Power device junction temperature online observer with unknown shell temperature and adaptive environmental disturbance

Abstract

The invention provides an environment disturbance self-adaptive power device junction temperature observer with unknown shell temperature, which takes the temperature measured by NTC (Negative Temperature Coefficient) in a power module as a feedback variable, takes the power device loss and the environment real-time temperature as known input, and takes the bottom shell temperature of the power device as unknown input. And according to a system temperature state space equation which does not contain thermal impedance between the radiator and the environment, designing an environment disturbance self-adaptive power device junction temperature observer with an unknown shell temperature. The method has the advantages that a sensor does not need to be installed on the bottom shell of the power module to measure the temperature of the bottom shell, thermal impedance parameters between the bottom shell of the power module and the environment do not need to be known, and the junction temperature of the power device is estimated through temperature feedback measured by the NTC in the module; junction temperature estimation failure caused by falling off and shifting of the shell temperature sensor in the operation process of the power electronic device is avoided, and meanwhile the method is suitable for random environments such as environment temperature, humidity, heat radiation and fluid velocity change and has environment disturbance self-adaptability.

Description

Technical Field

[0001] The present invention relates to the technical field of power device junction temperature monitoring, and in particular to an online junction temperature observer for a power device capable of self-adapting to environmental disturbances with unknown shell temperature. Background Art

[0002] As the core component of power electronic devices, the reliability of power devices is particularly important. Studies have shown that the failure of power electronic devices caused by power device failure accounts for as high as 34%, of which about 55% of power device failures are induced by thermal stress. At present, power electronic devices used in electric vehicles, electric aviation and other fields are required to have higher power density, smaller volume and weight. On the one hand, this requires the power devices to work in a high current density state, and on the other hand, it requires the power electronic devices to have a compact and compact structure, which makes the thermal safety operation of power devices face challenges. In order to avoid failure of power electronic devices due to overheating of power devices during operation, it is necessary to monitor the junction temperature of power devices online. When there is a risk of overheating of power devices, timely measures such as reducing losses or even blocking power device pulses can be taken to ensure the operational reliability of power electronic devices.

[0003] At present, there are three main types of power device junction temperature monitoring technology: direct sensor measurement method, thermal parameter method and method based on thermal network model. Direct sensor measurement method includes infrared thermal method, optical fiber temperature measurement method, chip integrated temperature sensor method, etc. Among them, infrared thermal method and optical fiber temperature measurement method need to destroy the power module package to measure the temperature of the semiconductor chip surface, and chip integrated temperature sensor method can only measure the temperature of a certain point on the chip surface instead of the junction temperature. The thermal parameter method uses the chip itself as a sensor, and measures the change of temperature-sensitive electrical parameters under the running state to reverse the real-time junction temperature of the power module. For example, on-resistance method, turn-on voltage method, turn-on delay method, turn-off delay method, current change rate method, etc. The thermal parameter method generally requires the addition of additional electrical signal measurement and processing circuits, which increases equipment cost and is susceptible to noise interference. Based on the method of thermal network model, the thermal impedance between the reference temperature and the chip is obtained by establishing a thermal network impedance model of the power device and its heat dissipation system, and then combined with the power loss and according to the definition of thermal impedance, the chip junction temperature is reversed and calculated. The method based on the thermal network model does not invade or interrupt the original equipment and does not require the addition of additional circuits. An online state observer can be established based on the thermal network model, which is suitable for engineering applications.

[0004] The existing power module temperature observer establishes an equivalent thermal network model of the power module and its heat dissipation system, takes the ambient temperature as the boundary condition, and takes the power device loss as the input to observe the junction temperature online. In reality, the environment in which the power electronic device is located may change, especially in the field of electric aviation. With different flight conditions, the temperature, humidity, thermal radiation, etc. of the environment are prone to change. The convection thermal resistance and radiation power between the heat sink and the environment in the power electronic device are nonlinear variables that change with the environment. The original power module temperature observer is based on a known and unchanging thermal network model. When the external environment changes, it will no longer be applicable and does not have the adaptability to external environmental disturbances. Summary of the invention

[0005] The purpose of the present invention is to address the deficiencies in the prior art and propose an online observer for the junction temperature of a power device that is adaptive to environmental disturbances when the shell temperature is unknown. The shell temperature of the power module is regarded as an unknown input quantity, and an online estimation of the junction temperature of the power device is achieved by establishing a state observer under unknown input.

[0006] In order to solve the above technical problems, the technical solution of the present invention is:

[0007] A power device junction temperature observer with unknown shell temperature and adaptive to environmental disturbances is designed with the following steps:

[0008] Step 1: Calculate the average power device loss P as an input variable of the observer based on the switching frequency, duty cycle, on-current of the power device, the switching loss curve provided in the data sheet, and the device working mode in the power electronic device; air As another input variable of the observer. These two variables form the input vector U = [P, T air ] T .

[0009] Step 2: According to the instantaneous thermal impedance curve in the power module product manual, obtain the thermal resistance and thermal capacitance parameters r1, r2, c1, c2 of the second-order Foster thermal network from the device chip to the module bottom shell in the power module by fitting; obtain the thermal resistance parameter r1 from the device chip to the corresponding point of the built-in temperature sensor NTC in the power module by the cooling curve method. b , the thermal resistance and thermal capacitance parameters r3 and c3 of the first-order Foster thermal network from the corresponding point of the built-in temperature sensor NTC to the bottom shell of the module; the temperature of the bottom shell of the power module is set to x4, and the unknown thermal impedance between the bottom shell of the power module and the external environment is set to Z4. According to the actual heat conduction path of the power module, the above thermal resistance, thermal capacitance and thermal impedance are connected to form a power module Foster thermal network model including a heat dissipation system.

[0010] Step 3: Take the real-time temperatures x1, x2 and x3 of the Foster thermal network nodes as state variables and establish the state vector X = [x1, x2, x3] T ; Take the corresponding point temperature measured by the built-in temperature sensor NTC of the power module as the measurable variable y; suppose the shell temperature x4 of the power module is the unknown input, and write the system temperature state space equation that does not include the thermal impedance Z4 between the radiator and the environment according to the thermal network model established above and its parameters, and determine the state matrix A, known input matrix B, unknown input matrix D, and output matrix C.

[0011] Step 4: According to the conditions of system state observability and linear observability, design the constant coefficient observation matrix L; according to the condition that the observer convergence is independent of the unknown input x4, design the constant coefficient matrix K of the temperature observer; according to the condition that the temperature observer can converge under any input U, design the constant coefficient matrix G of the temperature observer; according to the condition that the temperature observer converges independently of the current temperature state X, design the constant coefficient matrices F and H of the temperature observer.

[0012] Step 5: Multiply the observer coefficient matrix G by the known input vector U to obtain vector 1, multiply the measurable variable y by the observer coefficient matrix H to obtain vector 2, and multiply the observer state vector 5 by the observer coefficient matrix F to obtain vector 3; then add vector 1, vector 2, and vector 3 to obtain vector 4; obtain the observer state vector 5 through the integral link of vector 4; multiply the measurable variable y by the observer coefficient matrix K to obtain vector 6, and then add the observer state vector 5 and vector 6 to obtain the observer output vector V e =[v1,v2] T , where V1 is the junction temperature of the semiconductor device in the power module to be observed.

[0013] The present invention has the following characteristics and beneficial effects:

[0014] By adopting the above technical scheme, it is only necessary to use the temperature feedback measured by the NTC inside the module to estimate the junction temperature of the power device online, and there is no need to set a sensor for detecting the temperature of the bottom shell of the module, thereby avoiding the failure of junction temperature estimation caused by the shell temperature sensor falling off and shifting during the operation of the power electronic device; the present invention does not need to know the information such as the convective thermal resistance and the amount of thermal radiation that changes with the external environment. In a random environment such as changes in ambient temperature, humidity, thermal radiation and fluid velocity, the junction temperature of the power device can be effectively estimated online by using the junction temperature observer of the present invention, thereby improving the reliability of power electronic devices in a changeable environment such as electric aviation and aircraft. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0016] Figure 1 Schematic diagram of a temperature observer of a power device that is adaptive to environmental disturbances when the shell temperature is unknown according to an embodiment of the present invention.

[0017] Figure 2 This is a Foster thermal network model of a power module including a heat dissipation system in an embodiment of the present invention.

[0018] Figure 3 MATLAB / simulink simulation waveform for the average loss of power devices.

[0019] Figure 4 MATLAB / simulink simulation waveforms for the real-time temperature of the power module and the temperature estimated by the proposed temperature observer.

[0020] Figure 5 Tair MATLAB / simulink simulation waveforms of power device loss P and ambient temperature.

[0021] Figure 6 This is the MATLAB / simulink simulation waveform of the convection thermal resistance between the heat sink and the environment.

[0022] Figure 7 MATLAB / simulink simulation waveforms for the real-time temperature of the power module and the temperature estimated by the proposed temperature observer. DETAILED DESCRIPTION

[0023] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0024] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and the like are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, features defined as "first", "second", and the like may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0025] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood by specific circumstances.

[0026] The present invention provides a power device junction temperature observer capable of self-adapting to environmental disturbances when the shell temperature is unknown, and the steps are as follows:

[0027] Step 1: Calculate the power device loss P based on the power device switching frequency, duty cycle, on-current, and switching loss curve provided in the data sheet as an input value of the observer. The power device loss P is the average loss P of the power MOSFET in one switching cycle, which is calculated by the following formula:

[0028] P=P c,mos +P sw,mos +P c,diode +P rr,diode

[0029] P c,mos =R dson (T j )I 2

[0030]

[0031] P c,diode =u do I+R d I 2

[0032] P rr,diode =V dc Q rr f sw

[0033] Among them, P c,mos and P sw,mos are the conduction and switching losses of the power device MOSFE, P c,diode and P rr,diode are the conduction loss and reverse recovery loss of the power MOSFET body diode respectively; R dson (T j ) is the on-state resistance, obtained from the data sheet, I is the device on-state current; V dc is the input voltage of the device, usually the DC bus voltage; f sw is the switching frequency of the device; t on is the device turn-on time and t off is the device turn-off time, obtained from the device manual; R d is the body diode on-state resistance, u d0 is the body diode forward voltage drop at zero current, obtained from the device manual; Q rr is the body diode reverse recovery capacity, obtained from the device manual.

[0034] The external ambient temperature information T measured by the ambient temperature sensor air As another input variable of the observer, the observer input vector U = [P, T air ] T .

[0035] Step 2: According to the instantaneous thermal impedance curve in the power module product manual, obtain the thermal resistance and thermal capacitance parameters r1, r2, c1, c2 of the second-order Foster thermal network from the device chip to the module bottom shell in the power module by fitting; obtain the thermal resistance parameter r1 from the device chip to the corresponding point of the built-in temperature sensor NTC in the power module by the cooling curve method. b , the thermal resistance and thermal capacitance parameters r3 and c3 of the first-order Foster thermal network from the corresponding point of the built-in temperature sensor NTC to the bottom shell of the module; assuming that the temperature of the bottom shell of the power module is x4, and the thermal impedance between the bottom shell of the power module and the external environment is Z4.

[0036] According to the actual situation, connect the above thermal resistance, thermal capacitance and thermal impedance to form Figure 1 The Foster thermal network model of the power module including the heat dissipation system is shown. Specifically,

[0037] In the second-order Foster thermal network from the device chip to the module bottom shell in the power module, the thermal resistance r1 and the thermal capacitance c1 are connected in parallel, the thermal resistance r2 and the thermal capacitance c2 are connected in parallel, and the second-order Foster thermal networks from the device chip to the module bottom shell in the two power modules are connected in series;

[0038] The thermal resistance r3 and the thermal capacitance c3 in the first-order Foster thermal network from the corresponding point of the built-in temperature sensor NTC to the module bottom shell are connected in parallel. The thermal resistance r b Connect the first-order Foster thermal network in series with the corresponding point of the built-in temperature sensor NTC to the bottom shell of the module;

[0039] The output side of the device loss P is connected to the second-order Foster thermal network from the device chip in the power module to the module bottom shell and the thermal resistance r from the device chip in the power module to the corresponding point of the built-in temperature sensor NTC. b ;

[0040] The output of the second-order Foster thermal network from the device chip to the module bottom shell in the power module and the first-order Foster thermal network from the corresponding point of the built-in temperature sensor NTC to the module bottom shell is connected to the thermal impedance Z4 between the power module bottom shell and the external environment; the other end of the thermal impedance Z4 between the power module bottom shell and the external environment is connected to the external environment temperature information T air .

[0041] Among them, x1, x2, x3 and x4 are the voltages of nodes ①, ②, ③ and ④ respectively, which represent the junction temperature of the device in the power module, the temperature of a certain point in the power module, the temperature of the location of the NTC in the power module and the temperature of the bottom shell of the power module respectively.

[0042] In this embodiment, the thermal network parameters of the power module used obtained through step 2 are:

[0043] r1=1.71, r2=3.59, r3=11.27, r b =3.33, c1=37.41, c2=1.71,

[0044] c3=4.10, Z4 unknown.

[0045] Step 3: Take the real-time temperatures x1, x2 and x3 of the Foster thermal network nodes as state variables and establish the state vector X = [x1, x2, x3] T ; Take the corresponding point temperature measured by the built-in temperature sensor NTC of the power module as the measurable feedback variable y; Let the shell temperature x4 of the power module be the unknown input, according to Figure 1 The established thermal network model and its parameters establish the system state matrix A, known input matrix B, unknown input matrix D, and output matrix C, and establish the following system state space equation.

[0046]

[0047] y=CX

[0048]

[0049] In this embodiment:

[0050]

[0051] Step 4: According to the system observability and line observability conditions, design the observation matrix L:

[0052] Observable condition: rank([C, L] T )=rank(C)+rank(L)

[0053] Linear performance condition: rank([O A,C L] T )=rank(O A,C )

[0054] Where rank(.) represents the rank of the matrix, O A,C represents the observable matrix of the system, and L is the observation matrix involved.

[0055] In this embodiment, in order to ensure the system performance and line performance, the design

[0056] Step 5: Substitute the observation matrix L, output matrix C and unknown input matrix D into the equation (L-KC)D=0. According to this equation, design the constant coefficient matrix K of the temperature observer so that the temperature observer output vector V e =[V1,V2] T The error convergence of is independent of the unknown input x4, which makes the temperature observer adaptive to random environmental disturbances.

[0057] In this embodiment, in order to make the temperature observer have the adaptability to random environmental disturbances, K is designed to be [11.0009, 11.0009] T .

[0058] Step 6: Substitute the observation matrix L, output matrix C, input matrix B and the coefficient matrix K designed in step 5 into the calculation formula of the constant coefficient matrix G of the temperature observer: G = (L-KC) B. This ensures that the output vector V of the temperature observer is e =[V1,V2] T The error can converge to zero under any input U.

[0059] In this embodiment, in order to ensure the convergence of the temperature observer, G=[0.8814, 0; 0.8547, 0] is designed.

[0060] Step 7: Substitute the system matrix A, output matrix C, and the coefficient matrices L and K designed in steps 5 and 6 into the following equations, and design the constant coefficient matrices F and H of the temperature observer to ensure that the temperature observer can converge to zero under any finite temperature state X.

[0061]

[0062] Where λ(.) represents the characteristic root of the matrix. In this example, to ensure the convergence of the temperature observer under any finite temperature state X, we design:

[0063]

[0064] Step 8: Figure 1 As shown, the observer coefficient matrix G is multiplied by the known input vector U to obtain vector 1, the measurable variable y is multiplied by the observer coefficient matrix H to obtain vector 2, and the observer state vector 5 is multiplied by the observer coefficient matrix F to obtain vector 3; then vector 1, vector 2 and vector 3 are added to obtain vector 4; vector 4 is passed through an integral link to obtain the observer state vector 5; the measurable variable y is multiplied by the observer coefficient matrix K to obtain vector 6, and then the observer state vector 5 and vector 6 are added to obtain the observer output vector V e =[V1,V2] T , where V1 is the junction temperature of the semiconductor device in the module to be observed.

[0065] In this embodiment, a converter simulation model based on the SiC MOSFET half-bridge module and its thermal network model is built in matlab / simulink, and a comparison chart of the average loss of SiC MOSFET during converter operation, the real-time temperature of the SiC MOSFET half-bridge module, and the temperature estimated by the proposed temperature observer is obtained through simulation.

[0066] pass Figure 3 and Figure 4 It can be seen that Figure 4 In the figure, the black solid lines are the actual junction temperature x1 of the power module, the actual temperature x2 of a point in the power module and the temperature x3 measured by the NTC of the power module, and the red dotted lines are the estimated temperatures v1 and v2 output by the temperature state observer. Figure 3is the average power loss of the power device when the converter is running in the simulation. Since the output power of the converter changes at t=400 seconds, t=600 seconds and t=800 seconds, the average power device loss also changes at t=400 seconds, t=600 seconds and t=800 seconds. From the simulation results, it can be seen that the junction temperature estimated by the proposed temperature observer is basically consistent with the actual junction temperature of the power device after t=100 seconds, which means that the error of the proposed temperature observer converges to zero after a finite time when the initial temperature, module shell temperature and thermal impedance between the module bottom shell and the environment are unknown. From the simulation results, it can be seen that after the output of the proposed temperature observer converges, even when the device power loss changes, the junction temperature estimated by the temperature observer remains basically consistent with the actual junction temperature of the power device, proving that the error convergence of the proposed temperature observer is independent of power loss.

[0067] In this embodiment, a converter simulation model based on the SiC MOSFET half-bridge module and its thermal network model is built in matlab / simulink. A comparison chart of the real-time temperature of the SiC MOSFET half-bridge module and the temperature estimated by the proposed temperature observer is obtained through simulation when the thermal impedance between the module bottom shell and the environment changes due to the change of ambient temperature during converter operation.

[0068] pass Figure 5 and Figure 6 and Figure 7 It can be seen that Figure 5 is the ambient temperature and average loss of power devices when the converter is running in the simulation. Figure 6 To simulate the change in convective thermal resistance between the heat sink and the environment, the external wind speed suddenly changes at 150 seconds. Figure 7 In the figure, the black solid lines are the actual junction temperature x1 of the power module and the actual temperature x2 of a certain point in the power module, respectively, and the red dotted lines are the estimated temperatures v1 and v2 output by the temperature state observer. It can be seen that due to the nonlinear relationship between the convective thermal resistance between the heat sink and the environment and the external ambient temperature and external wind speed, when the external environment of the converter changes, the thermal impedance between the module bottom shell and the environment changes nonlinearly. From the simulation results, it can be seen that sudden changes in the external environment and device losses will cause large changes in the junction temperature of the power device. During the changes in the external environment and device losses, the estimated junction temperature of the proposed temperature observer is basically consistent with the actual junction temperature of the power device after a short period of time. This shows that when the initial temperature, module shell temperature and thermal impedance between the module bottom shell and the environment are unknown, the proposed temperature observer converges to zero after a finite time error, which proves that the error convergence of the proposed temperature observer is independent of changes in the external environment and has adaptability to external environmental disturbances.

[0069] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, various changes, modifications, substitutions and variations of these embodiments including components are made without departing from the principles and spirit of the present invention, and still fall within the scope of protection of the present invention.

Claims

1. An online observer for junction temperature of a power device with unknown shell temperature and adaptive to environmental disturbances, characterized in that: The steps include: Step 1: Calculate the power device loss P based on the power device switching frequency, duty cycle, on-current and switching loss curve, and measure the external ambient temperature information T through an ambient temperature sensor. air , and then establish the observer input vector U = [P, T air ] T ; Step 2: Establish a Foster thermal network model of a power module including a heat dissipation system by using the acquired parameters according to the actual situation, wherein the parameters include thermal resistance, thermal capacitance and thermal impedance; Step 3: According to the established Foster thermal network model of the power module including the heat dissipation system and its parameters, a system state matrix A, an input matrix B, an unknown input matrix D, and an output matrix C are established, and then a system state space equation is established; Step 4: Design the observation matrix L according to the observability and linear observability conditions of the Foster thermal network model of the power module of the cooling system; Step 5: Construct the constraint conditions of the observer. The observation matrix L, the output matrix C, and the input matrix B are used to obtain the coefficient matrix K, the constant coefficient matrix G, the constant coefficient matrix F, and the constant coefficient matrix H through the constraint conditions. Step 6: Input vector U = [P, T air ] T As the input of the observer, the observer output vector V is obtained by combining the observation matrix L, the output matrix C, the input matrix B, the coefficient matrix K, the constant coefficient matrix G, the constant coefficient matrix F and the constant coefficient matrix H. e =[V1,V2] T , where V1 is the junction temperature of the semiconductor device in the module to be observed.

2. The on-line observer for power device junction temperature with unknown shell temperature and adaptive to environmental disturbance according to claim 1, characterized in that: The switching loss curve can be obtained directly from the data sheet.

3. The on-line observer for power device junction temperature with unknown shell temperature and adaptive to environmental disturbance according to claim 1, characterized in that: In step 2, according to the instantaneous thermal impedance curve in the product manual of the power module, the thermal resistance r1, thermal resistance r2, thermal capacitance c1 and thermal capacitance c2 of the second-order Foster thermal network from the device chip to the module bottom shell in the power module are obtained by fitting; and the thermal resistance parameter r from the device chip to the corresponding point of the built-in temperature sensor NTC in the power module is obtained by the cooling curve method. b , the thermal resistance r3 and thermal capacitance parameter c3 of the first-order Foster thermal network from the corresponding point of the built-in temperature sensor NTC to the bottom shell of the module; assuming that the temperature of the bottom shell of the power module is x4, and the thermal impedance between the bottom shell of the power module and the external environment is Z4.

4. The on-line observer for power device junction temperature with unknown shell temperature and adaptive to environmental disturbance according to claim 3, characterized in that: The power module Foster thermal network model including the heat dissipation system includes a second-order Foster thermal network from the device chip to the module bottom shell in two series-connected power modules and a first-order Foster thermal network from the corresponding point of the built-in temperature sensor NTC to the module bottom shell; the input side of the second-order Foster thermal network from the device chip to the module bottom shell in one power module is the first node, the second node is between the second-order Foster thermal networks from the device chip to the module bottom shell in the two power modules, the input side of the first-order Foster thermal network from the corresponding point of the built-in temperature sensor NTC to the module bottom shell is the third node, and the output side of the first-order Foster thermal network from the corresponding point of the built-in temperature sensor NTC to the module bottom shell is the fourth node.

5. The on-line observer for power device junction temperature with unknown shell temperature and adaptive to environmental disturbances according to claim 4, characterized in that: The voltages of the first node, the second node, the third node and the fourth node respectively represent the junction temperature of the device in the power module, the temperature of a certain point in the power module, the temperature of the position where the NTC is located in the power module and the temperature of the bottom shell of the power module.

6. The on-line observer for power device junction temperature with unknown shell temperature and adaptive to environmental disturbances according to claim 4, characterized in that: In step 3, the real-time temperatures of the first node, the second node, and the third node are recorded as x1, x2, and x3 as state variables, thereby establishing a state vector X=[x1, x2, x3] T ; The corresponding point temperature measured by the built-in temperature sensor NTC of the power module is taken as the measurable feedback variable y; the shell temperature x4 of the power module is set as the unknown input.

7. The on-line observer for power device junction temperature with unknown shell temperature and adaptive to environmental disturbances according to claim 6, characterized in that: The system state matrix A: Input matrix B: Unknown input matrix D: Output matrix C:

8. The on-line observer for power device junction temperature with unknown shell temperature and adaptive to environmental disturbances according to claim 6, characterized in that: The system state space equation is expressed as follows:

9. The on-line observer for power device junction temperature with unknown shell temperature and adaptive to environmental disturbances according to claim 6, characterized in that: In step 4, when designing the observation matrix L: Observable condition: rank([C, L] T )=rank(C)+rank(L) Linear performance objective conditions: rank([O A,C L] T )=rank(O A,C ).

10. The on-line observer for power device junction temperature with unknown shell temperature and adaptive to environmental disturbances according to claim 6, characterized in that: The constraints in step 5 include: The first constraint: (L-KC)D=0 Substitute the observation matrix L, output matrix C and unknown input matrix D into the first constraint condition, and obtain the constant coefficient matrix K of the temperature observer according to the equation; Second constraint: G = (L - KC) B Substitute the observation matrix L, output matrix C, input matrix B and coefficient matrix K into the constant coefficient matrix G of the temperature observer; The third constraint is: Substitute the system matrix A, output matrix C, observation matrix L and constant coefficient matrix K into the third constraint condition and design the constant coefficient matrices F and H of the temperature observer.

11. The on-line observer of power device junction temperature with unknown shell temperature and adaptive to environmental disturbance according to claim 10, characterized in that: In step 6, the observer coefficient matrix G is multiplied by the known input vector U to obtain vector 1, the measurable variable y is multiplied by the observer coefficient matrix H to obtain vector 2, and the observer state vector 5 is multiplied by the observer coefficient matrix F to obtain vector 3; then vector 1, vector 2 and vector 3 are added to obtain vector 4; vector 4 is passed through an integral link to obtain observer state vector 5; the measurable variable y is multiplied by the observer coefficient matrix K to obtain vector 6, and then the observer state vector 5 and vector 6 are added to obtain the observer output vector V e =[V1,V2] T , where V1 is the junction temperature of the semiconductor device in the module to be observed.

Images