Method for implementing temperature monitoring on power devices by means of thermal observer

Abstract

Disclosed in the present invention is a method for implementing temperature monitoring on power devices by means of a thermal observer. The method comprises: mounting sensors on the upper surface of a radiator and at an NTC in a power module, so as to construct a sensor network to implement information collection with respect to heating situations of different power devices; fitting a thermal network on the basis of a thermal response curve, and extracting thermal resistance and thermal capacitance from a model to establish a thermal impedance matrix, and forming a state-space equation for a thermal sensor network; establishing a thermal observer by using a Gillijn De Moor filter algorithm, so as to estimate losses of the power devices, and realize thermal information observation with respect to operations of the devices; and fitting a thermal network model of the devices and the environment by means of the thermal network, so as to realize real-time tracking of junction temperatures of the power devices. The present invention can be flexibly expanded on the basis of the number of devices and a module structure, thereby realizing device temperature monitoring with high safety, good real-time performance and high accuracy.

Description

A method for realizing temperature monitoring of power devices by thermal observer Technical Field

[0001] The present invention relates to the technical field of power electronics, and in particular to a method for realizing temperature monitoring of a power device using a thermal observer. Background Art

[0002] In recent years, with the rapid development of electric vehicles, new energy power generation, data center power supply, mining and smelting, and other fields, the global converter market has experienced rapid expansion. Research and exploration of large-capacity power conversion equipment has become a key support for overcoming technical challenges and future industry expansion. According to statistics, 34% of equipment reliability issues are caused by power device failures, while overheating and aging caused by temperature stress account for 55% of the total failure causes. Therefore, device junction temperature monitoring is the core key to addressing device damage and ensuring stable and reliable equipment operation. Currently, power device junction temperature monitoring technologies can be divided into two categories: sensor-based junction temperature monitoring and thermal model-based junction temperature monitoring.

[0003] References [1] IGBT Module with User Accessible On-chip Current and Temperature Sensors (Applied Power Electronics Conference and Exposition (APEC), Orlando, FL, USA, 2012, pp. 176-181, doi: 10.1109 / APEC.2012.6165816.) and [2] Comparison of Junction Temperature Evaluations in a Power IGBT Module Using an IR Camera and Three Thermosensitive Electrical Parameters (IEEE Trans. on Industry Applications, vol. 49, no. 4, pp. 1599-1608, July-Aug. 2013) respectively arrange electrical sensors and optical sensors to collect corresponding operating data, and combine the mapping relationship between the collected data and temperature to deduce the junction temperature distribution of the device. Among them, the method based on electrical sensors in reference [1] does not need to destroy the original package and has a faster response speed, which has great advantages in terms of safety and real-time performance. However, further research is needed on how to ensure the comprehensiveness of temperature information captured by the sensor method and improve its performance in aging and bandwidth issues.

[0004] Reference [3] A Lumped Thermal Model Including Thermal Coupling and Thermal Boundary Conditions for High-Power IGBT Modules (IEEE Trans. on Power Electron., vol. 33, no. 3, pp. 2518-2530, March 2018) proposes a lumped thermal model that parameterizes the circuit structure based on data fitting, abstracting it into a thermal resistance network to analyze the thermal characteristics of the circuit. However, in actual applications, the parameters in the thermal resistance model are affected by changes in heat dissipation conditions and processing errors. Under the cumulative effects of factors such as material aging, material temperature changes, processing errors, and loss errors, the final calculation error of the thermal resistance model method can exceed 50%.

[0005] Summary of the Invention

[0006] Given the aforementioned limitations, junction temperature observation based on either sensor or thermal resistance models alone cannot meet the requirements for real-time, accurate monitoring. Therefore, this invention provides a method for power device temperature monitoring using a thermal observer. This method estimates intermittent power loss through the thermal observer and, combined with a thermal network model, enables tracking of intermittent power loss.

[0007] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0008] The present invention discloses a method for monitoring the temperature of a power device based on a thermal observer, the method comprising the following steps:

[0009] S1: Install a sensor network, which includes n+1 sensors, where the first n sensors are used to collect temperature information of n power device groups respectively, where n is a natural number greater than 0, and the last sensor is used to collect a reference temperature, to build a sensor thermal network model;

[0010] S2: Fitting the sensor thermal network model based on the temperature information of the power device group collected by the sensor network and the reference temperature, and extracting the thermal capacitance and thermal resistance information of the thermal network model;

[0011] S3: establishing a sensor thermal network state space equation based on the thermal resistance and thermal capacity information obtained in step S2, and then discretizing the established sensor thermal network state space equation to obtain a discrete sensor thermal network state space equation;

[0012] S4: Based on the discrete sensor thermal network state space equation, the power loss of the power device group is estimated using the Gillijn De Moore filter algorithm;

[0013] S5: A device-environment thermal network model is obtained by using Foster thermal network fitting, and a device-environment thermal network state space equation is established. Based on the power loss obtained in step S4, real-time tracking of the junction temperature of the power device is achieved.

[0014] As a preferred embodiment of the present invention, the power device group is located in a power module, the power module is mounted on a radiator, the n sensors are fixed to the upper surface of the radiator, and the center position of the power device located in the center of the n power device group is vertically projected onto the upper surface of the radiator, which is the specific installation position of the n sensors; the power module is also integrated with a thermistor, and a sensor is fixed in the center of the surface of the thermistor. The sensor on the surface of the thermistor is used to measure and determine the reference temperature of the sensor thermal network model.

[0015] Based on the above technical solution, the beneficial effects of the present invention compared with the prior art are:

[0016] (1) The present invention fully combines the advantages of temperature sensing and thermal resistance model methods, using the high accuracy of temperature sensors and the low latency of thermal resistance models to achieve real-time and accurate estimation of semiconductor device power loss;

[0017] (2) The present invention utilizes sensor data to reduce the impact of solder layer aging and material parameter temperature dependence on the loss observer, has strong robustness against aging, and improves the practicality of the observer throughout the life cycle of the converter;

[0018] (3) The present invention suppresses the uncertainty of loss while ensuring safety without destroying the module packaging. By adjusting the gain matrix of input quantity and state quantity, the minimum bias estimation of local temperature rise and unknown power loss is achieved, the junction temperature estimation error is reduced, and multi-dimensional real-time monitoring of power semiconductor devices is realized, which is conducive to ensuring the stable and healthy operation of large-capacity power conversion equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. 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 creative work.

[0020] FIG1 is a flow chart of a method for monitoring power device temperature based on a thermal observer according to the present invention;

[0021] FIG2 is a schematic diagram of a temperature sensor layout according to an embodiment of the present invention;

[0022] FIG3 is a schematic diagram of a sensor thermal network according to an embodiment of the present invention;

[0023] FIG4 is a schematic diagram of a thermal observer based on the Gillijn De Moore filter algorithm according to an embodiment of the present invention;

[0024] FIG5 is a schematic diagram of a device-environment thermal network according to an embodiment of the present invention;

[0025] FIG. 6 is a diagram showing power loss and temperature observation results according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The present invention will be further described and illustrated below in conjunction with specific embodiments. The embodiments are merely illustrative of the present disclosure and do not limit its scope. The technical features of the various embodiments of the present invention may be combined accordingly, provided that there is no conflict between them.

[0027] As shown in FIG1 , the method for implementing power device temperature monitoring using a thermal observer according to the present invention includes the following steps:

[0028] S1. Install sensors on the upper surface of the heat sink and the NTC (thermistor) inside the power module to form a sensor network to collect information on the heating conditions of different heat sources, that is, different power device groups;

[0029] S2. Fit the thermal network based on the collected information and construct a heat flow transmission path from the sensor network to the environment. The voltage, current, capacitance, and resistance in the path correspond to the temperature, loss, thermal capacitance, and thermal resistance information in the thermal network model respectively.

[0030] S3. Constructing a thermal impedance matrix based on the thermal resistance and thermal capacitance information to establish a sensor thermal network state space equation, and then discretizing the established sensor thermal network state space equation to obtain a discrete sensor thermal network state space equation;

[0031] S4. Use the Gillijn De Moore filter algorithm, an extension of the Kalman filter, to estimate the power device loss and realize the thermal information observation of the device operation;

[0032] S5. Fit the heat flow transmission path from the device to the environment through the thermal network, establish the device-environment thermal network state space equation, calculate the temperature changes caused by different heat sources based on the measured ambient temperature and the power loss obtained by the thermal observer, and accumulate the temperature rise to achieve online tracking of the power device junction temperature.

[0033] According to the above method, the present invention constructs a thermal observer to achieve real-time observation of the junction temperature of different heat sources.

[0034] The power device is located in a power module, which includes n power device groups. The power device groups include several power devices of the same type connected in parallel. The junction temperatures of the power devices in the same power device group can be considered to be the same. In this embodiment, the power devices are the same devices corresponding to each bridge arm in the half-bridge sub-module.

[0035] The present invention installs a total of n+1 sensors, all of which are temperature sensors. A schematic diagram of the temperature sensor layout of a specific embodiment of the present invention is shown in Figure 2. The power module is located above the heat sink. The power module contains 4 (n=4) power device groups. Each power device group contains 3 power devices connected in parallel. The power device located in the middle of the 3 power devices has its center position at the vertical projection of the upper surface of the heat sink (as shown in the left half of Figure 2), which is the installation position of the n sensors. The remaining 1 sensor is fixed in the center of the NTC surface integrated in the module. The sensor layout of the present invention ensures the safety of the measurement process while ensuring the accuracy and comprehensiveness of the measurement data.

[0036] A schematic diagram of a sensor thermal network model according to a specific embodiment of the present invention is shown in FIG3 . The thermal resistance network unit fits the heat flow transmission path from the sensor to the environment through the thermal network, including the temperature rise effect caused by different heat sources. The heat flow transmission path caused by the temperature response at sensor i caused by power device group j is abstracted as a fourth-order Foster network, represented by the thermal resistance R SiPj,1 、R SiPj,2 、R SiPj,3 、R SiPj,4 and heat capacity C SiPj,1 、C SiPj,2 、C SiPj,3 、C SiPj,4 n groups of fourth-order Foster networks constitute the heat flow transmission path from sensor i to the reference point, and n groups of heat flow transmission paths from a single sensor to the environment constitute the sensor thermal network.

[0037] The form of the thermal network model of the sensor network can be abstracted as follows:

[0038] Where, T Si (t) is the temperature at time t measured by the i-th sensor corresponding to the power device group, Zth SiPj (t)(i=1,2,…,n;j=1,2,…,n) is the thermal impedance between the i-th sensor and the j-th power device group at time t, P j (t) is the power loss of the jth power device group at time t, T NTC The temperature measured by the sensor at the thermistor inside the module is used as the reference temperature.

[0039] According to the transient thermal response curve obtained from the sensor network, the extraction formula of the parameters of the thermal network model is:

[0040] Where R SiPj,r is the rth order thermal resistance between the i-th sensor and the j-th power device group, that is, the required thermal resistance information, C SiPj,r is the rth order thermal capacitance between the i-th sensor and the j-th power device group, i.e. the required thermal capacitance information, τ SiPj,r is the rth order time constant of the i-th sensor and j-th power device group, and m is the order of the thermal network model. In this embodiment, k=4.

[0041] As a preferred embodiment of the present invention, the state space equation for modeling a continuous sensor thermal network is expressed as: y(t)=Cx(t)

[0042] Where x(t) is the state quantity of the sensor thermal network state space equation at time t, that is, the temperature rise on each order heat capacity in the thermal network model. is the differential of x(t), d(t) is the input of the sensor thermal network state space equation at time t, that is, the power loss in the thermal network model, y(t) is the output of the sensor thermal network state space equation at time t, that is, the temperature measured by the sensor in the thermal network model, A, B, and C are the state transfer matrix, input matrix, and output matrix of the sensor thermal network state space equation, that is, the thermal impedance matrix.

[0043] The expressions of state quantity x(t), input quantity d(t), output quantity y(t), state transfer matrix A, input matrix B, and output matrix C are:

[0044] Among them, u SiPj,r (t) is the temperature rise at time t on the r-th order heat capacity of the i-th sensor and j-th power device group, m is the order of the thermal network model, P j (t) is the power loss of the jth power device group at time t, T Si (t) is the temperature at time t measured by the sensor corresponding to the i-th power device group, τ SiPj,r is the r-th order time constant of the i-th sensor and j-th power device group, C SiPj,r is the rth order thermal capacitance between the i-th sensor and the j-th power device group.

[0045] The established continuous sensor thermal network state space equation is discretized with a fixed sampling time step, and the process noise w and measurement noise v are included in the model. The state space expression of discrete thermal information observation is established as:k+1 =A d x k +B d d k +w k y k =C d x k +v k

[0046] Where, process noise w k With the measurement noise v k is the zero-mean random white noise with no correlation; x k 、x k+1 are the state quantities at the kth moment and the k+1th moment, d k is the input at the kth moment, y k is the output at the kth moment, A d 、B d 、C d They are the state transfer matrix, input matrix, and output matrix in the discrete thermal network state space equation respectively.

[0047] A schematic diagram of a thermal observer based on the Gillijn De Moore filter algorithm according to a specific embodiment of the present invention is shown in FIG4 . This observer uses device power loss as input, device temperature rise as a state variable, and the observer's measured temperature as output. The Gillijn De Moore filter algorithm is used to modify the gain matrix, achieving unbiased minimum variance estimation of unknown state variables and online tracking of power device losses. The specific method for estimating the unknown input power loss is as follows:

[0048] The expression of the established recursive filter is:

[0049] Where, is the estimate of the input quantity of the sensor thermal network state space equation at the k-1th moment, the input quantity of the sensor thermal network state space equation is the power loss of the power device group, y k is the output at the kth moment, is the prior estimate of the state quantity when the input quantity at the kth moment is unknown, are the minimum variance unbiased estimates of the state quantity based on the input quantity estimation results at the k-1th moment and the kth moment, is the unbiased estimate of the state quantity based on the input quantity estimation result at the kth moment, M k and K k are the gain matrices of the input quantity and state quantity at the kth moment respectively.

[0050] In order to ensure the minimum unbiased estimation of input and state quantities, the input gain matrix M k And the state quantity gain matrix K k The calculation method is:

[0051] in, is the error covariance matrix at the k-1th moment, P k-1|k-2 、P k|k-1 are the prior covariance matrices at the k-1th moment and the kth moment, M k-1 are the input gain matrices at the k-1th moment, K k-1 are the state quantity gain matrices at the k-1th moment, Q k-1 =E[w k-1 w k-1 T ] is the process noise w at the k-1th moment k-1 The covariance matrix, R k-1 =E[v k-1 v k-1 T ] is the measurement noise v at the k-1th moment k-1 The covariance matrix, R k =E[v k v k T ] is the measurement noise v at the kth moment k The covariance matrix of n is the identity matrix of size n, and the superscript T indicates the matrix transpose.

[0052] The junction temperature thermal network diagram of the junction temperature calculation unit in a specific embodiment of the present invention is shown in Figure 5. The thermal resistance network unit fits the heat flow transmission path from the device to the environment through the thermal network. The heat flow transmission path caused by the temperature response of the target observation position i caused by the power device group j is abstracted as a fourth-order Foster network, which is represented by the thermal resistance R ja,SiPj,1 、R ja,SiPj,2 、R ja,SiPj,3 、R ja,SiPj,4 and heat capacity C ja,SiPj,1 、C ja,SiPj,2 、C ja,SiPj,3 、C ja,SiPj,4 n groups of fourth-order Foster networks constitute the heat flow transmission path from the target observation position i to the environment, and n groups of heat flow transmission paths from a single target observation position to the environment constitute the sensor thermal network.

[0053] Establish the corresponding device-environment thermal network state space equation and calculate the power device junction temperature based on the power device group loss. The calculation method is: y ja (t) = C ja x ja (t)

[0054] Where x ja (t) is the state quantity of the device-environment thermal network state space equation, is x ja The differential of (t), that is, the temperature rise of each order heat capacity in the device-environment thermal network model, d(t) is the input quantity of the device-environment thermal network state space equation, that is, the power loss in the thermal network model, y ja (t) is the output of the device-environment thermal network state space equation, that is, the junction temperature, A ja 、B ja 、C ja They are the state transfer matrix, input matrix, and output matrix in the device-environment thermal network state space equation, respectively.

[0055] State quantity x ja , input d, output y ja , state transfer matrix A ja , input matrix B ja , output matrix C ja The expression is:

[0056] Among them, u ja,dipj,r (t) is the temperature rise at time t between the i-th junction temperature point and the r-th order thermal capacitance of the j-th power device group, m ja is the order of the device-environment thermal network model, P j (t) is the power loss of the jth power device group at time t, T di (t) is the junction temperature of the i-th power device group at time t, τ ja,diPj,r is the rth order time constant between the i-th junction temperature point and the j-th power device group, C ja,diPj,r is the r-th order thermal capacitance between the i-th junction temperature point and the j-th power device group; the junction temperature point is the position of the power device observed by this thermal observer.

[0057] The device-environment thermal network state space equation is solved, and the output value is the temperature of each power device group at each moment.

[0058] Figure 6 shows the power loss and temperature observation results for a specific embodiment of the present invention. The experiment used a 1700V / 450A IGBT module (model 2MBI450VN-170) secured to an aluminum air-cooled heat sink via thermal pads. Sensors were installed according to the aforementioned method, and a thermal observer was constructed. During the experiment, a DC source alternately supplied currents of 0A, 60A, 0A, and 90A to the upper-arm diode of the IGBT module. The thermal observer estimated the diode power loss and the temperature of the diode and IGBT. The estimated results largely coincided with the actual measured results, demonstrating the feasibility and accuracy of the present invention.

[0059] The above-described embodiments merely illustrate several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. Persons skilled in the art will readily appreciate that variations and modifications may be made without departing from the scope of the present invention, all of which fall within the scope of protection of the present invention.

Claims

1. A method for implementing temperature monitoring of a power device using a thermal observer, characterized in that: The method comprises the following steps: S1: Install a sensor network, which includes n+1 sensors, where the first n sensors are used to collect temperature information of n power device groups respectively, where n is a natural number greater than 0, and the last sensor is used to collect a reference temperature, to build a sensor thermal network model; S2: Fitting the sensor thermal network model based on the temperature information of the power device group collected by the sensor network and the reference temperature, and extracting the thermal capacitance and thermal resistance information of the thermal network model; S3: establishing a sensor thermal network state space equation based on the thermal resistance and thermal capacity information obtained in step S2, and then discretizing the established sensor thermal network state space equation to obtain a discrete sensor thermal network state space equation; S4: Based on the discrete sensor thermal network state space equation, the power loss of the power device group is estimated using the Gillijn De Moore filter algorithm; S5: A device-environment thermal network model is obtained by using Foster thermal network fitting, and a device-environment thermal network state space equation is established. Based on the power loss obtained in step S4, real-time tracking of the junction temperature of the power device is achieved.

2. The method for implementing power device temperature monitoring using a thermal observer according to claim 1, wherein: The power device group includes a plurality of power devices of the same type connected in parallel.

3. The method for realizing temperature monitoring of power devices by a thermal observer according to claim 1, characterized in that: In step S1, the power device group is located in a power module, the power module is mounted on a heat sink, the first n sensors are fixed to the upper surface of the heat sink, and the specific installation position of each sensor is the vertical projection of the center position of the power device located in the center of the corresponding power device group on the upper surface of the heat sink; the power module also has an integrated thermistor, and the last sensor is fixed to the center of the surface of the thermistor to measure and determine the reference temperature of the sensor thermal network model.

4. The method for implementing temperature monitoring of a power device using a thermal observer according to claim 1, wherein: In step S1, the sensor thermal network model is expressed as: Where, T Si (t) is the temperature at time t measured by the i-th sensor corresponding to the power device group, Zth SiPj (t) (i=1,2,…,n;j=1,2,…,n) is the thermal impedance between the i-th sensor and the j-th power device group at time t, P j (t) is the power loss of the jth power device group at time t, T NTC is the base temperature.

5. The method for implementing temperature monitoring of a power device using a thermal observer according to claim 2, wherein: In step S2, the expression of the sensor thermal network model is: Where R SiPj,r is the rth order thermal resistance between the i-th sensor and the j-th power device group, that is, the required thermal resistance information, C SiPj,r is the rth order thermal capacitance between the i-th sensor and the j-th power device group, i.e. the required thermal capacitance information, τ SiPj,r is the rth-order time constant of the i-th sensor and j-th power device group, and m is the order of the thermal network model; The temperature information of the power device group collected by the sensor network and the reference temperature are used to calculate the expression of the sensor thermal network model, that is, the thermal resistance and thermal capacitance information R are obtained. SiPj,r and C SiPj,r .

6. The method for implementing temperature monitoring of a power device using a thermal observer according to claim 1, wherein: In step S3, the sensor thermal network state space equation is established as: y(t)=Cx(t) Where x(t) is the state quantity of the sensor thermal network state space equation at time t, that is, the temperature rise on each order heat capacity in the thermal network model. is the differential of x(t), d(t) is the input of the sensor thermal network state space equation at time t, that is, the power loss in the sensor thermal network model, y(t) is the output of the sensor thermal network state space equation at time t, that is, the temperature measured by the sensor in the sensor thermal network model, A, B, and C are the state transfer matrix, input matrix, and output matrix of the sensor thermal network state space equation, respectively; In the state space equation of the sensor thermal network, the expressions of the state quantity x(t), input quantity d(t), output quantity y(t), state transfer matrix A, input matrix B, and output matrix C are respectively: Among them, u SiPj,r (t) is the temperature rise at time t on the r-th order heat capacity of the i-th sensor and j-th power device group, m is the order of the sensor thermal network model, P j (t) is the power loss of the jth power device group at time t, T Si (t) is the temperature at time t measured by the sensor corresponding to the i-th power device group, τ SiPj,r is the r-th order time constant of the i-th sensor and j-th power device group, C SiPj,r is the rth order thermal capacitance between the i-th sensor and the j-th power device group.

7. The method for implementing temperature monitoring of a power device using a thermal observer according to claim 6, wherein: In step S3, the established sensor thermal network state space equation is discretized to obtain a discrete sensor thermal network state space equation. The specific method is: the established continuous sensor thermal network state space equation is discretized with a fixed sampling time step, and the process noise w is k and measurement noise v k The discrete sensor thermal network state space equation is contained in the sensor thermal network state space equation. The expression of the discrete sensor thermal network state space equation is: k+1 =A d x k +B d d k +w k y k =C d x k +v k Where, process noise w k With the measurement noise v k is the zero-mean random white noise with no correlation; x k 、x k+1 are the state quantities at the kth moment and the k+1th moment, d k is the input at the kth moment, y k is the output at the kth moment, A d 、B d 、C d They are the state transfer matrix, input matrix, and output matrix in the discrete thermal network state space equation respectively.

8. The method for implementing temperature monitoring of a power device using a thermal observer according to claim 1, wherein: In step S4, the Gillijn De Moore filter algorithm is an extended form of Kalman filtering. The power loss of the power device group is estimated using the Gillijn De Moore filter algorithm based on the discrete sensor thermal network state space equation. The specific method is: A recursive filter is established based on the Gillijn De Moore filter algorithm, which is an extension of the Kalman filter. The expression of the recursive filter is: Where, is the estimate of the input quantity of the sensor thermal network state space equation at the k-1th moment, the input quantity of the sensor thermal network state space equation is the power loss of the power device group, y k is the output at the kth moment, is the prior estimate of the state quantity when the input quantity at the kth moment is unknown, are the minimum variance unbiased estimates of the state quantity based on the input quantity estimation results at the k-1th moment and the kth moment, is the unbiased estimate of the state quantity based on the input quantity estimation result at the kth moment, M k and K k are the gain matrices of the input and state quantities at the kth moment, A d 、B d 、C d They are the state transfer matrix, input matrix, and output matrix in the discrete thermal network state space equation respectively.

9. The method for implementing temperature monitoring of a power device using a thermal observer according to claim 8, wherein: Input gain matrix M k And the state quantity gain matrix K k The calculation method is: in, is the error covariance matrix at the k-1th moment, P k-1|k-2 、P k|k-1 are the prior covariance matrices at the k-1th moment and the kth moment, M k-1 are the input gain matrices at the k-1th moment, K k-1 are the state quantity gain matrices at the k-1th moment, Q k-1 =E[w k-1 w k-1 T ] is the process noise w at the k-1th moment k-1 The covariance matrix, R k-1 =E[v k-1 v k-1 T ] is the measurement noise v at the k-1th moment k-1 The covariance matrix, R k =E[v k v k T ] is the measurement noise v at the kth moment k The covariance matrix of n is the identity matrix of size n, E[.] represents the computational expectation, and the superscript T represents the matrix transpose.

10. The method for implementing temperature monitoring of a power device using a thermal observer according to claim 1, wherein: In step S5, the device-environment thermal network model is characterized by a device-environment thermal network state space equation, and the device-environment thermal network state space equation is: y ja (t) = C ja x ja (t) Where x ja (t) is the state quantity of the device-environment thermal network state space equation, is x ja The differential of (t), that is, the temperature rise of each order heat capacity in the device-environment thermal network model, d(t) is the input quantity of the device-environment thermal network state space equation, that is, the power loss in the device-environment thermal network model, y ja (t) is the output of the device-environment thermal network state space equation, that is, the junction temperature, A ja 、B ja 、C ja are the state transfer matrix, input matrix, and output matrix in the device-environment thermal network state space equation respectively; In the device-environment thermal network state space equation, the state quantity x ja (t), input d(t), output y ja (t), state transfer matrix A ja , input matrix B ja , output matrix C ja The expression is: Among them, u ja,dipj,r (t) is the temperature rise at time t between the i-th target observation position and the r-th order heat capacity of the j-th power device group, m ja is the order of the device-environment thermal network model, P j (t) is the power loss of the jth power device group at time t, T di (t) is the junction temperature of the i-th power device group at time t, τ ja,diPj,r is the r-th order time constant between the i-th target observation position and the j-th power device group, C ja,diPj,r is the r-th order heat capacity between the i-th target observation position and the j-th power device group; the target observation position is the center of the power device finally observed; Solve the device-environment thermal network state space equation, and the output value is the junction temperature of each power device at each moment.

Images