CN114442694A - Self-calibration silicon carbide motor controller junction temperature estimation method - Google Patents

Abstract

The invention discloses a self-calibration silicon carbide motor controller junction temperature estimation method, which comprises the steps of analyzing and extracting an influence factor with the largest influence on junction temperature calculation according to a silicon carbide junction temperature calculation formula, constructing a junction temperature fitting formula, determining undetermined parameters in the junction temperature fitting formula according to infrared test data of a silicon carbide black module, calibrating the junction temperature fitting formula by using an NTC (negative temperature coefficient) sensor to improve estimation precision, completing self calibration of a silicon carbide junction temperature estimation result of a motor controller, and obtaining a silicon carbide junction temperature estimation value T_{j-calibration}. The method can feed back the junction temperature of the silicon carbide chip in real time, and uses an NTC sensor arranged in a motor controller to calibrate the temperature.

Description

Self-calibration silicon carbide motor controller junction temperature estimation method

Technical Field

The invention relates to a calculation method, in particular to a self-calibration silicon carbide motor controller junction temperature estimation method.

Background

The rapid development in the fields of electric vehicles, new energy power generation, smart grids and the like puts higher and higher requirements on the performance and reliability of power semiconductor devices. For an electric automobile, the application environment of the power semiconductor device is more severe due to the uncertainty of the external environment and the operation condition.

The third generation semiconductor materials represented by silicon carbide have the characteristics of high forbidden band width, high thermal conductivity, high breakdown field strength and the like, and become preferred materials of high-voltage, high-temperature and high-frequency power devices. The silicon carbide chip has smaller volume and thinner size, and brings stronger heat aggregation effect and higher current density. Therefore, chip temperature monitoring of silicon carbide power devices is important to reliable operation thereof. An electric vehicle motor controller based on a silicon carbide semiconductor device is generally provided with one or more Negative Temperature Coefficient (NTC) temperature sensors on a ceramic copper-clad plate (DBC) at a position close to a silicon carbide chip.

From the position, a certain distance exists between the NTC temperature sensor and the silicon carbide chip, and the real temperature of the silicon carbide chip cannot be reflected. Therefore, the value of the NTC temperature sensor can only be used as a reference temperature, and the temperature of the silicon carbide chip can represent the real temperature of the chip.

The interpretation of the prior art and the junction temperature monitoring method mainly focus on four aspects: circuit detection methods, thermal network methods, temperature estimation methods, and data driven methods.

The circuit detection method is to obtain electric signals such as blocking leakage current, conducting voltage drop and the like by using an additional detection circuit. And then calculating the junction temperature of the power semiconductor according to a junction temperature calculation formula. The extra detection circuit requires the arrangement of collecting points on the collector and emitter of the semiconductor, which changes the original circuit structure and increases the cost, and at the same time, the unpredictable influence of electromagnetic compatibility (EMC) is also introduced.

The thermal network method abstracts a stacked structure of the power semiconductor modules into a series-parallel network of thermal resistance and thermal capacity. And establishing a heat network equation for heat transfer of the power semiconductor module according to the heat transfer relation between the power semiconductor module and the cooling system. And substituting the loss of the power semiconductor into the established heat network equation to obtain the junction temperature of the power semiconductor. The method solves the problem that a cooling system and a cooling condition are not considered in the temperature prediction, so that the method is closer to the actual working condition. However, since the thermal resistance and the thermal capacity are not constants but nonlinear values with temperature changes, the thermal network equation is not completely accurate.

The temperature estimation method is to construct a loss model by using parameters such as output current, output frequency, bus voltage and the like of the power semiconductor module. And estimating the junction temperature of the power semiconductor through various self-adaptive algorithms. The method has higher requirement on the real-time performance of the operation of the motor controller chip and can occupy a large amount of computing resources of a processor.

The data driving method is to collect a great deal of sample data by means of an infrared camera, an optical fiber sensor, a thermocouple and the like. And a functional relation among the current, the voltage, the frequency and the junction temperature of the power semiconductor is constructed in a data fitting or neural network training mode. This method is too dependent on the sample size and if there is a gross deviation in the sample size, the measurement will be inaccurate. However, if the actual temperature collected by the NTC sensor is fully utilized, the reliability of the power semiconductor junction temperature estimation is greatly improved.

The methods have the problems that the original circuit structure is changed, the cost is increased, the requirement on the real-time performance of the operation of a motor controller chip is high, a large amount of computing resources of a processor are occupied, the measurement result is inaccurate and the like.

Disclosure of Invention

The invention provides a self-calibration silicon carbide motor controller junction temperature estimation method, which aims to feed back the junction temperature of a silicon carbide chip in real time and carry out temperature calibration by using an NTC sensor built in a motor controller.

The general idea of the invention is as follows:

and analyzing and extracting an influence factor which has the largest influence on the junction temperature calculation based on the silicon carbide junction temperature calculation formula, and constructing a junction temperature fitting formula. And determining undetermined parameters in a junction temperature fitting formula based on the infrared test data of the silicon carbide black module. And calibrating the temperature fitting formula by using the NTC sensor to improve the estimation accuracy. In the invention, data required by the junction temperature fitting formula are fed back by the existing sensor and the main control chip of the motor controller, and no additional sensor or hardware circuit structure is required to be added.

According to the derivation analysis of the junction temperature formula, the formula for fitting the junction temperature of the silicon carbide chip can be obtained as follows:

T_j＝f(K₁,K₂,K₃,K₄,K₅,f_s,i,V_dc) (1)

from the above analysis, it can be seen that the junction temperature T with silicon carbide_jThe three relevant factors are load current i and bus voltage V_dcSwitching frequency f_s. The junction temperature T can be obtained through a silicon carbide junction temperature test_jLoad current i, bus voltage V_dcSwitching frequency f_sThe measured data of (a). Then using the formula (1) to calculate the junction temperature T_jFor output, the load current i and the bus voltage V are used_dcSwitching frequency f_sFitting the data to the input, K in the formula can be determined₁To K₅And (4) the coefficient. Thereby realizing the on-line real-time estimation of the junction temperature of the silicon carbide.

The invention uses the NTC temperature sensor for calibrating the real-time estimation result of the silicon carbide junction temperature. Unlike silicon carbide junction temperature, which cannot be directly measured by a sensor, an NTC temperature sensor can acquire and feed back a temperature value (the temperature of the region where the NTC sensor is located) in real time. In the above-mentioned silicon carbide junction temperature test, the junction temperature T is recorded simultaneously_jMeasured temperature T with NTC_{NTC-actual measurement}And again applying the formula (1) as the NTC temperature T_{NTC-actual measurement}For output, the load current i and the bus voltage V are used_dcSwitching frequency f_sFitting the data to the input, M in the formula can be determined₁To M₅And (4) the coefficient. Thereby realizing real-time estimation of NTC temperature and obtaining estimation result T_{NTC estimation}The formula is as follows:

T_{NTC estimation}＝f(M₁,M₂,M₃,M₄,M₅,f_s,i,V_dc) (2)

Estimating the temperature T by the NTC_{NTC estimation}Measured temperature T with NTC_{NTC-actual measurement}By comparison, the deviation caused by the estimation formula can be obtained:

T_{NTC deviation}＝T_{NTC estimation}-T_{NTC-actual measurement} (3)

Applying the deviationAt junction temperature T_jCompensating the estimation result to finish the self calibration of the silicon carbide junction temperature estimation result of the motor controller, wherein the formula is as follows:

T_{j-calibration}＝T_j+T_{NTC deviation} (4)。

Based on the above thought, the technical scheme of the invention is as follows:

a self-calibration silicon carbide motor controller junction temperature estimation method comprises the following steps:

step 1, data acquisition: collecting load current i and bus voltage V_dcSwitching frequency f_sNTC measured temperature T_{NTC-actual measurement}；

Step 2, recording the actual measurement temperature T of the NTC_{NTC-actual measurement}；

Meanwhile, step 3, calculating the NTC estimated temperature, wherein the formula is as follows:

T_{NTC estimation}＝f(M₁,M₂,M₃,M₄,M₅,f_s,i,V_dc) (2)

Wherein M is₁To M₅Is the NTC temperature estimation coefficient;

and meanwhile, step 4, calculating the estimated junction temperature of the silicon carbide, and adopting a silicon carbide chip junction temperature fitting formula, wherein the formula is as follows:

T_j-estimation＝f(K₁,K₂,K₃,K₄,K₅,f_s,i,V_dc)

Wherein, K₁To K₅Is the junction temperature estimation coefficient;

step 5, estimating the temperature T of the NTC_{NTC estimation}Measured temperature T with NTC_{NTC-actual measurement}And comparing to obtain the deviation brought by an estimation formula, wherein the formula is as follows:

T_{NTC deviation}＝T_{NTC estimation}-T_{NTC-actual measurement} (3)

Applying the deviation to the junction temperature estimate T_j-estimationThe self calibration of the silicon carbide junction temperature estimation result of the motor controller is completed, and the estimated value T of the silicon carbide junction temperature is obtained_{j-calibration}The formula is as follows:

T_{j-calibration}＝T_j-estimation+T_{NTC deviation} (4)。

Further, the junction temperature estimation coefficient K of the step 4₁To K₅The method is to pass the junction temperature test of silicon carbide in advance, utilize the formula (1),

T_j＝f(K₁,K₂,K₃,K₄,K₅,f_s,i,V_dc) (1)

by NTC temperature T_{NTC-actual measurement}For output, the load current i and the bus voltage V are used_dcSwitching frequency f_sFitting data for input, and calculating to obtain T_jThe actual junction temperature of the silicon carbide obtained by collection is obtained.

Further, the NTC temperature estimation coefficient M₁To M₅The silicon carbide junction temperature test is passed in advance by using a formula (2)

T_{NTC estimation}＝f(M₁,M₂,M₃,M₄,M₅,f_s,i,V_dc) (2)

By NTC temperature T_{NTC estimation}For output, the load current i and the bus voltage V are used_dcSwitching frequency f_sAnd fitting data for input and calculating to obtain the target.

Further, data required by the junction temperature fitting formula are fed back by the motor controller, the sensor and the main control chip.

In conclusion, according to the method, the influence factor which has the greatest influence on the junction temperature calculation is analyzed and extracted according to the silicon carbide junction temperature calculation formula, the junction temperature fitting formula is constructed, the undetermined parameter in the junction temperature fitting formula is determined according to the infrared test data of the silicon carbide black module, the NTC sensor is used for calibrating the junction temperature fitting formula to improve the estimation precision, the self calibration of the silicon carbide junction temperature estimation result of the motor controller is completed, and the estimated value T of the silicon carbide junction temperature is obtained_{j-calibration}. The method can feed back the junction temperature of the silicon carbide chip in real time, and carry out temperature calibration by using the NTC sensor arranged in the motor controller, thereby reducing the heat loss of the power semiconductor device, improving the system efficiency of the driving motor and simultaneously ensuring that the power semiconductor device is designedSafe and reliable operation in the life cycle.

Drawings

Fig. 1 is a flow chart of silicon carbide junction temperature estimation through a silicon carbide junction temperature test.

Fig. 2 is a schematic diagram of on-board real-time estimation of silicon carbide junction temperature (including self-calibration).

Detailed Description

The invention is further described with reference to the accompanying drawings in which:

the silicon carbide junction temperature estimation shown in fig. 1 is calibrated during the design of the silicon carbide power module, and then applied to the real-time silicon carbide junction temperature estimation on the vehicle of fig. 2.

Table 1 is a table of experimental parameters of junction temperature of silicon carbide.

In the process of carrying out the silicon carbide junction temperature test, the bus voltage, the motor rotating speed and the torque can be set according to the table 1, and the load current, the switching frequency, the silicon carbide junction temperature and the NTC temperature under the working condition are recorded. Wherein, the load current, the bus voltage and the switching frequency respectively correspond to the input parameters of the load current i and the bus voltage V required by the silicon carbide junction temperature estimation working process_dcSwitching frequency f_s(ii) a The values in table 1 are examples, and the test data volume can be selectively increased or decreased according to the accuracy requirement of the silicon carbide junction temperature estimation.

Step 101: the silicon carbide junction temperature test was conducted, and the test was conducted according to the working conditions specified in table 1.

Table 1:

step 102: data acquisition: the actual junction temperature T of the silicon carbide is acquired by means of an infrared thermometer, an optical fiber temperature sensor, a silicon carbide chip built-in temperature sensor and the like_jLoad current i, bus voltage V_dcSwitching frequency f_sNTC temperature T_{NTC-actual measurement}By deduction and analysis according to a junction temperature formula, the method can obtainThe equation to the fitting for silicon carbide chip junction temperature is:

T_j＝f(K₁,K₂,K₃,K₄,K₅,f_s,i,V_dc) (1)

step 103: obtaining an NTC actual mining temperature: NTC temperature T_{NTC-actual measurement}The motor controller directly collects the signals and then applies the signals to the next calculation, and the formula is as follows:

T_{NTC estimation}＝f(M₁,M₂,M₃,M₄,M₅,f_s,i,V_dc) (2)

Step 104: according to the formula (2), the load current i and the bus voltage V are calculated_dcSwitching frequency f_sNTC temperature T_{NTC-actual measurement}Fitting by substituting formula, and calculating to obtain NTC temperature estimation coefficient M₁To M₅。

Step 105: according to the formula (1), the load current i and the bus voltage V are calculated_dcSwitching frequency f_sSilicon carbide junction temperature T_jFitting by substituting into a formula, and calculating to obtain a junction temperature estimation coefficient K₁To K₅。

Step 106: obtaining an NTC estimated temperature: the NTC temperature estimation coefficient M determined in step 104₁To M₅Substituting the formula (2) to calculate the NTC temperature T estimated in real time_{NTC estimation}。

Step 107: obtaining estimated junction temperature of silicon carbide: k determined according to step 105₁To K₅The coefficient, formula (1) is applied to the real-time estimation T of the IGBT junction temperature in the next step_j-estimation。

Referring to fig. 2, the present embodiment is a silicon carbide junction temperature estimation process with self-calibration compensation:

step 201: NTC temperature T is directly collected by motor controller_{NTC-actual measurement}And load current i, bus voltage V_dcSwitching frequency f_sAnd the like, and is used for the next calculation.

Step 202: NTC estimated temperature: from the previously determined coefficient M₁To M₅Applying equation (2) to the NTC temperature estimate to obtain T_{NTC-evaluationCalculating out}。

Meanwhile, step 203: estimating junction temperature of silicon carbide: determining the coefficient K from the preceding₁To K₅Using the formula (1) for the estimation of the junction temperature of silicon carbide to obtain T_j-estimation。

Step 204: for T_{NTC estimation}Filtering is performed to avoid inaccurate estimation due to fluctuation of input signal such as voltage, current, frequency, etc., and to make T_{NTC estimation}And T_{NTC-actual measurement}The following property between them is better.

The specific form of the filter is as follows:

Y(n)＝a·X(n)+(1-a)·Y(n-1)

a denotes a filter coefficient.

Y (n) represents the output value of this filtering

Y (n-1) represents the last filtered output value.

Step 205: and (4) subtracting the NTC estimated temperature from the NTC measured temperature according to the formula (3) to obtain an offset value. T is_{NTC deviation}The bias value can be more accurate for T_j-estimationSelf-calibration of (2).

The formula is as follows: t is_{NTC deviation}＝T_{NTC estimation}-T_{NTC-actual measurement} (3)

This deviation value will be used in self-calibration of the estimated junction temperature of the silicon carbide.

Step 206: and (3) compensation calibration: according to the formula (4), T_{NTC deviation}For junction temperature T of silicon carbide_j-estimationTo obtain a more accurate estimate of the silicon carbide junction temperature T_{j-calibration}。

The formula is as follows:

T_{j-calibration}＝T_j-estimation+T_{NTC deviation} (4)。

The data required by the junction temperature fitting formula are fed back by the motor controller, the sensor and the main control chip, and the sensor does not need to be additionally arranged or the hardware circuit structure does not need to be changed.

Claims (4)

1. A self-calibrating silicon carbide motor controller junction temperature estimation method, comprising the steps of:

step 1, data acquisition: collecting load current i and bus voltage V_dcSwitching frequency f_sNTC measured temperature T_{NTC-actual measurement}；

Step 2, recording the actual measurement temperature T of the NTC_{NTC-actual measurement}；

Meanwhile, step 3, calculating the NTC estimated temperature, wherein the formula is as follows:

T_{NTC estimation}＝f(M₁,M₂,M₃,M₄,M₅,f_s,i,V_dc) (2)

Wherein, M₁To M₅Is the NTC temperature estimation coefficient;

and meanwhile, step 4, calculating the estimated junction temperature of the silicon carbide, and adopting a silicon carbide chip junction temperature fitting formula:

T_j-estimation＝f(K₁,K₂,K₃,K₄,K₅,f_s,i,V_dc) (1)

Wherein, K₁To K₅Is the junction temperature estimation coefficient;

step 5, estimating the temperature T of the NTC_{NTC estimation}Measured temperature T with NTC_{NTC-actual measurement}And comparing to obtain the deviation brought by an estimation formula, wherein the formula is as follows:

T_{NTC deviation}＝T_{NTC estimation}-T_{NTC-actual measurement} (3)

Applying the deviation to the junction temperature estimate T_j-estimationThe self calibration of the silicon carbide junction temperature estimation result of the motor controller is completed, and the estimated value T of the silicon carbide junction temperature is obtained_{j-calibration}The formula is as follows:

T_{j-calibration}＝T_j-estimation+T_{NTC deviation} (4)。

2. The self-calibrating silicon carbide motor controller junction temperature estimation method of claim 1, wherein the junction temperature estimation coefficient K of step 4₁To K₅The method is to pass a silicon carbide junction temperature test in advance and utilize a formula,

T_j＝f(K₁,K₂,K₃,K₄,K₅,f_s,i,V_dc) (1)

by NTC temperature T_{NTC-actual measurement}For output, the load current i and the bus voltage V are used_dcSwitching frequency f_sFitting data for input, and calculating to obtain T_jThe actual junction temperature of the silicon carbide is collected.

3. The self-calibration silicon carbide motor controller junction temperature estimation method as claimed in claim 2, wherein in the step 1, the actual junction temperature T of the silicon carbide is collected by an infrared thermometer, an optical fiber temperature sensor and a silicon carbide chip built-in temperature sensor_j。

4. The self-calibrating silicon carbide motor controller junction temperature estimation method of claim 1, wherein the NTC temperature estimation coefficient M₁To M₅The silicon carbide junction temperature test is passed in advance by using a formula (2)

T_{NTC estimation}＝f(M₁,M₂,M₃,M₄,M₅,f_s,i,V_dc) (2)

By NTC temperature T_{NTC estimation}For output, the load current i and the bus voltage V are used_dcSwitching frequency f_sAnd fitting data for input and calculating to obtain the target.

Images