A five-phase permanent magnet motor fault diagnosis method based on adaptive sliding mode observation

Abstract

The application discloses a five-phase permanent magnet motor fault diagnosis method based on adaptive sliding mode observation, establishes a hybrid logic dynamic model of the system according to the topology structure of an inverter, the trigger signal state of a switching tube and the current path characteristics, and constructs a sliding mode observer based on the model; the adaptive sliding mode observer with fast convergence characteristics and strong chattering suppression capability is designed to obtain the state space expression of the system; the sliding mode observer is used to obtain the observation values of the phase currents, and the actual measurement values are compared with the observation values to obtain the phase current residual error signals; secondly, the current residual error signals of the phases are normalized and compared with the designed adaptive threshold value; if the current residual error of a certain phase exceeds the threshold value, it is determined that the open circuit fault of the power tube corresponding to the phase occurs.

Description

Technical Field

[0001] This application relates to the field of fault diagnosis and analysis technology, and in particular to a fault diagnosis method for a five-phase permanent magnet motor based on adaptive sliding mode observation. Background Technology

[0002] With the rapid development of electric vehicles, aerospace, wind power, and other fields, multiphase permanent magnet motors have become key power devices due to their advantages such as high power density, strong reliability, excellent energy efficiency, and low torque ripple. However, the various types of faults faced by motors under complex operating conditions can directly lead to performance degradation, soaring maintenance costs, and even system paralysis. Furthermore, as the use and operating conditions of motors change, they may encounter various fault problems. These faults not only affect the motor's performance and efficiency but may also lead to equipment downtime and increased maintenance costs. Taking a five-phase permanent magnet synchronous motor as an example, in actual operating conditions, the inverter of its electric drive system often experiences power transistor faults, capacitor faults, and interface faults. Among these, power transistor faults account for approximately 31% of inverter faults, mainly open-circuit and short-circuit faults. Short-circuit faults evolve rapidly, and fast-acting fuses are typically added to the converter to convert short-circuit faults into open-circuit faults. Long-term operation with faults will seriously affect the reliability of the five-phase permanent magnet synchronous motor drive system. Therefore, developing a high-precision and robust power transistor open-circuit fault diagnosis method is of significant engineering importance.

[0003] Currently, fault diagnosis methods for inverters are mainly divided into data-driven fault diagnosis methods, signal-based fault diagnosis methods, and model-based fault diagnosis methods. Data-driven fault diagnosis methods generally analyze historical offline data and current online data stored in the system to extract feature information suitable for fault diagnosis. However, this method requires a large dataset to train the established model to achieve good prediction accuracy, and the computational load required for the model is significant, making it difficult to deploy directly on hardware controllers such as DSPs. Signal-based fault diagnosis methods can be divided into current-based and voltage-based fault diagnosis methods, depending on the measurement. Current-based fault diagnosis methods often suffer from insufficient robustness, high false diagnosis rates, and slow response speeds; while voltage-based methods require additional hardware circuitry, leading to increased costs. Model-based fault diagnosis methods improve upon the shortcomings of traditional voltage-based fault diagnosis methods, generally not requiring additional hardware circuitry, but threshold selection becomes complex and cumbersome when faced with changes in operating conditions and working status. Especially during continuous fault operation, the reliability of the motor drive system will further decrease. Summary of the Invention

[0004] This application provides a fault diagnosis method for five-phase permanent magnet motors based on adaptive sliding mode observation, aiming to solve the problem that existing model-based fault diagnosis methods are complex and cumbersome in selecting thresholds due to changes in operating conditions and working status. In particular, the reliability of the motor drive system will further decrease when the fault continues to run.

[0005] In a first aspect, a fault diagnosis method for a five-phase permanent magnet motor based on adaptive sliding mode observation is provided, the method comprising:

[0006] Based on the inverter topology, switching transistor trigger signal status, and current path characteristics of the five-phase motor drive system, a hybrid logic dynamic model is established.

[0007] Based on the hybrid logic dynamic model, a sliding mode observer is established, and an adaptive sliding mode observer with fast convergence characteristics and strong chatter suppression capability is designed to obtain the state space expression of the five-phase motor drive system.

[0008] The adaptive sliding mode observer is used to obtain the observed values ​​of the current in each phase of the five-phase motor drive system, and the actual measured values ​​are compared with the observed values ​​to obtain the residual signals of the current in each phase.

[0009] The residual signals of each phase current are normalized and compared with the designed adaptive threshold; if the residual of the target phase current exceeds the threshold, it is determined that the corresponding power transistor of the target phase has an open circuit fault.

[0010] Optionally, in the above scheme, the hybrid logic dynamic model is: in, The derivative of the current in the five-phase stator winding; A = diag {- R / L , - R / L , - R / L , - R / L , - R / L}, R It is the resistance of the corresponding phase. L It is the inductance of the corresponding phase; B 1= diag {-1 / L -1 / L -1 / L -1 / L -1 / L}; B 2= U dc δ, , , D i ( i =1,…,10) represents the switching signal of the i-th power transistor. σ k ( k =a,b,c,d,e) represent the direction of current in each phase, and the direction of current flowing into the motor windings is defined as positive. σ k =1), outflow is negative ( =1), σ k =0) U dc This is the DC bus voltage; i =[ i a i b i c i d i e ] T Corresponding to the five-phase stator winding current; e =[ e a e b e c e d e e ] T Corresponding to five opposite electromotive forces;

[0011] The five-phase motor drive system d 1 axis relative to α The lead angle of the axis is defined as θ The expression for the back electromotive force is: ;

[0012] Where ω is the angular velocity, It is a permanent magnet flux chain.

[0013] Optionally, in the above scheme, the adaptive sliding mode observer is: ;in, This represents the derivative of the adaptive sliding mode observer with respect to the estimated value of the output current. This represents the estimated value of the output current by the adaptive sliding mode observer. B Represents the input matrix, u represents Input variables, This represents the estimated value of the output variable. f ( S) represents the adaptive reaching rate, and the sliding surface is defined as , y Indicates the output variable. C This represents a unit coefficient matrix.

[0014] Optionally, in the above scheme, the observed values ​​of the current in each phase of the five-phase motor drive system are obtained through the adaptive sliding mode observer, and the actual measured values ​​are compared with the observed values ​​to obtain the residual signals of the current in each phase. This specifically includes the following steps:

[0015] Construct the root mean square primitive equation ,in, ;

[0016] Record fault detection factors T n ( n = a, b, c, d, e )express n Measured phase current i n Compared with current observations The difference of the root mean square function: ;

[0017] Constructing fault detection variables based on current shape factors G n The large-amplitude current residual is smoothly mapped to the [0,1] interval using a normalization function; where, ;

[0018] Based on the estimated current value of the adaptive sliding mode observer, the adaptive threshold T for the fault detection variable is designed as follows: , G a , G b , G c , G d , G e Let represent the fault detection variables for phases a, b, c, d, and e respectively, and k represent the gain.

[0019] Combining fault detection variables G n And an adaptive threshold T, to obtain the open-circuit fault detection variable. D n ( n = a, b, c, d, e )for: ;

[0020] Fault detection variables when no fault occurs G nIt will not exceed the adaptive threshold T; however, when the target power transistor in the motor drive system experiences an open-circuit fault, the corresponding value of the target power transistor will be... G n Exceeding T.

[0021] Optionally, in the above scheme, the single-tube open-circuit fault determination includes:

[0022] When only B Mutually D 3. The phase current equation for an open-circuit fault is: , for B Mutually D 3. The derivatives of the phase currents of each phase after an open-circuit fault occurs; for B Mutually D 3. Phase currents after an open-circuit fault occurs;

[0023] If an open-circuit fault occurs in the upper arm power transistor, the positive half-wave of the corresponding phase's output current will be missing, resulting in a residual greater than zero between the estimated and actual values ​​of the corresponding phase's current. If an open-circuit fault occurs in the lower arm power transistor, the negative half-wave of the corresponding phase's output current will be missing, resulting in a current residual less than zero. The current polarity characteristics are extracted as fault location variables. , ;

[0024] Variables detected by open-circuit fault D n and fault location variables The combination of these variables yields the target fault diagnosis variables. Flag n ;in, .

[0025] Optionally, in the above scheme, the double-tube open-circuit fault includes a single-phase double-tube open-circuit fault and an out-of-phase double-tube open-circuit fault.

[0026] Optionally, in the above scheme, the single-phase dual-tube open-circuit fault determination is as follows:

[0027] In the case of a single-phase dual-tube fault, select the open-circuit fault detection variable. D n and fault location variables exist Changes between them serve as auxiliary variables for fault diagnosis. , .

[0028] Optionally, the above scheme further includes: after determining the fault, based on the residual current characteristics under different open-circuit fault modes, locating the specific faulty switch by analyzing the residual signal, and immediately updating the fault identifier to the corresponding fault code, thereby completing the accurate location of the power transistor open-circuit fault.

[0029] Secondly, a fault diagnosis system for a five-phase permanent magnet motor based on adaptive sliding mode observation is provided, the system comprising:

[0030] Model building module: Based on the inverter topology, switching transistor trigger signal status and current path characteristics of the five-phase motor drive system, a hybrid logic dynamic model is established;

[0031] Design module: used to establish a sliding mode observer based on the hybrid logic dynamic model, design an adaptive sliding mode observer with fast convergence characteristics and strong chatter suppression capability, and obtain the state space expression of the five-phase motor drive system;

[0032] Acquisition module: used to acquire the observed values ​​of the current of each phase of the five-phase motor drive system through the adaptive sliding mode observer, and compare the actual measured values ​​with the observed values ​​to obtain the residual signals of the current of each phase;

[0033] Normalization processing module: used to normalize the residual signals of each phase current and compare them with the designed adaptive threshold;

[0034] Judgment module: If the target phase current residual exceeds the threshold, it is determined that the power transistor corresponding to the target phase has an open circuit fault; wherein, the target phase current residual is any one of the phase current residual signals.

[0035] Compared with the prior art, this application has at least the following beneficial effects:

[0036] This application, based on further analysis and research of existing technical problems, recognizes that existing model-based fault diagnosis methods are complex and cumbersome in selecting thresholds due to changes in operating conditions and working status. Especially concerning is the further degradation of the reliability of the motor drive system during continuous fault operation. This solution establishes a hybrid logic dynamic model of the system, considering the inverter's topology, the trigger signal state of the switching transistors, and the current path characteristics, and constructs a sliding mode observer based on this model. By designing an adaptive sliding mode observer with fast convergence characteristics and strong chatter suppression capabilities, the state-space expression of the system is obtained. The observed values ​​of the current in each phase are acquired using this sliding mode observer, and the actual measured values ​​are compared with the observed values ​​to obtain the residual signals of the current in each phase. Next, the residual signals of the current in each phase are normalized and compared with the designed adaptive threshold; if the residual current of a certain phase exceeds the threshold, an open-circuit fault is determined for the corresponding power transistor. Finally, after fault determination, based on the residual current characteristics under different open-circuit fault modes, the specific faulty switching transistor is further located by analyzing the residual signals, and the fault identifier is immediately updated to the corresponding fault code, completing the accurate location of the power transistor open-circuit fault. This invention employs an adaptive gain and adaptive threshold strategy, which reduces the complexity of parameter settings, enhances the robustness of the system, improves fault diagnosis efficiency, and does not require additional hardware costs. Attached Figure Description

[0037] Figure 1 A flowchart illustrating a fault diagnosis method for a five-phase permanent magnet motor based on adaptive sliding mode observation, provided in one embodiment of this application;

[0038] Figure 2 This is a topology diagram of a five-phase permanent magnet motor drive system provided in one embodiment of this application;

[0039] Figure 3 A schematic diagram of a fault diagnosis method for a five-phase permanent magnet motor drive system based on an adaptive sliding mode observer provided in one embodiment of this application;

[0040] Figure 4 A simulation diagram of an open-circuit fault in the power transistor of the upper arm of a five-phase permanent magnet motor drive system A, provided in one embodiment of this application;

[0041] Figure 5 This is a simulation diagram of a single-phase dual-tube open-circuit fault in a five-phase permanent magnet motor drive system provided in one embodiment of this application. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0043] In one embodiment, such as Figure 1 As shown, a fault diagnosis method for a five-phase permanent magnet motor based on adaptive sliding mode observation is provided, including the following steps:

[0044] Based on the inverter topology, switching transistor trigger signal status, and current path characteristics of the five-phase motor drive system, a hybrid logic dynamic model is established.

[0045] Based on the hybrid logic dynamic model, a sliding mode observer is established, and an adaptive sliding mode observer with fast convergence characteristics and strong chatter suppression capability is designed to obtain the state space expression of the five-phase motor drive system.

[0046] The adaptive sliding mode observer is used to obtain the observed values ​​of the current in each phase of the five-phase motor drive system, and the actual measured values ​​are compared with the observed values ​​to obtain the residual signals of the current in each phase.

[0047] The residual signals of each phase current are normalized and compared with the designed adaptive threshold; if the residual of the target phase current exceeds the threshold, it is determined that the corresponding power transistor of the target phase has an open circuit fault.

[0048] In this embodiment, the hybrid logic dynamic model is: ;in, The derivative of the current in the five-phase stator winding; A = diag {- R / L , - R / L , - R / L , - R / L , - R / L}, R It is the resistance of the corresponding phase. L It is the inductance of the corresponding phase; B 1= diag {-1 / L -1 / L -1 / L -1 / L -1 / L}; B 2= U dc δ, , , D i ( i =1,…,10) represents the switching signal of the i-th power transistor. σk ( k =a,b,c,d,e) represent the direction of current in each phase, and the direction of current flowing into the motor windings is defined as positive. σ k =1), outflow is negative ( =1), σ k =0) U dc This is the DC bus voltage; i =[ i a i b i c i d i e ] T Corresponding to the five-phase stator winding current; e =[ e a e b e c e d e e ] T Corresponding to five opposite electromotive forces;

[0049] The five-phase motor drive system d 1 axis relative to α The lead angle of the axis is defined as θ The expression for the back electromotive force is: ;

[0050] Where ω is the angular velocity, It is a permanent magnet flux chain.

[0051] In this embodiment, the five phases of the five-phase permanent magnet motor are phase a, phase b, phase c, phase d, and phase e. The resistance values ​​of each phase of the five-phase permanent magnet motor are the same, and the inductance values ​​of each phase are also the same.

[0052] In this embodiment, the adaptive sliding mode observer is: .

[0053] in, This represents the derivative of the adaptive sliding mode observer with respect to the estimated value of the output current. This represents the estimated value of the output current by the adaptive sliding mode observer. B Represents the input matrix, u represents Input variables, This represents the estimated value of the output variable. f( S ) represents the adaptive reaching rate, and the sliding surface is defined as , y Indicates the output variable. C This represents a unit coefficient matrix.

[0054] In this embodiment, the observed values ​​of the current in each phase of the five-phase motor drive system are obtained through the adaptive sliding mode observer, and the actual measured values ​​are compared with the observed values ​​to obtain the residual signals of the current in each phase. Specifically, the process includes the following steps:

[0055] Construct the root mean square primitive equation ,in, Record the fault detection factors. T n ( n = a, b, c, d, e )express n Measured phase current i n Compared with current observations The difference of the root mean square function: .

[0056] Constructing fault detection variables based on current shape factors G n The large-amplitude current residual is smoothly mapped to the [0,1] interval using a normalization function; where, .

[0057] Based on the estimated current value of the adaptive sliding mode observer, the adaptive threshold T for the fault detection variable is designed as follows: , G a , G b , G c , G d , G e Let represent the fault detection variables for phases a, b, c, and d respectively, and k represent the gain.

[0058] Combining fault detection variables G n And an adaptive threshold T, to obtain the open-circuit fault detection variable. D n ( n = a, b, c, d, e )for: ;

[0059] Fault detection variables when no fault occurs G nIt will not exceed the adaptive threshold T; however, when the target power transistor in the motor drive system experiences an open-circuit fault, the corresponding value of the target power transistor will be... G n Exceeding T.

[0060] It should be noted that, G a This represents the fault detection variable for phase a. G b This represents the fault detection variable for phase b. G c This represents the fault detection variable for phase c. G d This represents the fault detection variable for phase d. G e This represents the fault detection variable for phase e.

[0061] In this embodiment, the open-circuit fault of the target corresponding power transistor specifically includes single-transistor open-circuit fault and double-transistor open-circuit fault.

[0062] In this embodiment, the single-tube open-circuit fault determination includes:

[0063] When only B Mutually D 3. The phase current equation for an open-circuit fault is:

[0064] ;

[0065] in, for B Mutually D 3. The derivatives of the phase currents of each phase after an open-circuit fault occurs. for B Mutually D 3. Phase currents of each phase after an open-circuit fault occurs.

[0066] If an open-circuit fault occurs in the upper arm power transistor, the positive half-wave of the corresponding phase's output current will be missing, resulting in a residual greater than zero between the estimated and actual values ​​of the corresponding phase's current. If an open-circuit fault occurs in the lower arm power transistor, the negative half-wave of the corresponding phase's output current will be missing, resulting in a current residual less than zero. The current polarity characteristics are extracted as fault location variables. , .

[0067] Variables detected by open-circuit fault D n and fault location variables The combination of these variables yields the target fault diagnosis variables. Flag n , .

[0068] In this embodiment, the double-tube open-circuit fault includes a single-phase double-tube open-circuit fault and an out-of-phase double-tube open-circuit fault.

[0069] In this embodiment, the single-phase dual-tube open-circuit fault is determined as follows:

[0070] In the case of a single-phase dual-tube fault, select the open-circuit fault detection variable. D n and fault location variables exist Changes between them serve as auxiliary variables for fault diagnosis. .

[0071] in, .

[0072] In this embodiment, the method further includes: after determining the fault, based on the residual current characteristics under different open-circuit fault modes, locating the specific faulty switch by analyzing the residual signal, and immediately updating the fault identifier to the corresponding fault code, thereby completing the accurate location of the power transistor open-circuit fault.

[0073] This embodiment establishes a hybrid logic dynamics (MLD) model of the system based on the inverter's topology, switching transistor trigger signal states, and current path characteristics. An adaptive sliding mode observer with fast convergence and strong chatter suppression capabilities is designed to obtain the system's state-space expression. The sliding mode observer is used to acquire the observed values ​​of each phase current, and the actual measured values ​​are compared with the observed values ​​to obtain the residual signals of each phase current. The residual signals of each phase current are normalized and compared with a designed adaptive threshold. If the residual current of a phase exceeds the threshold, an open-circuit fault is determined for the corresponding power transistor. Finally, after fault determination, based on the residual current characteristics under different open-circuit fault modes, the residual signals are analyzed to further locate the specific faulty switching transistor, and the fault identifier is immediately updated to the corresponding fault code, completing the accurate location of the power transistor open-circuit fault. This invention employs an adaptive gain and adaptive threshold strategy, reducing the complexity of parameter settings, enhancing system robustness, improving fault diagnosis robustness, and eliminating the need for additional hardware costs.

[0074] In one embodiment, such as Figure 2 As shown, the system is powered by a hybrid energy source consisting of a battery and a supercapacitor. The permanent magnet motor system consists of an inverter and a five-phase permanent magnet motor. A fault diagnosis method for a five-phase permanent magnet motor system based on an adaptive sliding mode observer is provided. This method improves the robustness of fault diagnosis through adaptive gain and reduces the complexity of parameter setting through adaptive threshold, thereby improving the overall fault diagnosis efficiency.

[0075] This embodiment presents a fault diagnosis method for a five-phase permanent magnet motor drive system based on an adaptive sliding mode observer. The steps of this method are as follows:

[0076] S1. Based on the inverter's topology, the state of the switching transistor's trigger signal, and the characteristics of the current path, establish a hybrid logic dynamic (MLD) model of the system.

[0077] S2. Based on the MLD model, establish a sliding mode observer, design an adaptive sliding mode observer with fast convergence characteristics and strong chattering suppression capability, and obtain the state space expression of the system.

[0078] S3. Obtain the observed values ​​of the current in each phase of the system through the sliding mode observer, and compare the actual measured values ​​with the observed values ​​to obtain the residual signals of the current in each phase;

[0079] S4. Normalize the residual current signals of each phase and compare them with the designed adaptive threshold. If the residual current of a certain phase exceeds the threshold, it is determined that the corresponding power transistor has an open-circuit fault. Finally, after the fault is determined, based on the residual current characteristics under different open-circuit fault modes, the specific faulty switch transistor is further located by analyzing the residual signal, and the fault identifier is immediately updated to the corresponding fault code, thus completing the accurate location of the power transistor open-circuit fault.

[0080] Preferably, the hybrid logic dynamic MLD model of the inverter in step S1:

[0081] Equation (1);

[0082] In the formula u sn , I s and e s These represent the voltage, current, and back electromotive force of each phase, respectively. L s and R s These represent the inductance and resistance of each phase, respectively.

[0083] According to Kirchhoff's voltage law, we can obtain:

[0084] Equation (2);

[0085] In the formula u sg This represents the voltage between the AC output terminal and the DC bus grounding point. u ng This represents the potential difference between the load neutral point and the DC bus ground.

[0086] Taking a five-phase permanent magnet synchronous motor as an example, since the motor adopts a star connection, the vector sum of the currents in each phase is zero; at the same time, assuming that the back electromotive force of the five-phase windings has the same amplitude and the phases are equally spaced at 72°, its back electromotive force remains balanced. Combining equation (1), we can obtain:

[0087] Equation (3);

[0088] Substituting equation (2) into equation (3), we can obtain the potential difference between the load neutral point and the DC bus grounding point as follows:

[0089] Equation (4);

[0090] Substituting equation (2) into equation (4), we get:

[0091] Equation (5);

[0092] In the formula: .

[0093] Therefore, the inverter and motor can be considered as a hybrid system with continuous state evolution driven by discrete switching actions. To facilitate the establishment of the system's MLD model, let... D i ( i =1,…,10) represents the switching signal of the i-th power transistor. The power transistor is defined as follows when it is turned on. D i =1, defined when the power transistor is turned off. D i =0. At the same time, let... σ k ( k =a,b,c,d,e) represent the direction of current in each phase, and the direction of current flowing into the motor windings is defined as positive. σ k =1), outflow is negative ( =1), σ k =0). Based on the current direction of each phase and the corresponding switching signal of the switching transistor, the phase voltage of each phase can be determined. Taking the B-phase winding as an example, combined with the above logic variables, the voltage can be obtained. u bg The mathematical expression:

[0094] Equation (6);

[0095] In the formula , .

[0096] Similarly, the voltage between the AC output terminal of other windings and the DC bus grounding point can be expressed as:

[0097] Equation (7);

[0098] Substituting equation (7) into equation (5) yields the system's discrete output vector:

[0099] Equation (8);

[0100] In the formula:

[0101] Substituting equation (8) into equation (1), we obtain the hybrid logic dynamic model of the five-phase PMSM inverter as follows:

[0102] Equation (9);

[0103] In the formula: A = diag {- R / L , - R / L , - R / L , - R / L , - R / L}, R It is the resistance of the corresponding phase. L It is the inductance of the corresponding phase; B 1= diag {-1 / L -1 / L -1 / L -1 / L -1 / L}; B 2=( U dc / L ) δ, U dc This is the DC bus voltage; i =[ i a i b i c i d i e ] T Corresponding to the five-phase stator winding current; e =[ e a e b e c e de e ] T The corresponding five back electromotive forces.

[0104] For a five-phase PMSM, it is usually... d 1 axis relative to α The lead angle of the axis is defined as θ At this point, the expression for the back electromotive force can be expressed as:

[0105] Equation (10);

[0106] In the formula ω Angular velocity, ψ f It is a permanent magnet flux chain.

[0107] Preferred, such as Figure 3 As shown, a sliding mode observer is established based on the MLD model. By designing an adaptive sliding mode observer with fast convergence characteristics and strong chattering suppression capability, the state-space expression of the system is obtained. Step S2 is as follows:

[0108] Traditional power-law reaching-law sliding mode observers typically use a constant as the reaching law, but suffer from slow reaching speeds when far from the sliding surface. To address this, an adaptive gain-based reaching-law sliding mode observer is designed. When the estimation error is small, the sliding mode gain is automatically reduced to decrease the observer's over-response to small fluctuations in the subsystem; when the estimation error increases, the sliding mode gain is increased to reduce the additional error caused by current distortion in non-faulty phases, thereby improving the observer's accuracy in fault diagnosis. The adaptive gain-based reaching law proposed in this chapter is as follows:

[0109] Equation (11);

[0110] In the formula, ; >0; α >1; k This is the observer gain.

[0111] Assume there exists a positive definite matrix. P Make If this holds true, then the residual dynamic equation is asymptotically stable and can be expressed as:

[0112] Equation (12);

[0113] In the formula: .

[0114] The Lyapunov functions selected are as follows:

[0115] Equation (13);

[0116] Differentiating equation (15) yields:

[0117] Equation (14);

[0118] From equation (12), we can see that the linear term retains negative definiteness. At the same time, because of the nonlinear sliding mode term P >0, R ( s )>0, | s | α >0, and tanh ( s )and s The same sign ensures that this term is also non-positive. Substituting the above analysis results into (14), we get:

[0119] Equation (15);

[0120] In summary, when there exists a positive definite matrix P such that... At the time of its establishment, Therefore, it can be concluded that the adaptive sliding mode observer designed by equation (13) has asymptotic stability.

[0121] Preferably, S3 specifically includes the following steps:

[0122] Under fault conditions, the varying degrees of distortion between the actual current and the observed current will cause a difference in their root mean square functions. Using this difference to replace the current residual allows for more sensitive detection of fault occurrences, while avoiding misdiagnosis caused by inverter system fluctuations, thus improving the accuracy and discriminative power of fault detection.

[0123] The root mean square primitive function is:

[0124] Equation (16);

[0125] Record fault detection factors T n ( n = a, b, c, d, e )express n Measured phase current i n With current observations The difference of the root mean square function:

[0126] Equation (17);

[0127] To reduce the computational complexity, a fault detection variable based on the current shape factor is constructed, and a normalization function is used to smoothly map the large-amplitude current residual to the [0,1] interval.G n It can effectively eliminate the influence of dimensional inconsistencies and suppress interference from abnormal mutations, thereby significantly improving the stability of fault detection indicators and the sensitivity to early faults.

[0128] Equation (18);

[0129] Meanwhile, to enhance the robustness of detection, avoid false alarms and false negatives, and reduce the false diagnosis rate, an adaptive threshold for the fault detection variable is designed using the estimated current value from an adaptive sliding mode observer. T for:

[0130] Equation (19);

[0131] Combining fault detection variables G n and adaptive threshold T The open-circuit fault detection variable Dn (n=a,b,c,d,e) can be obtained as follows:

[0132] Equation (20);

[0133] As can be seen from equation (22), the fault detection variable is determined when no fault occurs. G n It will not exceed the adaptive threshold T; however, when an open-circuit fault occurs in a power transistor in the motor drive system, its corresponding G n It will exceed T.

[0134] like Figure 4 , 5 As shown, based on the existing fault detection of motor drive system, the fault characteristics contained in the residual between the observer estimated current and the actual current under fault conditions are analyzed, and a fault location quantity based on current residual is constructed to achieve accurate location of open circuit faults in specific power transistors.

[0135] Single-tube open circuit fault:

[0136] For example, if an open-circuit fault occurs in the power transistor of bridge arm A, the positive half-wave of the output current of phase A will be missing, resulting in a residual greater than zero between the estimated and actual values ​​of the current in phase A. Based on this, fault location variables... It is set to 1. This shows that an open-circuit fault in the power transistor will cause a regular change in the polarity of the residual current, providing a valid basis for diagnosing this type of fault. Through the fault detection variable Da and the fault location variable... and auxiliary variables Since 'a' is 0, the fault number can be determined by looking up the table.

[0137] Dual-tube open circuit fault:

[0138] Two-tube open-circuit faults can be divided into single-phase two-tube open-circuit faults and out-of-phase two-tube open-circuit faults. The latter is essentially a special case of single-tube fault, so it will not be analyzed in detail; while single-phase two-tube open-circuit faults will exhibit different characteristics from single-tube faults.

[0139] For example, in the case of a single-phase two-tube fault in phase A, the fault detection variable Da is 1 when the fault is detected. Simultaneously, in the case of a single-phase two-tube fault, although the actual phase current is zero, the observer output still maintains a sinusoidal oscillation. This is mainly because the back electromotive force caused by the continuous rotation of the rotor is not affected by the stator phase failure. As the disturbance component directly input into the observer model, it always acts on the observed current value according to a sinusoidal law, and the residual exhibits a sinusoidal change. Furthermore, the polarity of the residual current changes with open-circuit faults, therefore... It will continuously jump between 1 and -1, serving as an auxiliary variable for fault diagnosis. At this point, 'a' is 1. Ultimately, by looking up the table, we can determine that the fault number is 11.

[0140] Preferably, S4 specifically includes the following steps:

[0141] Based on the existing fault detection of motor drive system, this paper analyzes the fault characteristics contained in the residual between the observer estimated current and the actual current under fault conditions, and constructs a fault location quantity based on current residual to achieve accurate location of open circuit faults in specific power transistors.

[0142] Single-tube open circuit fault:

[0143] by B Mutually D Taking an open-circuit fault as an example, the voltage after the fault can be obtained by combining the switching signal and the current flow direction. The analysis results are shown in Table 1.

[0144] Table 1: B Mutually D 3. When an open circuit fault occurs, u′ bo Truth table:

[0145]

[0146] Further derivation shows that when only B Mutually D 3. When an open-circuit fault occurs, the logic variables of each phase can be represented as:

[0147] Equation (21);

[0148] Substituting equation (23) into equation (9), we get only BMutually D 3. The phase current equation for an open-circuit fault is:

[0149] Equation (22);

[0150] As can be seen from the topology changes in the motor drive system, if an open-circuit fault occurs in the upper arm power transistor, the positive half-wave of the output current of the corresponding phase will be missing, resulting in a residual greater than zero between the estimated and actual values ​​of the current in that phase. Similarly, if an open-circuit fault occurs in the lower arm power transistor, the negative half-wave of the output current of the corresponding phase will be missing, resulting in a current residual less than zero. Therefore, an open-circuit fault will cause a regular change in the polarity of the residual current. This polarity characteristic can be extracted as a fault location variable, defined as follows:

[0151] Equation (23);

[0152] Through fault detection variables D n and fault location variables By combining these, the final fault diagnosis variables can be obtained. Flag n for:

[0153] Equation (24);

[0154] Dual-tube open circuit fault:

[0155] Two-tube open-circuit faults can be divided into single-phase two-tube open-circuit faults and out-of-phase two-tube open-circuit faults. The latter is essentially a special case of single-tube fault, so it will not be analyzed in detail; while single-phase two-tube open-circuit faults will exhibit different characteristics from single-tube faults.

[0156] Taking a single-phase double-tube fault in phase B of the motor drive system as an example, under the single-phase double-tube fault condition, although the actual phase current of the missing phase is zero, the observer output still maintains sinusoidal oscillation. This is mainly because the back electromotive force caused by the continuous rotation of the rotor is not affected by the stator phase failure. As the disturbance component directly input into the observer model, it always acts on the observed value of the current in a sinusoidal manner, and the residual shows a sinusoidal change. At the same time, it can be seen from equation (25) that the polarity of the residual current will change with the open circuit fault, therefore It will continuously jump between 1 and -1. Therefore, a fault detection variable is selected. D n as well as Will Changes between them serve as auxiliary variables for fault diagnosis. n is:

[0157] Equation (25);

[0158] Table 2 summarizes the fault characteristics of residual current under different open-circuit faults:

[0159]

[0160] This embodiment proposes a fault diagnosis method for a five-phase permanent magnet motor drive system based on an adaptive sliding mode observer. First, a hybrid logic dynamics (MLD) model of the system is established, considering the inverter topology, switching transistor trigger signal states, and current path characteristics. A sliding mode observer is then constructed based on this model. By designing an adaptive sliding mode observer with fast convergence and strong chatter suppression capabilities, the state-space expression of the system is obtained. The observed values ​​of the current in each phase are acquired using this sliding mode observer, and the actual measured values ​​are compared with the observed values ​​to obtain the residual signals of the current in each phase. Second, the residual signals of the current in each phase are normalized and compared with a designed adaptive threshold. If the residual current in a certain phase exceeds the threshold, an open-circuit fault is determined in the corresponding power transistor. Finally, after fault determination, based on the residual current characteristics under different open-circuit fault modes, the specific faulty switching transistor is further located by analyzing the residual signals, and the fault identifier is immediately updated to the corresponding fault code, completing the accurate location of the power transistor open-circuit fault. This invention employs an adaptive gain and adaptive threshold strategy, which reduces the complexity of parameter settings, enhances the robustness of the system, improves fault diagnosis efficiency, and does not require additional hardware costs.

[0161] In one embodiment, a fault diagnosis system for a five-phase permanent magnet motor based on adaptive sliding mode observation is provided, the system comprising:

[0162] Model building module: Based on the inverter topology, switching transistor trigger signal status and current path characteristics of the five-phase motor drive system, a hybrid logic dynamic model is established;

[0163] Design module: used to establish a sliding mode observer based on the hybrid logic dynamic model, design an adaptive sliding mode observer with fast convergence characteristics and strong chatter suppression capability, and obtain the state space expression of the five-phase motor drive system;

[0164] Acquisition module: used to acquire the observed values ​​of the current of each phase of the five-phase motor drive system through the adaptive sliding mode observer, and compare the actual measured values ​​with the observed values ​​to obtain the residual signals of the current of each phase;

[0165] Normalization processing module: used to normalize the residual signals of each phase current and compare them with the designed adaptive threshold;

[0166] Judgment module: If the target phase current residual exceeds the threshold, it is determined that the power transistor corresponding to the target phase has an open circuit fault; wherein, the target phase current residual is any one of the phase current residual signals.

[0167] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A method for fault diagnosis of five-phase permanent magnet motor based on adaptive sliding mode observation, characterized in that, The method comprises: According to the topology structure of the inverter of the five-phase motor driving system, the switching tube trigger signal state and the current path characteristics, a hybrid logic dynamic model is established; According to the hybrid logic dynamic model, a sliding mode observer is established, an adaptive sliding mode observer with fast convergence characteristics and strong chattering suppression capability is designed, and a state space expression of the five-phase motor driving system is obtained; The observation value of each phase current of the five-phase motor driving system is obtained through the adaptive sliding mode observer, and the actual measured value is compared with the observation value to obtain a residual error signal of each phase current; Specifically, the following steps are included: Constructing a root mean square primitive equation wherein, ; a malfunction detection factor T n ( n = a, b, c, d, e ) represents n a phase current measured value i n a difference of a root mean square function of the current observation value ;​ Constructing fault detection variables based on current shape factors G n and a normalization function is used to map the large amplitude current residual to the interval [0, 1] smoothly; wherein, ; The adaptive threshold T of the fault detection variable is designed according to the estimated current value of the adaptive sliding mode observer as follows: , G a , G b , G c , G d , G e respectively represent the fault detection variables of the abcde phases, and k represents a gain. Combining the fault detection variable G n and the adaptive threshold T, the open circuit fault detection variable D n ( n = a, b, c, d, e ) is: ; Fault detection variable in case of no fault G n Will not exceed the adaptive threshold T; while when open circuit fault occurs in the target power tube in the motor drive system, the corresponding G n Exceeds T; The residual error signal of each phase current is normalized and compared with the designed adaptive threshold value; If the target phase current residual error exceeds the threshold value, it is determined that the target phase corresponding power tube has an open circuit fault, wherein the target phase current residual error is any one of the residual error signals of each phase current.

2. The method of claim 1, wherein, The hybrid logic dynamic model is: ; wherein, derivative of the five-phase stator winding current; A diag R L R L R L R L R L , R is the resistance of the corresponding phase, L is the inductance of the corresponding phase; B 1= diag L L L L L ; B 2= U dc δ, , , D i i =1,...,10) represents the switch signal of the i-th power tube, σ k k =a,b,c,d,e) represents the direction of each phase current, and the direction is positive when the current flows into the motor winding σ k =1), and negative σ k =0) U dc is the DC bus voltage; i [ i a i b i c i d i e ] T corresponds to the five-phase stator winding current; e [ e a e b e c e d ​​​​​​​​​​​​​​​​​​ e e ] T corresponding five-phase back emf; The five-phase motor drive system of claim 1, wherein the five-phase motor drive system is configured to d 1 axis relative to α the lead angle of the axis is defined as θ , and the back electromotive force is expressed as: ; where ω is the angular velocity, is the permanent flux linkage.

3. The method of claim 1, wherein, the adaptive sliding mode observer is: ; wherein, denotes a derivative of the adaptive sliding mode observer estimate of the output current, denotes the adaptive sliding mode observer estimate of the output current, B denotes an input matrix, u represents input variable, denotes an estimate of an output variable, f S denotes an adaptive reaching rate, the sliding mode surface is defined as , y denotes an output variable, C denotes a unit coefficient matrix.​ 4. The method of claim 1, wherein, The target phase corresponding power tube open circuit fault specifically includes single tube open circuit fault and double tube open circuit fault.

5. The method of claim 4, wherein, The single tube open circuit fault determination includes: When only B Phase D 3 The open-circuit fault phase current equation is: , For B Phase D The derivative of each phase current after the open-circuit fault of phase 3; For B Phase D Each phase current after the open-circuit fault of phase 3 If the upper bridge arm power tube has an open circuit fault, the positive half wave of the output current of the corresponding phase will be missing, resulting in a residual error between the estimated value and the actual value of the current of the corresponding phase being greater than zero; if the lower bridge arm power tube has an open circuit fault, the negative half wave of the output current of the corresponding phase will be missing, resulting in the current residual error being less than zero; the current polarity feature is extracted as a fault positioning variable , ; By combining the open-circuit fault detection variable D n and the fault location variable , the target fault diagnosis variable Flag n ; wherein .

6. The method of claim 5, wherein, The double tube open circuit fault includes single-phase double tube open circuit fault and different-phase double tube open circuit fault.

7. The method of claim 6, wherein, The single-phase double tube open circuit fault determination is: In the single-phase double-pipe fault case, open-circuit fault detection variables are selected D n and fault location variables are changed between as auxiliary fault diagnosis variables , .

8. The method of claim 1, wherein, The method further comprises: after determining the fault, according to the residual error current characteristic law under different open circuit fault modes, the residual error signal is analyzed to locate to the specific fault switching tube, and the fault identification is immediately updated to the corresponding fault code to complete the accurate positioning of the power tube open circuit fault.

9. A five-phase permanent magnet motor fault diagnosis system based on adaptive sliding mode observation, characterized in that, The system comprises: A model establishing module: according to the topology structure of the inverter of the five-phase motor driving system, the switching tube trigger signal state and the current path characteristics, a hybrid logic dynamic model is established; A design module: for establishing a sliding mode observer according to the hybrid logic dynamic model, designing an adaptive sliding mode observer with fast convergence characteristics and strong chattering suppression capability, and obtaining a state space expression of the five-phase motor driving system; An acquisition module: for obtaining the observation value of each phase current of the five-phase motor driving system through the adaptive sliding mode observer, and comparing the actual measured value with the observation value to obtain a residual error signal of each phase current; Specifically for: Constructing a root mean square primitive equation wherein, ; a malfunction detection factor T n ( n = a, b, c, d, e ) represents n a phase current measured value i n a difference value of a root mean square function of the current observation value ;​ Constructing fault detection variables based on current shape factors G n and a normalization function is used to map the large amplitude current residual to the interval [0, 1] smoothly; wherein, ; The adaptive threshold T of the fault detection variable is designed according to the estimated current value of the adaptive sliding mode observer as follows: , G a , G b , G c , G d , G e respectively represent the fault detection variables of the abcde phases, and k represents a gain. Combining the fault detection variables G n and the adaptive threshold T, the open-circuit fault detection variable D n ( n = a, b, c, d, e ) is: ; Fault detection variable in case of no fault G n does not exceed the adaptive threshold T; while when an open circuit fault occurs in the target power tube in the motor drive system, the corresponding G n exceeds T; A normalization processing module: for normalizing the residual error signal of each phase current and comparing it with the designed adaptive threshold value; A determination module: if the target phase current residual error exceeds the threshold value, it is determined that the target phase corresponding power tube has an open circuit fault; wherein the target phase current residual error is any one of the residual error signals of each phase current.

Images