Method for detecting demagnetization of a plurality of permanent magnets fitted on a rotor of an electrical machine

Abstract

The present invention relates to a method for detecting demagnetization of at least one permanent magnet of a rotor of an electrical machine, comprising acquiring (ACQ) the voltages, currents and position and / or speed of the rotor, along with at least one temperature (for example that of the winding or that of the stator or rotor), and wherein two different estimates of the magnetic flux of the permanent magnets and of the temperature of the permanent magnets and of their standard deviation are carried out in order to detect demagnetization of the permanent magnets (Dem) by way of comparison (COMP).

Description

[0001]METHOD FOR DETECTING DEMAGNETIZATION OF A PLURALITY OF PERMANENT MAGNETS EQUIPPED IN A ROTOR OF AN ELECTRIC MACHINE Technical field The present invention relates to the field of monitoring and controlling electrical machines, in particular for synchronous electrical machines with smooth / salient poles and synchro-reluctant machines. These electrical machines find application in particular in the field of motor vehicles. Conventionally, an electrical machine comprises a rotor (moving part) and a stator (fixed part). The rotor is usually housed inside the stator. Generally, the stator is annular in shape and is housed inside a tubular support to be fixed there. The stator comprises magnetic flux generators, generally electrical windings. These windings are powered by a plurality (conventionally three) of electrical phases, in order to generate a rotating magnetic field.Furthermore, depending on the type of electrical machine, especially for salient-pole synchronous and synchro-reluctant electrical machines, the rotor may include permanent magnets. During operation of such an electrical machine, the stator windings are passed through with an electric current to generate the magnetic field necessary to drive the rotor in rotation. In the field of electrical machines, the permanent magnet-assisted synchro-reluctant machine is today increasingly developed in many industrial applications due to its power factor, efficiency, compactness and ability to operate at high speeds. Thanks to the permanent magnet in the rotor construction, the power factor is improved and stator losses are considerably reduced compared to synchro-reluctant machines.Furthermore, the reluctance torque created in such a machine limits the need for permanent magnets, which are still expensive, and therefore makes this type of machine cheaper to produce. However, the performance of such electrical machines can be degraded by the demagnetization of permanent magnets, which is often caused by high temperature during operation of the electrical machine or by the coupling of electrical and thermal stresses. Demagnetization of permanent magnets can lead to reduced torque production, torque ripples and vibrations, copper losses (Joule losses), and excessive heat. In extreme cases, unmonitored demagnetization can cause complete damage to the electrical machine; this can lead to a major risk, especially during operation. Therefore, the topic of real-time demagnetization detection is becoming increasingly important.Prior Art For this problem, many methods for analyzing permanent magnet demagnetization have been developed, and have focused on local demagnetization and also on global demagnetization of permanent magnets. These methods can be divided into two main groups: direct methods via measurement, or indirect methods via estimation techniques. Direct methods via measurement detect the demagnetization state by direct measurement with Gaussmeters or Hall effect sensors. The following papers illustrate direct methods: J. Hong, D. Hyun, and S. B. Lee, “Automated monitoring of magnet synchronous motors at standstill,” IEEE Trans. Ind. Appl., 46 (4): 1397 - 1405, 2010. IEC 62.2-2004. “IEEE guide for diagnostic field testing of electric power apparatus-electrical machinery,” 2004. D. Reigosa, D. Fernández, Y. Park, AB Diez, SB Lee and F.Briz, “Detection of Demagnetization in Permanent Magnet Synchronous Machines Using Hall-Effect Sensors”, IEEE Transactions on Industry Applications, vol. 54, Issue 4, pp: 3338 – 3349, 2018. However, these methods require dedicated instrumentation, thus the disassembly of the electrical machine or specific fabrication for the installation of the instrumentation. In addition, these methods generate additional costs related to the devices and the installation. The second group of methods analyzes demagnetization using back-electromotive force, or by high-frequency signal injection, or by stator flux observers. The following papers illustrate indirect methods by estimation techniques: Y. Gritli, C. Rossi, D. Casadei, L. Zarri, F.Filippetti, “Demagnetization diagnosis for permanent magnet synchronous motors based on advanced wavelet analysis”, Proceedings of the 2012 International Conference on Electrical Machines, Marseille, France, pp.2397–2403, 2012. J. Hong, “Detection and classification of rotor demagnetization and eccentricity faults for PM synchronous motors”. IEEE Trans. Ind. Appl.48, pp.923–932, 2012. X. Xiao, C. Chen, M. Zhang, “Magnet demagnetization observation or permanent magnet synchronous motor”, Proceedings of the International Conference on Electrical Machines and Systems, pp.3216–3219, 2008. Y. Min, W. Huang, J. Yang, Y. Zhao, “On-line estimation of permanent-magnet flux and temperature rise in stator winding for PMSM”, Proceedings of the 22nd International Conference on Electrical Machines and Systems (ICEMS), 2019.However, these approaches suffer from the consequences of injecting high-frequency signals, or lack robustness due to the use of a single estimation source. Summary of the invention The present invention aims to detect in real time a demagnetization of the permanent magnets of an electrical machine, in a robust, simple and inexpensive manner, thus ensuring good performance of the electrical machine (without reduction of torque production, with limited ripples and vibrations of the torque, with limited copper losses and limited heat induced by the demagnetization of the permanent magnets).For this purpose, the invention relates to a method for detecting a demagnetization of at least one permanent magnet of a rotor of an electrical machine, for which the voltages, the currents and the position and / or the speed of the rotor are acquired, as well as at least one temperature (for example that of the winding or that of the stator or the rotor), and in which two different estimations of the magnetic flux of the permanent magnets as well as the temperature of the permanent magnets and their standard deviation are implemented, to detect a demagnetization of the permanent magnets. Thus, the method implements two estimations of the magnetic flux of the permanent magnets as well as the temperature of the permanent magnets, therefore the method uses two estimation sources, which makes the demagnetization detection robust.In addition, such a method uses only conventionally measured parameters of the electrical machine, so the invention does not require any specific instrumentation, which makes the invention simple and inexpensive. The invention further relates to a method and a system for controlling and / or monitoring demagnetization of the magnets of an electrical machine. The invention relates to a method for detecting demagnetization of a plurality of permanent magnets equipping a rotor of an electrical machine, said electrical machine comprising a stator provided with a winding and said rotor provided with said plurality of permanent magnets. For this method, the following steps are implemented: a. Currents and voltages in phases of said winding of said stator are acquired, as well as the position and / or the rotational speed of said rotor; b. A temperature of said winding and / or a temperature of said stator is acquired; c.A first magnetic flux of said permanent magnets and a first standard deviation of said first magnetic flux are determined by applying at least one Kalman filter to a magnetic flux model of said electrical machine, said magnetic flux model being applied to the acquired currents and voltages and to the acquired position or rotational speed of the rotor; d. A temperature of said permanent magnets and a standard deviation of said temperature of said permanent magnets are determined by applying a Kalman filter to a thermal model of said electrical machine, said thermal model of said electrical machine being applied to the acquired currents, to the acquired position or speed of the rotor and to the acquired temperature of said acquired winding or stator; e.A second magnetic flux of said permanent magnets and a second standard deviation of said second magnetic flux of said permanent magnets are determined by a magnetothermal model of said permanent magnets from said determined temperature of said permanent magnets; and f. Said first magnetic flux is compared with said second magnetic flux, and a demagnetization of said permanent magnets is detected if the absolute value of said comparison is greater than a threshold dependent on said first and second standard deviations of magnetic flux of said permanent magnets. According to one embodiment, said magnetic flux model of said electrical machine is composed of a stator magnetic flux model and a magnetic flux model of said permanent magnets.Advantageously, said first magnetic flux of said permanent magnets, and said first standard deviation of the magnetic flux of said permanent magnets are determined by applying a linear Kalman filter to said magnetic flux model of said permanent magnets, said magnetic flux model of said permanent magnets being applied to the acquired currents and voltages and to the acquired rotor position or rotational speed, as well as to the magnetic flux of said stator determined by an unscented Kalman filter applied to said stator magnetic flux model. Advantageously, said stator magnetic flux model is defined by a state model with a state equation of the form:. shape measurement: = ^I^I^^ and f = ^f^f^^, ϕ^, ϕ^ being the magnetic flux of the stator in the Park frame, Vd, Vq the voltages in the phases of the stator in the Park frame, respectively the direct voltage and the quadrature voltage, Id, Iq the currents in the phases of the stator in the Park frame, respectively the direct current and the quadrature current, ω the rotation speed of the rotor, ϕ ^^ the first magnetic flux of the permanent magnets, R the stator resistance, f d , f q the nonlinear functions linking the stator current components to the magnetic flux components in the Park frame. According to one aspect of the invention, said magnetic flux model of the permanent magnets is defined by a state model with a state equation of the form: ϕ^ ^^ = 0 and a measurement equation y^ = + RI^^ with ^ = V^, ϕ^ being the magnetic flux of the direct stator in the Park frame, V qthe quadrature voltages in the stator phases in the Park frame, I q the quadrature currents in the stator phases in the Park frame, ω the rotor rotation speed, ϕ ^^ the magnetic flux of the permanent magnets, R the stator resistance. According to one implementation, said thermal model is defined by a state model with an equation of the form: X^ = A^ω^X + Bu^t^ ety = CX with X a vector of the temperatures of the different components of the electric machine, u a vector of the Joule losses of the different components of the electric machine and the temperatures of the cooling components of the electric machine, and y = T%&said acquired temperature, A^ω^ = matrix B is a function of the phase current of the electric machine and the rotation speed of the rotor, and C the output matrix. According to one embodiment option, said magnetothermal model is written: ϕ^^^T'^^, β)^ =ϕ* ^1 − β)^T'^^ − T*^^ with ϕ^^^T'^^, β)^ the second magnetic flux of the permanent magnets, T'^^ the determined temperature of said permanent magnets, β) a thermal coefficient of said permanent magnets, T * an ambient temperature, ϕ * the magnetic flux of said permanent magnets at room temperature T * .According to one embodiment, said threshold S is written S a positive coefficient, preferably between 2 and 5 and preferably equal to 3, σ representing the standard deviation, ϕ'^^ the first magnetic flux of the permanent magnets, ϕ^^^T'^^, β)^ the second magnetic flux of the permanent magnets.Furthermore, the invention relates to a method for controlling and / or monitoring an electrical machine, said electrical machine comprising a stator provided with a winding and a rotor provided with a plurality of permanent magnets. For this method, the following steps are implemented: a. Demagnetization of the permanent magnets is detected by means of the method for detecting demagnetization of the permanent magnets according to one of the preceding characteristics; and b. Said electrical machine is controlled and / or monitored as a function of the detection of demagnetization of the permanent magnets.Furthermore, the invention relates to a system for controlling and / or monitoring an electrical machine comprising an electrical machine and a controller for implementing the control and / or monitoring method according to one of the preceding characteristics. Other characteristics and advantages of the method and the system according to the invention will appear on reading the following description of non-limiting examples of embodiments, with reference to the appended figures described below. List of figures Figure 1 illustrates the steps of the method according to a first embodiment of the invention. Figure 2 illustrates the steps of the method according to a second embodiment of the invention. Figure 3 illustrates the steps of the method according to a third embodiment of the invention. Figure 4 illustrates, for one example, a curve of the torque as a function of time. Figure 5 illustrates, for the example of Figure 4, the rotation speed of the rotor as a function of time.Figure 6 illustrates, for the example of Figures 4 and 5, the first estimation of the magnetic flux of the permanent magnets as a function of time. Figure 7 illustrates, for the example of Figures 4 to 6, the estimation of the temperature of the permanent magnets as a function of time. Figure 8 illustrates, for the example of Figures 4 to 7, the comparison of the magnetic flux estimations of the permanent magnets to detect a demagnetization of the permanent magnets. Description of the embodiments The present invention relates to a method for detecting, in real time, a demagnetization of the permanent magnets of a rotor of an electrical machine. The electrical machine comprises a rotor and a stator, the latter being equipped with windings connected to several electrical phases, for example to three electrical phases (or four, five, six, nine or twelve, etc. phases) to generate a magnetic field allowing the rotation of the rotor.The rotor comprises at least two permanent magnets for generating a magnetic field. According to one aspect of the invention, the electrical machine may be a synchronous electrical machine with smooth / salient poles or a synchro-reluctant machine. Indeed, the method is particularly suitable for these types of electrical machine, on the one hand, because the models implemented are well representative of these types of electrical machine, and because the detection of demagnetization of the permanent magnets allows the control and / or monitoring of such an electrical machine. Preferably, the electrical machine may be a synchro-reluctant machine assisted by permanent magnets. Indeed, this type of electrical machine requires precise detection of the demagnetization of the permanent magnets, in particular for their control, for their diagnosis, and for their monitoring.The method for real-time detection of demagnetization of permanent magnets comprises the following steps: 1) Acquisition of currents, voltages, position and / or rotational speed 2) Acquisition of a temperature of the winding or stator 3) Determination of a first magnetic flux of the permanent magnets 4) Determination of a temperature of the permanent magnets 5) Determination of a second magnetic flux of the permanent magnets 6) Detection of demagnetization Steps 1 and 2 are independent, and can be carried out in this order, in reverse order or simultaneously. In addition, step 3 can be carried out before, simultaneously or after steps 4 and 5. The steps of the method can be implemented by computer means, in particular by means of a controller / estimator of the electrical machine. These steps will be detailed in the rest of the description.Figure 1 describes, schematically and in a non-limiting manner, the steps of the method according to one embodiment of the invention. Firstly, ACQ currents, voltages in the phases of the stator, and the position and / or rotational speed of the rotor are acquired, as well as a temperature of the electrical machine. Then, in parallel, a first magnetic flux is determined by means of a magnetic flux model MFM of the permanent magnets, and a second magnetic flux is determined by means of a thermal model MTH of the electrical machine and a magnetothermal model MMT of the permanent magnets. The two estimates of the magnetic flux of the permanent magnets are compared COMP to deduce a demagnetization (or an absence of demagnetization) of the permanent magnets Dem. In the remainder of the description, the derivatives with respect to time are indicated by a point.The notations indexed by the mention _6 (direct) or _7 (quadrature) mean that the quantities are expressed in the Park frame (rotating frame linked to the rotor) via in particular the acquisition of the rotor position. In addition, the notations indicated with a circumflex accent designate a value estimated by a Kalman filter. In addition, the values ​​in the initial state are indicated with a 0 (t or k = 0, t being the continuous time and k the discrete time). 1) Acquisition of currents, voltages and position and / or rotational speed During this step, we acquire continuously and in real time: - The currents in the phases of the stator of the electric machine, - The voltages at the terminals of the phases of the stator of the electric machine, and - The position and / or the electrical rotational speed of the rotor of the electric machine (we recall that the electrical rotational speed can be deduced from the position of the rotor of the electric machine, in particular by the derivative with respect to time).According to one embodiment of the invention, the currents, and / or the voltages and / or the position and / or the rotational speed can be acquired from measurements. In other words, this step can comprise the measurement of at least one of the acquired signals. The measurement makes it possible to improve the accuracy of the signals (currents, voltages, position and / or rotational speed), in fact, they then correspond to those actually present within the electrical machine. These measurements can be carried out by voltage and current sensors instrumenting the electrical machine, so as to measure the currents and voltages in the phases of the electrical machine, and / or by an angular position or angular speed sensor of the rotor, so as to measure the rotational speed of the rotor.For example, for the position or angular velocity, sinusoidal signals generated by a position sensor can be acquired, and the angular position and rotational speed of the rotor can be deduced from them. Alternatively, the currents, and / or voltages and / or rotational speed can be acquired from a controller / estimator of the electric machine. This configuration makes it possible to limit the instrumentation of the electric machine. Advantageously, the voltage and current signals acquired in the phases can be transformed into voltages and current in the Park frame (rotating frame linked to the rotor), so as to determine the direct current, the quadrature current, the direct voltage, and the quadrature voltage.According to one embodiment of the invention, the electrical rotation speed ωe of the rotor can be determined from the mechanical rotation speed of the rotor, using the formula: 89 = :8, with N the number of pairs of poles of the electrical machine and ω the mechanical rotation speed of said rotor. According to an implementation of this embodiment, the rotation speed and / or the mechanical position of the rotor can be estimated, by any method known to those skilled in the art. For example, the rotation speed and / or the mechanical position can be estimated from a phase-locked loop (PLL) type method. Alternatively, the method for estimating the rotation speed and / or the mechanical position can be in accordance with those described in patent application FR 2984637.Alternatively, the rotational speed and / or the mechanical position of the rotor can be measured by means of a position / angular speed sensor placed on the electrical machine. Alternatively, the electrical rotational speed can be determined directly. 2) Acquisition of a temperature During this step, a temperature of the stator winding and / or a temperature of the stator is acquired. This acquired temperature allows the application of the thermal model to determine the temperature of the permanent magnets. According to one embodiment of the invention, the temperature of the winding and / or a temperature of the stator can be acquired by means of a temperature sensor, arranged in the winding or on the rotor. Such measurements are inexpensive and conventional for electrical machines. Alternatively, the temperature of the winding and / or a temperature of the stator can be estimated, by any method known to those skilled in the art.3) Determination of a first magnetic flux In this step, a first magnetic flux of the permanent magnets and a first standard deviation of the first magnetic flux are determined (estimated), by applying at least one Kalman filter to a magnetic flux model of the permanent magnets. The magnetic flux model is applied to the currents, voltages, and rotor speed acquired in step 1. The first magnetic flux of the permanent magnets is the magnetic flux estimated by the at least one Kalman filter applied to the magnetic flux model of the permanent magnets. In other words, it is a first estimate of the magnetic flux of the permanent magnets. The standard deviation is an estimate of the probability dispersion of the magnetic flux. The standard deviation can be defined as the square root of the variance of the magnetic flux.The magnetic flux model of the permanent magnets is a dynamic model that links the currents, voltages, speed to the magnetic flux of the permanent magnets. The model is called dynamic because it is a function of the rotational speed of the rotor and it includes the transients of its variables. According to one embodiment of the invention, the magnetic flux model can be composed of a stator magnetic flux model (also called a dynamic model of the stator magnetic flux) and a permanent magnet magnetic flux model. Preferably, these models are interconnected for a robust and accurate determination of the first magnetic flux of the permanent magnets. Figure 2 illustrates, schematically and in a non-limiting manner, the steps of the method according to this embodiment. The steps identical to the embodiment of Figure 1 are not detailed again.For this embodiment, the magnetic flux model MFM comprises a stator magnetic flux model MFS and a permanent magnet magnetic flux model MFA, the stator magnetic flux model MFS having as input the output of the permanent magnet magnetic flux model MFA and vice versa. Thus, the stator magnetic flux models MFS and the permanent magnet magnetic flux model MFA are interconnected. The dynamic model of the stator magnetic flux is a state representation of the electrical machine. It is recalled that in systems theory (and in automation), a state representation makes it possible to model a dynamic system in a matrix form, using state variables. This representation may be linear or not, continuous or discrete.The representation allows to determine the internal state and outputs of the system at any future instant if we know the state at the initial instant and the behavior of the input variables that influence the system. The dynamics of the magnetic flux of permanent magnets ϕ. ^^ is slow and can be neglected compared to those of the stator magnetic flux components ϕ ^ and ϕ ^ . This is why we can separate the estimation into two separate models: one for the two stator magnetic flux components ϕ^, ϕ^ having fast dynamics, and another for the permanent magnet magnetic flux ϕ ^^ having slow dynamics. The dynamics of the stator magnetic flux can be expressed by the following differential equation: ϕ^ ^ = V^ − RI^ + ωϕ^ With R the resistance of the stator of the electric machine, ω the electric rotation speed, ;6 the direct magnetic flux, ;7 the magnetic flux in quadrature, the time derivative of the direct magnetic flux, ; ^ 7the time derivative of the quadrature magnetic flux, ϕ ^^ the magnetic flux of the permanent magnets, Id the direct current, Iq the quadrature current, V d the forward voltage, V q the quadrature voltage. Moreover, the system of equations of the whole set of measurements can be described by: With fd and fq being nonlinear functions for the states ϕ^, ϕ^ and ϕ^^ (the nonlinear functions are known from tables or maps of the electric machine, especially for synchro-reluctant machines), V R the stator resistance voltage which can be given by Thus, in accordance with one implementation of the invention, the magnetic flux model of the stator can be defined by a state model with an equation of state of the form: For a synchronous-reluctant machine, the inductances in the d (direct) and q (quadrature) axes are also functions of the current components. The previous system is therefore linear for the state and non-linear for the measurement. The dynamic model of the magnetic flux of permanent magnets is a state representation of the electric machine. It is recalled that in systems theory (and in automatic control), a state representation makes it possible to model a dynamic system in a matrix form, using state variables. This representation can be linear or not, continuous or discrete. The representation makes it possible to determine the internal state and the outputs of the system at any future instant if we know the state at the initial instant and the behavior of the input variables that influence the system. The dynamics of the flux of permanent magnets can be expressed by an equation having extremely slow dynamics: ϕ ^ ^^ = 0Thus, in accordance with one implementation of the invention, the magnetic flux model of permanent magnets can be defined by a state model with a state equation of the form: a measurement equation = V^.For this implementation of the invention, the first magnetic flux of the permanent magnets and the first standard deviation of the first magnetic flux of the permanent magnets can be determined by applying a linear Kalman filter to the magnetic flux model of the permanent magnets, the magnetic flux model of the permanent magnets being applied to the currents, voltages acquired from the phases of the electric machine, and to the acquired rotor rotation speed, as well as to the magnetic flux of the stator determined by an unscented Kalman filter applied to the stator magnetic flux model.In other words, the magnetic flux of permanent magnets can be determined by applying a linear Kalman filter to the magnetic flux model of permanent magnets, the magnetic flux model of permanent magnets being applied to: - The currents in the acquired stator phases, - The voltages in the acquired stator phases, - The acquired rotor rotation speed, and - The stator magnetic flux determined by an unscented Kalman filter applied to the stator magnetic flux model (an unscented Kalman filter is applied because the stator magnetic flux model is nonlinear) and to the acquired signals. Applying the Kalman filter allows obtaining a state observer. In the discrete context, it is a recursive estimator: to estimate the current state, only the estimation of the previous state and the current measurements are necessary.The linear Kalman filter consists of two phases: a prediction and an update (to correct the predicted state in order to obtain a more accurate estimate). The unscented Kalman filter (UKF) is a filtering algorithm that uses a system model to estimate the current hidden state of a system, and then corrects the estimate using available sensor measurements. The unscented Kalman filter can be applied to a nonlinear system. The philosophy of the UKF differs from the extended Kalman filter (another type of Kalman filter that can be applied to a nonlinear system) in that it uses the unscented transform to directly approximate the mean and covariance of the target distribution. The unscented Kalman filter can consist of the state prediction and measurement correction steps, both of which are preceded by a preliminary step for calculating the "sigma points".Sigma points are a set of samples calculated in such a way that the mean and covariance information can be accurately propagated in the space of a nonlinear function. Such determination of the magnetic flux of permanent magnets using Kalman models and filters is a stochastic method, which is stable in a noisy or discontinuous environment. According to one embodiment of the invention, the stator magnetic flux model (in its discrete version) can depend on the magnetic flux of the permanent magnets determined at the previous time step. Since the dynamics of the stator magnetic flux is faster than the dynamics of the magnetic flux of the permanent magnets and the dynamics of the stator resistance, this embodiment allows the robustness and accuracy of the determination to be maintained.This embodiment allows the two interconnected models to be applied for real-time determination of the magnetic flux of the permanent magnets. Furthermore, according to an embodiment option of the invention, the stator magnetic flux estimator can be executed first. Thus, the permanent magnet magnetic flux estimator can be implemented with the stator magnetic flux determined at the current time step. According to an implementation of the invention, the stator magnetic flux can be determined at a higher frequency than the permanent magnet magnetic flux determination frequency. Since the stator magnetic flux dynamics are faster than the permanent magnet magnetic flux dynamics and the stator resistance dynamics, this embodiment allows the robustness and accuracy of the determination to be maintained.This implementation allows to apply the two interconnected models, for real-time determination of the magnetic flux of permanent magnets. The stator magnetic flux observer is based on the discrete stator magnetic flux model and can be defined as follows: X^k^ = A^^k − 1^X^k − 1^ + B^^k − 1^u^k − 1^ + ^^k − 1^y^^k^ = f ^XB, ϕ'^^^k^^ + η^k^where A^ = e3DE and B^ = F3D* eE^G^Gdτ with A the transition matrix of the stator magnetic flux model equation of state, T. J the discretization step of the model, ^ ^ k ^ and η ^ k ^ are zero-mean white noise with Q covariance matrices L and R L , respectively. The u^k^ command can be described by: For the rest, we use the following notations: •X'^k|k − 1^ is the estimate of X at time tB knowing its value at time tB2^.• ' X^k|k^ is the estimate of X at time t Bfrom the measurements at time t B . •P^k|k − 1^ is the covariance matrix of the error of the estimate at time tB knowing its value at time t B2^ . • P ^ k|k ^ is the covariance matrix of the error of the estimate at time t B from the measurements at time t B . The Unscented Kalman Filter (UKF) technique is applied with two steps, after an initialization step at k=0, of the state vector and the state of the covariance matrix. - First step: Prediction of the state at time k X'^k|k − 1^ = A^^k^'X^k − 1|k − ​​1^ + B^^k^u^k^P^k|k − 1^ = A^^k^P^k − 1|k − ​​1^A3^^k^ + B^^k^QLB3 ^^k^ - Second step: Correction of the state with the measurement at time k: K^k^ = PRSPS2 S^ X'^k|k^ = ' X^k|k − 1^ + K^k^^y^^k^ − mS ^ After the prediction step, the distribution of X '^k^ is given by a Gaussian distribution U^VW , XW^ with mL = X'^k|k − 1^ and PL = P^k|k − 1^ . The sigma points associated with the mean mx and the matrix Px can, for example, be calculated as follows: With b = c^^d + e^ − dWith µ a scalar parameter determining the dispersion of the sigma points, and κ a secondary rescaling parameter, and n=2, and Si is a square root of Px of the Gaussian distribution U^VW , XW^. The sigma points propagate in the measurement model in the following form, for all i between 0 and 2n:y^Z^k^ = f ^ξZ, ϕ'^^^k^^The next step can be to calculate the predicted mean my, the predicted covariance of the measurement Pyy and the cross-covariance of the measurement state Pxy in the following way: with j k l and I k m the weights of the predicted mean and predicted covariance, and n W the measurement noise covariance matrix. Where K ^ k ^is the correction gain at time k, m S the average of measurements calculated via the sigma points, P SS the measurement covariance and P RS the cross-covariance between the state and the measurement. The calculation procedure for these terms is detailed in patent application WO2022 / 228923 using a specific “scentless transformation” technique. Note that the term ϕ'^^^k^ is a parameter estimated by the observer of the magnet flux presented later. For the model of the magnetic flux of permanent magnets, we can consider the discretized form of the model: where ^ / ^ k ^ and η / ^k^ are zero-mean white noise with covariance matrices Q / and R / , respectively. For this model, we use the linear Kalman filter (KF) with two steps to perform: - First step: Prediction at time kϕ'^^ ^k|k − 1^ = ϕ'^^ ^k − 1|k − ​​1^Po ^k|k − 1^ = Po ^k − 1|k − ​​1^ + Q / - Second step: Update with the measurement at time k: ϕ'^^^k|k^ = ϕ'^^^k|k − 1^ + K / ^k^ ^y^h^k^ − z / ^k^^Po^k|k^ = Po^k|k − 1^ − Ko^k^So^k^Ko^k^3With y^h^k^ = V^^k^So^k^ = ω^k^^Po^k|k − 1^ + R / zo^k^ = ω^k^ϕ'^^^k|k − 1^ + ^ω^k^ϕ'^^k^ + R'^k^I^^k^^We can notice that ϕ ' ^ ^k^ is a term coming from the stator magnetic flux observer. Therefore, based on this estimation model, one can estimate in addition to the first magnetic flux of the permanent magnets ϕ ' ^^ , the first standard deviation of the error of its estimation 4) Determining the temperature of the permanent magnets In this step, a temperature of the permanent magnets and a standard deviation of the temperature of the permanent magnets are determined by applying a Kalman filter to a thermal model of the electric machine. The thermal model of the electric machine is applied to the currents acquired in step 1, to the rotor speed acquired in step 1 and to the winding temperature or the stator temperature acquired in step 2. The thermal model of the electric machine links the currents, the rotor speed, the winding temperature or the stator temperature to the temperature of the permanent magnets. The standard deviation is an estimate of the probability dispersion of the temperature of the permanent magnets. The standard deviation can be defined as the square root of the variance of the temperature of the permanent magnets. According to one embodiment, the electric machine can be decomposed by several nodes.In this case, the thermal model of the machine can be written from a plurality of thermal equations, each thermal equation translating the heat exchanges between several components of the electric machine. For example, the components can be the cooling circuit, the winding, the stator yoke, the permanent magnets, the rotor, the rotor yoke, the inner ring ("inner ring electric machines" generally refer to electric machines where the rotor winding is located inside the stator winding, these motors are also called wound rotor motors), the rotor axis. In a non-limiting manner, the thermal model can be written:mCpwx T^wx = Q&yyz^wx − Qwx^JSmCpJS T^JS = Qwx^JS + PJS − QJS^&& − QJS^J{mCp&& T^&& = Q%&^&& − QJS^&& + P&& − Q&&^J{. mCp|ST^|S = Q|^|S − Q|S^Z|mCpZ|T^Z| = Q|S^Z| − QZ|^|JmCp|JT^|J = QZ|^|J − Q|J^yZzmCp%&T^%& = P%& − Q%&^&&Where mCpZ, PZ, TZ are, respectively, heating capacity, Joule loss and temperature of the ith element, with i in ~wj, sy, cc, st, PM, r, ry, ir, rs, ec^ representing, respectively, water jet (water cooling circuit), stator yoke, copper coil, stator teeth, permanent magnet, rotor, rotor yoke, inner ring, rotor shaft, coil head. Note that the Joule losses PZ are a function of current and speed, i.e. PZ =fZ^I^, I^, ω^ and are predetermined by 3D maps. Alternatively, the thermal model can be more or less detailed (taking into account additional equations for other components of the electrical machine). The example illustrated in this example provides a good compromise between precision and real-time determination.In addition, QZx is the heat exchange between two elements ~i, j^ , which belong to ~cool, wj, sy, cc, st, PM, r, ry, ir, rs, ec, oil^. We can consider two new components “cool” and “oil” representing respectively the cooling (stator cooling water) and the oil (rotor cooling). The heat exchange Q. Zx can be calculated as follows: Where G Zx is the heat transfer coefficient between components i and j, which depends on the motor speed ω. It is remarkable that the only temperature measured is that of the winding T %& (alternatively the stator body temperature). In addition, the model has two controllable temperatures which are the cooling water temperatures T & for the stator and oil T yZz for the rotor. We can write, in the form of an equation of state, the thermal model:X^ = A^ω, I^, ^7^X + Bu^t^ y = CXWhere X = ^T 3wx, TJS, T&&, TJ{, T^^, T|, T|S, TZ|, T|J, T%& ^u = ^T&yyz, PJS, P&&, PJ{, P^^, P|, P%&, TyZz ^3and y = T%& are, respectively, the states, inputs and measurements of the thermal model. The state matrices are, directly, obtained by rewriting the thermal model in the form of a state representation. Note that matrix A depends on the electrical rotation speed of the electric machine and its currents and has a size of 10x10 for the considered example of the thermal model (the size can be adjusted according to the decomposition of the thermal model), while matrix B has a size of 10x8 for the considered example of the thermal model (the size can be adjusted according to the decomposition of the thermal model). The output matrix can be written in this case C = [00000 00001], since the temperature of the stator coil head T %&is the only measurement available in the state vector X. For this embodiment, one can implement a Kalman filter as the temperature observer, which is based on the thermal model, and is defined as follows:X'^ = A'X + Bu + K^y − Cx^^ Where X' is the estimate of the thermal state X, P = E ^^X − X'^^X − X'^3^ is the covariance of the error of the estimate, Q3 and R3 are respectively the covariance matrices of the noises of the state and the measurements and K is the observer correction matrix. This gain is calculated as follows: K = PC3 R2^3 The initial values ​​of the observer are as follows: It should be noted that the dynamics of the thermal model is very slow; which implies a rather slow convergence for the covariance matrix P and the correction gain K of the Kalman filter. For this reason, we can introduce and use a steady-state Kalman form, during which the covariance matrix P has a very slow dynamics (P^ = 0) calculated in steady-state by the following Riccati equation: It can be noted that the solution of the error covariance matrix P is a function of the transition matrix A, the measurement matrix C, the covariance matrices of the state and measurement noises (Q3 and R3 respectively) of the global system. Since the matrix A depends on the speed, the solution of the matrix P therefore depends on the speed as well. In practice, the covariance matrix P can be calculated offline, and it can be mapped as a function of the electrical machine speed, and then the correction gain can be calculated by the formula K = PC3R2^3 to realize the observer. Finally, the temperature T can be estimated ' ^^ and the standard deviation of the error of the estimate q ^st of the temperature of the rotor permanent magnets from the estimated state X '. 5) Determination of the second magnetic flux In this step, a second magnetic flux of the permanent magnets and a standard deviation of the second magnetic flux are determined (estimated) by a magnetothermal model of the permanent magnets from the temperature of the permanent magnets determined in step 4. A second magnetic flux is called a second estimate of the magnetic flux of the permanent magnets, this second estimate of the magnetic flux of the permanent magnets being obtained from the temperature of the permanent magnets determined in step 4. Thus, the method determines the magnetic flux in two different ways, which allows robust detection of the demagnetization of the permanent magnets. According to an implementation of the invention, the magnetothermal model is written: the second magnetic flux of the permanent magnets, T'^^ the determined temperature of said permanent magnets, β) a thermal coefficient of said permanent magnets, T* an ambient temperature, ϕ * the magnetic flux of said permanent magnets at room temperature T * . Preferably, the thermal coefficient β ) permanent magnets can be estimated, as well as the standard deviation of the thermal coefficient Advantageously, the thermal coefficient β )can be estimated from the magnetic characteristics of the permanent magnets used in the rotor of the electric machine. 6) Demagnetization detection In this step, the first magnetic flux of the permanent magnets determined in step 3 is compared to the second magnetic flux of the permanent magnets determined in step 5. Then, demagnetization of the permanent magnets is detected if the absolute value of the comparison (in other words the absolute value of the difference) is greater than a threshold depending in particular on the first standard deviation of the magnetic flux, and the standard deviation of the temperature of the permanent magnets. In other words, demagnetization of the permanent magnets is detected if > ^ with ;)^^ the first magnetic flux of the permanent magnets, ϕ^^^T'^^, β)^ the second magnetic flux of the permanent magnets, and S the threshold. Indeed, if the difference is greater than the threshold, then the difference is not in the confidence interval, which implies the demagnetization of the permanent magnets. According to one aspect, the threshold S can be written as S = a positive coefficient, preferably between 2 and 5 and preferably equal to 3, σ representing the standard deviation, ϕ'^^ the first magnetic flux of the permanent magnets, ϕ^^^T'^^, β)^ the second magnetic flux of the permanent magnets. For example, the margin of uncertainty can be calculated via the standard deviation of the estimated flux difference between ϕ'^^ and ϕ^^^T'^^, β)^ , i.e. ϕ'^^ − ϕ^^^T'^^, β)^ from the magnetothermal model, and knowledge of ϕ'^^, T'^^, β), σ / '01, σ3'01, obtained during steps 3 to 5. By applying the propagation uncertainty approach, we can obtain: With g defined by g^ϕ'^^, T'^^, β)^ = ϕ'^^ − ϕ^^^T'^^, β)^ =Then, based on the calculated standard deviation, we can analyze the inconsistency between the two estimated information ϕ ' ^^ and T ' ^^ The main idea is to use the "three sigma rule": determine whether the estimated flux difference between ϕ'^^ and ϕ^^^'T^^, β)^ is in the confidence interval (here defined with a positive coefficient 3) The invention also relates to a method for controlling and / or monitoring an electrical machine. The method for controlling and / or monitoring implements the method for detecting demagnetization of the permanent magnets according to the invention and an additional step of controlling and / or monitoring the electrical machine as a function of the detected demagnetization. Thus, for this method for controlling and / or monitoring an electrical machine, the following steps can be implemented: a) Demagnetization of the permanent magnets is detected by means of the method for determining the stator temperature according to any one of the variants or combinations of variants described above; and b) The electrical machine is controlled and / or monitored as a function of the detection of demagnetization of the permanent magnets.In other words, the method for controlling the electrical machine comprises the following steps: 1) Acquisition of the currents, voltages and the position and / or rotation speed 2) Acquisition of the temperature of the winding or stator 3) Determination of the first magnetic flux 4) Determination of the temperature of the permanent magnets 5) Determination of the second magnetic flux 6) Detection of demagnetization 7) Control and / or monitoring of the electrical machine Figure 3 describes, schematically and in a non-limiting manner, the steps of the method according to a third embodiment of the invention. The steps previously described for Figure 2 are not detailed again. This embodiment further comprises an additional step of control CON and / or monitoring of the electrical machine as a function of the detection of demagnetization Dem.For example, the control of the electric machine can be based on a high-performance direct torque control method, particularly suitable for synchronous electric machines with salient poles. This can be, for example, a control strategy of the MTPA or MTPV type (respectively from the English "Maximum Torque per Ampere" which can be translated as maximum torque per ampere, and "Maximum Torque per Voltage" which can be translated as maximum torque per volt). Monitoring the demagnetization of the permanent magnets can consist of an alert (for example, an indicator light), an increase in rotor cooling (for example, by means of fluid circulation), or a reduction in the performance of the electric machine (reduction of the speed and / or torque) so as to reduce the rotor temperature.Furthermore, the invention relates to a system for controlling and / or monitoring an electrical machine, in particular a synchronous electrical machine, suitable for applying the method as described above. Such an electrical machine control system may comprise means for controlling the electrical machine comprising means for detecting the demagnetization of the permanent magnets and means for controlling or monitoring the electrical machine. The means for detecting the demagnetization of the permanent magnets detect the demagnetization of the permanent magnets from the current and voltage signals and the rotational speed of the rotor, as well as the stator temperature and / or the winding temperature. These are the currents and voltages of each of the three phases of the electrical machine.The control or monitoring means apply voltages to the terminals of the electric machine as a function of the torque in order to ensure a torque setpoint for the electric machine. Advantageously, the control and / or monitoring system may be a controller comprising computer means. This method and this control and / or monitoring system may be used for an electric machine on board a vehicle, in particular on board an electric or hybrid motor vehicle. However, the control system described is not limited to this application and is suitable for all applications of electric machines, including stationary applications. As goes without saying, the invention is not limited to the sole embodiments of the methods and the system described above by way of example; on the contrary, it embraces all variant embodiments. The characteristics and advantages of the method according to the invention will appear more clearly on reading the application examples below. For this application example, the operation of an electrical machine is simulated. The simulation is carried out for a 10 kW synchronous-reluctant machine with four pairs of poles and powered by 350 V DC. The machine operates with a torque of 150 Nm and a speed that varies from 500 to 14000 rpm to cover all operating zones: MTPA (maximum torque per ampere) zone from t = 0 s to t = 30 s and from t = 70 s to t = 80 s, and MTPV (maximum torque per volt) zone from t = 30 s to t = 70 s by imposing the maximum torque that can be obtained for the effective voltage available. Figure 4 illustrates for this example the curve of the torque Te in Nm as a function of time T in s, and figure 5 illustrates for this example the curve of the rotation speed ω in rpm as a function of time T in s.These two figures thus show the different operating ranges of the electrical machine. The first magnetic flux of the permanent magnets is then determined, as well as the first standard deviation of the first magnetic flux, in accordance with an embodiment of step 3 of the method according to the invention. Figure 6 illustrates the first magnetic flux of the permanent magnets ΦPM in Wb as a function of time T in s. This figure plots the reference curve obtained by the simulation ΦREF, as well as the first estimated magnetic flux ΦEST and the confidence interval defined by 3σ (3 times the standard deviation – three sigma rule). Note that the reference and estimated curves are superimposed. Consequently, this step allows an accurate estimation of the magnetic flux of the permanent magnets.The temperature of the permanent magnets is also determined, as well as the standard deviation of the temperature of the permanent magnets, in accordance with an embodiment of step 4 of the method according to the invention. Figure 7 illustrates the temperature of the permanent magnets TPM in °C as a function of time T in s. This figure plots the reference curve obtained by the simulation T. REF , as well as the first estimated magnetic flux T ESTand the confidence interval defined by 3σ (3 times the standard deviation – three sigma rule). Note that the reference and estimated curves are superimposed. Therefore, this step allows for an accurate estimation of the temperature of the permanent magnets. In addition, a faster increase in the temperature of the magnets is observed when the motor speed increases. On the other hand, the standard deviation of the error in the flux estimation of the magnets decreases when the speed increases. Then, from this temperature estimation, a second magnetic flux of the permanent magnets is determined according to an embodiment of step 5. Then, the difference between the first and second magnetic fluxes of the permanent magnets is calculated. A threshold S is further determined based on the standard deviations according to an embodiment of step 6.Figure 8 illustrates the difference between the two magnetic flux estimates ΔΦ in % as a function of time T in s, as well as the threshold S. ΔΦ must be between -S and +S. When ΔΦ leaves this envelope, demagnetization is observed. We note that from 42s, the comparison ΔΦ becomes greater than the threshold S, and at this moment we detect demagnetization of the permanent magnets.

Claims

Claims 1. Method for detecting a demagnetization of a plurality of permanent magnets equipping a rotor of an electrical machine, said electrical machine comprising a stator provided with a winding and said rotor provided with said plurality of permanent magnets, characterized in that the following steps are implemented: a. Currents and voltages in phases of said winding of said stator are acquired (ACQ), as well as the position and / or the rotational speed of said rotor; b. A temperature of said winding and / or a temperature of said stator is acquired (ACQ); c. A first magnetic flux of said permanent magnets and a first standard deviation of said first magnetic flux are determined by applying at least one Kalman filter to a magnetic flux model (MFM) of said electrical machine, said magnetic flux model being applied to the acquired currents and voltages and to the acquired position or rotational speed of the rotor; d.A temperature of said permanent magnets and a standard deviation of said temperature of said permanent magnets are determined by applying a Kalman filter to a thermal model (MTH) of said electrical machine, said thermal model of said electrical machine being applied to the acquired currents, to the acquired rotor position or speed and to the acquired temperature of said acquired winding or of said stator; e. A second magnetic flux of said permanent magnets and a second standard deviation of said second magnetic flux of said permanent magnets are determined by a magnetothermal model (MMT) of said permanent magnets from said determined temperature of said permanent magnets; and f.Said first magnetic flux is compared (COMP) with said second magnetic flux, and a demagnetization (Dem) of said permanent magnets is detected if the absolute value of said comparison is greater than a threshold dependent on said first and second standard deviations of magnetic flux of said permanent magnets.

2. Method for detecting a demagnetization according to claim 1, wherein said magnetic flux model of said electrical machine is composed of a stator magnetic flux model (MFS) and a magnetic flux model of said permanent magnets (MFA).

3. Method for detecting a demagnetization according to claim 2, wherein said first magnetic flux of said permanent magnets is determined, and said first. standard deviation of the magnetic flux of said permanent magnets by applying a linear Kalman filter to said magnetic flux model of said permanent magnets, said magnetic flux model of said permanent magnets (MFA) being applied to the acquired currents and voltages and to the acquired rotor position or rotational speed, as well as to the magnetic flux of said stator determined by an unscented Kalman filter applied to said stator magnetic flux model (MFS).

4. A method for detecting demagnetization according to one of claims 2 or 3, wherein said stator magnetic flux model (MFS) is defined by a state model with a state equation of the form: X ^ = A^ω^X + u avec X = ^ϕ^ V^ − RI^ 0 ω ϕ^^ , u = ^V^ − RI^ − ωϕ^^^ , A^ω^ = ^ −ω 0^ et une measurement equation of the form: = ^I^ I^^ et f = ^f^ f^^, ϕ^, ϕ^ étant le flux magnétique duin the Park frame, Vd, Vq the voltages in the stator phases in the Park frame, respectively the direct voltage and the quadrature voltage, Id, Iq the currents in the stator phases in the Park frame, respectively the direct current and the quadrature current, ω the rotor rotation speed, ϕ ^^ the first magnetic flux of the permanent magnets, R the stator resistance, f d , f q the non-linear functions linking the stator current components to the magnetic flux components in the Park frame.

5. Method for detecting demagnetization according to one of claims 2 to 4, wherein said permanent magnet magnetic flux model (MFA) is defined by a state model with a state equation of the form: ϕ ^ ^^ = 0 et une équation de mesure y = étant le flux magnétique du stator directin the Park frame, Vq the quadrature voltages in the stator phases in the Park frame, Iq the quadrature currents in the stator phases in the Park frame, ω the rotor rotation speed, ϕ ^^ the magnetic flux of the permanent magnets, R the resistance of the stator.

6. Method for detecting demagnetization according to one of the preceding claims, in which said thermal model (MTH) is defined by a state model with an equation of the form: X ^ = A^ω^X + Bu^t^ et y = CX avec X un vecteur des températures des différents composants de la machine electrical, u a vector of the Joule losses of the different components of the electrical machine and the temperatures of the cooling components of the machine électrique, et y = T%& ladite température acquise, A^ω^ = la matrice B est function of the current of the phases of the electric machine and the rotation speed of the rotor, and C the output matrix.

7. Method for detecting demagnetization according to one of the preceding claims, in which said magnetothermal model (MMT) is written: ϕ^^^T'^^, β)^ le deuxième flux magnétique d es aimants permanents, T'^^ la température déterminée desdits aimants permanents, β ) un coefficient thermique desdits aimants permanents, T* une température ambiante, ϕ* le flux magnétique desdits aimants permanents à la température ambiante T*.

8. Method for detecting demagnetization according to one of claims p récédentes, dans lequel ledit seuil S s’écrit S = -^σ / '012 / 01^3'01,4'^ avec K1 un positive coefficient, preferably between 2 and 5 and preferably worth 3, σ r eprésentant l’écart-type, ϕ'^^ le premier flux magnétique des aimants permanents, ϕ ^^^T'^^, β)^ le deuxième flux magnétique des aimants permanents.

9. Method for controlling and / or monitoring an electrical machine, said electrical machine comprising a stator provided with a winding and a rotor provided with a plurality of permanent magnets, characterized in that the following steps are implemented: a. Demagnetization of the permanent magnets is detected by means of the method for detecting demagnetization of the permanent magnets according to one of the preceding claims; and b. Said electrical machine is controlled and / or monitored (CON) as a function of the detection of demagnetization of the permanent magnets.

10. System for controlling and / or monitoring an electrical machine comprising an electrical machine and a controller for implementing the control and / or monitoring method according to claim 9.

Images