Method for monitoring transient flow field and performance of scroll compressor based on graph data mining

Abstract

The application provides a scroll compressor transient flow field and performance monitoring method based on graph data mining, relates to the technical field of scroll compressor operation monitoring, and aims to quickly obtain a high spatial resolution scroll compressor transient flow field and performance index, thereby providing a strong reference basis for subsequent control strategy formulation. The method comprises the following steps: based on different operation condition samples and flow field data corresponding to the operation condition samples, a training data set represented by a graph data structure is constructed; a scroll compressor transient flow field and performance monitoring model is trained by using the training data set, so that a trained scroll compressor transient flow field and performance monitoring model is obtained; and operation condition data of the collected scroll compressor is input into the trained scroll compressor transient flow field and performance monitoring model, so that a transient flow field result and a performance index result of the scroll compressor are obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of scroll compressor operation monitoring, and particularly relates to a scroll compressor transient flow field and performance monitoring method based on graph data mining. BACKGROUND

[0002] A heat pump air conditioning system is a vehicle air conditioning system with high energy efficiency. A scroll compressor is a power source of a vehicle heat pump air conditioning system, which drives refrigerant circulation in the system to realize heating or cooling of a passenger cabin. To ensure that the temperature of the passenger cabin meets the comfort needs of passengers, the scroll compressor of the vehicle heat pump air conditioning system needs to frequently operate in variable conditions according to external environmental conditions and vehicle operating conditions, which induces aggravation of internal flow separation of the scroll compressor, deterioration of the flow field, and even instability. Therefore, accurate monitoring of the internal flow field state and performance of the scroll compressor is of great significance to ensure safe and efficient operation of the vehicle heat pump air conditioning system.

[0003] In related technologies, scroll compressor flow field monitoring technology is usually performed by sensors arranged at key positions of the scroll compressor, and performance monitoring is mainly obtained by indirect calculation of directly measured sensor data. Considering factors such as interference with the flow field, physical feasibility, and additional costs, it is not feasible to install dense sensors. Therefore, this sensor monitoring technology can only obtain extremely sparse flow field data, has poor interpretability, and cannot comprehensively and accurately reflect the internal flow field structure of the entire scroll compressor. Although the computational fluid dynamics (CFD) method, for example, solving the Reynolds-averaged Navier-Stokes equation (RANS), can simulate the complex three-dimensional strong nonlinear flow in the scroll compressor and obtain complete flow field information, high-precision CFD analysis of the internal flow of the scroll compressor is an extremely time-consuming task and cannot be applied to the operation stage.

[0004] Therefore, how to quickly obtain high spatial resolution scroll compressor transient flow field and performance indicators is a technical problem to be solved. SUMMARY

[0005] In view of the above problems, the embodiments of the present application provide a scroll compressor transient flow field and performance monitoring method based on graph data mining, so as to overcome the above problems or at least partially solve the above problems.

[0006] In a first aspect, the embodiments of the present application disclose a scroll compressor transient flow field and performance monitoring method based on graph data mining, which comprises:

[0007] construct a training data set represented by a graph data structure based on different operating condition samples and flow field data corresponding to the operating condition samples, each graph data structure representation in the training data set comprising: global features representing operating conditions and performance indicators, local features representing transient flow field node information, and node connection relationships of the transient flow field;

[0008] train a scroll compressor transient flow field and performance monitoring model using the training data set to obtain a trained scroll compressor transient flow field and performance monitoring model having a mapping relationship between operating condition data and each node transient flow field, and a mapping relationship with performance indicators;

[0009] input the collected operating condition data of the scroll compressor into the trained scroll compressor transient flow field and performance monitoring model to obtain the transient flow field result and the performance indicator result of the scroll compressor.

[0010] Optionally, constructing a training data set represented by a graph data structure based on different operating condition samples and flow field data corresponding to the operating condition samples comprises:

[0011] construct a sample library, each sample in the sample library comprising an operating condition sample, a transient flow field sample, and a performance indicator sample;

[0012] construct a graph data structure representation of each sample in the sample library to obtain a training data set.

[0013] Optionally, constructing a sample library comprises:

[0014] based on a range of operating boundary conditions, obtain different operating condition samples, the operating condition samples comprising operating boundary conditions composed of suction pressure, suction temperature, discharge pressure, and rotational speed;

[0015] perform fluid mechanics simulation analysis on the transient flow field inside the scroll compressor using the operating condition samples, and use the simulation analysis results including the corner, unit information, and node information as the transient flow field sample;

[0016] perform performance indicator calculation according to the simulation analysis results to obtain a performance indicator sample, the performance indicator sample comprising the instantaneous suction mass flow rate, instantaneous discharge temperature, isentropic efficiency, and volumetric efficiency of the scroll compressor.

[0017] Optionally, constructing a graph data structure representation of each sample in the sample library comprises:

[0018] the operating boundary conditions in the operating condition sample, the corner in the transient flow field sample, and the performance indicators in the performance indicator sample are used as global features;

[0019] The node information of each node in the transient flow field sample is taken as a local feature, and the node information includes node three-dimensional coordinates and a flow field;

[0020] According to the cell information and the node information in the transient flow field sample, a node connection relationship is determined.

[0021] Optionally, the operating condition data includes rotation angle data and operating boundary condition data; the collected operating condition data of the scroll compressor is input to the trained scroll compressor transient flow field and performance monitoring model to obtain the transient flow field and performance index result of the scroll compressor, including:

[0022] According to the rotation angle data and the operating boundary condition data, a corresponding target graph data structure representation is found from the training data set, and corresponding node three-dimensional coordinate data and node connection relationship data are read from the target graph data structure representation;

[0023] The rotation angle data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data are input to the trained scroll compressor transient flow field and performance monitoring model to obtain the transient flow field and performance index result of the scroll compressor.

[0024] Optionally, the scroll compressor transient flow field and performance monitoring model includes a transient flow field monitoring network and a performance monitoring network connected in series;

[0025] The rotation angle data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data are input to the trained scroll compressor transient flow field and performance monitoring model to obtain the transient flow field result and performance index result of the scroll compressor, including:

[0026] The rotation angle data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data are input to the transient flow field monitoring network to obtain the transient flow field result of the scroll compressor;

[0027] The transient flow field result of the scroll compressor, the rotation angle data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data are input to the performance monitoring network to obtain the performance index result.

[0028] Optionally, the transient flow field monitoring network includes a first input layer, a first input encoding layer, a plurality of first graph attention layers, and a first output layer;

[0029] inputting the corner data, the operation boundary condition data, the node three-dimensional coordinate data and the node connection relationship data into the transient flow field monitoring network to obtain a transient flow field result of the scroll compressor, including:

[0030] inputting the corner data, the operation boundary condition data, the node three-dimensional coordinate data and the node connection relationship data through the first input layer;

[0031] encoding the corner data and the operation boundary condition data to each node through the first input encoding layer, and mapping the corner data, the operation boundary condition data and the node three-dimensional coordinate data into a target range to obtain a first encoding result;

[0032] extracting features of a correlation degree between adjacent nodes in the first encoding result through the multiple first graph attention layers to obtain a feature extraction result;

[0033] mapping the feature extraction result to a first target output dimension through the first output layer to obtain the transient flow field result of the scroll compressor.

[0034] Optionally, the performance monitoring network comprises a second input layer, a second input encoding layer, multiple second graph attention layers, multiple hierarchical pooling layers, multiple global pooling layers and a second output layer;

[0035] inputting the transient flow field result of the scroll compressor, the corner data, the operation boundary condition data, the node three-dimensional coordinate data and the node connection relationship data into the performance monitoring network to obtain a performance index result, including:

[0036] inputting the transient flow field result of the scroll compressor, the corner data, the operation boundary condition data, the node three-dimensional coordinate data and the node connection relationship data through the second input layer;

[0037] encoding the transient flow field result of the scroll compressor, the corner data, the operation boundary condition data, the node three-dimensional coordinate data and the node connection relationship data through the second input encoding layer to obtain a second encoding result;

[0038] processing the second encoding result in sequence through the multiple second graph attention layers, the multiple hierarchical pooling layers and the multiple global pooling layers to obtain a feature processing result;

[0039] mapping the feature processing result to a second target output dimension through the second output layer to obtain the performance index result.

[0040] Optionally, the transient flow field is a variable-grid transient flow field or a non-variable-grid transient flow field.

[0041] In a second aspect, the application discloses a vortex compressor transient flow field and performance monitoring device based on graph data mining, which comprises:

[0042] The construction module is configured to construct a training data set represented by a graph data structure based on different operation condition samples and flow field data corresponding to the operation condition samples, wherein each graph data structure representation in the training data set comprises global features representing operation conditions and performance indicators, local features representing node information of the transient flow field, and node connection relationships of the transient flow field.

[0043] The training module is configured to train a vortex compressor transient flow field and performance monitoring model by using the training data set, so as to obtain a trained vortex compressor transient flow field and performance monitoring model, wherein the trained vortex compressor transient flow field and performance monitoring model has a mapping relationship between operation condition data and each node transient flow field, and a mapping relationship with performance indicators.

[0044] The input module is configured to input collected operation condition data of the vortex compressor into the trained vortex compressor transient flow field and performance monitoring model, so as to obtain a transient flow field result and a performance indicator result of the vortex compressor.

[0045] In a third aspect, the application discloses an electronic device, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the vortex compressor transient flow field and performance monitoring method based on graph data mining according to the first aspect of the application when executing the computer program.

[0046] In a fourth aspect, the application discloses a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the vortex compressor transient flow field and performance monitoring method based on graph data mining according to the first aspect of the application when executed by a processor.

[0047] In a fifth aspect, the application discloses a computer program product, which comprises a computer program, and the computer program implements the steps of the vortex compressor transient flow field and performance monitoring method based on graph data mining according to the first aspect of the application when executed by a processor.

[0048] The application has the following advantages:

[0049] In the embodiment of the present application, a training data set represented by a graph data structure is constructed based on different operating condition samples. Each graph data structure representation in the training data set includes global features representing operating conditions and performance indicators, local features representing node information of a transient flow field, and node connection relationships of the transient flow field. A training data set of internal transient full three-dimensional flow fields of scroll compressors and performance indicators of scroll compressors under variable grid or constant grid conditions is established. A scroll compressor transient flow field and performance monitoring model is trained using the training data set. The trained scroll compressor transient flow field and performance monitoring model has a mapping relationship between operating condition data and each transient node flow field, and a mapping relationship with performance indicators. Therefore, based on the trained scroll compressor transient flow field and performance monitoring model, the internal transient full three-dimensional flow field of the scroll compressor and the performance indicators of the scroll compressor under variable operating conditions can be quickly predicted, overcoming the slow response defect of the traditional CFD method. Moreover, the operating condition data of the scroll compressor collected by the sparse sensor is used to quickly predict the internal transient full three-dimensional flow field of the scroll compressor and the performance indicators of the scroll compressor, overcoming the low spatial resolution defect of the traditional sensor measurement data, and being suitable for industrial application scenarios where dense measurement points are difficult to arrange. BRIEF DESCRIPTION OF DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the description of the embodiments of the present application will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0051] Figure 1 is a step flow chart of a scroll compressor transient flow field and performance monitoring method based on graph data mining provided by the embodiments of the present application;

[0052] Figure 2 is a schematic diagram of a graph structure data representation provided by the embodiments of the present application;

[0053] Figure 3 is a step flow chart of another scroll compressor transient flow field and performance monitoring method based on graph data mining provided by the embodiments of the present application;

[0054] Figure 4 is a structure schematic diagram of a scroll compressor transient flow field and performance monitoring device based on graph data mining provided by the embodiments of the present application;

[0055] Figure 5 is a structure schematic diagram of an electronic device provided by the embodiments of the present application. DETAILED DESCRIPTION

[0056] To make the above-mentioned objectives, features, and advantages of this application more apparent and understandable, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0057] In related technologies, sensor-based scroll compressor flow field and performance monitoring techniques suffer from low spatial resolution. While CFD methods can obtain high spatial resolution flow field information, their time-consuming nature prevents their application during operation. Therefore, to overcome these limitations, this application provides a graph data mining-based method for monitoring the transient flow field and performance of scroll compressors. This method uses data mining techniques to extract useful information and knowledge implicit in samples from different operating conditions. Based on this information and knowledge, it can quickly predict the transient full three-dimensional flow field inside the scroll compressor and its performance indicators under varying operating conditions. This method can be applied to the flow field and performance monitoring of scroll compressors in automotive heat pump air conditioning systems during variable operating conditions. It can rapidly predict the transient full three-dimensional flow field inside the scroll compressor and its performance indicators using data collected from sparse sensors. Furthermore, the predicted transient full three-dimensional flow field and performance indicators of the scroll compressor have significant guiding significance for equipment operation status assessment, equipment energy efficiency assessment, and subsequent control system design.

[0058] The transient flow field and performance monitoring method of a scroll compressor based on graph data mining according to embodiments of this application will be described below with reference to the accompanying drawings.

[0059] Reference Figure 1 As shown, Figure 1 This is a flowchart illustrating the steps of a method for monitoring the transient flow field and performance of a scroll compressor based on graph data mining, as provided in an embodiment of this application. Figure 1 As shown, the method for monitoring the transient flow field and performance of a scroll compressor based on graph data mining may include steps S110 to S130:

[0060] Step S110: Based on different operating condition samples and the flow field data corresponding to the operating condition samples, construct a training dataset of graph data structure representations. Each graph data structure representation in the training dataset includes: global features representing operating conditions and performance indicators, local features representing transient flow field node information, and node connection relationships of the transient flow field.

[0061] The running condition sample refers to corresponding condition data under certain running boundary conditions, for example, the running condition sample includes data such as suction pressure, suction temperature, exhaust pressure, and rotating speed. The flow field data can be obtained based on fluid mechanics simulation of the running condition sample, and the flow field data includes transient flow field data and performance index data, wherein the transient flow field data refers to data such as rotation angle, unit information, and node information (including node three-dimensional coordinates and flow field); the performance index data refers to data such as instantaneous suction mass flow rate, instantaneous exhaust temperature, isentropic efficiency, and volumetric efficiency of the scroll compressor. A graph data structure representation is constructed according to each running condition sample and corresponding flow field data, and then a training data set composed of multiple graphs is obtained.

[0062] Step S120: training the scroll compressor transient flow field and performance monitoring model using the training data set to obtain a trained scroll compressor transient flow field and performance monitoring model, wherein the trained scroll compressor transient flow field and performance monitoring model has a mapping relationship between the running condition data and the node transient flow field, and a mapping relationship with the performance index.

[0063] The scroll compressor transient flow field and performance monitoring model is a neural network model for predicting the transient flow field and performance index of the scroll compressor. The scroll compressor transient flow field and performance monitoring model includes a transient flow field monitoring network and a performance monitoring network connected in series, wherein the transient flow field monitoring network is used to predict the transient flow field, and the performance monitoring network is used to predict the performance index.

[0064] Each graph data structure representation in the training data set is input into the scroll compressor transient flow field and performance monitoring model in sequence for training, so that the scroll compressor transient flow field and performance monitoring model learns the mapping relationship between the running condition data and the node transient flow field, and the mapping relationship with the performance index.

[0065] Step S130: inputting the collected running condition data of the scroll compressor into the trained scroll compressor transient flow field and performance monitoring model to obtain the transient flow field result and the performance index result of the scroll compressor.

[0066] The operation condition data of the scroll compressor is collected by sensors, that is, the change curves of the operation condition (rotation angle and operation boundary condition) in the operation process of the scroll compressor are collected by the sparsely deployed sensors. Since the trained scroll compressor transient flow field and performance monitoring model has a mapping relationship between the operation condition data and the transient flow field of each node, and a mapping relationship with the performance indicators, the transient flow field result and the performance indicator result of the scroll compressor can be obtained based on the trained scroll compressor transient flow field and performance monitoring model. The transient flow field result includes pressure, temperature, and velocity components in different directions, and the performance indicator result includes instantaneous suction mass flow rate, instantaneous discharge temperature, isentropic efficiency, and volumetric efficiency of the scroll compressor.

[0067] The technical scheme of the embodiment of the present application constructs a training data set represented by a graph data structure based on different operation condition samples. Each graph data in the training data set includes global features representing operation conditions and performance indicators, local features representing node information of a transient flow field, and node connection relationships of the transient flow field. The training data set of the internal transient full three-dimensional flow field of the scroll compressor and the performance indicators of the scroll compressor under variable grid or constant grid conditions is established. The scroll compressor transient flow field and performance monitoring model is trained using the training data set. The trained scroll compressor transient flow field and performance monitoring model has a mapping relationship between the operation condition data and the transient flow field of each node, and a mapping relationship with the performance indicators. Therefore, based on the trained scroll compressor transient flow field and performance monitoring model, the internal transient full three-dimensional flow field of the scroll compressor and the performance indicators of the scroll compressor under variable condition conditions can be quickly predicted, overcoming the slow response defect of the traditional CFD method. Moreover, the operation condition data of the scroll compressor collected by the sparse sensors is used to quickly predict the internal transient full three-dimensional flow field of the scroll compressor and the performance indicators of the scroll compressor, overcoming the low spatial resolution defect of the traditional sensor measurement data, and being suitable for application in industrial application scenarios where dense measurement points are difficult to arrange.

[0068] In the embodiment of the present application, the transient flow field is a variable grid transient flow field or a constant grid transient flow field. That is, the technical scheme of the embodiment of the present application can be used for flow field prediction under variable grid conditions and flow field prediction under constant grid conditions. Specifically, when flow field prediction under variable grid conditions is needed, the flow field data corresponding to the operation condition sample in step S110 is flow field data under variable grid conditions, and when flow field prediction under constant grid conditions is needed, the flow field data corresponding to the operation condition sample in step S110 is flow field data under constant grid conditions.

[0069] In combination with the above embodiments, in an embodiment, the application further provides a method for monitoring the transient flow field and performance of a scroll compressor based on graph data mining. In the method, the step S110 of "constructing a training data set represented by a graph data structure based on different operating condition samples and flow field data corresponding to the operating condition samples" specifically includes the following steps S110-1 and S110-2:

[0070] Step S110-1: Constructing a sample library, each sample in the sample library including an operating condition sample, a transient flow field sample, and a performance index sample.

[0071] In the application, the operating condition sample can be the operating condition data collected under different operating boundary conditions, and the transient flow field sample and the performance index sample are flow field data and performance index data obtained based on the operating condition sample, each operating condition sample corresponding to a transient flow field sample and a performance index sample.

[0072] Specifically, the sample library is constructed, including steps A1 to A3:

[0073] Step A1: Based on the operating boundary condition range, different operating condition samples are obtained, the operating condition samples including operating boundary conditions composed of suction pressure, suction temperature, discharge pressure, and rotational speed.

[0074] In the application, multiple sets of operating conditions are obtained based on the operating boundary condition range through Latin hypercube sampling, i.e., multiple different operating condition samples are obtained. In some embodiments, the different operating condition samples obtained are collected and stored to form an operating condition sample library.

[0075] The operating boundary condition range is constrained by the following four conditions: the first constraint is the minimum and maximum evaporation temperature, the second constraint is the maximum discharge temperature, the third constraint is the maximum pressure difference, and the fourth constraint is the maximum condensation temperature. That is, the boundary condition data of the different operating condition samples obtained will not exceed the above four constraints. The four constraints are flexibly set according to the actual operation of the scroll compressor.

[0076] Step A2: Using the operating condition samples, fluid mechanics simulation analysis is performed on the internal flow field transient of the scroll compressor, and the angle, element information, and node information in the simulation analysis results are taken as the transient flow field sample.

[0077] In the embodiments of the present application, for each operating condition sample, the transient simulation analysis of the internal flow field of the scroll compressor can be carried out by a fluid mechanics method, wherein the simulation analysis involves a dynamic mesh method. After obtaining the simulation analysis result, the rotation angle, element information and node information of the last cycle data in the simulation analysis result are taken as the transient flow field sample. Specifically, the rotation angle, element information and node information of the fluid domain when the crankshaft rotation angle is 0° to 360° are collected and stored as the transient flow field sample.

[0078] It can be understood that each operating condition sample corresponds to a transient flow field sample, and based on the plurality of transient flow field samples corresponding to different operating condition samples, a transient flow field sample library can be obtained.

[0079] Step A3: calculating the performance index based on the simulation analysis result to obtain a performance index sample, wherein the performance index sample includes the instantaneous suction mass flow rate, the instantaneous discharge temperature, the isentropic efficiency and the volumetric efficiency of the scroll compressor.

[0080] The instantaneous suction mass flow rate and the instantaneous discharge temperature of the scroll compressor are obtained by statistical analysis of the simulation analysis result, and the isentropic efficiency and the volumetric efficiency are calculated by the following formulas:

[0081] ,

[0082] ,

[0083] wherein, h out is the suction specific enthalpy at the compressor outlet; h in is the suction specific enthalpy at the compressor inlet; h out,s is the specific enthalpy at the compressor outlet calculated by an isentropic process; m ave is the time-averaged suction mass flow rate in one cycle; V s is the volume of the suction chamber after the suction is completed; ρ in is the suction density.

[0084] It can be understood that each operating condition sample corresponds to a performance index sample, and based on the plurality of performance index samples corresponding to different operating condition samples, a performance index sample library can be obtained.

[0085] Step S120-2: constructing a graph data structure representation of each sample in the sample library to obtain a training data set.

[0086] In the embodiments of the present application, for each sample corresponding to the operating condition sample, the transient flow field sample, and the performance index sample, global features, local features, and node connection relationships are collected and stored from the samples to obtain corresponding graph structure representations, and then a training data set is obtained according to the graph structure representations corresponding to all samples.

[0087] Specifically, the graph data structure representation of each sample in the sample library is constructed, including steps B1 to B3:

[0088] Step B1: The operating boundary conditions in the operating condition sample, the corners in the transient flow field sample, and the performance indicators in the performance index sample are taken as global features.

[0089] For example, the global features can be represented as: , where the suction pressure P in , the suction temperature T in , the exhaust pressure P ou t , and the rotational speed N r are operating boundary conditions, θ the corners are corners, and the instantaneous suction mass flow of the scroll compressor m in , the instantaneous exhaust temperature T out , the isentropic efficiency η is , and the volumetric efficiency η v are performance indicators target1.

[0090] Step B2: The node information of each node in the transient flow field sample is taken as local features, and the node information includes node three-dimensional coordinates and flow fields.

[0091] For any operating condition (or operating boundary condition) and corner θ , the node information V of the local feature V θ includes three-dimensional coordinates and flow fields, that is:

[0092] ,

[0093] wherein x i , y i , and z i are the three-dimensional coordinates of the first i node c θ , and the flow field target2 includes pressureP i ,temperature T i , x i Directional velocity components u i , y i Directional velocity components v i , z i Directional velocity components w i , N This represents the number of nodes in the grid.

[0094] Local feature V is node information V θ The set, namely: .

[0095] Step B3: Determine the node connection relationship based on the unit information and node information in the transient flow field sample.

[0096] For any operating condition and turning angle θ Nodes are numbered, and nodes with connections are called neighbors. Nodes in shared units are considered neighbors, i.e., node connection relationship data E. θ :

[0097] ,

[0098] Node connection relationships can be represented by node connection relationship data E. θ The set, i.e., the node connection relationship ,in, v i and v j For two nodes that are connected, e ij An edge is an edge that connects two nodes.

[0099] For example, Figure 2 This is a schematic diagram of a graph structure data representation provided in an embodiment of this application. Specifically, each graph data includes global features, local features, and node connection relationships. The global features include operating boundary conditions (i.e., intake pressure, intake temperature, exhaust pressure, and rotational speed), rotation angle, and performance indicators (i.e., the instantaneous intake mass flow rate, instantaneous exhaust temperature, isentropic efficiency, and volumetric efficiency of the scroll compressor). The local features include the three-dimensional coordinates of the nodes and the flow field (i.e., pressure, temperature, and velocity components in different directions).

[0100] The technical scheme of the embodiment of the present application is based on different operating condition samples and flow field data corresponding to the operating condition samples, constructs a training data set represented by a graph data structure, and establishes a training data set of internal transient full three-dimensional flow field of a scroll compressor and performance index of the scroll compressor under variable grid or constant grid conditions, so that a scroll compressor transient flow field and performance monitoring model can be trained based on the training data set subsequently, thereby establishing a mapping relationship between operating condition data and each node transient flow field and a mapping relationship between the performance index.

[0101] In combination with the above embodiments, in an embodiment, the embodiment of the present application further provides a scroll compressor transient flow field and performance monitoring method based on graph data mining. In the method, the operating condition data includes rotation angle data and operating boundary condition data, and the step S130 of "inputting the collected operating condition data of the scroll compressor into the trained scroll compressor transient flow field and performance monitoring model to obtain the transient flow field and performance index result of the scroll compressor" specifically includes the following steps S130-1 and S130-2.

[0102] The step S130-1 is to find corresponding target graph data structure representation from the training data set according to the rotation angle data and the operating boundary condition data, and read corresponding node three-dimensional coordinate data and node connection relationship data from the target graph data structure representation.

[0103] The step S130-2 is to input the rotation angle data, the operating boundary condition data, the node three-dimensional coordinate data and the node connection relationship data into the trained scroll compressor transient flow field and performance monitoring model to obtain the transient flow field and performance index result of the scroll compressor.

[0104] In the embodiment of the present application, the scroll compressor transient flow field and performance monitoring model not only predicts the scroll compressor transient flow field and performance index based on the rotation angle data and the operating boundary condition data, but also needs to predict based on the node three-dimensional coordinate data and the node connection relationship data representing the scroll compressor. Specifically, the change curves of the rotation angle and the operating boundary condition during the operation of the scroll compressor are collected by using the sparsely deployed sensors, that is, the rotation angle data and the operating boundary condition data are obtained. The node three-dimensional coordinate data and the node connection relationship data under the current operating condition are read from the corresponding graph data structure representation. Then, the data collected by the sensors and the read data are input into the trained scroll compressor transient flow field and performance monitoring model, and the transient flow field and performance index result of the scroll compressor under the corresponding condition can be quickly obtained.

[0105] Specifically, the scroll compressor transient flow field and performance monitoring model comprises a transient flow field monitoring network and a performance monitoring network connected in series; the turning angle data, the operating boundary condition data, the node three-dimensional coordinate data and the node connection relationship data are input into the trained scroll compressor transient flow field and performance monitoring model to obtain the transient flow field result and the performance index result of the scroll compressor, comprising steps C1 to C2:

[0106] Step C1: inputting the turning angle data, the operating boundary condition data, the node three-dimensional coordinate data and the node connection relationship data into the transient flow field monitoring network to obtain the transient flow field result of the scroll compressor.

[0107] Step C2: inputting the transient flow field result of the scroll compressor, the turning angle data, the operating boundary condition data, the node three-dimensional coordinate data and the node connection relationship data into the performance monitoring network to obtain the performance index result.

[0108] In the embodiments of the present application, the transient flow field monitoring network can be characterized as a mapping relationship between the first target parameters and the flow fields of each node, wherein the first target parameters include the turning angle, the operating boundary condition, the node three-dimensional coordinate and the node connection relationship. The performance monitoring network can be characterized as a mapping relationship between the second target parameters and the corresponding performance indexes, wherein the second target parameters include the turning angle, the operating boundary condition, the node three-dimensional coordinate, the node connection relationship and the transient flow field.

[0109] For example, the mapping relationship represented by the transient flow field monitoring network can be expressed as:

[0110] ,

[0111] The mapping relationship represented by the performance monitoring network can be expressed as:

[0112] ,

[0113] wherein, θ is the turning angle, B is the operating boundary condition, c θ is the node three-dimensional coordinate, E θ is the node connection relationship, target1 is the flow field variable (including pressure P , temperature T , velocity component in different directions, i.e. x direction velocity component u, y direction velocity component v and zdirectional velocity component w), target2 is a corresponding performance index (instantaneous suction mass flow of the scroll compressor under the relevant condition (second target parameter) m in instantaneous discharge temperature T out isentropic efficiency η is and volumetric efficiency η v ), and respectively are parameters to be learned by the transient flow field monitoring network and the performance monitoring network in the training process.

[0114] The transient flow field monitoring network and the performance monitoring network are deployed in series. The acquired corner data, running boundary condition data, node three-dimensional coordinate data and node connection relationship data are input into the scroll compressor transient flow field and performance monitoring model, so that the transient flow field and performance index results of the scroll compressor under the corresponding condition can be quickly obtained.

[0115] In a specific embodiment, the transient flow field monitoring network comprises a first input layer, a first input encoding layer, a plurality of first graph attention layers and a first output layer; the corner data, the running boundary condition data, the node three-dimensional coordinate data and the node connection relationship data are input into the transient flow field monitoring network to obtain the transient flow field results of the scroll compressor, specifically comprising:

[0116] The corner data, the running boundary condition data, the node three-dimensional coordinate data and the node connection relationship data are input through the first input layer; the corner data and the running boundary condition data are encoded onto each node through the first input encoding layer, and the corner data, the running boundary condition data and the node three-dimensional coordinate data are mapped into a target range to obtain a first encoding result; the correlation degree between adjacent nodes in the first encoding result is feature extracted through the plurality of first graph attention layers to obtain a feature extraction result; the feature extraction result is mapped to a first target output dimension through the first output layer to obtain the transient flow field results of the scroll compressor.

[0117] That is, the input global feature U (i.e. the corner data and the running boundary condition data) is encoded onto each node through the first input encoding layer, and the global feature U and the local feature V (i.e. the node three-dimensional coordinate data) are mapped into a target range.

[0118] For example, if the target range is [-1~1], the global feature U and the local feature V are mapped into the range of [-1~1] in the following way:

[0119] ,

[0120] ,

[0121] in, Represents the global features after mapping. This represents the local features after mapping, where max represents the maximum value operation and min represents the minimum value operation.

[0122] Thus, by processing the input data through the transient flow field monitoring network with the above structure, the transient flow field results of the scroll compressor are obtained.

[0123] Specifically, the performance monitoring network includes a second input layer, a second input encoding layer, multiple second graph attention layers, multiple hierarchical pooling layers, multiple global pooling layers, and a second output layer. The transient flow field results of the scroll compressor, the rotation angle data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data are input into the performance monitoring network to obtain the performance index results, specifically including:

[0124] The transient flow field results, rotation angle data, operating boundary condition data, node 3D coordinate data, and node connection relationship data of the scroll compressor are input through the second input layer. The transient flow field results, rotation angle data, operating boundary condition data, node 3D coordinate data, and node connection relationship data of the scroll compressor are encoded through the second input encoding layer to obtain a second encoding result. The second encoding result is processed sequentially using the multiple second graph attention layers, the multiple hierarchical pooling layers, and the multiple global pooling layers to obtain a feature processing result. The feature processing result is mapped to the second target output dimension through the second output layer to obtain the performance index result.

[0125] Thus, the performance monitoring network described above processes the input data to obtain performance index results.

[0126] In this embodiment, to improve the stability of the training process, both the transient flow field monitoring network and the performance monitoring network employ residual connections. Based on this, the transient flow field monitoring network and the performance monitoring network are trained using the training dataset. The loss function chosen is the mean squared error function, and the optimization algorithm is ADAM (an adaptive learning rate optimization algorithm).

[0127] Reference Figure 3 As shown, Figure 3 This is a flowchart illustrating another method for monitoring the transient flow field and performance of a scroll compressor based on graph data mining, provided in this application embodiment. Specifically, the method includes steps S310 to S340:

[0128] Step S310: Construct a sample library, each sample in the sample library comprising an operating condition sample, a transient flow field sample, and a performance index sample; wherein the operating condition sample comprises operating boundary conditions composed of suction pressure, suction temperature, discharge pressure, and rotational speed, the transient flow field sample comprises a corner, unit information, and node information, and the performance index sample comprises instantaneous suction mass flow, instantaneous discharge temperature, isentropic efficiency, and volumetric efficiency of the scroll compressor.

[0129] Step S320: Construct a graph data structure representation of each sample in the sample library to obtain a training data set; each graph data structure representation in the training data set comprises global features representing operating conditions and performance indexes, local features representing node information of a transient flow field, and node connection relationships of the transient flow field.

[0130] Step S330: Train a scroll compressor transient flow field and performance monitoring model using the training data set to obtain a trained scroll compressor transient flow field and performance monitoring model, which has a mapping relationship between operating condition data and each node transient flow field, and a mapping relationship with performance indexes.

[0131] Step S340: The scroll compressor transient flow field and performance monitoring model comprises a transient flow field monitoring network and a performance monitoring network connected in series, the corner data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data are input into the transient flow field monitoring network to obtain a transient flow field result of the scroll compressor; the transient flow field result of the scroll compressor, the corner data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data are input into the performance monitoring network to obtain a performance index result.

[0132] In the embodiments of the present application, a training data set considering the internal transient full three-dimensional flow field of a scroll compressor and the performance index training data set of the scroll compressor under a variable grid or a constant grid is established based on a sample library constructing a graph data structure representation of a training data set; a scroll compressor transient flow field and performance monitoring model is trained using the training data set, and the trained scroll compressor transient flow field and performance monitoring model can quickly predict the internal transient full three-dimensional flow field of the scroll compressor and the performance index of the scroll compressor under variable operating conditions, overcoming the slow response defect of the traditional CFD method; and the operating condition data of the scroll compressor collected by sparse sensors are used to quickly predict the internal transient full three-dimensional flow field of the scroll compressor and the performance index of the scroll compressor, overcoming the low spatial resolution defect of the traditional sensor measurement data, and being suitable for application in industrial application scenarios where dense measurement points are difficult to arrange.

[0133] The embodiment of the present application also provides a vortex compressor transient flow field and performance monitoring device based on graph data mining. Figure 4 As shown in the figure, Figure 4 It is a structural schematic diagram of a vortex compressor transient flow field and performance monitoring device based on graph data mining provided by the embodiment of the present application, and the device comprises:

[0134] The construction module 410 is configured to construct a training data set represented by a graph data structure based on different operating condition samples and flow field data corresponding to the operating condition samples, wherein each graph data structure representation in the training data set comprises global features representing operating conditions and performance indicators, local features representing node information of a transient flow field, and node connection relationships of the transient flow field.

[0135] The training module 420 is configured to train a vortex compressor transient flow field and performance monitoring model by using the training data set, so as to obtain a trained vortex compressor transient flow field and performance monitoring model, wherein the trained vortex compressor transient flow field and performance monitoring model has a mapping relationship between operating condition data and each node transient flow field, and a mapping relationship with performance indicators.

[0136] The input module 430 is configured to input collected operating condition data of a vortex compressor into the trained vortex compressor transient flow field and performance monitoring model, so as to obtain a transient flow field result and a performance indicator result of the vortex compressor.

[0137] In an optional embodiment, the construction module comprises:

[0138] The sample library construction module is configured to construct a sample library, wherein each sample in the sample library comprises an operating condition sample, a transient flow field sample, and a performance indicator sample.

[0139] The graph data construction module is configured to construct a graph data structure representation of each sample in the sample library, so as to obtain a training data set.

[0140] In an optional embodiment, the sample library construction module comprises:

[0141] The operating condition sample module is configured to obtain different operating condition samples based on an operating boundary condition range, wherein the operating condition sample comprises operating boundary conditions composed of suction pressure, suction temperature, discharge pressure, and rotating speed.

[0142] The transient flow field sample module is configured to perform fluid mechanics simulation analysis on internal flow field transient of a vortex compressor by using the operating condition sample, and take corner angle information, unit information, and node information in a simulation analysis result as a transient flow field sample.

[0143] A performance index sample module is configured to calculate performance indexes according to the simulation analysis result to obtain performance index samples, wherein the performance index samples include instantaneous suction mass flow, instantaneous discharge temperature, isentropic efficiency and volumetric efficiency of the scroll compressor.

[0144] In an alternative embodiment, the graph data construction module comprises:

[0145] A global feature module is configured to take the operating boundary condition in the operating condition sample, the rotation angle in the transient flow field sample and the performance index in the performance index sample as global features.

[0146] A local feature module is configured to take node information of each node in the transient flow field sample as local features, wherein the node information includes node three-dimensional coordinates and flow field.

[0147] A node relationship module is configured to determine node connection relationships according to the element information and the node information in the transient flow field sample.

[0148] In an alternative embodiment, the operating condition data includes rotation angle data and operating boundary condition data; and the input module comprises:

[0149] A data reading module is configured to find corresponding target graph data structure representations from the training data set according to the rotation angle data and the operating boundary condition data, and read corresponding node three-dimensional coordinate data and node connection relationship data from the target graph data structure representations.

[0150] A data input module is configured to input the rotation angle data, the operating boundary condition data, the node three-dimensional coordinate data and the node connection relationship data into the trained scroll compressor transient flow field and performance monitoring model to obtain the transient flow field and the performance index result of the scroll compressor.

[0151] In an alternative embodiment, the scroll compressor transient flow field and performance monitoring model comprises a transient flow field monitoring network and a performance monitoring network connected in series; and the data input module comprises:

[0152] A first input submodule is configured to input the rotation angle data, the operating boundary condition data, the node three-dimensional coordinate data and the node connection relationship data into the transient flow field monitoring network to obtain the transient flow field result of the scroll compressor.

[0153] A second input submodule is configured to input the transient flow field result of the scroll compressor, the rotation angle data, the operating boundary condition data, the node three-dimensional coordinate data and the node connection relationship data into the performance monitoring network to obtain the performance index result.

[0154] In an alternative embodiment, the transient flow field monitoring network comprises a first input layer, a first input encoding layer, a plurality of first graph attention layers, and a first output layer;

[0155] The first input sub-module is further configured to input the corner data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data through the first input layer, encode the corner data and the operating boundary condition data to each node through the first input encoding layer, and map the corner data, the operating boundary condition data, and the node three-dimensional coordinate data to a target range to obtain a first encoding result, perform feature extraction on the correlation between adjacent nodes in the first encoding result through the plurality of first graph attention layers to obtain a feature extraction result, and map the feature extraction result to a first target output dimension through the first output layer to obtain the transient flow field result of the scroll compressor.

[0156] In an alternative embodiment, the performance monitoring network comprises a second input layer, a second input encoding layer, a plurality of second graph attention layers, a plurality of hierarchical pooling layers, a plurality of global pooling layers, and a second output layer;

[0157] The second input sub-module is further configured to input the transient flow field result of the scroll compressor, the corner data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data through the second input layer, encode the transient flow field result of the scroll compressor, the corner data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data through the second input encoding layer to obtain a second encoding result, sequentially process the second encoding result through the plurality of second graph attention layers, the plurality of hierarchical pooling layers, and the plurality of global pooling layers to obtain a feature processing result, and map the feature processing result to a second target output dimension through the second output layer to obtain the performance index result.

[0158] The embodiments of the present application also provide an electronic device, referring to Figure 5 , Figure 5 is a structural schematic diagram of an electronic device provided by the embodiments of the present application. As shown in Figure 5 , the electronic device 500 comprises a memory 510 and a processor 520, the memory 510 and the processor 520 are communicatively connected through a bus, the memory 510 stores a computer program, the computer program can run on the processor 520, and further implements the steps of the scroll compressor transient flow field and performance monitoring method based on graph data mining.

[0159] The embodiment of the present application further provides a computer readable storage medium, which has a computer program stored thereon, and the computer program is executed by a processor to implement the steps of the scroll compressor transient flow field and performance monitoring method based on graph data mining.

[0160] The embodiment of the present application further provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement the steps of the scroll compressor transient flow field and performance monitoring method based on graph data mining.

[0161] Each of the embodiments in the specification is described in a progressive manner, and each embodiment focuses on the difference from other embodiments, and the same or similar parts between the embodiments can be referred to each other.

[0162] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the method and device according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal devices to produce a machine, so that the instructions executed by the computer or other programmable data processing terminal devices produce a device implemented in the flowcharts and / or block diagrams. Figure 1 The device that implements the function specified in one flow or multiple flows and / or blocks. Figure 1 The device that implements the function specified in one flow or multiple flows and / or blocks.

[0163] These computer program instructions can also be stored in a computer readable memory that can guide the computer or other programmable data processing terminal devices to work in a specific way, so that the instructions stored in the computer readable memory produce a product including instruction devices, which implement the flowcharts and / or block diagrams. Figure 1 The device that implements the function specified in one flow or multiple flows and / or blocks. Figure 1 The device that implements the function specified in one flow or multiple flows and / or blocks.

[0164] These computer program instructions can also be loaded into a computer or other programmable data processing terminal device, so that a series of operation steps are performed on the computer or other programmable terminal device to produce a computer implemented process, so that the instructions executed on the computer or other programmable terminal device provide steps for implementing the flowcharts and / or block diagrams. Figure 1 The device that implements the function specified in one flow or multiple flows and / or blocks. Figure 1 The device that implements the function specified in one flow or multiple flows and / or blocks.

[0165] While the preferred embodiments of the application have been described above, it will be recognized and understood that various modifications and changes can be made to the embodiments of the present application by those skilled in the art having the benefit of this detailed description without departing from the scope of the present application. It is therefore intended that the appended claims be construed to include all such embodiments and modifications as fall within the scope of the present application.

[0166] Finally, it should be noted that, in the specification, relational terms such as first and second, and the like, can be used solely to distinguish one entity or action from another entity or action without necessarily implying any actual relationship or order between such entities or actions. Also, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises a... " does not, without more constraints, exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the recited element.

[0167] The above describes in detail a vortex compressor transient flow field and performance monitoring method based on graph data mining provided by the present application. The principles and implementation manners of the present application are described by using specific examples. The above description of the embodiments is only used to help understand the method of the present application and its core idea. Meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation manners and application ranges can be changed. In summary, the content of the specification should not be understood as a limitation of the present application.

Claims

1. A method for monitoring the transient flow field and performance of a scroll compressor based on graph data mining, characterized in that, The method comprises: Based on different operating condition samples and the flow field data corresponding to the operating condition samples, a training data set represented by a graph data structure is constructed, each graph data structure representation in the training data set comprising: global features representing operating conditions and performance indicators, local features representing transient flow field node information, and node connection relationships of the transient flow field; A vortex compressor transient flow field and performance monitoring model is trained using the training data set to obtain a trained vortex compressor transient flow field and performance monitoring model, which has a mapping relationship between operating condition data and each node transient flow field, and a mapping relationship with performance indicators; The collected operating condition data of the vortex compressor is input into the trained vortex compressor transient flow field and performance monitoring model to obtain the transient flow field results and performance indicator results of the vortex compressor; The operating condition data of the vortex compressor collected is input into the trained vortex compressor transient flow field and performance monitoring model to obtain the transient flow field and performance indicator results of the vortex compressor, comprising: finding the corresponding target graph data structure representation from the training data set according to the rotation angle data and the operating boundary condition data, and reading the corresponding node three-dimensional coordinate data and node connection relationship data from the target graph data structure representation; the rotation angle data, the operating boundary condition data, the node three-dimensional coordinate data and the node connection relationship data are input into the trained vortex compressor transient flow field and performance monitoring model to obtain the transient flow field and performance indicator results of the vortex compressor; The vortex compressor transient flow field and performance monitoring model comprises a transient flow field monitoring network and a performance monitoring network connected in series; the transient flow field monitoring network comprises a first input layer, a first input encoding layer, a plurality of first graph attention layers and a first output layer; the performance monitoring network comprises a second input layer, a second input encoding layer, a plurality of second graph attention layers, a plurality of hierarchical pooling layers, a plurality of global pooling layers and a second output layer; The rotation angle data, the operating boundary condition data, the node three-dimensional coordinate data and the node connection relationship data are input into the trained vortex compressor transient flow field and performance monitoring model to obtain the transient flow field results and performance indicator results of the vortex compressor, comprising: inputting the rotation angle data, the operating boundary condition data, the node three-dimensional coordinate data and the node connection relationship data into the transient flow field monitoring network to obtain the transient flow field results of the vortex compressor; inputting the transient flow field results of the vortex compressor, the rotation angle data, the operating boundary condition data, the node three-dimensional coordinate data and the node connection relationship data into the performance monitoring network to obtain the performance indicator results.

2. The method of claim 1, wherein, Based on different operating condition samples and the flow field data corresponding to the operating condition samples, a training data set represented by a graph data structure is constructed, comprising: constructing a sample library, each sample in the sample library comprising an operating condition sample, a transient flow field sample, and a performance index sample; constructing a graph data structure representation of each sample in the sample library to obtain a training data set.

3. The method of claim 2, wherein, constructing a sample library, comprising: obtaining different operating condition samples based on an operating boundary condition range, the operating condition samples comprising operating boundary conditions composed of suction pressure, suction temperature, discharge pressure, and rotational speed; performing fluid mechanics simulation analysis on the internal flow field transient of the scroll compressor using the operating condition samples, and taking the turning angle, unit information, and node information in the simulation analysis result as the transient flow field sample; performing performance index calculation according to the simulation analysis result to obtain a performance index sample, the performance index sample comprising instantaneous suction mass flow rate, instantaneous discharge temperature, isentropic efficiency, and volumetric efficiency of the scroll compressor.

4. The method of claim 2, wherein, constructing a graph data structure representation of each sample in the sample library, comprising: taking the operating boundary conditions in the operating condition sample, the turning angle in the transient flow field sample, and the performance index in the performance index sample as global features; taking the node information of each node in the transient flow field sample as local features, the node information comprising node three-dimensional coordinates and flow field; determining node connection relationships according to the unit information and node information in the transient flow field sample.

5. The method of claim 1, wherein, inputting the turning angle data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data into the transient flow field monitoring network to obtain the transient flow field result of the scroll compressor, comprising: inputting the turning angle data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data through the first input layer; encoding the turning angle data and the operating boundary condition data to each node through the first input encoding layer, and mapping the turning angle data, the operating boundary condition data, and the node three-dimensional coordinate data to a target range to obtain a first encoding result; performing feature extraction on the correlation degree between adjacent nodes in the first encoding result through the plurality of first graph attention layers to obtain a feature extraction result; mapping the feature extraction result to a first target output dimension through the first output layer to obtain the transient flow field result of the scroll compressor.

6. The method of claim 1, wherein, inputting the transient flow field result of the scroll compressor, the turning angle data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data into the performance monitoring network to obtain the performance index result, comprising: inputting the transient flow field result of the scroll compressor, the turning angle data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data through the second input layer; encoding the transient flow field result of the scroll compressor, the turning angle data, the operating boundary condition data, the node three-dimensional coordinate data, and the node connection relationship data through the second input encoding layer to obtain a second encoding result; The second encoding result is processed by using the plurality of second graph attention layers, the plurality of hierarchical pooling layers and the plurality of global pooling layers in sequence to obtain a feature processing result. The feature processing result is mapped to a second target output dimension by the second output layer to obtain the performance index result.

7. The method according to any of claims 1 to 6, characterized in that The transient flow field is a variable-grid transient flow field or a constant-grid transient flow field.

8. A device for monitoring the transient flow field and performance of a scroll compressor based on graph data mining, characterized in that, The device comprises: The construction module is configured to construct a training data set represented by a graph data structure based on different operating condition samples and flow field data corresponding to the operating condition samples, each graph data structure representation in the training data set comprising global features representing operating conditions and performance indexes, local features representing node information of a transient flow field, and node connection relationships of the transient flow field; The training module is configured to train a vortex compressor transient flow field and performance monitoring model using the training data set to obtain a trained vortex compressor transient flow field and performance monitoring model, the trained vortex compressor transient flow field and performance monitoring model having a mapping relationship between operating condition data and each node transient flow field and a mapping relationship with performance indexes; The input module is configured to input collected operating condition data of a vortex compressor into the trained vortex compressor transient flow field and performance monitoring model to obtain a transient flow field result and a performance index result of the vortex compressor. The operating condition data includes rotation angle data and operating boundary condition data. The vortex compressor transient flow field and performance monitoring model comprises a transient flow field monitoring network and a performance monitoring network connected in series. The vortex compressor transient flow field and performance monitoring model comprises a transient flow field monitoring network and a performance monitoring network connected in series. inputting the rotation angle data, the operation boundary condition data, the node three-dimensional coordinate data and the node connection relationship data into the trained scroll compressor transient flow field and performance monitoring model to obtain transient flow field results and performance index results of the scroll compressor, including: inputting the rotation angle data, the operation boundary condition data, the node three-dimensional coordinate data and the node connection relationship data into the transient flow field monitoring network to obtain the transient flow field results of the scroll compressor; inputting the transient flow field results of the scroll compressor, the rotation angle data, the operation boundary condition data, the node three-dimensional coordinate data and the node connection relationship data into the performance monitoring network to obtain the performance index results.

Images