Adaptive direct power semiconductor control

Abstract

A method for controlling a traction inverter for a motor includes receiving a torque estimate from a torque meter connected to the traction inverter; receiving parameter data of a power semiconductor from a traction inverter semiconductor monitor; determining a health prediction of the power semiconductor based on the parameter data; determining an operating mode of the traction inverter based on the torque estimate, the parameter data of the power semiconductor, and the health prediction of the power semiconductor; and generating control commands based on the operating mode, wherein the control commands include a first gating control of gating the traction inverter, a second gating control of gating multiple gatings of the power semiconductor, or current control of current flowing to the power semiconductor.

Description

Technical Field

[0001] This invention relates to a traction converter including a power supply, and a method for controlling the traction converter. Specifically, the traction converter is controlled by dynamically switching multiple gates of the power semiconductors in the traction inverter of the traction converter. Background Technology

[0002] Modern electric and hybrid electric vehicles include complex power management systems, including traction converters. A traction converter may include one or more traction inverters that convert direct current (DC) from a power source into alternating current (AC) output. Each traction inverter may include a power module that combines multiple power semiconductors into a single package. The performance of the power module can have a direct impact on the performance of the electric motor in the electric and hybrid vehicle.

[0003] The efficiency of electric and hybrid vehicles can depend on the monitoring and control of the electrical power consumed or generated by different vehicle systems and subsystems. Dynamically controlling at least voltage and / or current peaks based on vehicle and driving characteristics can improve the efficiency of electric vehicles. For simplicity, "electric vehicle" will be used to refer to both pure electric vehicles and hybrid electric vehicles.

[0004] The traction inverter of an electric vehicle may include a power module comprising multiple power semiconductors. The reliability of the power module has a direct impact on the performance of the electric vehicle. A typical power module may be designed to manage all scenarios without considering environmental variables. In other words, the power semiconductor design may be over-designed to manage most driving behaviors. Because the static design of the power semiconductor cannot account for every driving behavior, damage to the power semiconductors in the traction inverter may occur during unconsidered driving conditions. For example, extreme driving behaviors in the real world, such as aggressive driving of electric vehicles, are not considered in the static design of the power semiconductor.

[0005] Extremely aggressive driving can frequently generate high torque commands to the electric propulsion system of electric vehicles. These high torque commands require relatively high current flows through the power semiconductors of the traction inverter. This high current flow leads to increased temperatures within the power semiconductor package. The power semiconductor package may include bonding wires, a copper-clad ceramic substrate, solder, sintering, and other materials. The use of different materials results in different coefficients of thermal expansion (CTE). These different CTEs generate stress on the power semiconductor package. The resulting stress causes damage to the power semiconductor, as well as the bonding wires, copper, ceramic, and other materials of the power module. This damage can reach levels that lead to power module failure.

[0006] As previously mentioned, power modules (such as those with a relatively large footprint) may be overdesigned to achieve the high reliability required for efficient electric vehicles. Overdesign can also include dederating power semiconductors, for example, reducing their lifetime consumption to 30%. These overdesign choices result in higher power module costs and consume more space within the traction converter. Even with overdesign, power modules can still fail during unexpected extreme driving behaviors.

[0007] An over-designed power module may not include mission profiles for unintended driving behaviors. Because traction converters cannot be designed to cover every driving scenario, the power semiconductors in the traction inverter may fail when the electric vehicle is performing unintended driving behaviors. Increasing the size of the traction converter to include extreme driving behaviors may further increase the cost and size of the traction converter, and still not cover every driving scenario.

[0008] Since the traction converter of an electric vehicle may fail during unexpected driving behavior, it is desirable to adaptively control the power management system of the electric vehicle based on system parameters.

[0009] The background section concerning adaptive control of power modules is intended only to provide a contextual overview of some current problems and is not intended to be exhaustive. Further contextual information will become apparent to those skilled in the art after reading the following detailed description. Summary of the Invention

[0010] According to an embodiment, a method for controlling a traction inverter for a motor includes receiving a torque estimate from a torque meter connected to the traction inverter; receiving parameter data of a power semiconductor from a traction inverter semiconductor monitor; determining a health prediction of the power semiconductor based on the parameter data; determining an operating mode of the traction inverter based on the torque estimate, the parameter data of the power semiconductor, and the health prediction of the power semiconductor; and generating control commands based on the operating mode, wherein the control commands include a first gating control of gating the traction inverter, a second gating control of gating multiple gatings of the power semiconductor, or current control of current flowing to the power semiconductor. Controlling the traction inverter based on received power management system parameters allows the motor to adapt to the current driving behavior experienced by the electric vehicle.

[0011] A power management system can monitor key parameters of the power semiconductors in a traction inverter to achieve the desired reliability. For example, the power management system can employ artificial intelligence (AI) algorithms to design robust and efficient power semiconductors for the traction inverter. Such a design avoids over-designing or oversized power semiconductors, requiring less cost and footprint compared to existing designs. Adaptive power management systems can prevent system failures or damage due to unpredictable driving behavior. AI algorithms can predict the potential failure date of the power semiconductors based on power management system parameters. The system can then alert electric vehicle users to maintenance needs based on the predicted failure date. AI algorithms can eliminate the need for over-designing power semiconductors.

[0012] An adaptive power management system can monitor key parameters of the power semiconductors in the traction inverter. AI algorithms can receive these parameters to predict the health of the power semiconductors and the driving behavior of the electric vehicle user. The power management system can then adjust the traction converter control based on these AI algorithm predictions, such as predicted health and driving behavior.

[0013] Adaptive power management systems eliminate the need for over-designed traction converters, leading to reduced cost and size. Furthermore, compared to traction converters that do not dynamically control the traction converter based on power semiconductor parameters, traction converters offer enhanced adaptability to unintended driving behaviors. Therefore, traction inverters do not require over-design or excessive size, and are more reliable and less expensive.

[0014] The issue arises when electric vehicles experience stress on their power semiconductors during extreme driving conditions. Traction converters may not be designed to withstand such extreme driving behavior, which could damage their power semiconductors. Over time, the traction converter's ability to handle extreme driving conditions may decrease. This problem can occur when electric vehicles enter the market and encounter extreme driving behaviors not considered during the design phase. Until now, oversized or over-designed power semiconductors have been the only way to address this issue.

[0015] According to an embodiment, the parameter data includes real-time data. This real-time data can change with variations in the current of the power semiconductor. This real-time data allows for adaptive control of the traction inverter based on rapidly changing parameters of the power semiconductor.

[0016] According to an embodiment, the parameter data includes long-term data. This long-term data may vary over the lifespan of the power semiconductors. This long-term data allows for adaptive control of the traction inverter based on parameters associated with the lifespan of the power semiconductors.

[0017] According to an embodiment, the real-time data includes the on-state voltage of the power semiconductors. Controlling the traction inverter based on the on-state voltage can reduce switching losses and prevent damage to the power semiconductors due to voltage spikes.

[0018] According to an embodiment, the real-time data includes the junction temperature of the power semiconductor. Controlling the traction inverter based on the junction temperature of the power semiconductor can prevent damage to the power semiconductor due to elevated temperatures.

[0019] According to an embodiment, long-term data includes the threshold voltage of the power semiconductor. Threshold voltage-based control of the traction inverter allows the turn-on voltage to be compared with the threshold voltage to ensure that the turn-on voltage remains within the desired voltage range.

[0020] According to an embodiment, long-term data includes the on-resistance of the power semiconductor. Controlling the traction inverter based on on-resistance allows for adjustment of variations in the on-resistance of the power semiconductor throughout its lifetime.

[0021] According to an embodiment, long-term data includes the thermal resistance of the power semiconductor. Thermal resistance-based control of the traction inverter allows for adjustment of variations in the thermal resistance of the power semiconductor throughout its lifespan.

[0022] According to an embodiment, determining the operating mode includes determining whether the parameter data is real-time data or long-term data. Determining whether the data is real-time or long-term data improves the efficiency of processing the received data.

[0023] According to one embodiment, the method also includes determining the rate of change of the real-time data. Determining the rate of change allows for rapid adjustments to the control of the traction inverter in response to rapidly changing driving behavior.

[0024] According to an embodiment, the method also includes determining the rate of change of long-term data. Determining the rate of change can allow for prediction of the end of the power semiconductor's lifetime.

[0025] According to an embodiment, the operating mode is also based on the rate of change of real-time data.

[0026] According to an embodiment, the operating mode is also based on the rate of change of long-term data.

[0027] Both real-time and long-term data are helpful for selecting operating modes. Real-time measurement of power semiconductor parameters provides information about rapidly changing parameters. Therefore, real-time data is easier to analyze driver behavior than long-term data because it can be analyzed to detect changes in power semiconductor parameters over short time periods. For example, a rapid increase in junction temperature during aggressive driving.

[0028] Measuring power semiconductor parameters over a long period requires fewer resources compared to real-time measurements. For example, an algorithm configured to determine operating modes requires less complexity to do so compared to analyzing real-time data, because long-term data has a lower rate of change compared to real-time data.

[0029] Determining the operating mode based on the rate of change of real-time and long-term data allows for adaptive control of the traction inverter based on driving behavior and the age of the traction inverter. Attached Figure Description

[0030] The present disclosure will now be described in more detail with reference to the accompanying drawings, which illustrate one embodiment of the present disclosure:

[0031] Figure 1 It is a perspective view of a power module that includes multiple power semiconductors;

[0032] Figure 2 The control architecture of the traction inverter is shown;

[0033] Figure 3 An exemplary graph showing real-time parameter measurements of a power semiconductor over time is provided;

[0034] Figure 4 An exemplary graph showing long-term parameter measurements of power semiconductors over time is shown;

[0035] Figure 5 An exemplary flowchart of an exemplary method for controlling a power management system of an electric vehicle is shown;

[0036] Figure 6 An exemplary flowchart of a method for controlling a power module of a traction converter is shown. Detailed Implementation

[0037] This disclosure relates to a method and apparatus for controlling a traction inverter of a traction converter. The method includes monitoring the power semiconductors of the traction inverter and generating control commands based on data received in response to the monitoring of the power semiconductors. The method can be configured to determine whether the received data is real-time data or long-term data. Alternatively, the method can be configured to classify the received data differently. Each classification can further include any number of subclasses. Classification can allow for finer control of the traction inverter. The method can be used to adaptively control the traction inverter based on parameters of changes in the power semiconductors of the traction inverter.

[0038] Figure 1 A power module 100 (such as a power module for a traction inverter) is shown. Figure 1This can be any of the different conventional power modules. The proposed method described below is applicable to use with any power module without requiring modification to the power module. Power module 100 may include a housing 102 and include a plurality of power semiconductors 104. Power semiconductors 104 may be located on a copper plate 106. Copper plate 106 may be further located on a ceramic plate 108. Copper plate 106 and ceramic plate 108 may form a copper-clad substrate of the power module. Power semiconductors 104 may be electrically coupled to copper plate 106 via bonding wire 110.

[0039] although Figure 1 The focus is on the two power semiconductors 104 and their relationship to the copper plate 106, ceramic 108, and bonding wire 110 within the power module 100. However, it will be clear to those skilled in the art that other power semiconductors, copper plates, ceramics, and bonding wires are illustrated. Furthermore, those skilled in the art will understand that the power module 100 may have other configurations of the power semiconductors 104, copper plate 106, ceramic 108, and bonding wire 110.

[0040] The traction inverter may include a power module (such as power module 100). Monitoring the power module, particularly the power semiconductor 104, during operation allows the electric vehicle's power management system to adapt to changes in the power module. For example, if the voltage or current changes, the power management system can control the power module to adapt to the changing parameters. Additionally, as the power module degrades during its lifespan, the power management system can predict future failure dates based on the rate of degradation of the measured parameters of the power module.

[0041] Figure 2 An example of a power management system 200 for an electric vehicle is shown. The power management system 200 may include a control architecture 202 electronically coupled to a power semiconductor monitor 250. The power semiconductor monitor 250 may also be electronically coupled to a power module 101, similar to... Figure 1 The power module 100. Specifically, a power semiconductor monitor 250 can be connected to the power semiconductor of the power module 101. The power semiconductor monitor 250 can be configured to measure one or more parameters of the power semiconductor.

[0042] An exemplary power module 101 may include a three-phase half-bridge topology. As shown, power module 101 may include six power semiconductors. Each power semiconductor of power module 101 may include a switch such as switches S1-S6. Furthermore, each power semiconductor may be configured to receive a gated control mode (GCM) signal and gated switch (GS) signals GS1-GS6. Each power semiconductor may be connected to a single power semiconductor monitor 250, or a single power semiconductor monitor 250 may be connected to all six power semiconductors. Power module 101 may be connected to a power source 220 (such as a battery). Power source 220 may provide direct current (DC) to power module 101. Power module 101 may convert the DC provided by power source 220 to alternating current (AC) and provide the AC to motor 240. The power semiconductors may include switch S6, which includes four power semiconductor dies S1 and S2 operating in parallel. 61 -S 64 Each of the power semiconductors in the power module can share the same gated driver design, as illustrated in detailed view 230. The gated drivers for the power semiconductors can receive GCM commands and signals generated by the pulse modulator. The four switches S in detailed view 230 can be controlled based on the received signals. 61 -S 64 Each of the power semiconductors in power module 101 can receive a separate signal and be controlled independently.

[0043] Each power semiconductor in power module 101 may also include a switch, such as four power semiconductor dies S arranged in parallel. 61 -S 64 As shown in detailed view 230. Four power semiconductor dies S 61 -S 64 Parallel connections are possible. GS 61 –GS 64 Control gate signal. When gate GS 61 –GS 64 When closed, this means all switches S 61 -S 64 Both are connected and disconnected. If the gate GS... 61 If the switch S is open while the other gates are closed, this means that switch S... 61 It will not connect or disconnect. Switch S only. 62 –S 64 Connect and disconnect in parallel.

[0044] Those skilled in the art will understand that other power module topologies can have more power semiconductors or switches. For example, there are three voltage level topologies including neutral point clamping inverters such as Advanced Neutral Point Clamping (ANCP) inverters, Improved Neutral Point Clamping (INCP) inverters, and T-type Neutral Point Clamping (TNCP) inverters. Other power module topologies utilizing power semiconductors can be monitored to control traction inverters.

[0045] The power semiconductor monitor 250 is configured to measure one or more power semiconductors (such as...) Figure 1 The power semiconductor 104) can be monitored for one or more parameters. The control architecture can define its own power semiconductor parameter categories for measurement and tracking over time. A custom power semiconductor monitor 250 design can include hardware and software designed to monitor key parameters of the power semiconductor. For example, the power semiconductor monitor 250 can be configured to monitor two main categories of power semiconductor parameters. The first category can be labeled as real-time parameters of the power semiconductor and includes the on-state voltage (VDS). ON ) and junction temperature (T) J The second category can be categorized as long-term parameters of power semiconductors and includes the threshold voltage (V). TH ), On-resistance (RDS) ON ) and thermal resistance (R TH ).

[0046] Each category can be defined by its relationship to the current flowing through the power semiconductor. For example, real-time parameters can change with the real-time variation of the current flowing through the power semiconductor. Conversely, long-term parameters can remain constant with changes in the current flowing through the power semiconductor. Alternatively, long-term parameters may change over the lifetime of the power semiconductor.

[0047] Control architecture 202 may include several components configured to analyze and process data measured by power semiconductor monitor 250. Control architecture 202 may include a data analysis module 204 configured to receive measurement data from power semiconductor monitor 250. Data analysis module 204 may include a processor and memory to process the data transmitted from power semiconductor monitor 250. The memory of data analysis module 204 may store artificial intelligence (AI) algorithms configured to take measured power semiconductor parameters as input. The processor of data analysis module 204 may execute the AI ​​algorithms to process the measured power semiconductor parameters.

[0048] The power semiconductor monitor 250 sends its measured power semiconductor data to the data analysis module 204, where a processor processes the measured power semiconductor data to determine and / or predict trends in the power semiconductor. Different AI or machine learning (ML) algorithms can be used to analyze the power semiconductor data. For example, stochastic gradient descent (SGD); perceptron algorithm; online passive-active (PA) algorithm; incremental k-means; incremental principal component analysis (PCA); cumulative sum control chart (CUSUM); exponentially weighted moving average (EWMA); recurrent neural network (RNN); streaming convolutional neural network (CNN); autoregressive integral moving average (ARIMA); and exponential smoothing (ETS) algorithms can process the measured power semiconductor parameters.

[0049] Any other AI or ML algorithm capable of determining trends in power semiconductors based on measurement data can be implemented in the data analysis module 204. The requirement for the AI ​​or ML algorithm is that it must be able to analyze the data in real time. In other words, the algorithm must be optimized for low latency to ensure real-time results, thereby ensuring the desired performance of the power management system.

[0050] The algorithm should also be scalable, adaptable, and efficient. In addition to the ability to process data in real time, it should also be able to manage large amounts of data in real time. The algorithm should be able to adapt to changes in data distribution without requiring further training. The algorithm should be able to handle large amounts of data while minimizing resource requirements.

[0051] Any of the aforementioned machine learning algorithms can be trained to monitor the power semiconductors of electric vehicles. For example, deep reinforcement learning algorithms or adaptive filtering algorithms can be used to monitor test electric vehicles. Electric test vehicles can be driven under customized test conditions. Customized test conditions can include extreme driving conditions. In this way, the traction inverter can be controlled for extreme driving conditions without over-designing the traction inverter's power module. AI algorithms will be trained to handle a wide variety of driving behaviors in real-world scenarios, including extreme driving behaviors.

[0052] The data analysis module 204 includes algorithms (such as those described above) that use power semiconductor parameter data from the power semiconductor monitor 250 to generate predictions and determinations of driving behavior. The output of the data analysis module 204 may include trends or predictions based on the power semiconductor data. For example, the data analysis module 204 may generate two main sets of power semiconductor data (such as real-time data and long-term data), resulting in different data types.

[0053] Both real-time and long-term data types can be subcategories. For example, real-time data types can include three subcategories, and long-term data types can include two subcategories, which will be referred to separately. Figure 3 and Figure 4 Further description. Those skilled in the art will understand that, Figure 3 and Figure 4 The data types shown are exemplary, and other data types can be configured.

[0054] Data analysis module 204 can transmit data types to driver behavior analysis module 206. Driver behavior module 206 may include a processor and memory. The memory of driver analysis module 206 can store instructions for analyzing the output of data analysis module 204 to generate control instructions for an electric vehicle power management system based on AI or ML algorithm predictions. The processor of driver analysis module 206 can execute the instructions stored in memory to generate control instructions. Those skilled in the art will understand that data analysis module 204 and driver behavior analysis module 206 can be combined, separated, or share the processor and memory.

[0055] The driver behavior analysis module 206 can generate instructions for controlling the power management system of an electric vehicle. For example, the driver behavior analysis module 206 can generate control instructions based on a control mode selection algorithm. (About...) Figure 5 The method for selecting the control mode based on the output of the data analysis module 204 is further described. Control commands may include commands for gating the power semiconductors of the power module 101 and for controlling the amount of torque generated by the torque control 212.

[0056] Each of the power semiconductor switches S1-S6 shares the same gating drive design, such as Figure 2 As shown. Therefore, assuming that for each power semiconductor in the power module, there are four power semiconductor dies S 61-64 Parallel operation. GS 61-64 Control for four power semiconductor dies S 61-64 The gating signal. When GS 61-64 When closed, this means all S 61-64 Both are connected and disconnected. If GS 61 Open and the rest of the GS 62-64 If closed, then this means S 61 It will not connect or disconnect. Only S 62-64 Connect and disconnect in parallel.

[0057] The driver behavior analysis module 206 can generate Gated Control Mode (GCM) instructions to the power module 101. GCM instructions can control power semiconductor gating for the power semiconductors used in the power module 101. For example, the instructions can switch the gating drive scheme between a normal switching mode when GCM is disabled and an adaptive switching mode when GCM is enabled, based on measured power semiconductor voltages. The adaptive switching mode helps control switching transients, including dv / dt (the derivative of voltage with respect to time), di / dt (the derivative of current with respect to time), peak switching voltage, and peak switching current, in order to control switching losses in the power semiconductors. Higher dv / dt and di / dt mean faster switching speeds, which means lower switching losses if peak switching voltage and peak switching current are lower. Lower switching losses contribute to better reliability performance of the power semiconductors.

[0058] The driver behavior analysis module 206 can generate a torque control mode (TCM) command to the torque control 212. The TCM command controls the torque required by the electric vehicle. If the required torque is too high or the power semiconductors are in poor health, the TCM command can reduce the torque output when TCM is enabled. For example, the TCM command can limit the torque output based on determined aggressive driving behavior, or reduce the torque output of aging power semiconductors. This ensures the safety of driving the electric vehicle. The TCM command can be applied to the current control system 216 to limit the current.

[0059] For example, torque control 212 can receive an input signal based on motor torque 207 and torque reference 210. Torque meter 208 can directly measure motor torque 207 and generate a signal including the measured machine torque. Alternatively, torque calculator can estimate machine torque based on power, speed, and / or current. The measured signals including torque reference 210 and motor torque 207 can be multiplied and used as further input to torque control 212. Torque control 212 can generate a torque control signal based on TCM instructions, motor torque measurement, and torque reference value. The torque control signal can be multiplied with current signal 214, where current signal 214 represents the current control signal provided by DC power module 101 to motor 240. Current control 216 takes the current control signal as input and generates an output to modulator 218. Modulator 218 generates a signal representing the current control signal and transmits it to the power semiconductor of power module 101.

[0060] The power semiconductor of power module 101 can adaptively control switches S1-S6 and the gating of the power semiconductor based on GCM commands and current control signals to regulate the current generated for motor 240.

[0061] Other signals can be generated by the power management system 200. For example, the power management system 200 can also control the distribution of cooling fluid to maintain the junction temperature of the power semiconductor.

[0062] The data analysis module 204 can be customized to classify the measured power semiconductor parameters into any number of categories and subcategories. For example, Figure 3 The first category (real-time data) and its three subcategories are illustrated. Figure 3 The charts in the diagram can represent any measured power semiconductor parameter.

[0063] These three subcategories can be divided into three data types. The first real-time data type 302 (data type 1) is labeled as mild normal. The first real-time data type 302 can represent mild driving in an electric vehicle with a healthy electrical module. Mild driving can be driving behavior close to the center of the range that the motor can manage. Figure 3 As shown, the first data type 302 includes power semiconductor data that varies between Dmax and Dmin with a slowly changing slope. Compared to other data types, Dmax-Dmin is within a relatively small range, and Dmax is within the maximum permissible limit.

[0064] The second real-time data type 304 (data type 2) is marked as aggressive normal. The second real-time data type 304 can represent aggressive driving in an electric vehicle with a healthy electrical module. Aggressive driving can be driving behavior that approaches the limits of what the motor can manage. For example... Figure 3 As shown, the second data type 304 includes power semiconductor data that varies between Dmax and Dmin with a rapidly changing slope. Compared to other data types, Dmax-Dmin has a relatively large range, and Dmax can be within the maximum permissible limit.

[0065] The third real-time data type 306 (data type 3) is marked as aged. The third real-time data type 306 can represent driving behavior in an aged electric vehicle. For example... Figure 3 As shown, the third data type 306 includes power semiconductor data that varies between Dmax and Dmin. However, compared to other data types, Dmax and Dmin increase at a relatively rapid rate.

[0066] Figure 4 The second category (long-term data) and its two subcategories are shown. Figure 4 The charts in the diagram can represent any measured power semiconductor parameter.

[0067] The first long-term data type 402 (data type 1) is marked as normal. The first long-term data type 402 can represent a healthy power module.

[0068] The second long-term data type 404 (data type 2) is marked as aging. The second long-term data type 404 can represent an aging power module that may be prone to failure.

[0069] Figure 5 A driver behavior analysis method 500 that can be executed by the driver behavior analysis module 206 is shown. Method 500 may have a first step 502, which categorizes the measured power semiconductor data. In this example, the data is determined to be either real-time data or long-term data. If step 502 determines that the measured power semiconductor data is real-time data, method 500 moves to step 510. At step 510, method 500 determines whether the driving behavior is mild and normal, for example, data type 302. If step 510 determines that the driving behavior is mild and normal, method 500 disables the gating control mode at step 512, disables the torque control mode at step 514, and ensures at least one gating of each power semiconductor in the power semiconductor of power module 101 remains open at step 516, and terminates the method. Step 516 may ensure that the gating GS of detailed view 230 remains open. 61 –GS 64 The state remains the same. In other words, it maintains the gating state (open or closed), and the power module operates without change. By choosing to maintain the gating state, this method avoids unnecessary gating control, such as closing an open gating or opening a closed gating. In other words, because method 500 determines that the driving is not extreme, it does not need to limit the current or adaptively switch the power semiconductor gating. If step 510 determines that the driving behavior is not mild and normal, proceed to step 520.

[0070] At step 520, method 500 determines whether the driving behavior is aggressively normal, such as data type 304. If step 520 determines that the driving behavior is aggressively normal, then method 500 determines at step 522 whether all gates of each power semiconductor are closed. If all gates are closed, method 500 moves to step 524 to determine whether a gating control mode is enabled. If step 522 determines that not all gates are closed, method 500 moves to step 526 and ensures that at least one gate of each of the power semiconductors in power module 101 remains open and ends the method. For example, step 526 does not change anything. If step 524 determines that a gating control mode is not enabled, method 500 moves to step 528 to enable the gating control mode and then moves to step 530. If step 524 determines that a gating control mode is enabled, method 500 moves to step 530. Step 530 determines whether the fastest switching speed has been reached. If step 530 determines that the fastest switching speed has not been reached, then method 500 ends. If step 530 determines that the fastest switching speed has been reached, method 500 proceeds to step 532 and enables torque control mode, then terminates method 500. In other words, when method 500 identifies aggressive driving conditions, it attempts to eliminate stress and damage to the power module by adaptively opening and closing gating as quickly as possible based on power semiconductor parameters, and / or by controlling the amount of torque supplied to the motor to limit the amount of torque on the motor. If step 520 determines that the driving behavior is not aggressive and normal, it proceeds to step 540.

[0071] At step 540, method 500 determines whether the driving behavior is aging, such as data type 306. If step 540 determines that the driving behavior is aging, method 500 proceeds to step 542. Step 542 determines whether all gating is closed. If step 542 determines that all gating is closed, method 500 proceeds to step 546. If step 542 determines that not all gating is closed, method 500 proceeds to step 544. Step 544 ensures that at least one gating of each of the power semiconductors in power module 101 remains open and proceeds to step 546. For example, step 544 does not change anything. Step 546 enables torque control mode and proceeds to step 548. Step 548 generates an alarm message indicating that the power semiconductor has failed or is close to failing, and method 500 ends. In other words, method 500 has determined that the power module is at or near a failure and limits the torque on the motor by limiting the current generated by the power module. If step 540 determines that the driving behavior is not aged, proceed to step 570 and take no action and end method 500.

[0072] If step 502 determines that the measured power semiconductor data is long-term data, method 500 proceeds to step 550. At step 550, method 500 determines whether the driving behavior is normal, for example, data type 402. If step 550 determines that the driving behavior is normal, method 500 proceeds to step 552 and disables all gating of the power semiconductor and proceeds to step 554. Step 554 disables the gating control mode and proceeds to step 556. Step 556 disables the torque control mode and ends method 500. In other words, because method 500 determines that the power module is responding normally, it does not need to limit the current or adaptively switch the power semiconductor gating. If step 550 determines that the driving behavior is abnormal, it proceeds to step 560.

[0073] At step 560, method 500 determines whether the driving behavior is aging, such as data type 404. If step 560 determines that the driving behavior is for an aging power module, method 500 moves to step 562 to determine whether all gating of the power semiconductors is closed. If step 562 determines that all gating of the power semiconductors is closed, method 500 moves to step 566. If step 562 determines that all gating of the power semiconductors is not closed, method 500 moves to step 564 and ensures that at least one gating of each of the power semiconductors in power module 101 remains open and ends the method and moves to step 566. For example, step 564 does not change anything. Step 566 enables torque control mode and moves to step 568. Step 568 generates an alarm message indicating that the power semiconductors have failed or are close to failing, and ends method 500. In other words, method 500 has determined that the power module is at or near a failure and attempts to limit the torque on the motor by limiting the current generated by the power module or by adaptively controlling the gating of the power semiconductors. If step 560 determines that the driving behavior is not aged, proceed to step 570 without taking any action and end method 500.

[0074] Those skilled in the art will understand that Figure 5 The steps shown can be combined, or method 500 can include other steps.

[0075] Figure 6 An exemplary method 600 for controlling a traction inverter is illustrated. For example... Figure 6As shown, method 600 includes receiving a torque estimate from a torque meter connected to the traction inverter at step 602; receiving parameter data of the power semiconductors of a traction inverter semiconductor monitor at step 604; determining a health prediction of the power semiconductors based on the parameter data at step 606; determining an operating mode of the traction inverter based on the torque estimate, the parameter data of the power semiconductors, and the health prediction of the power semiconductors at step 608; and generating control commands based on the operating mode at step 610, wherein the control commands include a first gating control of the traction inverter, a second gating control of a plurality of gatings of the power semiconductors, or current control of the current flowing to the power semiconductors.

[0076] While this disclosure has been described with reference to exemplary embodiments, those skilled in the art will understand that various changes can be made and elements can be substituted with equivalents without departing from the scope of this disclosure. Furthermore, many modifications can be made to adapt particular situations or materials to the teachings of this disclosure without departing from its essential scope. Therefore, it is intended that this disclosure be limited to the specific embodiments disclosed, but rather that it encompass all embodiments falling within the scope of the appended claims.

Claims

1. A method for controlling a traction inverter for a motor, comprising: Receive torque estimates from a torque meter connected to the traction inverter; Receive parameter data of the power semiconductor from the traction inverter semiconductor monitor; The health prediction of the power semiconductor is determined based on the parameter data; The operating mode of the traction inverter is determined based on the torque estimation, the parameter data of the power semiconductor, and the health prediction of the power semiconductor. as well as Control commands are generated based on the operating mode, wherein the control commands include a first gating control of the traction inverter, a second gating control of multiple gatings of the power semiconductor, or current control of the current flowing to the power semiconductor.

2. The method according to claim 1, wherein, The parameter data includes real-time data.

3. The method according to claim 1 or 2, wherein, The parameter data includes long-term data.

4. The method according to claim 2, wherein, The real-time data includes the on-state voltage of the power semiconductor.

5. The method according to claim 2 or 4, wherein, The real-time data includes the junction temperature of the power semiconductor.

6. The method according to claim 3, wherein, The long-term data includes the threshold voltage of the power semiconductor.

7. The method according to claim 3 or 6, wherein, The long-term data includes the on-resistance of the power semiconductor.

8. The method according to claim 3, 6 or 7, wherein, The long-term data includes the thermal resistance of the power semiconductor.

9. The method according to any one of the preceding claims, wherein, Determining the operating mode includes determining whether the parameter data is real-time data or long-term data.

10. The method of claim 9, further comprising determining the rate of change of the real-time data.

11. The method of claim 9 or 10, further comprising determining the rate of change of the long-term data.

12. The method according to claim 10, wherein, The operating mode is also based on the rate of change of the real-time data.

13. The method according to claim 11 or 12, wherein, The operating mode is also based on the rate of change of the long-term data.

Images