Motor multi-physical field agent model construction method, device and equipment and medium
By combining Latin hypercube sampling and secondary sampling of on-site operation data in the construction of motor multi-physics agent model, the problems of insufficient accuracy and high cost in the existing technology are solved, and efficient and accurate agent model construction is achieved.
Patent Information
- Application Number
- CN202411264509.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-10
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-09-10
AI Technical Summary
The existing sample space sampling method fails to make full use of on-site operation data in the construction of motor multi-physics agent model, resulting in insufficient accuracy of the constructed agent model, which increases the construction time and cost.
A regular sampling is performed using the Latin supercube sampling method, and secondary sampling is performed in combination with the on-site operating condition data. Through the root mean square error and average error judgment, the model is iterated repeatedly until the requirements are met, and a high-precision motor multi-physics agent is formed. Model.
The accuracy of sample space point acquisition is improved and the fit between on-site data is reduced, the error of multi-physics simulation is reduced, and the accuracy and efficiency of the proxy model is improved.
Smart Images

Figure CN119066879A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of model construction, and in particular to a method, device, equipment and medium for constructing a multi-physics field proxy model of a motor. Background Art
[0002] As the complexity of the model increases, the amount of calculation in multi-physics simulation continues to increase. In the field of digital twin intelligent operation and maintenance of hydropower and pumped storage unit motors, different methods need to be used to construct real-time simulation models for different models, so as to achieve dimensionality reduction or data fitting of numerical calculation orders, and realize real-time display of multi-physics fields of various components through visualization.
[0003] The surrogate model is an efficient mathematical approximation model. It has gradually developed into a new computing method because it can significantly improve computing efficiency. The sampling method studies how to construct a fitting function with the least error using the least sample points and can take into account the random error of the data, while the accuracy of the surrogate model depends on the distribution of the sample points.
[0004] Existing sample space sampling methods include orthogonal design, uniform design, Latin hypercube design and other methods. In the sampling process, the proxy model is often constructed through a single sampling. However, the distribution of a single sampling often has errors and uncertainties, and it is also impossible to optimize for areas with insufficient precision. Furthermore, the proxy model constructed after the initial sampling may not meet the needs of the site, so secondary sampling is required. Conventional secondary sampling is still based on the data of the Latin hypercube sampling points for addition, without combining it with the actual on-site operation data, and there is no accurate method to determine the most frequently used operating condition data for the newly added units. In this way, in the process of selecting the new sample space, it is not comprehensive to divide the sampling area based on experience without relying on on-site operation data, and it does not match the actual long-term operation interval of the power station, resulting in insufficient accuracy of the constructed proxy model, and it will also increase the time and cost of constructing the proxy model. Summary of the invention
[0005] In view of this, the present invention provides a method, device, equipment and medium for constructing a motor multi-physics field proxy model to solve the problem that in the process of selecting a new sample space, the sampling area is divided only by experience without relying on field operation data, which is incomplete and inconsistent with the actual long-term operation range of the power station, resulting in the constructed proxy model being insufficiently accurate, and the time and cost of constructing the proxy model will be increased.
[0006] In a first aspect, the present invention provides a method for constructing a multi-physics proxy model of a motor, the method comprising:
[0007] Obtain a field operating condition data set of the target motor; perform a sampling once in a preset sample three-dimensional space using the Latin hypercube sampling method and determine a first sample point set; establish an initial multi-physics proxy model of the motor based on the first sample point set after processing with a preset periodic boundary three-dimensional fluid-thermal coupling model; perform root mean square error and average error judgments on the initial multi-physics proxy model of the motor in turn; when the root mean square error meets the requirements and the average error does not meet the requirements, perform secondary sampling based on the field operating condition data set and determine a target additional sample point set; add the target additional sample point set to the first sample point set, and iterate the model training based on the added first sample point set until the root mean square error meets the requirements and the average error meets the requirements and a target proxy model of the motor multi-physics field is obtained.
[0008] The method for constructing a multi-physics proxy model of a motor provided by the present invention first performs a conventional sampling in a preset sample stereo space using the Latin hypercube sampling method, and then determines the newly added sample points in combination with the field operation condition data set and performs local secondary sampling, thereby improving the accuracy of the sample space point acquisition and the degree of fit with the field data. Finally, the model is repeatedly trained until the root mean square error meets the requirements and the average error meets the requirements to obtain the final trained multi-physics target proxy model of the motor. By adding new field data sample points for the accuracy improvement of the proxy model, the error of the data set constructed in the early stage of the proxy model is minimized and avoided, while increasing the fit with the field operation and reducing the error caused by the multi-physics simulation.
[0009] In an optional implementation, based on the first sample point set, after being processed by a preset periodic boundary three-dimensional fluid-thermal coupling model, an initial proxy model of the motor multi-physics field is established, including:
[0010] Based on the first sample point set, numerical simulation is performed using a preset periodic boundary three-dimensional fluid-thermal coupling model to obtain a first sample data set; and an initial proxy model of the motor multi-physics field is established using the first sample data set.
[0011] The method for constructing a motor multi-physics proxy model provided by the present invention can obtain a first sample data set corresponding to a first sample point set by performing numerical simulation on a preset periodic boundary three-dimensional fluid-thermal coupling model, and then construct a corresponding motor multi-physics initial proxy model based on the first sample data set.
[0012] In an optional implementation, the root mean square error and the average error of the initial proxy model of the motor multi-physics field are judged in sequence, including:
[0013] The root mean square error of the initial proxy model of the motor multi-physics field is judged; when the root mean square error does not meet the requirements, the model parameters are adjusted and the training is repeatedly iterated based on the adjusted model parameters until the root mean square error meets the requirements, and the average error of the initial proxy model of the motor multi-physics field is judged.
[0014] The motor multi-physics field proxy model construction method provided by the present invention can improve the model accuracy by adjusting the model parameters so that the root mean square error meets the requirement if the root mean square error does not meet the requirement.
[0015] In an optional implementation, after successively performing root mean square error and average error judgment on the motor multi-physics field initial proxy model, the method further includes:
[0016] When the root mean square error meets the requirements and the average error meets the requirements, the motor multi-physics field initial proxy model is determined as the motor multi-physics field target proxy model.
[0017] In an optional implementation, when the root mean square error meets the requirement and the average error does not meet the requirement, secondary sampling is performed based on the field operation condition data set and a target additional sample point set is determined, including:
[0018] When the root mean square error meets the requirements and the average error does not meet the requirements, the on-site initial operating range data set is determined based on the on-site operating condition data set; based on the on-site initial operating range data set, after processing by the empirical distribution function method, the initial newly added sample point set is determined; based on the initial newly added sample point set and the first sample point set, after processing by the metric space method, the target newly added sample point set is determined.
[0019] The motor multi-physics field proxy model construction method provided by the present invention, if the root mean square error of the constructed motor multi-physics field initial proxy model meets the requirements and the average error does not meet the requirements, then the new sample points are determined in combination with the field operating condition data set and local secondary sampling is performed, thereby improving the accuracy of sample space point acquisition and the degree of fit with the field data. Further, in combination with the first sample point set after the new sample points are added, the duplicate points of the two samples can be removed through the metric space method and the final target new sample point set can be determined, thereby improving the accuracy of sample space point acquisition and the degree of fit with the field data.
[0020] In an optional implementation, based on the on-site initial operating range data set, the initial additional sample point set is determined by processing with an empirical distribution function method, including:
[0021] The data in the on-site initial operating range data set are sorted to obtain the on-site target operating range data set; based on the on-site target operating range data set, multiple cumulative proportion values are obtained through empirical distribution function calculation; and the initial newly added sample point set is determined based on the multiple cumulative proportion values.
[0022] In a second aspect, the present invention provides a device for constructing a multi-physics field proxy model of a motor, the device comprising:
[0023] An acquisition module is used to acquire a field operating condition data set of a target motor; a first sampling determination module is used to perform a sampling in a preset sample three-dimensional space using a Latin hypercube sampling method and determine a first sample point set; a processing module is used to establish an initial multi-physics proxy model of the motor based on the first sample point set after processing by a preset periodic boundary three-dimensional fluid-thermal coupling model; a judgment module is used to perform root mean square error and average error judgments on the initial multi-physics proxy model of the motor in turn; a second sampling determination module is used to perform secondary sampling based on the field operating condition data set and determine a target additional sample point set when the root mean square error meets the requirements and the average error does not meet the requirements; a repetition module is used to add the target additional sample point set to the first sample point set, and iterate the model training according to the added first sample point set until the root mean square error meets the requirements and the average error meets the requirements and the target multi-physics proxy model of the motor is obtained.
[0024] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the method for constructing a multi-physics proxy model of a motor according to the first aspect or any corresponding embodiment thereof by executing the computer instructions.
[0025] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the method for constructing a motor multi-physics field proxy model according to the first aspect or any corresponding embodiment thereof.
[0026] In a fifth aspect, the present invention provides a computer program product, comprising computer instructions for causing a computer to execute the method for constructing a motor multi-physics proxy model according to the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0028] Figure 1 is a flow chart of a method for constructing a multi-physics proxy model of a motor according to an embodiment of the present invention;
[0029] Figure 2 is a flow chart of another method for constructing a multi-physics proxy model of a motor according to an embodiment of the present invention;
[0030] Figure 3 is a flow chart of another method for constructing a multi-physics proxy model of a motor according to an embodiment of the present invention;
[0031] Figure 4 is a flow chart of a method for constructing a multi-physics proxy model of a motor based on data sampling according to an embodiment of the present invention;
[0032] Figure 5 is a schematic diagram of a process for determining a newly added sample point according to an embodiment of the present invention;
[0033] Figure 6 is a structural block diagram of a device for constructing a multi-physics proxy model of a motor according to an embodiment of the present invention;
[0034] Figure 7 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION
[0035] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.
[0036] The embodiment of the present invention provides a method for constructing a multi-physics proxy model of a motor, which improves the accuracy of sample space point acquisition and the degree of fit with field data through a regular sampling and local secondary sampling. Furthermore, by adding field data sample points to improve the accuracy of the proxy model, the error of the data set constructed in the early stage of the proxy model is minimized, while increasing the fit with field operation and reducing the error caused by multi-physics simulation.
[0037] According to an embodiment of the present invention, an embodiment of a method for constructing a multi-physics proxy model of a motor is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.
[0038] In this embodiment, a method for constructing a multi-physics proxy model of a motor is provided, which can be used in electronic devices such as computers, mobile phones, tablet computers, etc. Figure 1 is a flowchart of a method for constructing a motor multi-physics field proxy model according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:
[0039] Step S101, obtaining a field operating condition data set of a target motor.
[0040] Specifically, the on-site operating condition initial data set may include a plurality of on-site operating condition data of the target motor in a normal operating state.
[0041] In this embodiment, the construction of the stator winding and core temperature field proxy model of the pumped storage motor is taken as an example, the inlet wind speed V, air temperature T1, stator winding heat source P1, and stator core heat source P2 are selected as independent variables, and the stator winding average temperature T2 and the stator core average temperature T3 are selected as dependent variables to form an initial data set of field operating conditions.
[0042] The stator winding heat source P1 can be calculated using the following equation (1):
[0043] P1=P copper +P stray (1)
[0044] Where: P copper represents the copper loss (unit: watt, W), which is calculated by the following relations (2) and (3):
[0045]
[0046]
[0047] Where: I rns Indicates the effective value of current (unit: ampere, A); R coil represents the winding resistance (unit: ohm, Ω); ρ represents the resistivity of the wire (unit: ohm·meter, Ω·m), which is related to the material and temperature; l represents the length of the wire (unit: meter, m); A represents the cross-sectional area of the wire (unit: square meter, m 2 ).
[0048] Further, P stray represents the stray loss (unit: watt, W), which is calculated by the following relations (4) to (6):
[0049]
[0050] B=μ·H (5)
[0051]
[0052] Where: k represents a proportional constant, which depends on the material and structure of the winding; B max represents the maximum value of magnetic flux density (unit: Tesla, T); f represents the operating frequency (unit: Hertz, Hz); V coil Indicates the winding volume (unit: cubic meter, m 3 ), B represents magnetic flux density, μ represents magnetic permeability (unit: Henry / meter, H / m), H represents magnetic field strength, N represents the number of turns of the coil, and I represents the effective length of the coil.
[0053] Furthermore, the stator core heat source P2 can be calculated using the following relationship (7):
[0054] P2=P h +P e (7)
[0055] Where: P h represents the hysteresis loss (unit: watt, W), as shown in the following equation (8); P e represents the eddy current loss (unit: Watt, W), as shown in the following equation (9):
[0056]
[0057]
[0058] Where: k1 represents the hysteresis loss coefficient of the material (unit: W / kg); V represents the volume of the core (unit: cubic meter, m 3 ), σ represents the electrical conductivity of the material (unit: Siemens / meter, S / m), and t represents the thickness of the core.
[0059] Step S102: In a preset sample stereo space, a Latin hypercube sampling method is used to perform a sampling operation and determine a first sample point set.
[0060] The preset sample three-dimensional space represents a global sample three-dimensional space composed of a data range set according to experience.
[0061] Specifically, Latin hypercube sampling is performed in the preset sample three-dimensional space to collect N sample point data, namely, N first sample points, and form a first sample point set.
[0062] In one example, taking the construction of the pumped storage motor stator winding and core temperature field proxy model as an example, according to experience, the data ranges of the three variables of wind speed, temperature, and current are set [V x ,V y ]、[T x ,T y ]、[I x ,Iy ] and establish the corresponding preset sample three-dimensional space.
[0063] Step S103 , based on the first sample point set, after being processed by a preset periodic boundary three-dimensional fluid-thermal coupling model, an initial proxy model of the motor multi-physics field is established.
[0064] Specifically, the preset periodic boundary three-dimensional fluid-thermal coupling model can perform numerical simulation for different operating conditions. Taking the construction of the proxy model of the stator winding and iron core temperature field of the pumped storage motor as an example, the preset periodic boundary three-dimensional fluid-thermal coupling model is a periodic boundary three-dimensional fluid-thermal coupling model based on the stator winding and iron core.
[0065] Specifically, a preset periodic boundary three-dimensional fluid-thermal coupling model is used to perform numerical simulation on a plurality of sample points included in the first sample point set and establish a corresponding motor multi-physics field initial proxy model.
[0066] Step S104 , performing root mean square error and average error judgment on the initial proxy model of the motor multi-physics field in turn.
[0067] Among them, the root mean square error (RMSE) and mean error (ME) are important indicators for evaluating the accuracy and reliability of the model.
[0068] Specifically, the accuracy and reliability of the model can be ensured by successively determining the root mean square error and the average error.
[0069] Step S105: when the root mean square error meets the requirement and the average error does not meet the requirement, secondary sampling is performed based on the field operating condition data set and a target additional sample point set is determined.
[0070] Specifically, if the root mean square error of the initial proxy model of the motor multi-physics field meets the requirements, but the average error does not meet the requirements, it is necessary to optimize the model by adding new sample points.
[0071] In this embodiment, secondary sampling can be performed through the field operation condition data set, which improves the accuracy of sample space point acquisition and the fit with field data. Furthermore, the final target new sample point set is determined through sampling and used for subsequent precision improvement of the proxy model, which minimizes and avoids the error of the data set constructed in the early stage of the proxy model, while increasing the fit with field operation and reducing the error caused by multi-physics field simulation.
[0072] Step S106, adding the target newly added sample point set to the first sample point set, and iterating the model training according to the added first sample point set until the root mean square error meets the requirements and the average error meets the requirements and the motor multi-physics field target proxy model is obtained.
[0073] Specifically, the determined target additional sample point set is added to the first sample point set to form a new sample point set. Furthermore, based on the obtained new sample point set, steps S103 to S105 are repeated to perform iterative model training until the root mean square error of the model meets the requirements and the average error meets the requirements to obtain the corresponding trained motor multi-physics field target proxy model.
[0074] The method for constructing a multi-physics proxy model for a motor provided in this embodiment first performs a conventional sampling in a preset sample stereo space using the Latin hypercube sampling method, and then determines the newly added sample points in combination with the field operation condition data set and performs local secondary sampling, thereby improving the accuracy of the sample space point acquisition and the degree of fit with the field data. Finally, the model is repeatedly trained until the root mean square error meets the requirements and the average error meets the requirements to obtain the final trained multi-physics target proxy model for the motor. By adding new field data sample points to improve the accuracy of the proxy model, the error of the data set constructed in the early stage of the proxy model is minimized and avoided, while increasing the fit with the field operation and reducing the error caused by multi-physics simulation.
[0075] In this embodiment, a method for constructing a multi-physics proxy model of a motor is provided, which can be used in electronic devices such as computers, mobile phones, tablet computers, etc. Figure 2 is a flowchart of a method for constructing a motor multi-physics field proxy model according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:
[0076] Step S201, obtain the on-site operating condition data set of the target motor. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.
[0077] Step S202: In the preset sample stereo space, a Latin hypercube sampling method is used to perform a sampling and determine a first sample point set. Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.
[0078] Step S203 , based on the first sample point set, after being processed by a preset periodic boundary three-dimensional fluid-thermal coupling model, an initial proxy model of the motor multi-physics field is established.
[0079] Specifically, the above step S203 includes:
[0080] Step S2031: Based on the first sample point set, numerical simulation is performed using a preset periodic boundary three-dimensional fluid-thermal coupling model to obtain a first sample data set.
[0081] Specifically, taking the construction of the temperature field proxy model of the stator winding and core of the pumped storage motor as an example, for the m operating points corresponding to the target sample point set, based on the periodic boundary three-dimensional fluid-heat coupling model of the stator winding and core, commercial software is used to realize the numerical simulation of the operating points, and the average temperature of the stator winding and stator core under different operating conditions is generated, and the corresponding target data set A′ is generated, as shown in the following relationship (10):
[0082]
[0083] Step S2032: using the first sample data set to establish an initial proxy model of the motor multi-physics field.
[0084] First, a convolutional neural network (CNN) is selected as the model training network.
[0085] Secondly, the number of network layers q and the number of neurons in each layer w are determined through experiments, the activation function Sigmoid is selected, the loss function mean square error (MSE) is determined, and the Adam optimization algorithm is selected to update the network weights.
[0086] Finally, the neural network is trained using the first sample data set, the network parameters are adjusted to minimize the loss function, and a trained initial proxy model of the motor multiphysics field is obtained.
[0087] Step S204 , performing root mean square error and average error judgment on the initial proxy model of the motor multi-physics field in turn.
[0088] Specifically, the above step S204 includes:
[0089] Step S2041, performing root mean square error judgment on the initial proxy model of the motor multi-physics field.
[0090] Specifically, determine whether the root mean square error (MSE) of the initial proxy model of the motor multi-physics field meets the requirements.
[0091] Step S2042, when the root mean square error does not meet the requirements, adjust the model parameters and iterate the training based on the adjusted model parameters until the root mean square error meets the requirements, and perform average error judgment on the motor multi-physics field initial proxy model.
[0092] Specifically, if the root mean square error does not meet the requirements, the network model parameters can be adjusted by adding regularization terms, using Dropout, adjusting the learning rate, etc., and step S2032 can be repeated based on the adjusted network model parameters until the root mean square error meets the requirements.
[0093] Furthermore, when the root mean square error meets the requirements, the secondary error judgment of the initial proxy model of the motor multi-physics field is continued, that is, the average error judgment is performed. The specific judgment process is as follows:
[0094] Randomly select n on-site working conditions and the corresponding output temperature values, and determine the output temperature value Y of the selected sample point n and the surrogate model predicted value Y ′ The average error value of n ave , whether it is 10% lower than the specified efficiency value.
[0095] If yes, it means the average error meets the requirement; if no, it means the average error does not meet the requirement.
[0096] Among them, the average error value ρ ave It can be calculated by the following relations (11) and (12):
[0097]
[0098] Step S205: When the root mean square error meets the requirement but the average error does not meet the requirement, secondary sampling is performed based on the field operation condition data set and the target additional sample point set is determined. Figure 1 Step S105 of the illustrated embodiment will not be described in detail here.
[0099] Step S206, add the target new sample point set to the first sample point set, and iterate the model training according to the added first sample point set until the root mean square error meets the requirements and the average error meets the requirements and the motor multi-physics field target proxy model is obtained. For details, please refer to Figure 1 Step S106 of the illustrated embodiment will not be described in detail here.
[0100] Step S207 , when the root mean square error meets the requirement and the average error meets the requirement, the motor multi-physics field initial proxy model is determined as the motor multi-physics field target proxy model.
[0101] Specifically, through judgment, if the root mean square error and the average error both meet the requirements, the established motor multi-physics field initial proxy model is directly used as the final motor multi-physics field target proxy model.
[0102] The method for constructing a multi-physics proxy model of a motor provided in this embodiment first performs a conventional sampling in a preset sample stereoscopic space using the Latin hypercube sampling method, and performs numerical simulation through a preset periodic boundary three-dimensional fluid-heat coupling model to obtain a first sample data set corresponding to a first sample point set, and then the corresponding motor multi-physics initial proxy model can be constructed based on the first sample data set. Further, if the root mean square error of the initial proxy model of the motor multi-physics field meets the requirements and the average error does not meet the requirements, the additional sample points are determined in combination with the field operation condition data set and local secondary sampling is performed, and the model is repeatedly trained until the root mean square error meets the requirements and the average error meets the requirements to obtain the final trained motor multi-physics target proxy model, which improves the accuracy of sample space point acquisition and the fit with the field data. At the same time, by adding new field data sample points for the precision improvement of the proxy model, the error of the data set constructed in the early stage of the proxy model is minimized, while the fit with the field operation is increased, and the error caused by the multi-physics simulation is reduced.
[0103] In this embodiment, a method for constructing a multi-physics proxy model of a motor is provided, which can be used in electronic devices such as computers, mobile phones, tablet computers, etc. Figure 3 is a flowchart of a method for constructing a motor multi-physics field proxy model according to an embodiment of the present invention. Figure 3 As shown, the process includes the following steps:
[0104] Step S301, obtain the on-site operating condition data set of the target motor. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.
[0105] Step S302: In the preset sample stereo space, a Latin hypercube sampling method is used to perform a sampling and determine a first sample point set. Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.
[0106] Step S303: Based on the first sample point set, after being processed by a preset periodic boundary three-dimensional fluid-thermal coupling model, an initial proxy model of the motor multi-physics field is established. Figure 2 Step S203 of the illustrated embodiment will not be described in detail here.
[0107] Step S304, perform root mean square error and average error judgment on the initial proxy model of the motor multi-physics field in turn. Figure 2 Step S204 of the illustrated embodiment will not be described in detail here.
[0108] Step S305: When the root mean square error meets the requirement and the average error does not meet the requirement, secondary sampling is performed based on the field operating condition data set and a target additional sample point set is determined.
[0109] Specifically, the above step S305 includes:
[0110] Step S3051: when the root mean square error meets the requirement and the average error does not meet the requirement, determine the on-site initial operation range data set based on the on-site operation condition data set.
[0111] Specifically, when the root mean square error meets the requirements and the average error does not meet the requirements, the initial field operation range data set can be determined by the following steps:
[0112] (1) Data on wind speed, temperature, and current are usually recorded in a time series format, which includes a timestamp and the corresponding temperature reading.
[0113] (2) The corresponding physical quantity values are extracted every 10 minutes as a node, and sampling is performed at equal time intervals to form a field operation range data set.
[0114] Step S3052, based on the on-site initial operating range data set, the initial newly added sample point set is determined by processing with the empirical distribution function method.
[0115] Among them, the empirical distribution function method is a statistical method used to estimate the probability distribution function of sample data. This method is based on the observation of sample data and constructs a cumulative distribution function by calculating the sum of the probability density of each data point in the sample data. This function is called the empirical distribution function (ECDF). The empirical distribution function is a non-parametric estimation method that is applicable to sample data of any distribution form and does not require any assumptions about the data. It is directly calculated through the observation of sample data and is an estimate of the true distribution function.
[0116] In some optional implementations, the above step S3052 includes:
[0117] Step a1, sorting the data in the on-site initial operating range data set to obtain the on-site target operating range data set.
[0118] Step a2, based on the on-site target operating range data set, multiple cumulative proportion values are obtained through empirical distribution function calculation.
[0119] Step a3: determining an initial set of newly added sample points based on multiple cumulative ratio values.
[0120] First, the data in the field initial operation range data set are sorted in ascending order.
[0121] Second, for each data point x i , calculates less than or equal to xi The ratio of the number of data points to the total number of data points is defined as the empirical distribution function, as shown in the following equation (13):
[0122]
[0123] Then, calculate F of adjacent data points n (x)Difference.
[0124] Finally, select data point F n (x) The top three points with the highest difference are taken as new sample points, and different parameter quantities are freely combined to form three groups of new sample points, which form the initial set of new sample points.
[0125] Step S3053: Based on the initial newly added sample point set and the first sample point set, a target newly added sample point set is determined by processing using a metric space method.
[0126] Among them, the metric space defines a distance function on a set, so that the distance between any two elements in the set can be calculated through this distance function.
[0127] Specifically, the metric space method is used to determine duplicate points. Define a distance metric (such as Euclidean distance) d(p,q), as shown in the following relationship (14):
[0128]
[0129] Where: p and q represent two points in three-dimensional space respectively.
[0130] Furthermore, a threshold is set. If the distance between the new sample in the initial new sample point set and any sample point in the known first sample point set is less than the threshold, they are considered to be duplicates. At this time, a point in the initial new sample point set obtained by Latin hypercube sampling is deleted, and the final target new sample point set N is obtained. n .
[0131] Further, we get the final target set of newly added sample points m = N + N n .
[0132] Step S306, add the target new sample point set to the first sample point set, and iterate the model training according to the added first sample point set until the root mean square error meets the requirements and the average error meets the requirements and the motor multi-physics field target proxy model is obtained. For details, please refer to Figure 1 Step S106 of the illustrated embodiment will not be described in detail here.
[0133] The method for constructing a multi-physics proxy model for a motor provided in this embodiment first performs a conventional sampling in a preset sample stereoscopic space using the Latin hypercube sampling method, and establishes an initial proxy model for the multi-physics of the motor. If the root mean square error of the constructed initial proxy model for the multi-physics of the motor meets the requirements and the average error does not meet the requirements, then the new sample points are determined in combination with the field operating condition data set and local secondary sampling is performed, thereby improving the accuracy of the sample space point acquisition and the degree of fit with the field data. Furthermore, in combination with the first sample point set after the new sample points are added, the duplicate points of the two samples can be removed through the metric space method and the final target new sample point set can be determined, thereby improving the accuracy of the sample space point acquisition and the degree of fit with the field data.
[0134] In one example, a method for constructing a multi-physics proxy model of a motor based on data sampling is provided. The specific process is as follows: Figure 4 As shown in the figure. The two error requirement judgments are the root mean square error and the average error judgment respectively; the process of determining the new sample points is as follows: Figure 5 shown.
[0135] The method for constructing a multi-physics proxy model of a motor based on data sampling provided in this example has the following features:
[0136] Beneficial effects:
[0137] 1. Improve the accuracy of proxy model datasets:
[0138] The most common operating conditions of the unit are determined through on-site data. Through a routine sampling and a secondary sampling of local key conditions, and the removal of duplicate points, the accuracy of sample space point collection and its fit with on-site data are improved.
[0139] 2. Enhance the accuracy of proxy models and achieve adaptive optimization:
[0140] By adjusting the proxy model parameters and adding new field data sample points, the accuracy of the proxy model is improved, minimizing the errors in the data set constructed in the early stage of the proxy model, while increasing the fit with field operations and reducing the errors caused by multi-physics field simulation.
[0141] Therefore, through the technical solution of this method, the construction of the unit equipment proxy model becomes more efficient and accurate. The beneficial effects of this method are not only reflected in improving work efficiency and accuracy, but these beneficial effects will promote the intelligent and digital process in the field of hydropower and pumped storage engineering, and provide support for the innovation and development of engineering management.
[0142] In this embodiment, a motor multi-physics proxy model construction device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.
[0143] This embodiment provides a motor multi-physics field proxy model construction device, such as Figure 6 As shown, the device comprises:
[0144] The acquisition module 601 is used to acquire a field operating condition data set of a target motor.
[0145] The first sampling determination module 602 is used to perform a sampling in a preset sample three-dimensional space by using a Latin hypercube sampling method and determine a first sample point set.
[0146] The processing module 603 is used to establish an initial proxy model of the motor multi-physics field based on the first sample point set after processing by a preset periodic boundary three-dimensional fluid-thermal coupling model.
[0147] The judgment module 604 is used to judge the root mean square error and the average error of the initial proxy model of the motor multi-physics field in turn.
[0148] The second sampling determination module 605 is used to perform secondary sampling based on the field operating condition data set and determine a target additional sample point set when the root mean square error meets the requirement and the average error does not meet the requirement.
[0149] The repetition module 606 is used to add the target new sample point set to the first sample point set, and iterate the model training according to the added first sample point set until the root mean square error meets the requirements and the average error meets the requirements and the motor multi-physics field target proxy model is obtained.
[0150] In some optional implementations, the processing module 603 includes:
[0151] The simulation submodule is used to perform numerical simulation based on the first sample point set using a preset periodic boundary three-dimensional fluid-thermal coupling model to obtain a first sample data set.
[0152] A submodule is established, which is used to establish an initial proxy model of the motor multi-physics field using the first sample data set.
[0153] In some optional implementations, the determination module 604 includes:
[0154] The first judgment submodule is used to judge the root mean square error of the initial proxy model of the motor multi-physics field.
[0155] The adjustment submodule is used to adjust the model parameters when the root mean square error does not meet the requirements and repeatedly iterate the training based on the adjusted model parameters until the root mean square error meets the requirements, and to make an average error judgment on the initial proxy model of the motor multi-physics field.
[0156] In some optional embodiments, the device further comprises:
[0157] The determination module is used to determine the motor multi-physics field initial proxy model as the motor multi-physics field target proxy model when the root mean square error meets the requirements and the average error meets the requirements.
[0158] In some optional implementations, the second sampling determination module 605 includes:
[0159] The determination submodule is used to determine the field initial operation range data set based on the field operation condition data set when the root mean square error meets the requirement and the average error does not meet the requirement.
[0160] The first processing submodule is used to determine an initial set of newly added sample points based on an initial on-site operating range data set through processing using an empirical distribution function method.
[0161] The second processing submodule is used to determine a target new sample point set based on the initial new sample point set and the first sample point set through metric space method processing.
[0162] In some optional implementations, the first processing submodule includes:
[0163] The sorting unit is used to sort the data in the on-site initial operating range data set to obtain the on-site target operating range data set.
[0164] The calculation unit is used to obtain a plurality of cumulative proportion values through empirical distribution function calculation based on the on-site target operation range data set.
[0165] The determination unit is used to determine an initial set of newly added sample points based on a plurality of cumulative proportion values.
[0166] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.
[0167] The motor multi-physics field proxy model building device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.
[0168] The embodiment of the present invention also provides a computer device having the above Figure 6 The apparatus for building a multiphysics proxy model of an electric motor is shown.
[0169] See also Figure 7 , Figure 7 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 7 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 7 A processor 10 is taken as an example.
[0170] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.
[0171] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.
[0172] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0173] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.
[0174] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.
[0175] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.
[0176] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.
[0177] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.
Claims
1. A method for constructing a multi-physics proxy model of a motor, characterized in that: The method comprises: Obtaining the on-site operating condition data set of the target motor; In the preset sample three-dimensional space, a Latin hypercube sampling method is used to perform a sampling and determine a first sample point set; Based on the first sample point set, an initial proxy model of the motor multi-physics field is established after being processed by a preset periodic boundary three-dimensional fluid-thermal coupling model; Performing root mean square error and average error judgment on the motor multi-physics field initial proxy model in turn; When the root mean square error meets the requirements but the average error does not meet the requirements, secondary sampling is performed based on the field operation condition data set and the target additional sample point set is determined; The target newly added sample point set is added to the first sample point set, and model training is iterated repeatedly according to the added first sample point set until the root mean square error meets the requirements and the average error meets the requirements and the motor multi-physics field target proxy model is obtained.
2. The method according to claim 1, characterized in that Based on the first sample point set, after being processed by a preset periodic boundary three-dimensional fluid-thermal coupling model, an initial proxy model of the motor multi-physics field is established, including: Based on the first sample point set, numerical simulation is performed using the preset periodic boundary three-dimensional fluid-thermal coupling model to obtain a first sample data set; The motor multi-physics initial proxy model is established using the first sample data set.
3. The method according to claim 1, characterized in that The root mean square error and average error of the motor multi-physics field initial proxy model are judged in turn, including: Performing root mean square error judgment on the motor multi-physics field initial proxy model; When the root mean square error does not meet the requirements, the model parameters are adjusted and the training is iterated repeatedly based on the adjusted model parameters until the root mean square error meets the requirements, and an average error judgment is performed on the motor multi-physics field initial proxy model.
4. The method according to claim 3, characterized in that After successively performing root mean square error and average error judgment on the motor multi-physics field initial proxy model, the method further includes: When the root mean square error meets the requirement and the average error meets the requirement, the motor multi-physics field initial proxy model is determined as the motor multi-physics field target proxy model.
5. The method according to claim 1, characterized in that When the root mean square error meets the requirements but the average error does not meet the requirements, secondary sampling is performed based on the field operation condition data set and the target additional sample point set is determined, including: When the root mean square error meets the requirement and the average error does not meet the requirement, determining the on-site initial operation range data set based on the on-site operation condition data set; Based on the on-site initial operation range data set, an initial additional sample point set is determined by processing with an empirical distribution function method; Based on the initial newly added sample point set and the first sample point set, the target newly added sample point set is determined through metric space method processing.
6. The method according to claim 5, characterized in that Based on the initial operating range data set of the site, the initial additional sample point set is determined through processing by the empirical distribution function method, including: Sorting the data in the on-site initial operating range data set to obtain an on-site target operating range data set; Based on the on-site target operating range data set, a plurality of cumulative proportion values are obtained through empirical distribution function calculation; The initial newly added sample point set is determined based on the multiple accumulated proportion values.
7. A motor multi-physics field proxy model construction device, characterized in that: The device comprises: An acquisition module, used to acquire a field operating condition data set of a target motor; A first sampling determination module is used to perform a sampling in a preset sample three-dimensional space using a Latin hypercube sampling method and determine a first sample point set; A processing module, used for establishing an initial proxy model of the motor multi-physics field based on the first sample point set and processed by a preset periodic boundary three-dimensional fluid-thermal coupling model; A judgment module, used for sequentially judging the root mean square error and the average error of the motor multi-physics field initial proxy model; The second sampling determination module is used to perform secondary sampling based on the field operation condition data set and determine the target additional sample point set when the root mean square error meets the requirement and the average error does not meet the requirement; A repetitive module is used to add the target newly added sample point set to the first sample point set, and iterate the model training according to the added first sample point set until the root mean square error meets the requirements and the average error meets the requirements and the motor multi-physics field target proxy model is obtained.
8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the motor multi-physics field proxy model construction method according to any one of claims 1 to 6 by executing the computer instructions.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the motor multi-physics field proxy model construction method according to any one of claims 1 to 6.
10. A computer program product, characterized in that The method comprises computer instructions, wherein the computer instructions are used to cause a computer to execute the method for constructing a multi-physics proxy model of a motor according to any one of claims 1 to 6.
Citation Information
Patent Citations
Proxy model optimization method and device based on multipoint adding, and computer equipment
CN112749495A
Proxy model optimization method suitable for CFD uncertainty quantification and related equipment
CN114757131A
Blade temperature calculation method under reversing turbine moving blade reversing state
CN116167198A
Construction method of reduced order model for calculating internal temperature field of electrical equipment
CN116522734A
Geothermal reservoir fracture network feature intelligent inversion method, device and equipment and storage medium
CN118171571A
Cited By
Motor multi-physical field agent model construction method and related device
CN120012617A
A method for constructing a multi-physics proxy model of a motor and related devices
CN120012617B
Method and device for constructing digital real-time simulation model of generator stator
CN120068668A
