Motor rotor magnetic steel temperature estimation method and system based on detection coil

Abstract

The method and system for estimating the temperature of the motor rotor magnetic steel based on the detection coil belong to the field of motor technology and solve the problem that the temperature cannot be estimated in real time when the motor is running and the estimation accuracy is low. The method of the present invention comprises: establishing a magnetic steel residual magnetism estimation model; during the operation of the motor, real-time acquisition of the motor current signal, the motor position signal and the terminal voltage signals of several detection coils surrounding the circumference of the motor; obtaining the direct-axis current and the quadrature-axis current according to the motor current signal and the motor position signal; obtaining the direct-axis flux and the quadrature-axis flux according to the terminal voltage signals of the several detection coils and the motor position signal; obtaining the estimated value of the permanent magnet residual magnetism by using the magnetic steel residual magnetism estimation model according to the acquired direct-axis current, quadrature-axis current, direct-axis flux and quadrature-axis flux; obtaining the estimated value of the motor rotor magnetic steel temperature. The present invention is suitable for online real-time estimation of the permanent magnet temperature of the motor rotor.

Description

Technical Field

[0001] The present application relates to the field of motor technology, and in particular to online real-time estimation of the temperature of a motor rotor permanent magnet, and is mainly aimed at background applications such as electric vehicles where the working conditions are complex and the permanent magnet temperature is critical and needs to be monitored. Background Art

[0002] In order to meet the power requirements of electric vehicles, drive motors often have the characteristics of high power density and high efficiency. However, the reliability and safety of such drive systems will also face severe challenges. On the one hand, the working conditions of electric vehicle drive motors are harsh, and they are frequently subjected to instantaneous overload shocks. The thermal load of the motor is usually high. On the other hand, in order to meet the requirements of waterproof and dustproof, automobile drive motors generally adopt a fully enclosed structure, which makes it difficult to design the heat dissipation and cooling structure of the motor, and easily leads to excessive temperature of the motor permanent magnet, and the internal heat cannot be effectively dissipated, and then demagnetization failure occurs, resulting in a significant reduction in the electromagnetic performance of the motor. Therefore, it is necessary to effectively monitor the temperature of the permanent magnet, which can not only improve the overall reliability level of the motor, but also accurate permanent magnet temperature estimation can make the motor control more accurate, and ultimately improve the performance of the entire electric drive system.

[0003] Traditional methods for monitoring the temperature of permanent magnets of motors are mainly carried out in three ways. First, the temperature of permanent magnets of motors is monitored by embedding thermocouples, thermometers and other temperature measuring elements in advance. However, due to the limitation of motor structure, this method is not conducive to sensor arrangement and is not suitable for motor operation status. Some studies have used wireless data transmission, but the structure is often complex and easily interfered by the strong magnetic field of the rotor. Second, a lumped parameter model is constructed through the thermal circuit model and the motor loss model, and the model parameters are adjusted through experience and experimental results. The model is implanted into the motor control program to estimate the motor rotor temperature in real time during motor operation. The disadvantage of this method is that the model parameters are difficult to accurately adjust, the estimation effect is heavily dependent on experience, and there is a great deal of randomness. Third, by measuring the motor back electromotive force, the magnetic flux of the permanent magnet is indirectly calculated, and the temperature of the permanent magnet of the motor is thus inferred. This method is limited by the multi-coupled nonlinear complex motor model. Under different motor operation conditions, the motor model will undergo significant changes. Therefore, the magnetic flux of the permanent magnet is calculated only through a simple voltage equation, and the temperature error in the inferred temperature is large. Summary of the invention

[0004] The purpose of the present invention is to solve the problem that the prior art cannot perform real-time temperature estimation when the motor is running and the estimation accuracy is low, and a method and system for estimating the temperature of a motor rotor magnet based on a detection coil are provided.

[0005] The present invention is implemented by the following technical solutions. On the one hand, the present invention provides a method for estimating the temperature of a motor rotor magnetic steel based on a detection coil, the method comprising:

[0006] Step 1: Establish a magnetic steel remanence estimation model, including:

[0007] Acquire measurement data of the motor under different permanent magnet remanence conditions and permanent magnet remanence, wherein the measurement data includes direct-axis current, quadrature-axis current, direct-axis flux linkage, and quadrature-axis flux linkage of the motor under different permanent magnet remanence conditions;

[0008] Using a multi-layer feedforward neural network, nonlinear fitting is performed on the measurement data and the remanent magnetization of the permanent magnet;

[0009] According to the nonlinear fitting, a magnetic steel residual magnetism estimation model is established;

[0010] Step 2: During the operation of the motor, the three-phase current signal of the motor, the motor position signal and the terminal voltage signals of several detection coils around the circumference of the motor are collected in real time;

[0011] Step 3, obtaining the direct-axis current and the quadrature-axis current according to the three-phase current signal of the motor and the motor position signal;

[0012] Obtaining direct-axis flux and quadrature-axis flux according to terminal voltage signals of the plurality of detection coils and motor position signals;

[0013] Step 4: according to the direct-axis current, quadrature-axis current, direct-axis flux linkage and quadrature-axis flux linkage obtained in step 3, using the magnetic steel residual magnetism estimation model, obtain an estimated value of the permanent magnet residual magnetism;

[0014] Step 5: Obtain an estimated value of the motor rotor magnetic steel temperature based on the estimated value of the permanent magnet remanence.

[0015] Further, in step 3, the direct-axis current and the quadrature-axis current are obtained according to the three-phase current signal of the motor and the motor position signal, which specifically includes: according to the motor position signal, the three-phase current signal of the motor is subjected to Clark transformation and Park transformation to obtain the direct-axis current and the quadrature-axis current.

[0016] Further, in step 3, the direct-axis flux and the quadrature-axis flux are obtained according to the terminal voltage signals of the plurality of detection coils and the motor position signal, specifically including: passing the terminal voltage signals of the three detection coils and the motor frequency calculated from the motor position signal through a second-order generalized integrator respectively to extract three fundamental wave signals;

[0017] Integrate the three fundamental wave signals to obtain the ABC three-phase flux;

[0018] The coordinate transformation is performed on the ABC three-phase flux to obtain the direct-axis flux and the quadrature-axis flux.

[0019] Furthermore, the method for calculating the motor frequency from the motor position signal specifically includes:

[0020] The motor speed n is obtained by taking the difference of the motor position signals obtained by two samplings and dividing it by the sampling time. The motor frequency is then obtained by the following formula:

[0021]

[0022] f is the motor frequency, n is the motor speed, and p is the number of motor pole pairs.

[0023] Furthermore, the ABC three-phase flux specifically includes the A-phase, B-phase, and C-phase fluxes of the motor, and the axes of the three detection coils are aligned with the A-phase, B-phase, and C-phase main magnetic fluxes, respectively.

[0024] Further, obtaining the ABC three-phase flux linkage according to the plurality of fundamental wave signals specifically includes:

[0025] According to the several fundamental wave signals, using the formula

[0026]

[0027] Among them, u sc is the fundamental signal of the voltage at the detection coil end, ψ is the flux linkage of one phase in the three-phase ABC motor, N1 is the number of turns of each coil in one phase winding in the three-phase ABC motor, N2 is the number of coils in one phase winding, a is the number of parallel branches of the motor winding, ψ is the flux linkage,

[0028] Get the ABC three-phase flux.

[0029] Furthermore, the multi-layer feedforward neural network adopts a three-layer network, the first layer is the input layer, and the number of input nodes is set to 4; the second layer is the hidden layer, and the number of nodes is 10, which is used to realize any nonlinear mapping between input and output; the third layer is the output layer, and the number of output nodes is 1, and the error is used to recursively correct the weights and thresholds of each node between layers layer by layer, and output is performed at the same time.

[0030] Furthermore, the magnetic steel residual magnetism estimation model is trained by the fastest gradient descent algorithm, the weights and bias coefficients of the multi-layer feedforward neural network are adjusted, and finally the corresponding relationship between the permanent magnet residual magnetism and the working state of the motor is constructed.

[0031] Furthermore, step 5 specifically includes:

[0032] According to the estimated value of the permanent magnet remanence, according to the formula

[0033]

[0034] Among them, B1 is the estimated value of the permanent magnet remanence, B0 is the remanence of the permanent magnet at room temperature, t0 is room temperature, α Br is the temperature coefficient of the remanence of the permanent magnet material changing with temperature,

[0035] Get the estimated value t1 of the motor rotor magnetic steel temperature.

[0036] In a second aspect, the present invention provides a motor rotor magnetic steel temperature estimation system based on a detection coil, the system comprising: an online acquisition module, a signal processing module, a magnetic steel residual magnetism estimation model and a temperature conversion module;

[0037] The online acquisition module is used to collect the three-phase current signal of the motor, the motor position signal and the terminal voltage signal of several detection coils around the circumference of the motor in real time during the operation of the motor;

[0038] The signal processing module is used to obtain the direct-axis current and the quadrature-axis current according to the three-phase current signal of the motor and the motor position signal;

[0039] It is also used to obtain the direct-axis flux and the quadrature-axis flux according to the terminal voltage signals of the plurality of detection coils and the motor position signal;

[0040] The magnetic steel residual magnetism estimation model is used to obtain an estimated value of the permanent magnet residual magnetism by using the magnetic steel residual magnetism estimation model according to the direct-axis current, quadrature-axis current, direct-axis flux linkage and quadrature-axis flux linkage obtained in the signal processing module;

[0041] The temperature conversion module is used to obtain an estimated value of the motor rotor magnetic steel temperature based on the estimated value of the permanent magnet remanence.

[0042] Beneficial effects of the present invention:

[0043] Firstly, the present invention measures the temperature based on the detection coil. When the motor is running, the information of the magnetic field is directly monitored, and relevant data for estimating the temperature can be collected in real time, thereby estimating the temperature in real time.

[0044] Secondly, the present invention performs temperature measurement based on the detection coil, which can directly monitor the information of the magnetic field and obtain the magnetic flux information more accurately. Moreover, since the detection coil is in an open circuit state, the terminal voltage is the back electromotive force, which can avoid the error caused by the coil resistance. The traditional indirect temperature measurement method based on the motor back electromotive force is limited by this point because the temperature will also affect the resistance value.

[0045] Finally, the present invention establishes the relationship between the motor flux, working point and permanent magnet remanence through offline training of a multi-layer feedforward neural network, and finally derives the permanent magnet remanence size and current temperature, which is of great significance to the reliability of the motor and high-precision motor control.

[0046] The present invention is suitable for online real-time estimation of the temperature of a permanent magnet of a motor rotor, and is mainly intended for background applications such as electric vehicles where the working conditions are complex and the permanent magnet temperature is critical and needs to be monitored. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solution of the present application, the drawings required for use in the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0048] Figure 1 It is a block diagram of the SOGI structure of the present invention;

[0049] Figure 2 It is a module block diagram of the online estimation system of the rotor magnetic steel temperature of the permanent magnet synchronous motor of the present invention;

[0050] Figure 3 Schematic diagram of the method for online estimation of the rotor magnetic steel temperature of a permanent magnet synchronous motor of the present invention

[0051] Figure 4 This is a detection coil arrangement diagram of an 8-pole 48-slot permanent magnet synchronous motor (single-layer winding, pitch 5) of the present invention. DETAILED DESCRIPTION

[0052] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0053] Implementation method 1: Figure 3 As shown, a method for estimating the temperature of a motor rotor magnetic steel based on a detection coil, the method comprising:

[0054] Step 1: Establish a magnetic steel remanence estimation model, including:

[0055] Acquire measurement data of the motor under different permanent magnet remanence conditions and permanent magnet remanence, wherein the measurement data includes direct-axis current, quadrature-axis current, direct-axis flux linkage, and quadrature-axis flux linkage of the motor under different permanent magnet remanence conditions;

[0056] Using a multi-layer feedforward neural network, nonlinear fitting is performed on the measurement data and the remanent magnetization of the permanent magnet;

[0057] According to the nonlinear fitting, a magnetic steel residual magnetism estimation model is established;

[0058] Step 2: During the operation of the motor, the three-phase current signal of the motor, the motor position signal and the terminal voltage signals of several detection coils around the circumference of the motor are collected in real time;

[0059] Step 3, obtaining the direct-axis current and the quadrature-axis current according to the three-phase current signal of the motor and the motor position signal;

[0060] Obtaining direct-axis flux and quadrature-axis flux according to terminal voltage signals of the plurality of detection coils and motor position signals;

[0061] It should be noted that the motor position signal is used to calculate the motor speed, thereby calculating the motor frequency, and then inputting it into SOGI for filtering; the position signal is also used to obtain the direct-axis current, quadrature-axis current and direct-axis flux, quadrature-axis flux through coordinate transformation of the three-phase current and three-phase flux.

[0062] Step 4: according to the direct-axis current, quadrature-axis current, direct-axis flux linkage and quadrature-axis flux linkage obtained in step 3, using the magnetic steel residual magnetism estimation model, obtain an estimated value of the permanent magnet residual magnetism;

[0063] Step 5: Obtain an estimated value of the motor rotor magnetic steel temperature based on the estimated value of the permanent magnet remanence.

[0064] In this embodiment, a method for online monitoring of the motor rotor magnetic steel temperature based on a detection coil is proposed. The magnetic field inside the motor is directly monitored through specially arranged detection coils. Accurate magnetic flux information can be obtained by combining the relationship between the back electromotive force of the detection coil and the winding parameters. The relationship between the motor magnetic flux, the operating point and the permanent magnet remanence is established through an offline trained multi-layer feedforward neural network. Finally, the permanent magnet remanence size and the current temperature are derived. The motor rotor magnetic steel temperature can be estimated in real time, while the estimation accuracy is improved.

[0065] It should be noted that the present embodiment can be provided with three detection coils, each of which is wound by a turn of copper wire and wound in a specified slot of the motor stator, with both ends connected to the sampling circuit for data collection. Figure 4 The detection coil terminal voltage contains the internal magnetic flux information of the motor, and its principle is Faraday's law of electromagnetic induction:

[0066]

[0067] Where N is the number of turns, Φ is the magnetic flux, and u is the back EMF.

[0068] This formula represents the basic principle of the detection coil. It can be proved that it is reasonable to use the signal of the detection coil to represent the three-phase signal of the motor.

[0069] Embodiment 2: This embodiment further limits the motor rotor magnetic steel temperature estimation method based on the detection coil described in embodiment 1. In this embodiment, in step 3, the direct-axis current and the quadrature-axis current are obtained according to the three-phase current signal of the motor and the motor position signal, and is further limited, specifically including:

[0070] According to the motor position signal, the three-phase current signal of the motor is subjected to Clark transformation and Park transformation to obtain the direct-axis current and the quadrature-axis current.

[0071] This embodiment provides a method for processing the motor's own signal, specifically:

[0072] First, the three-phase current of the motor is converted into dq-axis current through coordinate transformation:

[0073] Clark coordinate transformation:

[0074]

[0075] In the formula, i a 、i b 、i c are the A, B, and C phase currents in the motor three-phase coordinate system; i α 、i β are the α-axis and β-axis currents of the motor in the two-phase stationary coordinate system;

[0076] Park Transform:

[0077]

[0078] In the formula, i d 、i q They are the direct-axis current and quadrature-axis current in the motor rotating coordinate system respectively; θ is the motor position angle, that is, the motor position signal, which can be collected by the rotary transformer.

[0079] Implementation method three, this implementation method is a further limitation of the motor rotor magnetic steel temperature estimation method based on the detection coil described in implementation method one. In this implementation method, in step 3, the direct-axis flux and the quadrature-axis flux are obtained according to the terminal voltage signals of the plurality of detection coils and the motor position signal, and further limitations are made, specifically including:

[0080] The terminal voltage signals of the three detection coils and the motor frequency calculated from the motor position signal are respectively passed through a second-order generalized integrator to extract three fundamental wave signals;

[0081] Integrate the three fundamental wave signals to obtain the ABC three-phase flux;

[0082] The coordinate transformation is performed on the ABC three-phase flux to obtain the direct-axis flux and the quadrature-axis flux.

[0083] In this implementation, a method for processing the detection coil data is provided, specifically:

[0084] Extracting fundamental wave: SOGI (Second Order Generalized Integrator) is a mature tool, and its basic principle is as follows:

[0085] Input the original signal v and the specified frequency ω, and get two sets of signals with a phase difference of 90°. After filtering, the signal v d 、v q (v q It is the signal of another channel output by SOGI, with the same amplitude as vd, but with a phase shift of 90°).

[0086] like Figure 1 As shown, based on this block diagram, the voltage signal and fundamental frequency of the detection coil are input, and the output is v d That is the fundamental signal of the detection coil terminal voltage, which realizes filtering.

[0087] The principle of obtaining the magnetic flux by integrating the terminal voltage: According to Faraday's law of electromagnetic induction, formula (1),

[0088]

[0089] From equation (4), we can see that the differential of the flux is voltage. Therefore, the flux passing through the detection coil can be inversely calculated by integrating the voltage at the detection coil end.

[0090] It should be noted that the steps of converting the three-phase flux into the dq-axis flux are similar to the above-mentioned three-phase current through Clark transformation and Park transformation. The ABC three-phase flux is also converted into the dq-axis flux through these two coordinate transformations.

[0091] It should be noted that the ABC three-phase flux is the flux of phase A, phase B, and phase C of the motor, which is calculated by multiplying the flux of the three detection coils by a ratio.

[0092] Embodiment 4: This embodiment further limits the method for estimating the temperature of the motor rotor magnetic steel based on the detection coil described in Embodiment 3. In this embodiment, the method for calculating the motor frequency obtained by the motor position signal is further limited, specifically including:

[0093] The motor speed n is obtained by taking the difference of the motor position signals obtained by two samplings and dividing it by the sampling time. The motor frequency is then obtained by the following formula:

[0094]

[0095] f is the motor frequency, n is the motor speed, and p is the number of motor pole pairs.

[0096] This embodiment provides a specific method for obtaining the motor frequency. Subsequently, when the voltage signals at the three detection coil terminals are input into SOGI, the corresponding motor frequency needs to be input for filtering to obtain the fundamental wave signal.

[0097] Embodiment 5: This embodiment further limits the motor rotor magnetic steel temperature estimation method based on the detection coil described in Embodiment 3. In this embodiment, the ABC three-phase flux linkage is further limited, specifically including:

[0098] The ABC three-phase flux specifically includes the A-phase, B-phase, and C-phase fluxes of the motor, and the axes of the three detection coils are aligned with the A-phase, B-phase, and C-phase main magnetic fluxes, respectively.

[0099] In this embodiment, a method for obtaining the flux of the three phases ABC is provided. According to the special arrangement of the detection coils, the axes of the three coils are aligned with the main magnetic flux of the A phase, the B phase, and the C phase respectively, so that the flux information of the three coils can represent the flux of the three phases ABC.

[0100] Embodiment 6: This embodiment further limits the method for estimating the temperature of the motor rotor magnetic steel based on the detection coil described in Embodiment 3. In this embodiment, the method for obtaining the ABC three-phase flux linkage according to the several fundamental wave signals is further limited, specifically including:

[0101] According to the several fundamental wave signals, using the formula

[0102]

[0103] Among them, u sc is the fundamental signal of the detection coil terminal voltage, ψ is the flux linkage of one phase in the three-phase ABC motor, N1 is the number of turns of each coil in one phase winding in the three-phase ABC motor, N2 is the number of coils in one phase winding, a is the number of parallel branches of the motor winding, ψ is the flux linkage,

[0104] Get the ABC three-phase flux.

[0105] In this implementation, because it is necessary to consider the different number of turns of the detection coil and the motor winding (the different number of turns will affect the calculated flux), the total number of coils in one-phase winding and the number of parallel winding branches, the final expression of the flux from the detection coil terminal voltage to the one-phase flux of the motor is obtained as formula (5). This implementation can obtain accurate ABC three-phase flux, thereby improving the estimation accuracy of the motor rotor magnetic steel temperature.

[0106] Embodiment 7: This embodiment further limits the motor rotor magnetic steel temperature estimation method based on the detection coil described in Embodiment 1. In this embodiment, the multi-layer feedforward neural network is further limited, specifically including:

[0107] The multi-layer feedforward neural network adopts a three-layer network. The first layer is the input layer, and the number of input nodes is set to 4; the second layer is the hidden layer, and the number of nodes is 10, which is used to realize any nonlinear mapping between input and output; the third layer is the output layer, and the number of output nodes is 1. The error is used to recursively correct the weights and thresholds of each node between layers layer by layer, and output is performed at the same time.

[0108] This embodiment provides a design method for a multi-layer feedforward neural network. It is felt that the number of design nodes for the practical application of the present invention is consistent with the input and output dimensions and is convenient for nonlinear fitting, and thus a magnetic steel remanence estimation model that can achieve the purpose of the present invention is provided.

[0109] Embodiment 8: This embodiment further limits the method for estimating the temperature of the motor rotor magnetic steel based on the detection coil described in Embodiment 1. In this embodiment, the magnetic steel residual magnetism estimation model is further limited, specifically including:

[0110] The magnetic steel residual magnetism estimation model is trained by the fastest gradient descent algorithm, and the weights and bias coefficients of the multi-layer feedforward neural network are adjusted to finally construct the corresponding relationship between the permanent magnet residual magnetism and the working state of the motor.

[0111] This embodiment provides a training method for a magnetic steel residual magnetism estimation model, thereby improving the estimation accuracy of the model.

[0112] Embodiment 9: This embodiment further limits the method for estimating the temperature of the motor rotor magnetic steel based on the detection coil described in embodiment 1. In this embodiment, step 5 is further limited, specifically including:

[0113] Specifically include:

[0114] According to the estimated value of the permanent magnet remanence, according to the formula

[0115]

[0116] Among them, B1 is the estimated value of the permanent magnet remanence, B0 is the remanence of the permanent magnet at room temperature, t0 is room temperature, α Br is the temperature coefficient of the remanence of permanent magnet material changing with temperature, %,

[0117] Get the estimated value t1 of the motor rotor magnetic steel temperature.

[0118] In this embodiment, a temperature conversion module is constructed based on the conversion relationship between the magnet temperature and the residual magnetism change. By inputting the residual magnetism parameter identification result obtained in the previous step, that is, the estimated value of the permanent magnet residual magnetism, the real-time temperature of the motor rotor magnet is obtained, thereby improving the estimation accuracy of the real-time temperature of the motor rotor magnet.

[0119] It should be noted that the remanence B0 of the permanent magnet at room temperature is derived from the magnetic steel material handbook.

[0120] Embodiment 10: A motor rotor magnetic steel temperature estimation system based on a detection coil, the system comprising:

[0121] The online acquisition module is used to collect the three-phase current signal of the motor, the motor position signal and the terminal voltage signal of several detection coils around the circumference of the motor in real time during the operation of the motor;

[0122] The signal processing module is used to obtain the direct-axis current and the quadrature-axis current according to the three-phase current signal of the motor and the motor position signal;

[0123] It is also used to obtain the direct-axis flux and the quadrature-axis flux according to the terminal voltage signals of the plurality of detection coils and the motor position signal;

[0124] The magnetic steel residual magnetism estimation module is used to obtain an estimated value of the permanent magnet residual magnetism according to the direct-axis current, quadrature-axis current, direct-axis flux linkage and quadrature-axis flux linkage obtained in the signal processing module and using the magnetic steel residual magnetism estimation model;

[0125] The temperature conversion module is used to obtain an estimated value of the motor rotor magnetic steel temperature based on the estimated value of the permanent magnet remanence.

[0126] In this embodiment, if Figure 2 As shown, the motor rotor magnetic steel temperature estimation system based on the detection coil of this embodiment,

[0127] First, the system of this embodiment measures temperature based on the detection coil. When the motor is running, the magnetic field information is directly monitored, and relevant data for estimating temperature can be collected in real time, thereby estimating the temperature in real time.

[0128] Secondly, the system of this embodiment performs temperature measurement based on the detection coil, which can directly monitor the information of the magnetic field and obtain the magnetic flux information more accurately; and because the detection coil is in an open circuit state, the terminal voltage is the back electromotive force, which can avoid the error caused by the coil resistance. The traditional indirect temperature measurement method based on the motor back electromotive force is limited by this point because temperature will also affect the resistance value.

[0129] Finally, the system of this embodiment establishes the relationship between the motor flux, working point and permanent magnet remanence through offline training of a multi-layer feedforward neural network, and finally derives the permanent magnet remanence size and current temperature, which is of great significance to the reliability of the motor and high-precision motor control.

[0130] Implementation eleven, this implementation is a specific embodiment of a method for estimating the temperature of a motor rotor magnetic steel based on a detection coil as described above, and specifically includes:

[0131] Step 1, data collection: The three-phase current signal and motor position signal of the motor during operation, as well as the terminal voltage signal of the detection coil around the circumference of the motor are collected in real time through the current sensor, rotary transformer and sampling circuit, and all the collected data are used to build a database (for data management) in time series;

[0132] Step 2: Signal processing: Based on the data collected in step 1, the following data processing needs to be done in real time:

[0133] 1) Motor’s own signal processing:

[0134] 1. Process the three-phase current signal and motor position signal obtained in step 1 in the following order:

[0135] 2. The three-phase currents ABC are transformed by Clark and Park to obtain the direct axis current (d-axis current) i d and the quadrature axis current (q-axis current) i q ;

[0136] 3. Subtract the motor position signals obtained by two samplings and divide them by the sampling time to get the motor speed n, and then get the motor frequency by the following formula:

[0137]

[0138] f is the motor frequency, n is the motor speed, and p is the number of motor pole pairs.

[0139] 2) Detection coil data processing:

[0140] The terminal voltage signals of the three detection coils obtained in step 1 are processed in sequence as follows:

[0141] 1. The three detection coil signals and the motor frequency calculated from the motor position signal are respectively passed through SOGI (second-order generalized integrator) to extract the fundamental signal of each signal;

[0142] 2. Integrate the three fundamental wave signals to obtain the A-phase flux, B-phase flux and C-phase flux (i.e., ABC three-phase flux) respectively;

[0143] 3. Perform coordinate transformation (Clark transformation, Park transformation) on the three-phase flux to obtain the direct-axis flux (d-axis flux) and the quadrature-axis flux (q-axis flux).

[0144] Step 3: Magnetic steel residual magnetism estimation module construction: According to the data of motor dq axis current and dq axis flux linkage (i.e. motor direct axis current i d , quadrature axis current i q 、Direct axis flux ψ d 、cross-axis flux ψ q and the corresponding permanent magnet remanence Br), nonlinear fitting is performed through a multi-layer feedforward neural network (the input is direct-axis current, quadrature-axis current, direct-axis flux linkage, quadrature-axis flux linkage, and the output is permanent magnet remanence Br), training is performed through the fastest gradient descent algorithm, the weights and bias coefficients of the multi-layer feedforward neural network are adjusted, and finally the corresponding relationship between the permanent magnet remanence and the working state of the motor is constructed;

[0145] The multi-layer feedforward neural network adopts a three-layer network, the first layer is the input layer, because the input is 4 variables, namely: direct-axis current, quadrature-axis current, direct-axis flux, quadrature-axis flux, so the number of input nodes is set to 4, this layer is only used to transmit signals to the next layer, the second layer is the hidden layer, the number of nodes is 10, which is used to realize any nonlinear mapping between input and output, the third layer is the output layer, because there is only one output permanent magnet remanence Br, so the number of output nodes is 1, and the weights and thresholds of each node between layers are corrected layer by layer by using the error recursion, and the output is performed at the same time.

[0146] Step 4, estimation of rotor magnetic steel residual magnetism parameters: The data collected in step 1 is calculated in step 2 to obtain the dq axis current and dq axis flux of the motor in the current state, and input into the residual magnetism estimation module built in step 3 (the input is the motor direct axis current, quadrature axis current, direct axis flux, quadrature axis flux, and the output is the permanent magnet residual magnetism Br), and the magnetic steel residual magnetism parameters are identified to obtain the estimated value of the permanent magnet residual magnetism;

[0147] Step 5: Real-time estimation of rotor magnetic steel temperature: Based on the conversion relationship between magnetic steel temperature and residual magnetism change, a temperature conversion module is built. By inputting the residual magnetism parameter identification result obtained in step 4, the real-time temperature of the motor rotor magnetic steel is obtained, as follows:

[0148] The current temperature of the magnetic steel is obtained from the remanence of the permanent magnet:

[0149]

[0150] Among them, B1 is the estimated value of the permanent magnet remanence, that is, the remanence parameter identification result obtained in step 4, B0 is the remanence of the permanent magnet at room temperature, which comes from the magnetic steel material manual, t0 is room temperature, α Bris the temperature coefficient of the remanence of the permanent magnet material as it changes with temperature, %.

[0151] The working principle of the present invention is specifically given below to verify the technical effect of the present invention:

[0152] According to the specific detection coil arrangement (see Figure 4 , the three coils lead the outgoing wire segments from the inside of the motor), so that the detection coils can be arranged on the axis of the motor's three-phase main magnetic flux A, B, and C, and the magnetic field information can be directly monitored, as shown in formula (1). Based on the relationship between the detection coil and the winding parameters, the three-phase magnetic flux information (ψ A , B , C ), as shown in equation (5). Then, by coordinate transformation, we can obtain the direct-axis flux and the cross-axis flux (ψ d , q ).

[0153] In the dq rotating coordinate system, the voltage equation of the permanent magnet synchronous motor is:

[0154]

[0155] Where u d 、u q are the direct-axis and quadrature-axis voltages in the two-phase rotating coordinate system;

[0156] i d 、i q are the direct-axis and quadrature-axis currents in the two-phase rotating coordinate system;

[0157] ψ d , q are the direct-axis and quadrature-axis flux linkages in the two-phase rotating coordinate system;

[0158] R s is the resistance of one phase winding of the motor;

[0159] ω e is the electrical angular velocity of the motor.

[0160] Among them, the direct axis and quadrature axis flux ψ d , q and the AC and DC axis current i d 、i q The inductance parameter is directly related to the performance of the permanent magnet. Since the inductance parameter is determined by factors such as the motor structure, size parameters, and motor saturation state, it has strong nonlinearity, and there is cross-coupling between the AC and DC axes, that is, they affect each other, which makes it difficult to accurately express it with a mathematical formula. Therefore, the relationship between the flux linkage, current, and permanent magnet parameters (permanent magnet remanence size Br) is expressed as the following nonlinear function relationship:

[0161]

[0162] Therefore, the mapping relationship between flux linkage, current and permanent magnet remanence Br is established as shown in the following formula:

[0163] Br=g(i d ,i q ,ψ d ,ψ q ) (9)

[0164] Because the above relationship is highly nonlinear, the neural network is used to fit the input and output variables, and the current state of the motor (known i d 、i q , d , dq ), the permanent magnet remanence Br is calculated, and the permanent magnet temperature is the factor that has the greatest impact on the permanent magnet remanence. Therefore, based on the permanent magnet remanence Br, the size of the motor permanent magnet remanence can be inferred through formula (10), and the value of the permanent magnet temperature can be calculated accordingly.

[0165]

[0166] In the formula, B1 refers to the currently identified residual magnetism, B0 refers to the residual magnetism of the permanent magnet at room temperature, which is derived from the magnetic steel material manual, t0 refers to room temperature, and α Br is the temperature coefficient of the remanence of the permanent magnet material as it changes with temperature, %).

[0167] The following is how to obtain the above input variables:

[0168] ① Current: can be directly collected and calculated through steps 1 and 2;

[0169] ② Magnetic flux: By integrating and multiplying the terminal voltage signal of the detection coil, as shown in formula (5), the ABC three-phase magnetic flux can be obtained, and then the dq axis magnetic flux can be obtained through coordinate transformation.

[0170] The method and system of the present invention are based on temperature measurement of the detection coil, which can directly monitor the information of the magnetic field and obtain the magnetic flux information more accurately; and because the detection coil is in an open circuit state, the terminal voltage is the back electromotive force, which can avoid the error caused by the coil resistance. The traditional indirect temperature measurement method based on the motor back electromotive force is limited by this point because the temperature will also affect the resistance value.

[0171] The present invention proposes a method and system for online monitoring of the temperature of the motor rotor magnetic steel based on a detection coil. The method and system directly monitor the internal magnetic field of the motor through a specially arranged detection coil, and can obtain accurate flux information by combining the relationship between the back electromotive force of the detection coil and the winding parameters. Then, the relationship between the motor flux, the working point and the residual magnetism of the permanent magnet is established through a multi-layer feedforward neural network trained offline, and finally the residual magnetism size of the permanent magnet and the current temperature are derived, which is of great significance to the reliability of the motor and high-precision motor control.

Claims

1. A method for estimating the temperature of a motor rotor magnetic steel based on a detection coil, characterized in that: The method comprises: Step 1: Establish a magnetic steel remanence estimation model, including: Acquire measurement data of the motor under different permanent magnet remanence conditions and permanent magnet remanence, wherein the measurement data includes direct-axis current, quadrature-axis current, direct-axis flux linkage, and quadrature-axis flux linkage of the motor under different permanent magnet remanence conditions; Using a multi-layer feedforward neural network, nonlinear fitting is performed on the measurement data and the remanent magnetization of the permanent magnet; According to the nonlinear fitting, a magnetic steel residual magnetism estimation model is established; Step 2: During the operation of the motor, the three-phase current signal of the motor, the motor position signal and the terminal voltage signals of several detection coils around the circumference of the motor are collected in real time; Step 3, obtaining the direct-axis current and the quadrature-axis current according to the motor three-phase current signal and the motor position signal; Obtaining direct-axis flux and quadrature-axis flux according to terminal voltage signals of the plurality of detection coils and motor position signals; Step 4: according to the direct-axis current, quadrature-axis current, direct-axis flux linkage and quadrature-axis flux linkage obtained in step 3, using the magnetic steel residual magnetism estimation model, obtain an estimated value of the permanent magnet residual magnetism; Step 5: Obtain an estimated value of the motor rotor magnetic steel temperature according to the estimated value of the permanent magnet remanence; Step 5 specifically includes: According to the estimated value of the permanent magnet remanence, according to the formula Among them, B1 is the estimated value of the permanent magnet remanence, B0 is the remanence of the permanent magnet at room temperature, t0 is room temperature, α Br is the temperature coefficient of the remanence of the permanent magnet material changing with temperature, Get the estimated value t1 of the motor rotor magnetic steel temperature.

2. The method for estimating the temperature of a motor rotor magnetic steel based on a detection coil according to claim 1, characterized in that: In step 3, the direct-axis current and the quadrature-axis current are obtained according to the three-phase current signal of the motor and the motor position signal, which specifically includes: according to the motor position signal, the three-phase current signal of the motor is subjected to Clark transformation and Park transformation to obtain the direct-axis current and the quadrature-axis current.

3. The method for estimating the temperature of a motor rotor magnetic steel based on a detection coil according to claim 1, characterized in that: In step 3, the direct-axis flux and the quadrature-axis flux are obtained according to the terminal voltage signals of the plurality of detection coils and the motor position signal, which specifically includes: passing the terminal voltage signals of the three detection coils and the motor frequency calculated from the motor position signal through a second-order generalized integrator respectively to extract three fundamental wave signals; Integrate the three fundamental wave signals to obtain the ABC three-phase flux; The coordinate transformation is performed on the ABC three-phase flux to obtain the direct-axis flux and the quadrature-axis flux.

4. The method for estimating the temperature of a motor rotor magnetic steel based on a detection coil according to claim 3, characterized in that: The method for calculating the motor frequency from the motor position signal specifically includes: The motor speed n is obtained by taking the difference of the motor position signals obtained by two samplings and dividing it by the sampling time. The motor frequency is then obtained by the following formula: f is the motor frequency, n is the motor speed, and p is the number of motor pole pairs.

5. The method for estimating the temperature of a motor rotor magnetic steel based on a detection coil according to claim 3, characterized in that: The ABC three-phase flux specifically includes the A-phase, B-phase, and C-phase fluxes of the motor, and the axes of the three detection coils are aligned with the A-phase, B-phase, and C-phase main magnetic fluxes, respectively.

6. The method for estimating the temperature of a motor rotor magnetic steel based on a detection coil according to claim 3, characterized in that: Integrating the three fundamental wave signals to obtain the ABC three-phase flux specifically includes: According to the fundamental signal, using the formula Among them, u sc is the fundamental signal of the voltage at the detection coil end, ψ is the flux linkage of one phase in the three-phase ABC motor, N1 is the number of turns of each coil in one phase winding in the three-phase ABC motor, N2 is the number of coils in one phase winding, a is the number of parallel branches of the motor winding, ψ is the flux linkage, Get the ABC three-phase flux.

7. The method for estimating the temperature of a motor rotor magnetic steel based on a detection coil according to claim 1, characterized in that: The multi-layer feedforward neural network adopts a three-layer network, the first layer is the input layer, and the number of input nodes is set to 4; The second layer is the hidden layer with 10 nodes, which is used to realize any nonlinear mapping between input and output. The third layer is the output layer with 1 output node. The error is used to recursively correct the weights and thresholds of each node between layers and output them at the same time.

8. The method for estimating the temperature of a motor rotor magnetic steel based on a detection coil according to claim 1, characterized in that: The magnetic steel residual magnetism estimation model is trained by the fastest gradient descent algorithm, and the weights and bias coefficients of the multi-layer feedforward neural network are adjusted to finally construct the corresponding relationship between the permanent magnet residual magnetism and the working state of the motor.

9. A motor rotor magnetic steel temperature estimation system based on a detection coil based on the motor rotor magnetic steel temperature estimation method based on a detection coil according to claim 1, characterized in that: The system comprises: an online acquisition module, a signal processing module, a magnetic steel residual magnetism estimation module and a temperature conversion module; The online acquisition module is used to collect the three-phase current signal of the motor, the motor position signal and the terminal voltage signal of several detection coils around the circumference of the motor in real time during the operation of the motor; The signal processing module is used to obtain the direct-axis current and the quadrature-axis current according to the three-phase current signal of the motor and the motor position signal; It is also used to obtain the direct-axis flux and the quadrature-axis flux according to the terminal voltage signals of the plurality of detection coils and the motor position signal; The magnetic steel residual magnetism estimation module is used to obtain an estimated value of the permanent magnet residual magnetism according to the direct-axis current, quadrature-axis current, direct-axis flux linkage and quadrature-axis flux linkage obtained in the signal processing module and using the magnetic steel residual magnetism estimation model; The temperature conversion module is used to obtain an estimated value of the motor rotor magnetic steel temperature based on the estimated value of the permanent magnet remanence.

Images