Method for determining the state of health of the winding insulation on the stator of an electric machine

Abstract

The invention relates to a method for determining the state of health of the winding insulation on the stator of an electric machine based solely on one parameter of the stator of the electric machine, while estimating and weighting three main components of the thermal load on the winding insulation, namely temperature change (20), temperature change rate (13), and static aging (15).

Description

[0001] Method for determining the health of the winding insulation on the stator of an electrical machine

[0002] The invention relates to a method for determining the state of health of the winding insulation on the stator of an electrical machine based solely on a parameter of the stator of the electrical machine.

[0003] State of the art

[0004] The central approach of the forecast and health management system is to use data on usage and condition from the operation of a single system to detect malfunctions and failures at an early stage and define appropriate countermeasures. This allows decisions to be made while taking overall economic interests into account. For example, this allows costs to be reduced, risks to be minimized, usable operating hours to be maximized, and availability to be improved.

[0005] Electric vehicles use a purely electric drive system, known as an eDrive. The drive system of an electric vehicle includes an electric motor, a mechanical reduction gear, an inverter, and a motor controller.

[0006] The inverter inverts DC power from batteries into AC power, which is used to drive the motor. The motor delivers torque to the vehicle's drivetrain. The controller controls the power supplied to the motor by the inverter.

[0007] Compared to combustion engines, electric motors generally require less but more careful maintenance over their lifetime.

[0008] Therefore, it is critical to have a clear understanding of the historical, current, and predicted health of the electric motor. US 2021 / 0 175 835 A1 shows a method for determining the health of a drive. A controller estimates a motor torque using a model of the motor. The model estimates the motor torque based on the DC power and the motor speed. The controller is further configured to calculate the motor torque based on the sensed current and a magnetic flux of the motor. The controller is further configured to generate a first health state for the electric drive system based on a difference between the estimated and calculated motor torques.The controller is further configured to generate a second health state for at least one of the position sensor, the plurality of sensors, and the motor in response to the first health state being less than or equal to a first threshold. The controller is further configured to determine whether a fault exists based on a combination of the first health state and the second health state.

[0009] For example, the health indicators include the system-level health indicator, which indicates the performance of the electric drive system; the component-level health indicators, which indicate the performance of the current sensors; and the health indicator, which indicates the health of the motor windings.

[0010] However, some sensors are necessary for the process.

[0011] One of the components of an electric motor that wears out over time is the winding insulation. Thermal stress is the main cause of winding insulation deterioration.

[0012] CN 1 09 633 435 B describes a method for determining the thermal aging of winding insulation. However, this method only considers statistical aging based on the measured temperature of the windings. DE 34 43 276 A1 describes monitoring the temperature of the motor winding of an electrical machine, generating and integrating a failure rate, and predicting the remaining service life.

[0013] WO 2023 / 122 259 A1 shows a method in which the thermal load caused by temperature changes with effect on a drive train is shown.

[0014] Thermal loads, including static and dynamic loads acting on the winding insulation of the electrical machine, have not yet been comprehensively taken into account.

[0015] The object of the invention is to enable vehicle-side health monitoring of the electric machine of an electric drive using a software-based solution and simple sensors.

[0016] Description of the invention

[0017] The problem is solved by a method for determining the health status of the winding insulation on the stator of an electrical machine based solely on a parameter of the stator of the electrical machine, estimating and weighting three main components of the thermal stress on the winding insulation, namely temperature change, temperature change rate and static aging.

[0018] The task is solved using a data-driven method for monitoring the condition of the winding insulation by quantifying the effects of thermal stress on the insulation.

[0019] The parameter is at least one temperature measurement or at least one current measurement at the stator. The proposed method focuses exclusively on considering thermal stress by using temperature measurements or current measurements at the stator, while neglecting stress factors such as mechanical, electrical, and environmental influences. This neglect is due, on the one hand, to the lack of sensors for monitoring specific stress factors and, on the other, to the fact that thermal stress can be considered a manifestation of the aforementioned stress factors.

[0020] The purely vehicle-side monitoring of the aforementioned health indicator via many sensors is computationally intensive, which can be reduced by limiting it to simple temperature or current measurements.

[0021] The main components are combined into a comprehensive indicator.

[0022] Individual indicators are determined from the main components, temperature change and temperature change rate, on the basis of a global model.

[0023] The main component, static aging, is estimated using a life expectancy model and results in an indicator.

[0024] Indicators are weighted differently over the lifetime of the winding insulation and thus of the electrical machine.

[0025] The procedure is a component in a higher-level procedure for determining the health status of an electrical machine.

[0026] The method includes an output step in which the health monitor transmits information to a driver and / or a vehicle-external monitoring system. Description of the figures

[0027] Figure 1 shows a diagram showing the health status of the electrical machine over time t,

[0028] Figure 2 shows a thematic sequence of the method according to the invention, Figure 3 shows a diagram of the insulation resistance over a number of equivalent thermal cycles,

[0029] Figure 4 shows a diagram with the cumulative sum of the temperature change rate over the thermal cycles,

[0030] Figure 5 shows a diagram describing the predicted number of hours until the end of insulation life over a constant temperature of different insulation quality classes (A, B, F, H),

[0031] Figure 6 shows a graph of the estimated health status over time for the entire lifetime (the first data point is specific to each machine, as this defines the initial condition for the lifetime estimation and is determined from tests),

[0032] Figure 7 shows a diagram with the choice of weights for the algorithm over time,

[0033] Figure 8 shows schematically the dynamic weighting procedure.

[0034] The method according to the invention provides a health status indicator that estimates the resistance of the winding insulation.

[0035] The monitored thermal load is categorized into dynamic and static components. The information from these two components is combined using variable weights that reflect different operating modes based on the evolution of the stator temperature. Figure 1 schematically shows the health status of the electric machine, starting from its commissioning at time 0 until the end of its lifetime (EoL). The decreasing health status trend is visible in the middle line of the diagram. Two further trends are shown in the diagram. The EoLc trend follows a very conservative estimate, so the lifetime ends earlier. In an optimistic EoLo estimate, the end of life is delayed.

[0036] The diagram illustrates very schematically the fact that each engine is exposed to a unique combination of stress factors when operating in the field. Therefore, it exhibits a different health indicator than the planned one. In reality, under real operating conditions, the engine exhibits a trajectory that should lie between the optimistic and conservative scenarios. The end of life of each component is accepted as different from the planned one.

[0037] The process of the proposed method is illustrated in the block diagram of Figure 2. The procedure runs in two phases. The first phase involves testing the electric motor 4, and the second phase involves on-board monitoring 5. In the first phase of testing the electric motor 4, which is part of the powertrain development process in the automotive industry, samples of electric motors from series production are selected and tested to evaluate their durability and robustness.

[0038] In the test procedure known as “Power Thermal Cycle Endurance” PTCE test, the motor stator is subjected to thermal cycles 6 by applying current to the windings to heat them up and then immersing the stator in oil without current flow to cool them down. Typically, measurements in the temperature range of -20 to 150 degrees Celsius are sufficient, and more than 1,000 cycles are carried out on the test stator assembly. During these tests, intermediate measurements 7 are recorded regularly. This includes measurement data such as insulation resistance 10 and partial discharge input voltage 11. The intermediate measurements 7 are recorded regularly along with the number of thermal cycles 6. The intermediate measurements 7 are then compared with the respective threshold values ​​of an ideal machine. A model 8 is created for each test stator assembly tested.Through model aggregation, a global model 9 is created, which enables a more precise mapping between inputs consisting of temperature measurements and outputs consisting of representations of the aging process. The global model 8' is stored and used in the on-board monitoring 5.

[0039] In the on-board monitoring system 5, temperature measurements at various locations on the stator 30 are used as inputs for the process. The monitoring system is divided into three components: a thermal cycle counter 20, a temperature change rate component 13, and a static aging component 15.

[0040] The vehicle-side monitoring system 5 uses a thermal cycle counter 20 to calculate the temperature of the stator 30, which comes from sensor values ​​or thermal models, and uses this to determine the number of thermal cycles 6.

[0041] The thermal cycle counter 20 uses a rainfall counting algorithm to capture the actual thermal cycles.

[0042] The algorithm for counting rain flow is well known and is used in the analysis of fatigue data. More or less justified decompositions allow for the determination of the equivalent cyclic loads for each load.

[0043] Then, a weighted average of past thermal cycles is calculated and converted into equivalent thermal cycles 21 using a Coffin-Manson approach.

[0044] In other words, this approach performs the conversion of slow thermal cycles into the corresponding accelerated thermal cycles during the test phase. The converted equivalent thermal cycles 21 are then fed into the global model 8' to estimate the insulation resistance 10 based on the thermal cycles 6, Rins = f (n C yc). An indicator W1 is delivered to a health status monitor 40.

[0045] Figure 3 shows the estimated insulation resistance 10 in MOhm versus the number of equivalent thermal cycles 21 . The insulation resistance 10 decreases until it reaches a threshold value slightly greater than zero, which corresponds to the "EoL," the end of the service life.

[0046] The temperature change rate 13 continuously calculates the absolute value of the first derivative of the temperature T with respect to time t and accumulates it over time t in the accumulated thermal load process step 14.

[0047] Figure 4 shows the course of the accumulated thermal load 14 over the thermal cycles of a specific temperature change profile

[0048] An indicator W3 is then fed into the global model 9 to estimate the insulation resistance 10 based on the accumulated temperature change rate 13.

[0049] The data establishes the relationship between the accumulated thermal stress 14 and the insulation resistance 10, Rins = f( E IcTT / cftl ). An indicator W3 is fed to the health status monitor 40. Together, they ensure that the effects of both components of dynamic aging, namely thermal cycling 6 and temperature change rate 13, are taken into account.

[0050] Static aging 15 takes into account the effects of a constant temperature T on the deterioration of the winding insulation.

[0051] To determine the life expectancy, a representative life expectancy model based on the insulation class specified in IEC 60085:2007 is used. The solution derives an indicator W2 for the on-board health monitor 40, taking into account the insulation class (e.g., Class H).

[0052] Figure 5 shows different straight lines relating to different classes A, B, F, H and representing the hours of functionality over temperature.

[0053] This indicator W2 takes into account the ageing of the winding insulation at standstill, where dynamic ageing has no influence on the insulation resistance 10.

[0054] All of the above-mentioned important health indicators W1, W2, W3 are estimated simultaneously. To obtain a comprehensive indicator W gesTo obtain a reliable indicator for the health of the winding insulation, these three indicators are merged into a single indicator. However, since the specific weightings of the individual factors are not clear, the inventive solution uses a fusion approach with variable weightings, which is briefly explained in Figure 6. The figure shows the health of the winding insulation over time in hours. There is a comprehensive indicator Wges, which is formed from the additional indicators such as W1, W2, and W3. The static aging 14 generates the indicator W2, which progresses linearly over time.

[0055] The indicator W1, derived from the equivalent thermal cycles and the comparison with the global model, degrades over time with a pronounced step around 500 hours. The indicator W3, derived from the temperature change rate 14 compared with the global model 8', follows a similar pattern. The comprehensive indicator Wtotal lies between the two indicators W1 and W3.

[0056] Figure 8 shows the dynamic weighting approach. This involves determining the Pearson correlation coefficient, a method that measures the degree of linear relationship between two at least interval-scaled characteristics. This method does not depend on the units of measurement and is therefore dimensionless.

[0057] The first component of the formula up to the minus sign is a weighted average of the dynamic thermal aging factors, namely thermal cycles 6 and temperature change rate 13.

[0058] The second component of the formula is the consideration of static aging 15.

[0059] Hl - health indicator W ges , which summarizes the dynamic and static factors of thermal ageing,

[0060] WA- Dynamic weighting value at time t of the estimated deterioration of insulation resistance due to temperature variations Rins(.n--cyc),

[0061] R ins (n_cyc) - estimated insulation resistance due to thermal cycling,

[0062] WB- Dynamic weighting value at time t of the estimated deterioration of insulation resistance due to the temperature change rate,

[0063] Rins - estimated insulation resistance due to the temperature change rate,

[0064] Wc- Dynamic weighting value at time t of the consumed lifetime due to static aging,

[0065] LoE(T, t) - consumed lifetime due to static aging.

[0066] Figure 7 shows the dynamic development of the weights over the lifetime of the winding insulation for a specific motor. This development is specific to each machine due to different operating conditions and loads.

[0067] The core of the invention is the continuous, data-driven monitoring of three main components of the thermal stress on the winding insulation of an electric motor, namely temperature cycling, temperature change rate, and static aging, and their integration to establish a direct correlation with the physical properties of the winding insulation, such as insulation resistance.

[0068] The winding insulation health indicator can be used as part of a comprehensive approach to monitoring the health of the electric motor or the entire drive system, incorporating the principle of multifactorial aging.

[0069] The winding insulation health indicator can be used by vehicle manufacturers to gain insights into fleet management aimed at reducing the occurrence of early failures, identifying oversized winding insulation systems, and correlating driving habits with their impact on winding insulation health.

[0070] An alternative solution may investigate the integration of information on the current supplied to the stator windings during stator tests and establish a correlation between the insulation resistance in relation to the thermal cycles and the maximum current supplied during heating of the stator unit.

[0071] The proposed solution aims to perform on-board condition monitoring for the engine stator and is considered as a descriptive approach.

[0072] Descriptive analysis methods are a preliminary stage of data processing that creates a summary of historical data to obtain useful information and potentially prepare the data for further analysis. In the present solution, descriptive methods are used to quantify the health status. Assuming that each component exhibits either an optimistic or conservative deterioration characteristic under real-world operating conditions, the end-of-life for each component is expected to deviate from the design specifications.

[0073] The proposed method allows for its use as a software module and as part of a larger health monitoring management system. The winding insulation health monitor thus represents an important building block for a complete monitoring system. The information is communicated to the driver and user, as well as to an entity such as the vehicle manufacturer or fleet operator.

[0074] This simplifies fleet management. It also provides vehicle manufacturers with an opportunity to verify whether components are correctly dimensioned.

[0075] For the end user, the driver, this provides a better assessment of the car's health status, which is important information when purchasing a used car. The health monitor serves this purpose and can also display the historical development of the car's health status.

[0076] The health monitor allows maintenance intervals to be optimized and planned in advance without time pressure.

[0077] The totality of the information provides a picture of how driving habits affect health, so that the driver can extend the life of the vehicle by adjusting his driving style.

Claims

Claims 1 . Method for determining the health status of the winding insulation on the stator of an electrical machine based solely on a parameter of the stator of the electrical machine, by estimating and weighting three main components of the thermal stress on the winding insulation, namely temperature change (20), temperature change rate (13) and static aging (15).

2. Method according to claim 1, characterized in that the parameter is at least one temperature measurement (30) or at least one current measurement on the stator.

3. Method according to claim 1 or 2, characterized in that the main components are combined to form a comprehensive indicator (W ges ) are merged.

4. Method according to one of the preceding claims, characterized in that individual indicators (W1, W3) are determined from the main components, the temperature change (20) and the temperature change rate (13) on the basis of a global model (8') 5. Method according to one of the preceding claims, characterized in that the main component static aging (15) is estimated with a life expectancy model and results in an indicator (W2).

6. Method according to one of claims 1-5, characterized in that indicators (W1, W2, W3) are weighted differently from one another over the lifetime of the winding insulation and thus of the electrical machine.

7. Method according to one of the preceding claims, characterized in that the method is a component in a higher-level method for determining the state of health of an electrical machine and / or a vehicle.

8. Method according to one of the preceding claims, characterized in that the method comprises an output step in which the health monitor (40) transmits information to a driver and / or a vehicle external monitoring system.

Images