Zero offset determination method and device, equipment, computer readable storage medium

Abstract

Embodiments of the present application disclose a zero deviation determination method and device, equipment and computer readable storage medium. The method comprises: inputting relevant parameters of a motor and a current time resolver reading value into a trained forward calculation network, so that the trained forward calculation network outputs a resolver zero deviation value; wherein the relevant parameters comprise a rotor flux value and a stator winding temperature; the trained forward calculation network comprises a first neural network, and the first neural network is used for correcting the rotor flux value according to the stator winding temperature to obtain a corrected rotor flux value. The present application only needs to input the relevant parameters of the motor and the current time resolver reading value into the trained forward calculation network, without the need for a complicated manual detection calibration process, eliminating manual errors, thereby improving the accuracy of the resolver zero deviation value.

Description

Technical Field

[0001] This application relates to the field of vehicles, and specifically to a method, apparatus, device, and computer-readable storage medium for determining zero-position deviation. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs) are widely used in the drive systems of hybrid and pure electric vehicles due to their excellent performance. The speed detection and control of PMSMs are closely related to the rotor position. The deviation between the rotor's rotary transformer during installation and the actual rotor position is called the zero-position deviation. The rotor's mechanical angle zero-position deviation increases exponentially when converted to electrical angles. In the high-performance control of PMSMs, the detection accuracy of the rotary transformer's zero-position deviation directly affects the PMSM's performance, efficiency, and other indicators.

[0003] Currently, the method for determining the zero-point deviation of a rotary transformer requires manual calibration and testing, which is a rather cumbersome process and may result in human error, leading to inaccurate determination of the zero-point deviation of the rotary transformer. Summary of the Invention

[0004] To address the aforementioned technical problems, embodiments of this application provide a method, apparatus, device, and computer-readable storage medium for determining zero-position deviation, which facilitates and accurately determines the zero-position deviation of a rotary transformer.

[0005] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part from practice of this application.

[0006] According to one aspect of the embodiments of this application, a method for determining zero-position deviation is provided, comprising: inputting relevant parameters of a motor and the current reading value of a rotary transformer into a trained forward computing network, so that the trained forward computing network outputs the zero-position deviation value of the rotary transformer; wherein, the relevant parameters include a rotor flux linkage value and a stator winding temperature; the trained forward computing network includes a first neural network, the first neural network being used to correct the rotor flux linkage value according to the stator winding temperature to obtain a corrected rotor flux linkage value.

[0007] According to one aspect of the embodiments of this application, a zero-point deviation determination device is provided, comprising: a determination module configured to input relevant parameters of a motor and the current reading value of a rotary transformer into a trained forward computing network, so that the trained forward computing network outputs the zero-point deviation value of the rotary transformer; wherein the relevant parameters include a rotor flux linkage value and a stator winding temperature; the trained forward computing network includes a first neural network; the first neural network is used to correct the rotor flux linkage value according to the stator winding temperature to obtain a corrected rotor flux linkage value.

[0008] In another embodiment, the trained forward computation network further includes a second neural network for determining the zero-position deviation value of the rotary transformer based on the current reading of the rotary transformer, the corrected rotor flux linkage value, and other relevant parameters.

[0009] In another embodiment, the zero-point deviation determination device further includes: an initial feedforward computation network construction module configured to construct an initial feedforward computation network, the initial feedforward computation network including a first neural network and a second neural network; and a training module configured to input training samples into the initial feedforward computation network to obtain the trained feedforward computation network; wherein the training samples include a first sample for training the parameters in the first network and the second network; a second sample for optimizing the hyperparameters in the first network and the second network; and a third sample for evaluating the training results to determine whether the training of the initial feedforward computation network is complete.

[0010] In another embodiment, the training module further includes: an acquisition unit configured to acquire the theoretical relevant parameters of the motor corresponding to the theoretical zero-point deviation value of the rotary transformer; a first construction unit configured to construct a plurality of initial training samples by using the theoretical zero-point deviation value of the rotary transformer as a label and the theoretical relevant parameters of the motor as features; and a second construction unit configured to divide the plurality of initial training samples into a first sample, a second sample, and a third sample according to a preset ratio to construct the training samples.

[0011] In another embodiment, the training module includes: a detection unit configured to detect whether the initial zero-position deviation value output by the initial feedforward computation network matches the theoretical zero-position deviation value; a first detection result unit configured to indicate that the training of the initial feedforward computation network has not been completed if the initial zero-position deviation value fails to match the theoretical zero-position deviation value; and a second detection result unit configured to indicate that the training of the initial feedforward computation network has been completed if the initial zero-position deviation value successfully matches the theoretical zero-position deviation value, thereby obtaining the trained feedforward computation network.

[0012] In another embodiment, the relevant parameters further include bench-read torque value, motor speed, three-phase current, three-phase inductance, and stator winding resistance; the trained forward computation network further includes a second neural network; the determining module includes: a conversion unit configured to convert the three-phase current and the three-phase inductance into two-phase rotating current and two-phase rotating inductance in a rotating coordinate system; a correction unit configured to input the rotor flux linkage value and the stator winding temperature into the first neural network to obtain a corrected rotor flux linkage value; and a determining unit configured to input the current time-based rotary transformer reading value, the bench-read torque value, the motor speed, the corrected rotor flux linkage value, the two-phase rotating current, and the two-phase rotating inductance into the second neural network to obtain the zero-position deviation value of the rotary transformer output by the second neural network.

[0013] In another embodiment, the determination of the zero-position deviation further includes: a calculation module configured to average the zero-position deviation values ​​of the rotary transformer output by the second neural network to obtain an average zero-position deviation value; and to calculate the variance of the zero-position deviation values ​​of the rotary transformer output by the second neural network to obtain the variance of the zero-position deviation values; a value range determination module configured to determine the value range of the zero-position deviation values ​​based on the average zero-position deviation value and the variance; and a zero-position deviation value determination module configured to take the zero-position deviation values ​​within the value range as target zero-position deviation values, and to calculate the average target zero-position deviation value of the target zero-position deviation values ​​to determine the zero-position deviation value of the rotary transformer output by the trained forward computation network.

[0014] According to one aspect of the embodiments of this application, an electronic device is provided, including: a controller; and a memory for storing one or more programs, which, when executed by the controller, perform the above-described method for determining zero deviation.

[0015] According to one aspect of the embodiments of this application, a computer-readable storage medium is also provided, on which computer-readable instructions are stored, which, when executed by a computer's processor, cause the computer to perform the above-described method for determining zero-position deviation.

[0016] According to one aspect of the embodiments of this application, a computer program product or computer program is also provided, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the aforementioned method for determining zero-point deviation.

[0017] In the technical solution provided by the embodiments of this application, the relevant parameters of the motor and the current reading of the resolver are input into a trained forward computing network, so that the trained forward computing network outputs the zero-point deviation value of the resolver. The relevant parameters include the rotor flux linkage value and the stator winding temperature. The trained forward computing network includes a first neural network, which is used to correct the rotor flux linkage value based on the stator winding temperature to obtain the corrected rotor flux linkage value. This application only requires inputting the relevant parameters of the motor and the current reading of the resolver into the trained forward computing network, eliminating the need for tedious manual detection and calibration processes, thus preventing human error and improving the accuracy of the zero-point deviation value of the resolver.

[0018] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this application. Attached Figure Description

[0019] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. In the drawings:

[0020] Figure 1 This is a flowchart illustrating a method for determining zero-position deviation according to an exemplary embodiment of this application;

[0021] Figure 2 This is a schematic diagram of the structure of the trained forward computation network in the embodiments of this application;

[0022] Figure 3 Based on Figure 1 A flowchart of another method for determining zero-position deviation proposed in the illustrated embodiment;

[0023] Figure 4 Based on Figure 3 A flowchart of another method for determining zero-position deviation proposed in the illustrated embodiment;

[0024] Figure 5 This is a schematic diagram of the initial training sample collection process in an embodiment of this application;

[0025] Figure 6 This is a schematic diagram of the training process of the feedforward computing network in the embodiments of this application;

[0026] Figure 7 Based on Figure 4 A flowchart of another method for determining zero-position deviation proposed in the illustrated embodiment;

[0027] Figure 8 Based on Figure 1 A flowchart of another method for determining zero-position deviation proposed in the illustrated embodiment;

[0028] Figure 9 This is a schematic diagram of the processing flow of the trained forward computing network in the embodiments of this application;

[0029] Figure 10 Based on Figure 8 A flowchart of another method for determining zero-position deviation proposed in the illustrated embodiment;

[0030] Figure 11 This is a schematic diagram of the structure of a zero-position deviation determination device shown in an exemplary embodiment of this application;

[0031] Figure 12 This is a schematic diagram of the structure of a computer system for an electronic device, as illustrated in an exemplary embodiment of this application. Detailed Implementation

[0032] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0033] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0034] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0035] In this application, "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0036] Currently, there are three common methods for identifying the zero-point deviation of the resolver in a permanent magnet synchronous motor: First, bench calibration is used, requiring the motor to be mounted on a test bench, and then operators to manipulate the bench and related software to obtain the data. This process is complex and prone to human error. Second, a back-EMF sampling circuit is added to the motor controller to detect the no-load back EMF of the motor, and the zero-point deviation of the resolver is calculated from the no-load back EMF value. However, this requires additional hardware to be added to the motor controller, increasing its cost. Third, a self-learning method is used. When the motor is in no-load condition, a given d-axis current is applied to make the q-axis current zero, and the resolver zero-point deviation value is continuously adjusted until the motor reaches a constant speed. The zero-point deviation value of the resolver is recorded at this point. This method is greatly affected by external resistance and the adjustment process is slow, making it difficult to accurately determine the zero-point deviation of the resolver.

[0037] Please refer to the following first. Figure 1 , Figure 1 This is a flowchart illustrating a method for determining zero-position deviation according to an exemplary embodiment of this application. Figure 1 As shown, the method includes at least S110, which is described in detail below:

[0038] S110: Input the relevant parameters of the motor and the current reading of the rotary transformer into the trained forward computing network so that the trained forward computing network outputs the zero-point deviation value of the rotary transformer; wherein, the relevant parameters include the rotor flux linkage value and the stator winding temperature; the trained forward computing network includes a first neural network, which is used to correct the rotor flux linkage value according to the stator winding temperature to obtain the corrected rotor flux linkage value.

[0039] In this embodiment, the relevant parameters of the motor include not only the rotor flux linkage value and stator winding temperature, but also the bench-read torque value, motor speed, three-phase current, three-phase inductance, and stator winding resistance. That is, it is not limited to the parameters generated by the motor itself, but also includes parameters related to it that can be used to determine the zero-point deviation value of the rotary transformer.

[0040] The current rotation transformer reading is the angle value read by the rotation transformer at the current moment. It is used to perform a difference operation with the zero-position deviation angle value determined by the forward computation network, and the resulting difference is the rotation transformer reading value output by the forward computation network.

[0041] In this embodiment, the first neural network is used to correct the rotor flux linkage value. For example, the rotor flux linkage value and the stator winding temperature are input into the first neural network so that the first neural network outputs the corrected rotor flux linkage value. Then, the rotary transformer reading value is determined based on the corrected rotor flux linkage value and other relevant parameters. The difference between the rotary transformer reading value at the current time and the current angle value read by the rotary transformer is calculated. The difference is the rotary transformer reading value output by the forward calculation network.

[0042] This embodiment inputs relevant motor parameters and the current resolver reading into a trained forward computation network, enabling the network to output the resolver's zero-point deviation value. The relevant parameters include the rotor flux linkage and stator winding temperature. The trained forward computation network includes a first neural network, which corrects the rotor flux linkage based on the stator winding temperature to obtain a corrected value. This application only requires inputting the motor parameters and the current resolver reading into the trained forward computation network, eliminating the need for tedious manual calibration and thus preventing human error and improving the accuracy of the resolver's zero-point deviation value. Furthermore, this embodiment eliminates the need for additional hardware on the motor controller, such as sensors or sampling circuits, reducing the cost of modifying the controller. Additionally, the determination of the zero-point deviation is less affected by external resistance.

[0043] In another embodiment, such as Figure 2 As shown, Figure 2 This is a schematic diagram of the structure of the trained forward computation network in an embodiment of this application. The trained forward computation network includes a first neural network 210 and a second neural network 220. After correcting the rotor flux linkage value, the first neural network 210 inputs the corrected rotor flux linkage value and other relevant parameters into the second neural network 220. The second neural network 220 outputs the rotary transformer reading value, and then performs a difference operation between this zero-position deviation angle value and the current rotary transformer reading value. The resulting difference is the rotary transformer reading value output by the forward computation network.

[0044] This embodiment uses data as a driving force, designing a deep learning network to learn the mathematical relationship between motor output torque, zero-position deviation value, and motor dq-axis current, thereby determining the zero-position deviation of the rotary transformer. The designed forward computation network includes a first neural network and a second neural network. These two neural networks process different data and have different functions; their cooperation makes the zero-position deviation of the rotary transformer output by the forward computation network more accurate.

[0045] In another exemplary embodiment of this application, the process of constructing the feedforward computation network and training samples is further described; please refer to [link to relevant documentation] for details. Figure 3 , Figure 3 Based on Figure 1 The flowchart illustrates another method for determining zero-position deviation proposed in the illustrated embodiment. This method... Figure 1 The circuit S110 shown includes S310 to S320, which will be described in detail below:

[0046] S310: Construct the initial forward computation network, which includes a first neural network and a second neural network.

[0047] S320: Input the training samples into the initial feedforward computation network to obtain the trained feedforward computation network; wherein, the training samples include a first sample, used to train the parameters in the first network and the second network; a second sample, used to optimize the hyperparameters in the first network and the second network; and a third sample, used to evaluate the training results to determine whether the training of the initial feedforward computation network is complete.

[0048] The training samples in this embodiment include at least three types of samples, namely the first sample, the second sample, and the third sample in this embodiment.

[0049] For example, the three types of samples are training sub-samples for training, validation sub-samples for validation, and test sub-samples for testing. Gradient descent is used, and the parameters in the first neural network and the second neural network are trained using the training sub-samples. The hyperparameters in the first neural network and the second neural network are optimized using the validation sub-samples. The training results of the initial feedforward computation network are evaluated using the test sub-samples. If the test results do not meet the preset conditions, the initial feedforward computation network is improved again until its output test results meet the preset conditions.

[0050] This embodiment further illustrates how to construct and train the feedforward computation network to obtain a well-trained feedforward computation network. Specifically, three types of samples from the training dataset are used for training, optimization, and evaluation to ensure the computational accuracy of the trained feedforward computation network, thereby making the zero-position deviation value of the rotary transformer output by the trained feedforward computation network more accurate.

[0051] In another exemplary embodiment of this application, the process of acquiring sample data is further described, please refer to [link to relevant documentation]. Figure 4 , Figure 4 Based on Figure 3 A flowchart of another method for determining zero-position deviation proposed in the illustrated embodiment. This method, in... Figure 3 The S320 shown includes S410 to S430, which are described in detail below:

[0052] S410: Obtain the theoretical relevant parameters of the motor corresponding to the theoretical zero-point deviation value of the rotary transformer.

[0053] The theoretical zero-position deviation value refers to the zero-position deviation value with no standard error, and the corresponding motor-related parameters are theoretical related parameters.

[0054] An example is provided to illustrate how to obtain the theoretical parameters of the motor, such as... Figure 5 As shown, Figure 5 This is a schematic diagram of the initial training sample acquisition process in an embodiment of this application. The acquisition process includes steps S510 to S570, which are described in detail below:

[0055] S510: Install and test the motor.

[0056] S520: Bench calibration of the current zero point value of the rotary transformer.

[0057] S530: Drag the test motor to a random fixed speed.

[0058] S540: Randomly send q-axis target current commands and d-axis target current commands.

[0059] S550: Record relevant sample data.

[0060] S560: Detect whether the number of repetitions is greater than the preset number.

[0061] If the number of repetitions detected is greater than the preset number, it indicates that the collection of the initial training samples is complete, and the collection of sample data stops.

[0062] S570: If the number of repetitions detected is less than or equal to the preset number, the stator of the rotary transformer is randomly rotated and fixed, and step S520 is executed again.

[0063] When the number of repetitions is detected to be greater than the preset number, it indicates that the number of sample data obtained at this time has reached the required number of training samples.

[0064] S420: Using the theoretical zero-point deviation value of the rotary transformer as the label and the theoretical relevant parameters of the motor as the feature, multiple initial training samples are constructed.

[0065] The initial training samples are the initial data obtained. In this embodiment, the initial training samples are not directly used to train the initial feedforward computation network. It is necessary to classify the uses of multiple initial samples.

[0066] Since the theoretical zero-point deviation value of the rotary transformer is assumed to be a value with no deviation or extremely high precision, using it as a label can ensure that the zero-point deviation value of the rotary transformer output by the trained feedforward computing network is more accurate while being convenient and efficient in subsequent processes.

[0067] S430: Divide multiple initial training samples into first samples, second samples, and third samples according to a preset ratio to construct the training samples.

[0068] In this embodiment, multiple initial samples are divided into three types of samples according to their three different uses, and then divided according to a preset ratio. For example, there are 10 initial samples, which are divided into 6 first samples, 2 second samples, and 2 third samples in a ratio of 3:1:1, thus constructing training samples.

[0069] This embodiment illustrates how to construct training samples, which include a first sample, a second sample, and a third sample. The proportion of the three types of samples in the training samples is a preset proportion, so that the constructed training samples can better train the forward computing network, and the zero-position deviation value of the rotary transformer output by the trained forward computing network is more accurate.

[0070] In another exemplary embodiment of this application, the process of training the feedforward computation network is further described, please refer to [link to relevant documentation]. Figure 6 , Figure 6 This is a schematic diagram of the training process of the feedforward computing network in an embodiment of this application. The training process includes steps S610 to S630, which are described in detail below:

[0071] S610: Design and modify the forward computing network.

[0072] S620: Parameters for training the feedforward computation network.

[0073] S630: Check whether the training results meet the preset conditions.

[0074] If the training results meet the preset conditions, the training process ends, indicating that the feedforward computation network at this time is a well-trained feedforward computation network.

[0075] If the training results do not meet the preset conditions, execute S610 to redesign and modify the feedforward computation network.

[0076] The preset condition can be that the zero-point deviation value output by the forward computation network is the same as the theoretical zero-point deviation value, which means that the accuracy of the output result of the forward computation network meets the expected requirements, and thus the training of the forward computation network is completed.

[0077] For example, please refer to Figure 7 , Figure 7 Based on Figure 4 The flowchart illustrates another method for determining zero-point deviation proposed in the illustrated embodiment. This method includes steps S710 to S730 in S320, which will be described in detail below:

[0078] S710: Detects whether the initial zero-point deviation value output by the initial forward computation network matches the theoretical zero-point deviation value.

[0079] In this embodiment, the theoretical zero-point deviation value is used to detect whether the initial feedforward computation network has completed training. Based on the matching result between the initial zero-point deviation value output by the initial feedforward computation network and the theoretical zero-point deviation value, it is determined whether the training of the initial feedforward computation network has been completed.

[0080] S720: If the initial zero-position deviation value fails to match the theoretical zero-position deviation value, it indicates that the training of the initial feedforward computation network has not been completed.

[0081] If the initial zero-position deviation value fails to match the theoretical zero-position deviation value, it indicates that there is a deviation between the initial zero-position deviation value and the theoretical zero-position deviation value. The zero-position deviation value output by the initial feedforward computing network is not accurate, and its accuracy does not meet the expected requirements. The initial feedforward computing network needs to be trained again.

[0082] S730: If the initial zero-position deviation value is detected to match the theoretical zero-position deviation value, it indicates that the training of the initial feedforward computation network is completed, and a trained feedforward computation network is obtained.

[0083] If the initial zero-point deviation value successfully matches the theoretical zero-point deviation value, it indicates that the initial zero-point deviation value is the same as the theoretical zero-point deviation value, and the accuracy of the zero-point deviation value output by the initial forward computation network meets the expected requirements.

[0084] This embodiment further illustrates how to obtain a trained feedforward computation network. By detecting whether the initial zero-point deviation value output by the initial feedforward computation network matches the theoretical zero-point deviation value, it is determined whether the training of the initial feedforward computation network is complete, so as to ensure that the accuracy of the trained feedforward computation network meets the expected requirements.

[0085] Please see Figure 8 , Figure 8 Based on Figure 1 The flowchart illustrates another method for determining zero-position deviation proposed in the illustrated embodiment. The relevant parameters include bench-read torque values, motor speed, three-phase current, three-phase inductance, and stator winding resistance. The trained forward computation network also includes a second neural network. The method further includes steps S810 to S830 in S110, which will be described in detail below:

[0086] S810: Converts three-phase current and three-phase inductance into two-phase rotating current and two-phase rotating inductance in a rotating coordinate system.

[0087] This embodiment can directly convert three-phase current and three-phase inductance into two-phase rotating current and two-phase rotating inductance in a rotating coordinate system, or it can convert three-phase current and three-phase inductance into two-phase rotating current and two-phase rotating inductance in a rotating coordinate system in steps, such as... Figure 9 As shown, Figure 9 This is a schematic diagram of the processing flow of the trained forward computation network in an embodiment of this application. Specifically, the three-phase current and three-phase inductance are transformed into two-phase rotating current and two-phase rotating inductance in a rotating coordinate system through two-step transformations: CLARK transformation and PARK transformation.

[0088] First, the three-phase current i is transformed using the Clark transformation. a i b i c Three-phase inductor L a L b L c Convert each i to a stationary coordinate system α i β L α L β This involves converting the three-phase current and three-phase inductance into two-phase current and two-phase inductance in a stationary coordinate system; then, using the PARK transformation, the two phases of i in the stationary coordinate system are converted... α i β L α L β Convert to i in the rotating coordinate system d i q L d L q This involves transforming the two-phase current and two-phase inductance in a stationary coordinate system into a two-phase rotating current and two-phase rotating inductance in a rotating coordinate system.

[0089] S820: Input the rotor flux linkage value and stator winding temperature into the first neural network to obtain the corrected rotor flux linkage value.

[0090] like Figure 9 As shown, the rotor flux ψ f and stator winding temperature T s Input the first neural network so that the first neural network outputs the corrected rotor flux linkage ψ f* .

[0091] S830: Input the current value of the rotary transformer, the torque value read from the test bench, the motor speed, the corrected rotor flux linkage value, the two-phase rotating current and the two-phase rotating inductance into the second neural network to obtain the zero-position deviation value of the rotary transformer output by the second neural network.

[0092] like Figure 9 As shown, the current value θ of the rotary transformer is read.r Torque value T read from the test bench e Motor speed w r Corrected rotor flux linkage ψ f* dq current (i d and i q ) and dq inductor (L d and L q Input the second neural network so that the second neural network outputs the zero deviation value Δθ of the rotary transformer at the current moment.

[0093] This embodiment further illustrates that after processing the relevant data through the first neural network and the second neural network, the second neural network outputs the zero-position deviation value of the rotary transformer at the current moment. Through the distributed processing of the two neural networks, the data processing flow is accelerated, and the zero-position deviation value of the rotary transformer at the current moment can be obtained quickly.

[0094] The zero-position deviation value of the rotary transformer at the current moment can be used as the zero-position deviation value of the rotary transformer output by the trained forward computing network. However, it may have a large random error, making the zero-position deviation value of the rotary transformer at the current moment inaccurate.

[0095] Therefore, in another exemplary embodiment of this application, the zero-position deviation value of the rotary transformer output by the second neural network is further processed to obtain the zero-position deviation value of the rotary transformer output by the trained forward computation network. For details, please refer to [link to relevant documentation]. Figure 10 , Figure 10 Based on Figure 8 The flowchart illustrates another method for determining zero-point deviation proposed in the illustrated embodiment. This method further includes steps S1010 to S1030, which will be described in detail below:

[0096] S1010: The zero-position deviation value of the rotary transformer output by the second neural network is averaged to obtain the average zero-position deviation value; and the variance of the zero-position deviation value of the rotary transformer output by the second neural network is calculated to obtain the variance of the zero-position deviation value.

[0097] Within a preset time period, the second neural network outputs the zero-position deviation values ​​of multiple rotary transformers, and calculates their average value and variance.

[0098] S1020: Determine the range of zero-point deviation values ​​based on the average zero-point deviation value and the variance.

[0099] For example, the range of the zero-point deviation value is (μ-σ,μ+σ); where μ represents the average zero-point deviation value and σ represents the variance.

[0100] S1030: Take the zero-position deviation value within the range as the target zero-position deviation value, and calculate the average target zero-position deviation value to determine the zero-position deviation value of the rotary transformer output by the trained forward computing network.

[0101] For example, if μ is 5, σ is 2, and the range of zero-position deviation values ​​is (μ-σ=3,μ+σ=7), then the zero-position deviation values ​​within the range of (3,7) are taken as the target zero-position deviation values. If the zero-position deviation values ​​within the range of (3,7) include 4 and 5, then the average target zero-position deviation value is 4.5, that is, the zero-position deviation value of the rotary transformer output by the trained forward computation network is 4.5.

[0102] This embodiment further processes the zero-position deviation value of the rotary transformer output by the second neural network. It performs averaging and variance calculations on the zero-position deviation value, and determines the range containing the target zero-position deviation value based on the obtained average and variance. Then, it averages the target zero-position deviation values ​​within this range to obtain the zero-position deviation value of the rotary transformer output by the trained feedforward computing network. This avoids directly using the zero-position deviation value of the rotary transformer output by the second neural network as the zero-position deviation value of the rotary transformer output by the trained feedforward computing network, which could lead to random errors. Therefore, the zero-position deviation value of the rotary transformer output by the trained feedforward computing network is more accurate.

[0103] Another aspect of this application provides a device for determining zero-point deviation, such as... Figure 11 As shown, Figure 11 This is a schematic diagram illustrating the structure of a zero-position deviation determination device according to an exemplary embodiment of this application. The zero-position deviation determination device includes:

[0104] The determination module 1110 is configured to input the relevant parameters of the motor and the current reading value of the rotary transformer into the trained forward computing network, so that the trained forward computing network outputs the zero-point deviation value of the rotary transformer; wherein, the relevant parameters include the rotor flux linkage value and the stator winding temperature; the trained forward computing network includes a first neural network; the first neural network is used to correct the rotor flux linkage value according to the stator winding temperature to obtain the corrected rotor flux linkage value.

[0105] In another embodiment, the trained forward computation network further includes a second neural network for determining the zero-position deviation value of the rotary transformer based on the current rotary transformer reading, the corrected rotor flux linkage value, and other relevant parameters.

[0106] In another embodiment, the zero-point deviation determination device further includes:

[0107] The initial feedforward computation network building module is configured to build an initial feedforward computation network, which includes a first neural network and a second neural network.

[0108] The training module is configured to input training samples into an initial feedforward computation network to obtain a trained feedforward computation network; wherein the training samples include a first sample for training the parameters in the first network and the second network; a second sample for optimizing the hyperparameters in the first network and the second network; and a third sample for evaluating the training results to determine whether the training of the initial feedforward computation network is complete.

[0109] In another embodiment, the training module further includes:

[0110] The acquisition unit is configured to acquire the theoretical relevant parameters of the motor corresponding to the theoretical zero-point deviation value of the rotary transformer.

[0111] The first building unit is configured to use the theoretical zero-point deviation value of the rotary transformer as a label and the theoretical relevant parameters of the motor as features to build multiple initial training samples.

[0112] The second building unit is configured to divide multiple initial training samples into a first sample, a second sample, and a third sample according to a preset ratio to build training samples.

[0113] In another embodiment, the training module includes:

[0114] The detection unit is configured to detect whether the initial zero-point deviation value output by the initial forward computation network matches the theoretical zero-point deviation value.

[0115] The first detection result unit is configured to indicate that the training of the initial feedforward computation network has not been completed if the initial zero-position deviation value fails to match the theoretical zero-position deviation value.

[0116] The second detection result unit is configured to indicate that the training of the initial feedforward computation network is completed if the initial zero-position deviation value is successfully matched with the theoretical zero-position deviation value, so as to obtain the trained feedforward computation network.

[0117] In another embodiment, the relevant parameters also include bench-read torque values, motor speed, three-phase current, three-phase inductance, and stator winding resistance; the trained forward computation network also includes a second neural network; the determination module includes:

[0118] The conversion unit is configured to convert three-phase current and three-phase inductance into two-phase rotating current and two-phase rotating inductance in a rotating coordinate system.

[0119] The correction unit is configured to input the rotor flux linkage value and the stator winding temperature into the first neural network to obtain the corrected rotor flux linkage value.

[0120] The determining unit is configured to input the current time-at-time rotary transformer reading value, bench reading torque value, motor speed, corrected rotor flux value, two-phase rotating current and two-phase rotating inductance into the second neural network to obtain the zero-position deviation value of the rotary transformer output by the second neural network.

[0121] In another embodiment, the determination of the zero-point deviation further includes:

[0122] The calculation module is configured to average the zero-position deviation values ​​of the rotary transformer output by the second neural network to obtain the average zero-position deviation value; and to calculate the variance of the zero-position deviation values ​​of the rotary transformer output by the second neural network to obtain the variance of the zero-position deviation values.

[0123] The value range determination module is configured to determine the value range of the zero-point deviation value based on the average zero-point deviation value and the variance.

[0124] The zero-position deviation value determination module is configured to take the zero-position deviation value within the range as the target zero-position deviation value, and calculate the average target zero-position deviation value of the target zero-position deviation values ​​to determine the zero-position deviation value of the rotary transformer output by the trained forward computing network.

[0125] It should be noted that the zero-position deviation determination device provided in the above embodiments and the zero-position deviation determination method provided in the foregoing embodiments belong to the same concept. The specific way in which each module and unit performs operations has been described in detail in the method embodiments, and will not be repeated here.

[0126] Another aspect of this application provides an electronic device, including: a controller; and a memory for storing one or more programs, which, when executed by the controller, perform the methods described above.

[0127] Please see Figure 12 , Figure 12 This is a schematic diagram of the structure of a computer system for an electronic device, illustrating an exemplary embodiment of this application. It shows a schematic diagram of the structure of a computer system suitable for implementing the embodiments of this application.

[0128] It should be noted that, Figure 12 The computer system 1200 of the electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0129] like Figure 12As shown, the computer system 1200 includes a Central Processing Unit (CPU) 1201, which can perform various appropriate actions and processes, such as executing the methods described in the above embodiments, based on programs stored in Read-Only Memory (ROM) 1202 or programs loaded from storage portion 1208 into Random Access Memory (RAM) 1203. The RAM 1203 also stores various programs and data required for system operation. The CPU 1201, ROM 1202, and RAM 1203 are interconnected via a bus 1204. An Input / Output (I / O) interface 1205 is also connected to the bus 1204.

[0130] The following components are connected to I / O interface 1205: an input section 1206 including a keyboard, mouse, etc.; an output section 1207 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1208 including a hard disk, etc.; and a communication section 1209 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 1209 performs communication processing via a network such as the Internet. A drive 1210 is also connected to I / O interface 1205 as needed. Removable media 1211, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 1210 as needed so that computer programs read from them can be installed into storage section 1208 as needed.

[0131] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 1209, and / or installed from removable medium 1211. When the computer program is executed by central processing unit (CPU) 1201, it performs various functions defined in the system of this application.

[0132] It should be noted that the computer-readable medium shown in the embodiments of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable computer program. The transmitted data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0133] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0134] The units described in the embodiments of this application can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself.

[0135] Another aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for determining zero-point deviation. This computer-readable storage medium may be included in the electronic device described in the above embodiments, or it may exist independently without being assembled into the electronic device.

[0136] Another aspect of this application provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the zero-point deviation determination method provided in the various embodiments described above.

[0137] According to one aspect of the embodiments of this application, a computer system is also provided, including a Central Processing Unit (CPU), which can perform various appropriate actions and processes based on a program stored in read-only memory (ROM) or a program loaded from storage into random access memory (RAM), such as performing the methods described above. Various programs and data required for system operation are also stored in the RAM. The CPU, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0138] The following components are connected to the I / O interface: input components including keyboards, mice, etc.; output components including cathode ray tubes (CRTs), liquid crystal displays (LCDs), and speakers; storage components including hard drives; and communication components including network interface cards such as LAN (Local Area Network) cards and modems. The communication components perform communication processing via networks such as the Internet. Drives are also connected to the I / O interface as needed. Removable media, such as disks, optical discs, magneto-optical discs, semiconductor memories, etc., are installed on the drive as needed so that computer programs read from them can be installed into the storage components as required.

[0139] The above description is merely a preferred exemplary embodiment of this application and is not intended to limit the implementation of this application. Those skilled in the art can easily make corresponding modifications or alterations based on the main concept and spirit of this application. Therefore, the scope of protection of this application should be determined by the scope of protection claimed in the claims.

Claims

1. A method for determining zero-position deviation, characterized in that, include: The relevant parameters of the motor and the current reading of the rotary transformer are input into the trained forward computing network so that the trained forward computing network outputs the zero-point deviation value of the rotary transformer; wherein, the relevant parameters include the rotor flux linkage value and the stator winding temperature; the trained forward computing network includes a first neural network, which is used to correct the rotor flux linkage value according to the stator winding temperature to obtain the corrected rotor flux linkage value; The relevant parameters also include bench-read torque value, motor speed, three-phase current, three-phase inductance, and stator winding resistance; the trained forward computation network also includes a second neural network; the three-phase current and the three-phase inductance are converted into two-phase rotating current and two-phase rotating inductance in a rotating coordinate system; the current time-based rotary transformer reading value, the bench-read torque value, the motor speed, the corrected rotor flux linkage value, the two-phase rotating current, and the two-phase rotating inductance are input into the second neural network to obtain the zero-position deviation value of the rotary transformer output by the second neural network.

2. The method according to claim 1, characterized in that, Before inputting the relevant parameters of the motor and the current read value of the resolver into the trained forward computation network, the method further includes: Construct an initial feedforward computation network, which includes the first neural network and the second neural network; Training samples are input into the initial feedforward computation network to obtain the trained feedforward computation network; wherein, the training samples include a first sample for training the parameters in the first neural network and the second neural network; a second sample for optimizing the hyperparameters in the first neural network and the second neural network; and a third sample for evaluating the training results to determine whether the training of the initial feedforward computation network is complete.

3. The method according to claim 2, characterized in that, Before inputting the training samples into the initial feedforward computation network to obtain the trained feedforward computation network, the method further includes: Obtain the theoretical relevant parameters of the motor corresponding to the theoretical zero-point deviation value of the rotary transformer; Using the theoretical zero-point deviation value of the rotary transformer as a label and the theoretical relevant parameters of the motor as features, multiple initial training samples are constructed. Multiple initial training samples are divided into a first sample, a second sample, and a third sample according to a preset ratio to construct the training samples.

4. The method according to claim 3, characterized in that, The step of inputting training samples into the initial feedforward computation network to obtain a trained feedforward computation network includes: The initial zero-position deviation value output by the initial forward computation network is checked to see if it matches the theoretical zero-position deviation value. If the initial zero-position deviation value fails to match the theoretical zero-position deviation value, it indicates that the training of the initial feedforward computation network has not been completed. If the initial zero-position deviation value is detected to match the theoretical zero-position deviation value, it indicates that the training of the initial feedforward computing network is completed, so as to obtain the trained feedforward computing network.

5. The method according to claim 1, characterized in that, The method further includes: The zero-position deviation value of the rotary transformer output by the second neural network is averaged to obtain the average zero-position deviation value; and the variance of the zero-position deviation value of the rotary transformer output by the second neural network is calculated to obtain the variance of the zero-position deviation value. The range of the zero-point deviation value is determined based on the average zero-point deviation value and the variance. The zero-position deviation value within the specified range is taken as the target zero-position deviation value, and the average target zero-position deviation value is calculated to determine the zero-position deviation value of the rotary transformer output by the trained forward computing network.

6. A device for determining zero-position deviation, characterized in that, include: The determination module is configured to input relevant parameters of the motor and the current reading value of the resolver into a trained forward computing network, so that the trained forward computing network outputs the zero-point deviation value of the resolver; wherein, the relevant parameters include the rotor flux linkage value and the stator winding temperature; the trained forward computing network includes a first neural network; the first neural network is used to correct the rotor flux linkage value according to the stator winding temperature to obtain a corrected rotor flux linkage value; The relevant parameters also include bench-read torque value, motor speed, three-phase current, three-phase inductance, and stator winding resistance; the trained forward computation network also includes a second neural network; the three-phase current and the three-phase inductance are converted into two-phase rotating current and two-phase rotating inductance in a rotating coordinate system; the current time-based rotary transformer reading value, the bench-read torque value, the motor speed, the corrected rotor flux linkage value, the two-phase rotating current, and the two-phase rotating inductance are input into the second neural network to obtain the zero-position deviation value of the rotary transformer output by the second neural network.

7. An electronic device, characterized in that, include: Controller; A memory for storing one or more programs that, when executed by the controller, cause the controller to implement the method for determining zero-position deviation as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, It stores computer-readable instructions that, when executed by the computer's processor, cause the computer to perform the method for determining the zero-position deviation as described in any one of claims 1 to 5.

Images