A model-based fault detection method based on voltage ratio

Abstract

This invention proposes a model-predictive fault detection method based on voltage ratios. It directly processes the acquired signals, using the line voltage ratio as a key indicator for fault detection, simplifying the algorithm's complexity. Furthermore, it introduces a counting principle to improve the robustness and reliability of the fault detection method. This method also accurately locates the fault by determining the current polarity, improving fault diagnosis accuracy and reducing computational burden. In addition, this invention proposes a fault-tolerant control strategy for a two-phase four-switch topology based on model-predictive current control. By combining a permanent magnet synchronous motor (PMSM) with model predictive control (MPC) and considering the specific requirements and constraints of aerospace applications, it precisely controls the current, torque, and position of the PMSM, achieving high-performance control of the electric drive system, thereby improving control accuracy, response speed, and energy efficiency.

Description

Technical Field

[0001] This invention relates to the field of motor control and fault detection technology, specifically to a model-based fault detection method based on voltage ratio. Background Technology

[0002] With the application of rare-earth permanent magnet materials in the field of motors, the performance of permanent magnet synchronous motors (PMSMs) has been significantly improved. Currently, neodymium iron boron (NdFeB) and ferrite permanent magnet materials are widely used to manufacture motor rotors. The use of permanent magnets not only improves motor operating efficiency but also gives PMSMs advantages such as high power density, small size, light weight, and large starting torque. Therefore, PMSMs are widely used in fields requiring high precision and high reliability, and their application and development are inseparable from reliable fault diagnosis methods.

[0003] Fault diagnosis is a crucial aspect of PMSM drive systems. PMSM drive system faults include sensor failures, mechanical failures, and power converter failures. As a vital component of the PMSM drive system, transistors are among the most vulnerable devices due to their operating conditions. Statistics show that 80% of PMSM drive system failures are caused by transistor failures; therefore, power converter failures are the most common type of PMSM drive system fault.

[0004] For power converter fault detection, current mainstream methods can be categorized as follows: model-based methods, signal-based methods, and AI-based methods. Model-based methods rely heavily on model accuracy and parameter precision. AI-based methods require extensive data samples for training and the construction of high-quality training datasets, which necessitates numerous destructive experiments on the PMSM drive system, resulting in high costs. Signal-based methods can be divided into frequency domain and time domain methods. For frequency domain methods, some researchers use FFT to extract fault current features, but this method does not demonstrate high detection accuracy in dynamic situations. For time domain methods, some researchers utilize ACPV to transform the current signal for fault detection. While this method exhibits good robustness to changes in electrode speed and torque, the complex signal transformations inevitably impose a significant computational burden, impacting real-time detection performance. Summary of the Invention

[0005] To address the shortcomings of existing power converter fault detection methods in terms of robustness, computational burden, and real-time performance, this invention proposes a model prediction fault detection method based on voltage ratio. This method directly processes the acquired signals, using the line voltage ratio as a key indicator for fault detection, simplifying the algorithm's complexity. Furthermore, it introduces a counting principle to improve the robustness and reliability of the fault detection method. Moreover, this method accurately locates the fault by determining the current polarity, improving the accuracy of fault diagnosis and reducing the computational burden. In addition, this invention proposes a fault-tolerant control strategy for a two-phase four-switch topology based on model predictive current control. By combining a permanent magnet synchronous motor (PMSM) with model predictive control (MPC) and considering the specific requirements and constraints of aerospace applications, it precisely controls the current, torque, and position of the PMSM, achieving high-performance control of the electric drive system, thereby improving control accuracy, response speed, and energy efficiency.

[0006] The technical solution of this invention is as follows:

[0007] The aforementioned fault detection method based on voltage ratio model prediction includes the following steps:

[0008] Step 1: Use a voltage sensor to monitor the transistor voltage in the permanent magnet synchronous motor power converter and calculate the line voltage U between phases A and B. AB The line voltage U between phases BC and BC BC ;

[0009] Step 2: Determine the fault phase: Using the line voltage values ​​obtained in Step 1, calculate the ratio λ of the BC line voltage to the AB line voltage; if λ = -2, it indicates that the A-phase transistor has failed; if λ = 1, it indicates that the B-phase transistor has failed; if λ = -1 / 2, it indicates that the C-phase transistor has failed.

[0010] Step 3: Identify the faulty transistor: After identifying the faulty phase according to Step 2, integrate the current of the faulty phase for a set time. If the integral value is positive, it indicates that the lower bridge arm transistor of the faulty phase has an open circuit fault. If the integral value is negative, it indicates that the upper bridge arm transistor of the faulty phase has an open circuit fault.

[0011] Step 4: Perform inverter fault-tolerant control based on steps 2 and 3:

[0012] Construct the cost function:

[0013]

[0014] Where K is the set cost function scaling factor, and Flag is the fault diagnosis indicator. When there is no fault, Flag is 0. After a fault is detected, its value corresponds to the faulty transistor number. Specify the current value for the dq axis. For the predicted current value along the dq axis, δ[i s [This is for current constraint;]

[0015] The faulty phase and corresponding transistor are determined based on the magnitude of the cost function, and the faulty phase bridge arm is disconnected. At the same time, the fast thyristor connecting the motor neutral point and the power supply midpoint is closed, turning the three-phase six-switch inverter into a two-phase four-switch inverter. The four basic voltage vectors generated by the two-phase four-switch inverter are used to continue to control the PMSM.

[0016] Furthermore, the specific process of step 1 is as follows: The voltage across the four transistors in the permanent magnet synchronous motor power converter is monitored using a voltage sensor: U VT1 U VT2 U VT3 U VT6 U VT1 U is the voltage of the on-circuit transistor in phase A. VT2 U is the voltage of the C-phase downstream transistor. VT3 U is the voltage of the B-phase on-circuit transistor. VT6 This refers to the voltage of the B-phase downstream transistor.

[0017] Using formula

[0018]

[0019] The line voltage U between phases A and B is calculated. AB The line voltage U between phases BC and BC BC V dc Given a DC voltage.

[0020] Furthermore, in step 2, considering the voltage bias in the actual operation of the inverter, when judging λ, if λ∈(-2-Δσ,-2+Δσ), it indicates that the A-phase transistor has failed; if λ∈(1-Δσ,1+Δσ), it indicates that the B-phase transistor has failed; if λ∈(-1 / 2-Δσ,-1 / 2+Δσ), it indicates that the C-phase transistor has failed; Δσ is the set change amplitude.

[0021] Furthermore, in step 2, after filtering λ using the counting principle, λ is then judged.

[0022] Furthermore, in step 2, the judgment threshold Constant in the counting principle is determined according to the formula...

[0023]

[0024] The calculation yields, where T is the electrical period, T0... S The sampling period is k, and k is a set coefficient.

[0025] Furthermore, in step 3, the integration time for the current of the faulty phase is half an electrical cycle.

[0026] Furthermore, in step 4, the fault-tolerant control strategy also sets up a counter with an initial value of 0 to detect the number of faulty switches. When the amplitude of the cost function is close to 0, it is determined that a fault has occurred. The system counts the level changes of the cost function amplitude. When the level changes of the cost function amplitude, the counter is incremented by 1. When the counter value is less than 2, the system operates normally through fault-tolerant control. When the counter value is greater than or equal to 2, the power supply to the motor is cut off.

[0027] Furthermore, in step 4, after a fault occurs, the level of the cost function amplitude changes. When the cost function amplitude changes, the counter is incremented by 1. When the cost function amplitude changes again, and the new faulty transistor does not belong to the same bridge arm as the previously faulty transistor, the counter is incremented by 1 again, and the motor power supply is cut off.

[0028] Beneficial effects

[0029] The present invention has the following beneficial effects:

[0030] First, this method utilizes the line voltage ratio as a key indicator for fault detection. By comparing the line voltages of different phases, the presence of a faulty phase can be effectively detected. This voltage ratio-based fault detection method eliminates the need for complex signal transformations, simplifying the algorithm's complexity, and demonstrates good robustness under real-world operating conditions such as noise, speed variations, load changes, and motor parameter variations.

[0031] Secondly, this method incorporates a counting principle to improve the robustness of the fault detection method. By counting the line voltage ratio of the faulty phase over a period of time and comparing it with a pre-set threshold, the presence of the faulty phase can be determined. This counting principle effectively reduces sensitivity to real-time signal changes, improving the reliability and stability of the fault detection method.

[0032] Furthermore, this method precisely locates the fault by determining the polarity of the current, avoiding the complex calculations required in traditional methods. By monitoring changes in the current polarity of the faulty phase, the location of the fault can be determined quickly and accurately. This precise location capability not only improves the accuracy of fault diagnosis but also reduces the computational burden, making the method more efficient and practical.

[0033] Finally, this invention proposes a fault-tolerant control strategy that enables the motor to operate normally even when a local transistor open-circuit fault occurs, while ensuring motor safety and production continuity.

[0034] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0035] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0036] Figure 1 Flowchart for open-circuit fault detection in model predictive control of permanent magnet synchronous motor;

[0037] Figure 2 Schematic diagram of a normal inverter;

[0038] Figure 3 VT1 fault diagram;

[0039] Figure 4 VT3 fault diagram;

[0040] Figure 5 VT5 fault diagram;

[0041] Figure 6 Schematic diagram of the counting principle;

[0042] Figure 7 Fault-tolerant control topology diagram;

[0043] Figure 8 Voltage vector diagram;

[0044] Figure 9 System block diagram. Detailed Implementation

[0045] The embodiments of the present invention are described in detail below. These embodiments are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0046] like Figure 1 As shown, this embodiment proposes a model prediction fault detection method based on voltage ratio to address the requirements of open-circuit fault detection and fault-tolerant control in the power converter of a permanent magnet synchronous motor. The specific steps are as follows:

[0047] Step 1: Monitor the voltage across the four transistors in the permanent magnet synchronous motor power converter using a voltage sensor: U VT1 U VT2 U VT3 U VT6 U VT1 U is the voltage of the on-circuit transistor in phase A. VT2 U is the voltage of the C-phase downstream transistor. VT3 U is the voltage of the B-phase on-circuit transistor. VT6 This refers to the voltage of the B-phase downstream transistor.

[0048] Using Thevenin's theorem, the voltage equations in a three-phase voltage inverter are established, such as... Figure 3 , 4 5. The voltage equation (1) is given from the loop shown by the blue line; the voltage equation (2) is given from the loop shown by the red line:

[0049]

[0050] Where V dc Given a DC voltage; the line voltage U between phases A and B is calculated using formulas 1 and 2. AB The line voltage U between phases BC and BC BC .

[0051] Step 2: Determine the fault phase.

[0052] First, using the line voltage values ​​obtained in step 1, calculate the ratio λ of the BC line voltage to the AB line voltage.

[0053] like Figure 3 As shown, when the A-phase upper transistor VT1 experiences an open-circuit fault, and the current three-phase voltage inverter control switch state is (1,0,1) (i.e., the current three-phase voltage inverter control switch state is to turn on the A-phase upper transistor, the B-phase lower transistor, and the C-phase upper transistor, while the remaining transistors are open), then the line voltage ratio λ = U BC / U AB It can be obtained from the following formula:

[0054]

[0055] Obviously, when the A-phase transistor fails, λ = -2.

[0056] like Figure 4 As shown, when the upper-circuit transistor VT3 of phase B experiences an open-circuit fault, the line voltage ratio λ = U BC / U AB It can be obtained from the following formula:

[0057]

[0058] Obviously, when the B-phase transistor fails, λ = 1.

[0059] like Figure 5 As shown, when the C-phase upper transistor VT5 experiences an open-circuit fault, the line voltage ratio λ = U BC / U AB It can be obtained from the following formula:

[0060]

[0061] Obviously, when the C-phase transistor fails, λ = -1 / 2.

[0062] Based on the above analysis, it is clear that the fault phase can be determined by monitoring the ratio λ of the BC line voltage to the AB line voltage: if λ = -2, it indicates that the A-phase transistor has failed; if λ = 1, it indicates that the B-phase transistor has failed; and if λ = -1 / 2, it indicates that the C-phase transistor has failed.

[0063] However, in actual signal detection, we found that due to a slight voltage bias in the inverter during operation, λ does not precisely stay at the values ​​of -2, 1, and -1 / 2. Therefore, the effective range of λ needs to be expanded to (λ-Δσ, λ+Δσ) to ensure rapid detection. However, as the effective range of λ expands, the detection results are more susceptible to noise. Therefore, this embodiment introduces a counting principle to filter λ, greatly reducing the impact of noise on the results and ensuring detection accuracy in harsh environments.

[0064] Counting principle as Figure 6 As shown, the main criterion is the signal duration, i.e., the signal width. When the signal width meets the requirement, the signal is considered valid; otherwise, it is considered noise. The counter remains at zero until a change in the λ level is detected. When a change in level is detected, the counter starts counting. When the count value reaches the threshold Constant, the signal is valid; otherwise, the signal is filtered out. The threshold Constant is determined by the formula...

[0065]

[0066] The calculation yields, where T is the electrical period, T0... S Where λ is the sampling period and k is a set coefficient. After filtering using the counting principle, λ can effectively reflect the phase of the faulty transistor.

[0067] Step 3: Identify the faulty transistor.

[0068] After obtaining the fault phase, it is also necessary to determine whether the fault occurs in the upper or lower bridge arm. This step determines which upper or lower bridge arm the fault is in by integrating the current.

[0069] The filtered λ is used as the start signal. When λ changes from 0 to a certain characteristic value, the faulty phase is identified. The current of the faulty phase is then integrated. When the upper arm transistor of the inverter is open, the phase current of that phase only contains negative values ​​and 0. Conversely, when the fault occurs in the lower arm, the phase current only contains the positive half-cycle and 0. This characteristic is used to integrate the faulty phase current, as shown in the following formula.

[0070]

[0071] If the integral value is positive, it indicates that the lower bridge arm transistor of the faulty phase has an open-circuit fault; if the integral value is negative, it indicates that the upper bridge arm transistor of the faulty phase has an open-circuit fault. Since this scheme only integrates the current of the faulty phase and the integration time is half an electrical cycle, it avoids prolonged calculations that could worsen the fault and significantly reduces the computational burden.

[0072] Step 4: Implement fault-tolerant control based on the above fault detection principle.

[0073] To ensure that the motor can still operate smoothly without interrupting production under fault conditions, this invention proposes a fault-tolerant control strategy based on a two-phase four-switch inverter, based on the aforementioned fault detection principle:

[0074] During normal operation of the permanent magnet synchronous motor control system, the inverter operates as a three-phase six-switch system. The two power switches on the same phase bridge arm conduct alternately, supplying DC voltage source U... dc The output voltage is inverted into the AC power required by the motor and supplied to the motor. When a switch on one phase arm of the three-phase six-switch inverter fails, the faulty phase arm is immediately disconnected, and simultaneously the fast thyristor TR connecting the motor neutral point and the power supply neutral point is closed. The neutral-point-leaded two-phase four-switch fault-tolerant inverter reconfigures its topology as follows: Figure 7 As shown, the three-phase six-switch inverter is transformed into a two-phase four-switch inverter. Fault-tolerant control continues to apply effective control to the PMSM using the four basic voltage vectors generated by the two-phase four-switch inverter. Normal model predictive current control has eight alternative switching states, while switching to model predictive current fault-tolerant control has four alternative switching states.

[0075] Since the inverter control signals are limited, the key to implementing this fault-tolerant control strategy is how to determine whether a fault has occurred, which phase is faulty, and which switch in the faulty phase is faulty using only one signal. To this end, this invention integrates fault diagnosis indicators into a single cost function, facilitating the observation of whether a fault has occurred and its location. The cost function is:

[0076]

[0077] Where K is the cost function scaling factor, which is a pre-set, large value; in this embodiment, it is set to 10. 5 This allows for a more intuitive observation of diagnostic indicators after a fault occurs. Flag is a fault diagnostic indicator, and its value corresponds to the faulty transistor number. Specify the current value for the dq axis. For the predicted current value along the dq axis, δ[i s [This is for current constraint. In this embodiment, the transistor number is as follows] Figure 2 As shown in Table 1, the fault indication table is 0 when there is no fault.

[0078] Table 1 Fault Indication Table

[0079]

[0080]

[0081] The magnitude of the cost function is used as the criterion for judging the fault-tolerant control strategy:

[0082] When the permanent magnet synchronous motor control system is operating normally, λ = 0 and Flag = 0. At this time, the amplitude of the cost function is the corresponding current value. When an open-circuit fault occurs in a certain phase transistor, λ changes to the corresponding characteristic value. Since the scaling factor K of the cost function is set to a very large value, the first term of the cost function is approximately 0. At this time, the fault phase current has not yet been integrated, and the Flag value has not yet been determined. The Flag remains 0, so the amplitude of the cost function is close to 0. Therefore, when the amplitude of the cost function changes from the current value to a value close to 0, a fault is judged to have occurred. After the integration of the fault phase current is completed, the fault transistor is determined based on the integrated value, and the Flag value is determined according to the fault indication table. At this time, the amplitude of the cost function is approximately the Flag value. Thus, the fault phase and the corresponding transistor can be determined based on the amplitude of the cost function.

[0083] Figure 7 The diagram shows the fault-tolerant control topology for an open-circuit fault in transistor VT1, with a fast thyristor connecting the motor neutral point and the capacitor midpoint. Fault-tolerant control activates when the cost function amplitude becomes 1. First, the faulty phase is immediately disconnected, transforming the three-phase six-switch inverter into a two-phase four-switch inverter. The three-phase stator voltages are as follows:

[0084]

[0085] Figure 8 This isolates all the basic voltage vectors that the two-phase four-switch inverter can provide after isolating phase A. The number of basic voltage vectors the inverter can provide is reduced to four, and the three-phase current maintaining motor operation is changed from I... N I B I C composition.

[0086] In addition, the fault-tolerant control strategy also includes a counter with an initial value of 0 to detect the number of faulty switches. When the amplitude of the cost function approaches 0, a fault is detected. The system then counts the level changes in the amplitude of the cost function. When the amplitude of the cost function changes, the counter is incremented by 1. When the counter value is less than 2, the system operates normally through fault-tolerant control. When the counter value is greater than or equal to 2, the power supply to the motor is cut off.

[0087] Specifically, after a fault is detected, the Flag value is not yet determined because the faulty phase current has not yet been integrated; therefore, the Flag remains at 0, and the amplitude of the cost function is close to 0. Once the faulty phase current integration is complete, the faulty transistor is identified based on the integrated value, and the Flag value is determined according to the fault indication table. At this point, the amplitude of the cost function is approximately equal to the Flag value. When the amplitude of the cost function changes, the counter increments by 1. If a new transistor fault occurs, the specific value of the Flag changes, and the amplitude of the cost function changes again, the counter increments by 1 again, and the motor power supply is cut off. Furthermore, a further limitation can be added: if the new faulty transistor does not belong to the same bridge arm as the previously faulty transistor, the counter increments by 1 again, and the motor power supply is cut off.

[0088] As can be seen from this embodiment, the present invention does not require any signal transformation, the result output is more direct, eliminates intermediate links, does not cause any computational burden, and has strong real-time performance; it can quickly and accurately detect the fault location, effectively detect two transistors failing simultaneously, and has a considerable detection speed compared with existing detection methods; and it integrates a fault-tolerant control strategy, so that the motor can still operate normally when a partial transistor open circuit fault occurs, while ensuring motor safety and production continuity.

[0089] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention without departing from the principles and spirit of the present invention.

Claims

1. A model-based fault detection method based on voltage ratio, characterized in that: Includes the following steps: Step 1: Use a voltage sensor to monitor the transistor voltage in the permanent magnet synchronous motor power converter and calculate the line voltage U between phases A and B. AB The line voltage U between phases BC and BC BC ; Step 2: Determine the fault phase: Using the line voltage values ​​obtained in Step 1, calculate the ratio λ of the BC line voltage to the AB line voltage; if λ = -2, it indicates that the A-phase transistor has failed; if λ = 1, it indicates that the B-phase transistor has failed; if λ = -1 / 2, it indicates that the C-phase transistor has failed. Step 3: Identify the faulty transistor: After identifying the faulty phase according to Step 2, integrate the current of the faulty phase for a set time. If the integral value is positive, it indicates that the lower bridge arm transistor of the faulty phase has an open circuit fault. If the integral value is negative, it indicates that the upper bridge arm transistor of the faulty phase has an open circuit fault. Step 4: Perform inverter fault-tolerant control based on steps 2 and 3: Construct the cost function: Where K is the set cost function scaling factor, and Flag is the fault diagnosis indicator. When there is no fault, Flag is 0. After a fault is detected, its value corresponds to the faulty transistor number. Specify the current value for the dq axis. For the predicted current value along the dq axis, δ[i s [This is for current constraint;] The faulty phase and corresponding transistor are determined based on the magnitude of the cost function, and the faulty phase bridge arm is disconnected. At the same time, the fast thyristor connecting the motor neutral point and the power supply midpoint is closed, turning the three-phase six-switch inverter into a two-phase four-switch inverter. The four basic voltage vectors generated by the two-phase four-switch inverter are used to continue to control the PMSM.

2. The model-based fault detection method based on voltage ratio as described in claim 1, characterized in that: The specific process of step 1 is as follows: Voltage sensors are used to monitor the voltage across four transistors in the power converter of a permanent magnet synchronous motor: U VT1 U VT2 U VT3 U VT6 U VT1 U is the voltage of the on-circuit transistor in phase A. VT2 U is the voltage of the C-phase downstream transistor. VT3 U is the voltage of the B-phase on-circuit transistor. VT6 This refers to the voltage of the B-phase downstream transistor. Using formula The line voltage U between phases A and B is calculated. AB The line voltage U between phases BC and BC BC V dc Given a DC voltage.

3. The model-based fault detection method based on voltage ratio as described in claim 1, characterized in that: In step 2, considering the voltage bias in the actual operation of the inverter, when judging λ, if λ∈(-2-Δσ,-2+Δσ), it indicates that the A-phase transistor is faulty; if λ∈(1-Δσ,1+Δσ), it indicates that the B-phase transistor is faulty; if λ∈(-1 / 2-Δσ,-1 / 2+Δσ), it indicates that the C-phase transistor is faulty. Δσ is the set change amplitude.

4. The fault detection method based on voltage ratio model prediction according to claim 3, characterized in that: In step 2, after filtering λ using the counting principle, λ is then judged.

5. The model-based fault detection method based on voltage ratio as described in claim 4, characterized in that: In step 2, the judgment threshold Constant in the counting principle is determined according to the formula... The calculation yields, where T is the electrical period, T0... S The sampling period is k, and k is a set coefficient.

6. The fault detection method based on voltage ratio model prediction according to claim 1, characterized in that: In step 3, the integration time for the current of the faulty phase is half an electrical cycle.

7. The fault detection method based on voltage ratio model prediction according to claim 1, characterized in that: In step 4, the fault-tolerant control strategy also includes a counter with an initial value of 0 to detect the number of faulty switches. When the amplitude of the cost function approaches 0, it is determined that a fault has occurred. The system then counts the level changes in the amplitude of the cost function. When the amplitude of the cost function changes, the counter is incremented by 1. When the counter value is less than 2, the system operates normally through fault-tolerant control; when the counter value is greater than or equal to 2, the power supply to the motor is cut off.

8. The model-based fault detection method based on voltage ratio as described in claim 7, characterized in that: In step 4, after a fault occurs, the level of the cost function amplitude changes. When the cost function amplitude changes, the counter is incremented by 1. When the cost function amplitude changes again, and the new faulty transistor does not belong to the same bridge arm as the previously faulty transistor, the counter is incremented by 1 again, and the motor power supply is cut off.

Images