Motor angle detection sensor

Abstract

The application discloses a motor angle detection sensor, which comprises a coil winding, a core, a rotating shaft, an excitation module, a signal conditioning module and a signal processing module. The core is driven to rotate by the rotating shaft and comprises a ring-shaped first part and a spiral-shaped special-shaped part. The cross section of the special-shaped part is in the form of an equiangular solenoid, so that each angle in the rotation range of 0°-180° corresponds to a unique space pose and magnetic field distribution. The signal processing module analyzes the direct current voltage difference, the ambient temperature and the rotating speed based on an AI compensation model to determine the angle of the rotating shaft. The model is trained by a gradient boosting decision tree algorithm to fit the complex mapping relationship between the angle and multiple parameters, thereby simultaneously overcoming temperature drift, rotating speed fluctuation and nonlinear error, eliminating the measurement dead zone and improving the measurement accuracy and robustness under wide temperature range and high dynamic working conditions.

Description

Technical Field

[0001] This application relates to the field of motor detection technology, and more specifically, to a motor angle detection sensor. Background Technology

[0002] In the field of motor control, high-precision angle detection sensors are key components for achieving closed-loop control. While traditional resolvers and rotary variable differential transformers (RVDTs) are widely used, their cores typically employ a fan-shaped or cam-shaped structure, achieving angle detection through physical obstruction of the magnetic circuit. Figure 1 As shown in the diagram, the iron core has a symmetrical structure. After rotating it by a certain angle, the magnetic flux of the secondary coil at the corresponding position will coincide with that before. Therefore, there is a large measurement dead zone and a narrow linear measurement range (usually less than ±45°).

[0003] Existing signal demodulation methods rely on analog circuits to perform sine and cosine function calculations, which cannot dynamically compensate for temperature and rotational speed coupling interference, leading to frequent system calibrations. Therefore, a novel sensor structure is urgently needed to overcome the limitations of traditional magnetic circuit modulation and achieve large-angle and high-precision measurements. Summary of the Invention

[0004] The purpose of this application is to provide a motor angle detection sensor that solves the above-mentioned problems in the prior art and can obtain a larger angle measurement.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: a motor angle detection sensor, including a coil winding, an iron core, a rotating shaft, an excitation module, a signal conditioning module, and a signal processing module. The iron core is driven to rotate by the rotating shaft, which is connected to the motor under test. The rotating shaft includes a first part and an irregular part. The position of the first part corresponds to the primary coil, and the outer contour of the cross-section of the irregular part is an equiangular spiral.

[0006] The coil winding includes a primary coil and two secondary coils wound in the same manner. The first part of the iron core is positioned corresponding to the primary coil, and the two secondary coils are symmetrically arranged along the axis of rotation. The irregular part of the iron core is positioned corresponding to the two secondary coils, and the secondary coils are located on the outside of the iron core.

[0007] The excitation module is used to provide an excitation signal to the primary coil;

[0008] The signal conditioning module is used to extract the voltage values ​​of the two secondary coils and condition them into two DC voltage signals.

[0009] The signal processing module receives two DC voltage signals from the signal conditioning module and determines the angle of the motor shaft under test based on the preset correspondence between the difference between the two DC voltage signals and the rotation angle of the shaft.

[0010] The preset method for the correspondence between the difference of two DC voltage signals and the rotation angle of the shaft includes:

[0011] Step 1: Simulate the true value θ of the core rotation angle of the motor under test under different operating ambient temperatures T and different speeds W, and compare it with the measured value U of the difference between two DC voltage signals. Divide the dataset into:

[0012] Training set: 70%-80% of the data, used to train the model;

[0013] Validation set: 10%-15% of the data, used to evaluate model performance during training, adjust hyperparameters, and prevent overfitting;

[0014] Test set: 10%-15% of the data, used for the final evaluation of the model's generalization ability;

[0015] Step 2: Standardize the input feature X, X=[U,T,W];

[0016] Step 3: Input the training and validation set data into the gradient boosting decision tree model for training, and fit a function model F of θ with respect to T, W, and U. The optimization objective of the function model F is to minimize the predicted angle value. =F(U,T,W) is the loss function between the actual angle value θ;

[0017] Step 4: Evaluate the performance of the final model using test set data;

[0018] The training process of the gradient boosting tree model includes iteratively executing the following steps:

[0019] The k-th iteration process includes:

[0020] Calculate the current model F k-1 The negative gradient r on all training samples i ;

[0021] Fit a new regression decision tree f k To predict the negative gradient r i ;

[0022] Add the new decision tree to the model and update the formula as follows: F k =F k-1 +η·f k ;

[0023] Where η is the learning rate, a pre-defined hyperparameter with a value range of (0,1].

[0024] Wherein, the loss function is the mean squared error function:

[0025] ;

[0026] N is the number of samples in the training set, θ true,i This represents the i-th true value. i Let be the i-th predicted value.

[0027] Furthermore, in the initial state, the outer contour of the cross-section of the irregularly shaped portion of the iron core satisfies the following relationship:

[0028] ,

[0029] Where r0 is the inner diameter of the iron core, k is a constant, θ is the polar angle of the profile, and r(θ) is the length of the profile from the iron core axis.

[0030] Furthermore, it also includes a housing, end caps located at both ends of the housing, a stator, and a coil frame. The coil frame is used to wind the primary coil, the stator is used to wind the secondary coil, the coil frame is disposed in the stator, and the iron core passes through the coil frame and the stator.

[0031] This application embodiment uses a spiral iron core structure to achieve a unique and non-repeatable spatial pose and magnetic field distribution for each angle within a rotation range of 0° to 180°, eliminating the measurement dead zone problem of traditional sensors; it adopts a data-driven AI compensation algorithm to simultaneously overcome the combined interference of temperature drift, rotation speed fluctuation and iron core nonlinearity, significantly improving the measurement robustness of the system under wide temperature range and high dynamic conditions. Attached Figure Description

[0032] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a schematic diagram of the structure of a motor angle detection sensor provided in an embodiment of this application;

[0034] Figure 2 This is a schematic diagram of the coil winding structure of the motor angle detection sensor provided in an embodiment of this application;

[0035] Figure 3This is a schematic diagram showing the distribution of the core and coil of the motor angle detection sensor provided in an embodiment of this application. Detailed Implementation

[0036] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Unless otherwise defined, the technical or scientific terms used in this application should have the ordinary meaning understood by those skilled in the art. The words "first," "second," and similar terms used in this application do not indicate any order, quantity, or importance, but are only used to distinguish different components. The words "comprising" or "including," etc., mean that the element or object preceding the word covers the element or object listed after the word and its equivalents, but do not exclude other elements or objects. The words "connected," "coupled," or "connected," etc., are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up," "down," "left," "right," etc., are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0037] The principle of the motor angle detection sensor provided in this application embodiment is as follows: Figure 1-3 As shown, the sensor includes a housing 1, end caps 2 at both ends of the housing, a stator 3, a shaft 7, an iron core 4, and a coil frame 5. The coil frame 5 is used to wind the primary coil 9, and the stator is used to wind the secondary coil 10. The coil frame is located within the stator, and the iron core passes through the coil frame and the stator. A wire outlet hole 11 is provided on the housing. The sensor also includes an excitation module, a signal conditioning module, and a signal processing module. These modules are generally located outside the housing at the measuring end, which reduces the size of the sensor and the required installation volume. The excitation module, signal conditioning module, and signal processing module can be integrated onto a single PCB. Two bearings 8 are also provided inside the housing to support the shaft.

[0038] The iron core 4 is fixed on the rotating shaft 7 and rotates as driven by the shaft 7. The shaft 7 is connected to the motor shaft of the motor under test and rotates with the motor. The iron core includes a first part 4.1 and a shaped part 4.2. The first part has a ring-shaped cross-section and is positioned corresponding to the primary coil. The function of the first part is to concentrate magnetic field lines and enhance the magnetic field. The outer contour of the cross-section of the shaped part 4.2 is helical. Preferably, it is an equiangular helix. In the initial state, the outer contour of the cross-section of the shaped part of the iron core satisfies the following relationship:

[0039]

[0040] Where r0 is the inner diameter of the iron core, k is a constant, θ is the polar angle of the profile, and r(θ) is the length of the profile from the iron core axis. The polar angle is the angle between the line connecting a point on the profile to the center of the inner circle of the spiral and the line connecting the center of the spiral to the starting point of the spiral. In this embodiment, the starting point, ending point, and center of the inner circle of the spiral are on the same straight line.

[0041] The secondary coil consists of two secondary coils wound in the same way. The two secondary coils are symmetrically arranged along the axis of rotation, and the irregular part of the iron core corresponds to the position of the two secondary coils. The secondary coils are located outside the iron core, that is, the iron core does not pass through the secondary coils.

[0042] The excitation module is used to provide an excitation signal to the primary coil to generate a magnetic field. The excitation module can be an oscillation circuit, which generates a high-frequency sine wave signal by an ICL8038 chip. This sine wave signal is amplified by a Class AB push-pull circuit and then output as an oscillation signal.

[0043] The signal conditioning module extracts the voltage values ​​of the two secondary coils and conditions them into two DC voltage signals. Connected to the two secondary windings, the module acquires the voltage signal from each winding and conditions it into a DC signal. If the excitation module uses sinusoidal excitation, the alternating magnetic field generated by the primary coil will induce an amplitude-modulated signal of the same frequency in the secondary coil, making it impossible to extract the static value. To eliminate the influence of the excitation signal itself on the difference in the secondary coil values, synchronous demodulation can be used. By multiplying the secondary signal U with the excitation reference signal, the voltage information is demodulated from the high-frequency carrier wave. This demodulation then passes through a low-pass filter (LPF) to remove the 2ω high-frequency component, where ω is the angular frequency of the excitation signal. This is achieved using an analog multiplier chip (such as AD630) or a switching demodulation circuit, in conjunction with a low-pass filter whose cutoff frequency is much smaller than ω / (2π).

[0044] In this embodiment, the helical iron core is asymmetrical along any plane passing through its axis. Therefore, within the rotation range of [0°, 180°], after the iron core rotates any angle θ from its initial position, its spatial pose does not coincide with any other angle pose (i.e., there is no θ1 ≠ θ2 that results in the same iron core pose). Since the iron core does not pass through the secondary coil, its rotational motion affects the secondary coil through spatial magnetic field coupling. The rotation angle θ of the iron core forms a strict one-to-one mapping relationship with the magnetic field strength distribution around the two secondary coils. Furthermore, the rotation angle θ also forms a one-to-one mapping relationship with the voltage values ​​of the two secondary coils; therefore, the rotation angle θ can be obtained by analyzing the difference between the two voltage values.

[0045] The signal processing module can be a microcontroller with digital-to-analog / analog-to-digital conversion function, used to analyze the acquired digital or analog signals and determine the shaft angle of the motor under test based on the preset mapping relationship between the difference between two DC voltage signals and the shaft rotation angle.

[0046] The mapping relationship between the rotation angle θ and the voltage difference can be obtained through two methods. The first method uses physical calculations to calculate the magnetic field distribution of the iron core at different angles and then calculates the corresponding voltage value. This method requires extremely high precision in the manufacturing of the coil and the iron core. Therefore, this embodiment adopts AI fitting. AI fitting can avoid the tedious calculation process and can incorporate ambient temperature parameters and vibration parameters (vibration mainly depends on the motor speed) to compensate for accuracy errors caused by temperature and vibration. The specific AI fitting method is as follows;

[0047] Step 1: In the laboratory, simulate the true value θ of the core rotation angle and the measured value U of the difference between two DC voltage signals of the motor under test under different operating ambient temperatures T and different speeds W, and divide the dataset into:

[0048] Training set: 70%-80% of the data is used to train the model.

[0049] Validation set: 10%-15% of the data, used to evaluate model performance during training, adjust hyperparameters, and prevent overfitting.

[0050] Test set: 10%-15% of the data, used for the final evaluation of the model's generalization ability.

[0051] To improve the accuracy of the fitting, the motor under test can be placed directly in the laboratory to replicate the actual operating conditions of the sensor. For motors that are inconvenient to place in the laboratory, a regular motor can be used to simulate the operating conditions of the motor under test. The output shaft radius of the regular motor must be the same as that of the motor under test to allow for sensor installation. Furthermore, a higher-precision angle sensor is needed to obtain the actual rotation angle of the motor. High-precision temperature and speed sensors need to be installed in the laboratory to obtain the motor's operating temperature and speed.

[0052] Step 2: Because the numerical range and unit differences of U (which may be a few volts), T (which may be a few hundred Kelvin), and W (which may be a few thousand RPM) are huge, the data X, which is used as input feature, is standardized, X=[U,T,W], so that its mean is 0 and its standard deviation is 1.

[0053] The formula is: X_scaled=(X-μ) / σ, where μ is the mean of the features in the training set and σ is the standard deviation of the features in the training set. Standardization can accelerate the convergence process of the model and prevent certain features from dominating the entire training process due to excessively large values.

[0054] Step 3: Input the training and validation set data into the gradient boosting decision tree model for training, and fit a function model F of θ with respect to T, W, and U. The optimization objective of the function model F is to minimize the predicted angle value. =F(U,T,W) is the loss function between the actual angle value θ and the true angle value θ.

[0055] Gradient boosting decision trees consist of K decision trees (weak learners). The model's final prediction is a weighted sum of the predictions from all these trees.

[0056]

[0057] Let f be the predicted value of the iron core rotation angle, F be the gradient boosting decision tree model, and K be the total number of decision trees. k Let be the output function of the k-th decision tree.

[0058] The loss function measures the difference between the model's predictions and the actual values. The core of weight adjustment is minimizing the loss function.

[0059] The most commonly used loss function is mean squared error:

[0060]

[0061] Where N is the number of samples in the training set, representing the θ-th... true,i Let i be the i-th true value. The i-th predicted value;

[0062] Gradient boosting decision trees use the gradient boosting algorithm, which is an iterative, additive training process. Each new tree learns the residuals of all previous tree combinations.

[0063] Before the iteration begins, initialization is required, using a simple initial model as the starting point, such as all prediction results being a constant.

[0064] During iterative training, one tree is added sequentially, gradually improving the model. The specific process is as follows:

[0065] Calculate the negative gradient (pseudo residual):

[0066]

[0067] Among them, ri (k) F represents the pseudo residual of the i-th sample in the k-th iteration. k-1 For the cumulative model after the first k-1 iterations, θ true,i Let be the true angle value of the i-th sample.

[0068] New decision tree fitting, training a new tree fk Objective function:

[0069]

[0070] Where f is the regression decision tree model, and argmin optimization ensures that the new tree corrects the defects of the current model to the greatest extent.

[0071] Model update

[0072]

[0073] Where: η is the learning rate, and the hyperparameters satisfy 0 < η ≤ 1; F k This is the updated integrated model.

[0074] Training stops when the preset total number of trees K is reached, or when the loss on the validation set no longer decreases significantly.

[0075] Step 4: Evaluate the performance of the final model using test set data, including metrics such as mean absolute error, root mean square error, and coefficient of determination.

[0076] The pre-trained model from the laboratory is integrated into an embedded system or host computer software. During actual operation, the system reads the three values ​​(U, T, W) in real time based on the motor's built-in temperature and speed sensors. These three values ​​are then standardized using μ and σ calculated during training. The standardized vector is then input into the trained model F, which instantly outputs a high-precision angle prediction.

[0077] This application embodiment uses a spiral iron core structure to achieve a unique and non-repeatable spatial pose and magnetic field distribution for each angle within a rotation range of 0° to 180°, eliminating the measurement dead zone problem of traditional sensors; it adopts a data-driven AI compensation algorithm to simultaneously overcome the combined interference of temperature drift, rotation speed fluctuation and iron core nonlinearity, significantly improving the measurement robustness of the system under wide temperature range and high dynamic conditions.

[0078] Although preferred embodiments have been described in this application, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of this application.

[0079] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of the embodiments of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the claims in this application and their equivalents, then this application also intends to include these modifications and variations.

Claims

1. A motor angle detection sensor, comprising a coil winding, an iron core, a rotating shaft, an excitation module, a signal conditioning module, and a signal processing module, characterized in that: The iron core is driven to rotate by a rotating shaft, which is connected to the motor under test. The rotating shaft includes a first part and a special-shaped part. The position of the first part corresponds to the primary coil. The outer contour of the cross-section of the special-shaped part is an equiangular spiral. The coil winding includes a primary coil and two secondary coils wound in the same manner. The first part of the iron core is positioned corresponding to the primary coil, and the two secondary coils are symmetrically arranged along the axis of rotation. The irregular part of the iron core is positioned corresponding to the two secondary coils, and the secondary coils are located on the outside of the iron core. The excitation module is used to provide an excitation signal to the primary coil; The signal conditioning module is used to extract the voltage values ​​of the two secondary coils and condition them into two DC voltage signals. The signal processing module receives two DC voltage signals from the signal conditioning module and determines the angle of the motor shaft under test based on the preset correspondence between the difference between the two DC voltage signals and the rotation angle of the shaft. The preset method for the correspondence between the difference of two DC voltage signals and the rotation angle of the shaft includes: Step 1: Simulate the true value θ of the core rotation angle of the motor under test under different operating ambient temperatures T and different speeds W, and compare it with the measured value U of the difference between two DC voltage signals. Divide the dataset into: Training set: 70%-80% of the data, used to train the model; Validation set: 10%-15% of the data, used to evaluate model performance during training, adjust hyperparameters, and prevent overfitting; Test set: 10%-15% of the data, used for the final evaluation of the model's generalization ability; Step 2: Standardize the input feature X, X=[U,T,W]; Step 3: Input the training and validation set data into the gradient boosting decision tree model for training, and fit a function model F of θ with respect to T, W, and U. The optimization objective of the function model F is to minimize the predicted angle value. =F(U,T,W) is the loss function between the actual angle value θ; Step 4: Evaluate the performance of the final model using test set data; The training process of the gradient boosting tree model includes iteratively executing the following steps: The k-th iteration process includes: Calculate the current model F k-1 The negative gradient r on all training samples i ; Fit a new regression decision tree f k To predict the negative gradient r i ; Add the new decision tree to the model and update the formula as follows: F k =F k-1 +η·f k ; Where η is the learning rate, a pre-defined hyperparameter with a value range of (0,1]. Wherein, the loss function is the mean squared error function: ; N is the number of samples in the training set, θ true,i This represents the i-th true value. i Let be the i-th predicted value.

2. The motor angle detection sensor according to claim 1, characterized in that: In the initial state, the outer contour of the cross-section of the irregular part of the iron core satisfies the following relationship: ， Where r0 is the inner diameter of the iron core, k is a constant, θ is the polar angle of the profile, and r(θ) is the length of the profile from the iron core axis.

3. The motor angle detection sensor according to claim 1, characterized in that: It also includes a housing, end caps located at both ends of the housing, a stator, and a coil frame. The coil frame is used to wind the primary coil, and the stator is used to wind the secondary coil. The coil frame is disposed in the stator, and the iron core passes through the coil frame and the stator.

Images