may include: equipment number, equipment type, equipment production date, equipment commissioning date, equipment rated power, equipment stroke, equipment plunger specifications, equipment maximum pressure, equipment maximum displacement, equipment overall dimensions, equipment weight, etc.
[0056] The equipment status hot tags may include: equipment health status tag (tag value is normal or faulty), equipment current pressure status tag (tag values include: high, medium, low), equipment failure risk level, etc.
[0057] S103: Acquire target attribute data and target state data of the mechanical equipment based on the multi-source data information set obtained above.
[0058] The initial attribute data of the device is obtained from the multi-source data information set obtained above, that is, all relevant attribute fields of the device in the multi-source data information set obtained above are retrieved as the initial attribute data using the device unique identifier as an index.
[0059] Taking the fracturing truck-mounted plunger pump at a fracturing well site as an example, the initial attribute data may include (but is not limited to): equipment number, equipment type, equipment production date, equipment commissioning date, equipment rated power, equipment stroke, equipment plunger specifications, equipment maximum pressure, equipment maximum displacement, equipment dimensions, equipment weight...
[0060] The initial attribute data is standardized to obtain the target attribute data. According to a preferred embodiment of the present application, the standardization method can sort the initial attribute data by attention and importance based on the experience of experts in the field, and perform a comprehensive score between 0 and 1; and remove attribute data with a comprehensive score below 0.8, and use all attribute data with a comprehensive expert score greater than 0.8 as target attribute data.
[0061] Taking the fracturing truck-mounted plunger pump at a fracturing well site as an example, the target attribute data, that is, the initial attribute data with an expert score greater than 0.8 points based on expert experience, may include (but not limited to): equipment number, equipment type, equipment production date, equipment commissioning date, equipment rated power, equipment stroke, and equipment plunger specifications.
[0062] The device initial state data is obtained from the multi-source data information set obtained above, that is, using the device unique identifier as an index, all field data related to the device state of the device in the multi-source data information set obtained above are retrieved as the initial state data.
[0063] Taking the fracturing truck-mounted plunger pump at a fracturing well site as an example, the plunger pump status can be (but is not limited to) the health status (normal or faulty) of the hydraulic end pump valve of the plunger pump, and the status data corresponding to this status can include (but is not limited to): historical hydraulic end single cylinder vibration amplitude data, historical hydraulic end single cylinder temperature value data, and historical hydraulic end single cylinder pressure value data.
[0064] Furthermore, the initial state data is subjected to normalization processing. According to a preferred embodiment of the present application, the normalization processing includes performing MIN-MAX normalization processing on the initial state data, thereby obtaining the target state data.
[0065] S104: Generate an attribute label of the equipment based on the acquired target attribute data of the mechanical equipment. Specifically, for example, directly name and construct based on the field name of the target attribute data of the equipment; and store the attribute label data in a standard structured label database.
[0066] Taking the fracturing truck-mounted plunger pump at a fracturing well site as an example, the attribute labels may include (but are not limited to): equipment number label, equipment type label, equipment production date label, equipment commissioning date label, equipment rated power label, equipment stroke label, and equipment plunger specification label.
[0067] S105: Input the current target state data of the mechanical equipment into the trained neural network model to obtain the current state label of the mechanical equipment.
[0068] According to a preferred embodiment of the present application, a back propagation (BP) neural network is used. Of course, other neural networks can also be used.
[0069] According to the historical target state data of the device and the number of label values of the device state, initialize the initial values of the neural network, such as the number of network layers, the number of neurons in each layer, the nonlinear activation function of each layer, the connection weights, the number of single training samples, the number of sample training cycles, the learning rate, the loss function, and the optimizer.
[0070] According to a preferred embodiment of the present application, the initial network parameters of the neural network are set as follows:
[0071] Parameter name Parameter Value Network Layers 10 Number of neurons in each layer 20-50, the last layer is 2 Non-linear activation function Sigmod Connection weight Fixed random number Number of single training samples 100 Number of sample training cycles 800 Learning Rate 0.001 Loss Function MSE Optimizer Adam
[0072] The BP neural network is trained using training data.
[0073] The training process is explained using the fracturing truck-mounted plunger pump at a fracturing well site as an example.
[0074] The device status label to be classified can be (but is not limited to) the health status label of the hydraulic end pump valve of the device, and the label value is normal or faulty. The health status data of the hydraulic end pump valve of the device includes three indicators: historical hydraulic end single cylinder vibration amplitude, historical hydraulic end single cylinder temperature value, and historical hydraulic end single cylinder pressure value.
[0075] The example sample data is the historical operating data of the target equipment (i.e., the hydraulic end pump and valve of the equipment) in both normal and faulty states for 10 hours respectively. The original data is divided into 36,000 segments at intervals of 1 second, and the mean, standard deviation, root mean square amplitude, effective value, peak value, skewness, kurtosis, maximum value, margin value, waveform factor, and pulse index of each segment of each indicator data (i.e., historical target state data, including but not limited to historical hydraulic end single cylinder vibration amplitude, historical hydraulic end single cylinder temperature value, and historical hydraulic end single cylinder pressure value) are statistically calculated. 11 statistical indicator data.
[0076] The 11 statistical indicator data of the above-mentioned vibration amplitude, the 11 statistical indicator data of the single-cylinder temperature value, the 11 statistical indicator data of the single-cylinder pressure value, and the fault or normal data label value of the hydraulic end pump valve of the equipment (that is, the state classification label data obtained from the historical target state data of the equipment) are used as the columns of the matrix, here, there are a total of 11*3+1 columns; the 36,000 sample data obtained at intervals of 1s in the normal and faulty states are used as the rows of the matrix, a total of 36,000*2 rows, thereby obtaining a data matrix of size (36,000*2,11*3+1).
[0077] The data matrix obtained above is divided into training data set and test data set in a ratio of 9:1.
[0078] The training data set is input into the constructed BP neural network model. After the data is calculated in each layer of the network, it is output to the next layer of the neural network through the activation function. The last layer outputs the calculation result. The calculation result and the real data are input into the loss function. The loss function calculates the loss value. When the loss value is greater than the set threshold, the optimization function updates the connection weights of each layer of the network in the direction of reducing the loss value according to the back propagation value of the loss value in the gradient direction; when the loss function value is less than the set threshold, the neural network training ends, and the network structure and the information of neurons at all levels are saved, thereby obtaining a trained neural network.
[0079] The smaller the threshold is, the better. Preferably, it should be at least less than 0.05.
[0080] The current target state data in the above test data set is input into the trained neural network model, and the test results (i.e., the predicted label of the current state of the device) and the test accuracy are output. The test accuracy is obtained by calculating the AUC value based on the test results output based on the test data and its corresponding true state label. If the test accuracy is less than the set threshold, the model test is completed. If the test accuracy is greater than the threshold, the sample data is reselected and the above model training and testing are repeated.
[0081] Based on the trained neural network model, the current target state data of the device is input, and the current state classification of the device is predicted and output as the current state label of the device. Finally, the current state label of the device is stored in the standard structured label database.
[0082] Taking the fracturing truck-mounted plunger pump at a fracturing well site as an example, the sample data is based on the above experimental data and neural network training steps, and the current hydraulic end pump valve health status label of the target equipment is calculated to be faulty.
[0083] In addition, the health status of other equipment components or status labels of other equipment such as failure risk level can also be calculated using the above method.
[0084] In addition, the current state label of the equipment can also be obtained in a predefined way, that is, for the current target state data of the equipment, based on historical expert experience, the current state classification of the equipment is output as the current state label of the equipment. The current state label of the equipment is stored in the standard structured label database.
[0085] Taking the fracturing truck-mounted plunger pump at a fracturing well site as an example, the current status label of the equipment can be obtained in a predefined manner, but it is not limited to the current pressure status label of the equipment. The label values include: high, medium, and low. For example, α1 and β1 are preset as the upper and lower limits of the threshold, and the following label values are set in this predefined manner:
[0086] High: Current equipment pressure > α1
[0087] Middle: β1 <= current pressure of the equipment <= α1
[0088] Low: The current pressure of the device is < β1.
[0089] In addition, taking the fracturing vehicle-mounted plunger pump at the fracturing well site as an example, based on the above image label construction method, the following can be constructed: Figure 3The label system shown. As shown in the figure, the label system may include the following labels: transient parameter label class, such as, instantaneous pressure label, instantaneous temperature label, instantaneous sand ratio label, instantaneous speed label, etc.; operation status label class, such as, current pressure status of equipment, current temperature status of equipment, current sand ratio status of equipment, etc.; vibration feature class label, for example, power end vibration feature class label such as vibration signal time domain amplitude mean, vibration signal time domain amplitude absolute mean, vibration signal time domain amplitude variance, etc.; hydraulic end vibration feature class label such as vibration signal time domain amplitude mean, vibration signal time domain amplitude absolute mean, vibration signal time domain amplitude variance, etc.
[0090] S106: Generate a device profile of the mechanical device by using the attribute tag value of the generated attribute tag and the state tag value of the obtained current state tag.
[0091] According to a preferred embodiment of the present application, a device profile is constructed based on a tag value clustering method.
[0092] Get all attribute tag values and status tag values of mechanical equipment, create a set of all tag values X and a set of all equipment in the target area Y, and construct a co-occurrence matrix (X, Y). If equipment Y i With X j If the class label value is 0, the corresponding position value in the matrix is 1, otherwise it is 0. Here, 1 and 0 are just used as the corresponding situation marks of each device and each label value. The corresponding relationship between each device and the label value is used as the clustering feature to cluster labels with similar corresponding relationships into one category, and each type of label is used as a portrait of the device with this type of label.
[0093] Taking the fracturing truck-mounted plunger pump at a fracturing well site as an example, the co-occurrence matrix example is shown below:
[0094]
[0095] Using clustering algorithms such as k-medoid, various tags are clustered based on the above tag-device co-occurrence matrix, and finally multi-class tag clustering clusters of attribute tags of target attribute data and current status tags are obtained, where the number of clustering categories can be set as needed.
[0096] Compare all label values of each device with the label values contained in each label cluster, obtain the label cluster with the largest number of label value categories covered by the label values contained in each device, and take the label value under the corresponding label cluster as the current key label value of the device to generate a device profile for the device.
[0097] For example, the first category includes 10 tags, such as the pump valve health status (fault) tag, the crosshead health status (healthy) tag, and the pump valve pressure status (high) tag. The second category includes another 15 tags. Device A has a total of 100 current tags. In order to highlight the key points, some tags in the portrait of device A are selected. If the 100 tags of device A include 20 tags of the first category, that is, all 10 tags of the first category, and only 5 tags of the second category, then the tag values in the first category are the most concentrated and important tags of device A at the current moment, that is, the current key tags mentioned above. Therefore, the portrait of device A is generated with the first category tags.
[0098] Furthermore, according to a preferred embodiment of the present application, the device portrait is visualized as a portrait word cloud diagram, such as Figure 4 As shown in the figure, the operating status of the equipment can be easily obtained from the figure, so as to facilitate the understanding of the dynamics, operating status and health status of the equipment.
[0099] Figure 5 A schematic flow chart of a method for generating a mechanical equipment image according to a specific embodiment of the present application is shown.
[0100] As shown in the figure, all data information related to mechanical equipment is obtained according to the method described above, so as to construct an equipment portrait data source.
[0101] Next, according to the method described above, data fusion is performed on the acquired equipment portrait data source, that is, all data information related to the mechanical equipment, so as to obtain a multi-source data information set.
[0102] According to the obtained multi-source data information set, the device tags are classified, for example, the device attribute tags and device status tags mentioned above.
[0103] Furthermore, the obtained device attribute label data and device status label data are preprocessed to obtain a device target attribute label and a device target status label.
[0104] Generate a device profile based on the device target attribute label and the device target status label.
[0105] Among the labels including attribute labels and status labels, the current status label of the device can be obtained through the above-constructed neural network model, and of course, the current status label of the device can also be obtained through a preset threshold according to the experience of experts in this field. According to actual needs, the above two methods can be selected or used in combination.
[0106] Finally, a device portrait is constructed using the obtained device target attribute labels and device target status labels, for example, by using a label value clustering method, and these data are presented using a visualization method (eg, a word cloud diagram).
[0107] According to this application, by applying customer portrait technology to plunger pump equipment, a relatively complete plunger pump equipment portrait labeling system is constructed.
[0108] Moreover, the labeling system constructed in the present application includes not only static attributes and instantaneous parameters, but also parameter status evaluation and maintenance status information, and provides a method for constructing current status labels of equipment based on neural networks, so as to facilitate obtaining labels such as the current health status of the equipment, and finally obtain a portrait of the equipment, which is helpful to realize plunger pump fault detection, and can timely and accurately grasp the current operating status of the equipment, so as to facilitate control of the entire production process.
[0109] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0110] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A method for generating a mechanical equipment image, characterized in that: The mechanical equipment portrait generation method comprises: Obtain all data information related to the mechanical equipment; Performing data fusion on the data information related to the mechanical equipment to obtain a multi-source data information set; Acquire target attribute data and target state data of the mechanical equipment based on the multi-source data information set; generating an attribute label of the mechanical equipment based on the acquired target attribute data; Inputting the current target state data of the mechanical equipment into the trained neural network model to obtain the current state label of the mechanical equipment, wherein the current state label includes the state label value of the mechanical equipment; Generate a device profile of the mechanical device by using the acquired attribute tag value of the attribute tag and the acquired state tag value of the current state tag, in, Data fusion is performed on the data information related to the mechanical equipment to obtain a multi-source data information set, including: Set a unique identifier for each device; Acquire a first library table having a unique identifier for each device and a second library table associated with the first library table; and The first library table and the second library table are associated with each other by the unique identifier, and wherein: Generating a device profile of the mechanical device by using the acquired attribute tag value of the attribute tag and the acquired state tag value of the current state tag, including: Based on the attribute tag value and the current state tag value of the mechanical equipment, construct all tag value sets and equipment sets; Constructing a co-occurrence matrix based on the tag value set and the device set; Based on the co-occurrence matrix, obtaining a label cluster of the attribute label and the current state label by a clustering algorithm; and Based on the tag cluster, the current key tag value of the equipment is obtained to generate the equipment portrait of the mechanical equipment. And among them, Acquiring the current key tag value of the equipment based on the tag clustering cluster and generating the equipment portrait of the mechanical equipment includes: Compare all label values of each device with the obtained label values contained in the label cluster; and A first label cluster with the largest number of label value categories in the label clusters is obtained, in which label values contained in each device cover, and the label value corresponding to the first label cluster is set as the current key label value of the device.
2. The method for generating a mechanical equipment image according to claim 1, characterized in that: The mechanical equipment portrait generation method further includes: Initializing a neural network model based on the target state data of the mechanical equipment and the number of target state label values of the target state data; generating a plurality of statistical indicators based on the target status data; and A data matrix including the multiple statistical indicators and the state classification label data of the target state data is input into the initialized neural network model to train the neural network model.
3. The method for generating a mechanical equipment image according to claim 1, characterized in that: Acquiring target attribute data and target state data of the mechanical equipment based on the multi-source data information set includes: Acquire initial attribute data and initial state data based on the multi-source data information set; Based on the attention and / or importance of the initial attribute data, acquiring the target attribute data; The initial state data is normalized to obtain the target state data.
4. The method for generating a mechanical device image according to any one of claims 1 to 3, characterized in that: The mechanical equipment portrait generation method further includes: The generated device portrait is visualized as a portrait word cloud diagram.
5. The method for generating a mechanical device image according to any one of claims 1 to 3, characterized in that: The mechanical equipment portrait generation method further includes: The state tag value of the current target state data of the mechanical equipment is compared with the set value to obtain the current state tag of the mechanical equipment.
6. The method for generating a mechanical device image according to claim 1, characterized in that: The neural network model includes a back-propagation neural network model.
7. The method for generating a mechanical device image according to claim 2, characterized in that: The statistical indicators include: mean, standard deviation, root amplitude, effective value, peak value, skewness, kurtosis, maximum value, margin value, waveform factor, and pulse index.
8. The method for generating a mechanical device image according to claim 5, characterized in that: The status tag values include: normal, fault, high, medium and low.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.
