WO2017114001A1

WO2017114001A1 - Predictive control-based open-circuit fault diagnosis method for matrix converter switch

Info

Publication number: WO2017114001A1
Application number: PCT/CN2016/105673
Authority: WO
Inventors: 彭涛; 但汉兵; 邓慧; 孙尧; 粟梅; 杨建�; 韩华; 朱奇; 王春生; 桂卫华
Original assignee: 中南大学
Priority date: 2015-12-28
Filing date: 2016-11-14
Publication date: 2017-07-06
Also published as: CN105548792B; CN105548792A

Abstract

A predictive control-based open-circuit fault diagnosis method for matrix converter switch comprises the following steps: when a matrix converter functions normally, establishing a relationship model between an output voltage and an input voltage, and between an input current and a load current (S1); determining a count of combined switch states of the matrix converter (S2); establishing a state-space model to obtain predicted values of the load current, the input current and the input voltage in next sampling period (S3, S4); determining an evaluation function (S5); adopting, in each sampling period, a finite control set-model predictive control (FCS-MPC) strategy, to select, among the combined switch states, a switch state in which an evaluation function value is minimum as a switch state for a next sampling period (S6); establishing a state-space model to obtain a variation rule of a fault phase load current after occurrence of an open-circuit fault (S7); and performing online monitoring of the load current, and performing, according to the switch state selected by the FCS-MPC strategy, a fault diagnosis to identify a location of a breakdown switch (S8). The invention enables a diagnosis of an open-circuit fault in time.

Description

Matrix converter open-circuit fault diagnosis method based on predictive control

Technical field

The invention relates to the field of circuit fault diagnosis, in particular to a matrix converter open-circuit fault diagnosis method based on predictive control.

Background technique

In recent years, with the rapid development of power electronics technology, many new power converters have appeared. In many AC-DC converters, the matrix converter can realize the transformation of all AC parameters (phase number, phase, amplitude, frequency). And with its unique advantages, it has attracted the attention of a large number of scholars. Compared with the traditional converter, the matrix converter has the following advantages: no intermediate DC energy storage, four-quadrant operation, compact power supply design, excellent input current waveform and output voltage waveform, energy bidirectional flow, volume Small, lightweight and free to control the input power factor. In addition, since the matrix converter has no electrolytic capacitor, it has a long life and can be widely applied to fields requiring high volume and weight, such as aerospace, military, and medical.

However, the switching device of the matrix converter system is under high-frequency modulation for a long time, the heat is more serious, the switching loss is large, and it is also prone to failure. When an open circuit fault occurs in a switching device, if it cannot be found and processed in time, it is easy to damage the entire matrix converter system. Especially after long-term operation, it will seriously affect the running performance of the system and the service life of the equipment. Therefore, fault diagnosis and fault tolerance for matrix converter systems are essential.

Predictive control has been applied to power electronic converters in recent years because it has many advantages, such as fast dynamic response, clear thinking, and simple constraints, which can be flexibly applied to many systems. However, the idea of predictive control for matrix converter fault diagnosis and fault tolerance has not been seen yet. Therefore, it is necessary to design a matrix converter open-circuit fault diagnosis method based on predictive control, timely diagnose the occurrence of system faults and take corresponding fault-tolerant measures to improve the reliability and service life of the matrix converter system.

Summary of the invention

The object of the present invention is to provide a matrix converter open-circuit fault diagnosis method based on predictive control, which solves the technical problem that the switching device of the matrix converter cannot be found in time when an open circuit fault occurs.

To achieve the above object, the present invention provides a matrix converter switch open circuit fault diagnosis method based on predictive control, comprising the following steps:

S1: in the case that the matrix converter works normally, the output voltage, the input voltage, the input current and the load current of the matrix converter are collected according to the sampling period, and a relationship model between the output voltage and the input voltage, the input current and the load current is established;

S2: determining the number of all switch combination states of the matrix converter if the constraint is satisfied;

S3: establishing a state space model of the load transformer load end, and obtaining a predicted value of the load current in the next sampling period;

S4: establishing a state space model of the input filter of the matrix converter, and discretizing, obtaining a predicted value of the input current and the input voltage of the next sampling period;

S5: determining an evaluation function;

S6: using a finite set model predictive control strategy, in each sampling period, selecting a switch state that minimizes the evaluation function value in all switch combination states as a switch state of the next sampling period;

S7: Establish a state space model of the equivalent circuit under the open circuit fault condition, and obtain the change law of the fault phase load current after the open circuit fault of the switch;

S8: On-line monitoring of the load current. When the load current continues to be zero, the fault state is determined according to the finite set model prediction control switch state, and the position of the fault switch is identified.

As a further improvement of the invention:

Preferably, after the step S8 is completed, the method further includes:

S9: after identifying the position of the fault switch, remove the switch state related to the action of the fault switch in all switch combination states, and determine the number of all remaining switch combination states;

S10: re-determining the evaluation function under the specific fault in all the remaining switch combination states;

S11: Using a finite set model predictive control strategy, in each sampling period, a switch combination state that minimizes the evaluation function value is selected among all remaining switch combination states as a switch state of the next sampling period.

Preferably, the relationship between the output voltage and the input voltage, input current and load current is as follows:

Where u _oA , u _oB and u _oC are the output voltages of the three phases A, B and C respectively; u _ea , u _eb and u _ec are the input voltages of the three phases A, B and C respectively; i _ea , i _eb and i _ec is the input current of three phases A, B and C respectively; i _oA , i _oB and i _oC are the load currents of three phases A, B and C respectively; S is the switching state matrix; the element S _{XY in} S (X ∈{A, B, C}, Y∈{a, b, c}) indicates the switching state of the switch, S _XY =1 indicates that the switch S _XY is closed, and S _XY =0 indicates that the switch S _XY is off. .

Preferably, the constraints include:

S201: the input end is not short-circuited, and the output end is continuous;

S202: at any time, among the three bidirectional switches connected to the same phase output, one and only one switch is turned on, and the other two switches are turned off;

S203: satisfying non-negativeity of duty ratios of the respective switches;

Under the above constraints, each element in the switch state matrix S satisfies:

The state of all switch combinations in the normal operation of the matrix converter is 3*3*3=27.

Preferably, step S3 comprises the following steps:

S301: Establish a state space model of the matrix converter load end as follows:

Where i _o and u _o are the load current and the output voltage, respectively; L and R represent the inductance and resistance of the load, respectively;

S302: Define a switching period T _s , and obtain a predicted value i _o ^k+1 of the load current in the next sampling period according to the forward Euler's formula:

Where i _o ^k and u _o ^k are the load current value and the output voltage value at the current sampling time, respectively; the coefficients F ₁ and F ₂ are:

Preferably, step S4 comprises the following steps:

S401: Establish a filter state space model at the input end of the matrix converter as follows:

Where u _s , u _e , i _s and i _e represent the supply voltage, the filter capacitor voltage, the supply current and the input current, respectively; R _i , L _i and C _i represent the filter resistor, the power supply and the filter inductor, and the filter capacitor, respectively;

S402: Combining (9) and (10) to obtain:

among them:

S403: discretizing (11) to obtain:

In the middle

with

Representing the filter capacitor voltage and the supply current value in the next sampling period;

with

The power supply voltage, filter capacitor voltage, power supply current, and input current value at the current sampling time are respectively indicated; and:

S404: Substituting (15) into equation (14) to obtain an input current and a filter capacitor voltage for the next sampling period:

Where E ₂ , E ₃ , D ₂ and D ₃ are elements of the matrix G, and E ₁ , E ₄ , D ₁ and D ₄ are elements of the matrix H, by R _i , L _i , C _i and the sampling period T _s decided.

Preferably, step S5 includes the following steps:

S501: determining an evaluation function related to an input current of the matrix converter;

The reference value of the matrix converter input current is:

Where φ is the phase angle of the input voltage,

Reference input current

The magnitude and have:

Where U _sm is the magnitude of the supply voltage u _s ; η is the efficiency of the matrix converter, obtained by the following formula:

P _o and P _in represent the output power and input power, respectively, which are obtained by the following formula:

Where I ^* _om represents the magnitude of the reference load current i ^* _o ; the expected input current i _{s has a} power factor of 1, so θ is 0;

S502: determining the predictive control on the input current i _s reaches unity input power factor is desirable to evaluate the performance as a function of g _1:

_Wherein, Δi sα and _Δi sβ represent the input current two-phase stationary coordinate system in the next sampling period of a predicted value with a reference value i _s ^p i α and β components _s ^* deviation, obtained by the following equation:

In the formula, T _abctoαβ is a transformation matrix of a three-phase stationary coordinate system converted into a two-phase rotating coordinate system, which is obtained by:

S503: Determine an evaluation function related to the matrix converter load current in the predictive control:

The evaluation function g ₂ defining the expected value of the load current predicted to achieve the tracking reference load current is:

_Wherein, Δi oα and _Δi oβ represent load current two-phase stationary coordinate system in the next sampling period of a predicted value with a reference value i _o ^p i _o ^* deviation α and β components, obtained by the following formula:

S504: Define the total evaluation function as:

g=λ*g ₁ +g ₂ (28)

Where λ is a weighting factor, indicating the importance of the performance index of the input current in the control of the entire matrix converter system;

Then, in step S6, specifically: during the operation of the matrix converter, the finite set model prediction control strategy is adopted, that is, the g value of the total evaluation function of all switch combination states is calculated in each sampling period, and the g value is selected to be minimized. A switching state, as the switching state of the next sampling period, achieves the operation of achieving the desired dynamic performance and steady state performance specifications of the system.

Preferably, step S7 includes the following steps:

S701: It is assumed that the following conditions S7011 to S7013 are established:

S7011: The sampling period is T _s ;

S7012: The input voltage period is divided into six sectors in which I, II, III, IV, V and VI are sequentially arranged, the input voltage phase angle is in the sector V and is about to enter the sector VI; the reference current value i of the A output phase ^* _oA and the actual value i _oA are positive values;

S7013: S _Aa has an open circuit fault. During the k ^th sampling period T _s ^k , the predictive controller always turns on the S _Aa turn-on signal in order to reduce the deviation between the actual value of the A output phase current and the reference value;

Then, the A output phase is open, and the A phase load current i _oA flows through the clamp circuit. The circuit state space model is described as follows:

Where u _N is the voltage between the load neutral point and the power ground point;

The voltage of the output phase is expressed as follows:

Where U _cp is the capacitance voltage of the clamp circuit, and u _max is the maximum value of the input three-phase voltage;

Since the phase A load current flows through the clamp circuit, U _{cp is} much larger than u _max , which is expressed as:

Where u _min represents the minimum value of the input three-phase voltage;

According to the equations (29) and (30), the rate of change of the phase A load current i _oA is obtained as follows:

Available from (31):

Substituting (33) into (32) gives:

When i _oA >0 (34)

S703: Construct energy function v:

The rate of change is expressed as:

So, get:

When i _oA <0 (37)

It can be known from equations (34) and (37) that when the fault switch satisfies S _Aa =1, the amplitude of the phase A load current will become smaller and smaller, so that the deviation between the actual value and the reference value becomes more and more Large; because S _{Aa is} in the sector VI and I connected to the A corresponding output of the maximum input voltage, the predictive controller always turns on the switch S _Aa , however, S _Aa cannot be turned on because of the fault, the result is the A phase load current amplitude The value decreases and S _Aa =1 for several consecutive sampling periods; the amplitude of the phase A load current gradually decreases to zero within a certain period of time;

S704: When the input voltage phase is in the sectors III and IV, the reference current value i ^* _oA and the actual value _{ioA of the} A output phase are both negative values, and the energy function change rate at this time is expressed as:

The amplitude of the phase A load current will also decrease to zero for a certain period of time.

Preferably, step S8 includes the following steps:

S801: When the duration T _zero of the load current remains _zero reaches the time required for fault diagnosis, the fault state is determined according to the finite set model prediction control switch state, the fault switch position, the fault condition and the input voltage period and the load are identified. The current fundamental period is related to T _o and the duration of the fault is expressed as:

Wherein T _in and T _o represent the input voltage fundamental period and the load current fundamental period, respectively;

The time T _discharge required to define the load current of the fault phase to zero by the clamp circuit, and the time T _{zero at} which the fault phase current remains zero is expressed as:

T _zero =T _fault -T _discharge (39)

The troubleshooting time needs to be met:

k _f T _s >T _ε (40)

Where k _f is the sensitivity coefficient of fault diagnosis; T _s is the zero-crossing of the reference current of the output phase under normal operation (the reference current of the output phase is less than the threshold ε); the threshold ε is to reduce the false alarm rate of the fault diagnosis. a small positive value set;

Comprehensive (38) ~ (40), the time required for fault diagnosis should meet:

S802: Perform fault diagnosis according to the limited set model predictive control switch state, identify the fault switch position, and _assume that the open fault of the _{SA Aa} , the logic of the fault switch positioning may be expressed as:

If |i _oA | ≤ ε and S _Aa =1 are satisfied in several consecutive sampling periods, then S _Aa has an open circuit fault, namely:

Preferably, when S _{Aa is} open in step S9, the constraint of the operation of the matrix converter is:

Therefore, the remaining switch combination states are 2*3*3=18; when the other switches are open, and so on;

In step S10, in all the remaining switch combination states, the evaluation function related to the input current of the matrix converter is not considered, that is, the value of λ in the total evaluation function of (28) is zero, and the evaluation function under the specific fault is re-determined as :

g ^F = g ₂ (44)

In step S11, the matrix converter with fault during operation, the finite set of model predictive control method, i.e., calculated at each sampling period g ^F values for all remaining combinations of state switches, selecting the minimum value which the switching g ^F The combined state is used as the switching state of the next sampling period; the fault-tolerant operation that satisfies the dynamic and steady-state performance indexes of the system after the fault is realized.

The invention has the following beneficial effects:

1. The predictive control based matrix converter open-circuit fault diagnosis method of the present invention monitors the three-phase load current in real time, and determines whether the output of the matrix converter system is faulty by judging that the output phase current remains zero, and predicts the controller according to the time. The switch state locates the fault switch. The invention can diagnose the open circuit fault in time, so that the controller can process in time, thereby prolonging the service life.

2. The predictive control based matrix converter open-circuit fault diagnosis method of the present invention can further perform fault-tolerant processing on faults. When an open-circuit fault occurs in a switch of the matrix converter, a new switch state combination is formed by using the remaining normal switches. Then, continue to use the predictive control idea to select the most appropriate switch state, so that the matrix converter system continues to operate in the case of a single switch open fault. The invention can avoid device damage caused by open circuit failure and further prolong the service life of the device.

In addition to the objects, features and advantages described above, the present invention has other objects, features and advantages. The invention will now be described in further detail with reference to the accompanying drawings.

DRAWINGS

The accompanying drawings, which are incorporated in the claims In the drawing:

1 is a flow chart showing a method for predicting open-circuit fault of a matrix converter based on predictive control according to a preferred embodiment of the present invention;

2 is a topological structural view of a matrix converter in accordance with a preferred embodiment of the present invention;

3 is a flow chart showing a model predictive control strategy of a matrix converter in accordance with a preferred embodiment of the present invention;

4 is an experimental waveform diagram of input/output voltage and input/load current of a matrix converter under normal operating conditions in accordance with a preferred embodiment of the present invention;

5 is a schematic diagram showing a period of a fundamental voltage period of an input voltage of a matrix converter according to a preferred embodiment of the present invention; wherein, Voltage is a voltage; and Sector is a sector;

6 is a schematic diagram showing an equivalent circuit of a matrix converter switch SAa under a fault condition according to a preferred embodiment of the present invention;

7 is a flow chart of the open circuit fault diagnosis of the matrix converter of the preferred embodiment of the present invention;

8 is a flow chart of a matrix converter switch open fault fault tolerant strategy in accordance with a preferred embodiment of the present invention;

9 is a schematic diagram showing a comparison of load current waveforms before and after adding diagnostic and fault-tolerant measures under the condition that an open circuit fault occurs in a matrix converter of a preferred embodiment of the present invention.

detailed description

The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

Referring to FIG. 1, a method for predicting open-circuit fault of a matrix converter based on predictive control according to an embodiment of the present invention includes the following steps:

S5: determining an evaluation function;

S8: On-line monitoring of load current, predictive control selection based on finite set model when load current continues to zero The switch status is diagnosed and the position of the fault switch is identified.

In the above steps, by monitoring the three-phase load current in real time, it is judged that the output phase current remains zero (in practical applications, keeping zero can be judged by whether its duration exceeds the threshold), detecting whether the matrix converter system has a fault, and according to At that time, the controller's switch state was positioned to locate the fault switch, which can diagnose the open circuit fault in time.

In practical applications, based on the above steps, the predictive control based matrix converter open-circuit fault diagnosis method according to another preferred embodiment of the present invention increases the fault-tolerant operation steps for optimization:

That is, after the above step S8 is completed, the following steps are performed:

The preferred embodiment can further perform fault tolerance processing on the fault. When an open circuit fault occurs in a switch of the matrix converter, the remaining normal switch is used to form a new switch state combination, and then the predictive control idea is used to select the most suitable switch state. The matrix converter system continues to operate in the event of a single switch open circuit failure. The embodiment of the invention can avoid device damage caused by open circuit failure and further prolong the service life of the device.

The above embodiment is applied to the matrix converter of the topology shown in FIG. 2, and the open-circuit fault of the switch S _{Aa of the} matrix converter is taken as an example to further analyze the present invention:

When the matrix converter is in normal operation, its main circuit topology is shown in Figure 2. The process of predicting the control switch state by the matrix converter finite set model is shown in Figure 3. The experimental parameters are shown in Table 1.

Table 1. Experimental parameter table

The predictive control based matrix converter open-circuit fault diagnosis method according to the embodiment of the invention comprises the following steps:

S1: In the case of normal operation of the matrix converter, the output voltage, the input voltage, the input current and the load current of the matrix converter are collected according to the sampling period, and the relationship between the output voltage and the input voltage, the input current and the load current is established. The type is as follows:

S2: Determine the number of all switch combination states of the matrix converter if the constraints are met. The constraints are set up to ensure the safe operation of the matrix converter. Constraints include the following:

S201: the input end is not short-circuited, and the output end is continuous;

S203: satisfying non-negativeity of duty ratios of the respective switches;

S3: Establish a state space model of the load transformer load end to obtain a predicted value of the load current in the next sampling period. Specifically, the following steps are included:

S4: Establish a state space model of the input filter of the matrix converter. After discretization, obtain the predicted value of the input current and the input voltage for the next sampling period. Includes the following steps:

S402: Combining (9) and (10) to obtain:

among them:

S403: discretizing (11) to obtain:

In the middle

with

S5: Determine the evaluation function, which specifically includes the following steps:

S501: Determine an evaluation function related to an input current of the matrix converter.

The reference value of the matrix converter input current is:

Where φ is the phase angle of the input voltage,

Reference input current

The magnitude and have:

S504: Define the total evaluation function as:

g=λ*g ₁ +g ₂ (28)

Where λ is the weighting factor, indicating the importance of the performance index of the input current in the control of the entire matrix converter system.

S6: In the operation of the matrix converter, the finite set model predictive control strategy is adopted, and the g value of the total evaluation function of all switch combination states is calculated in each sampling period, and a switch state in which the g value can be minimized is selected as The switching state of the next sampling period achieves the operation of the system with the desired dynamic performance and steady state performance specifications. Figure 4 shows the input/output voltage and input/load current experimental waveforms of the matrix converter under normal operating conditions. As can be seen from Figure 4, the use of a predictive control strategy to select the appropriate switching state can sine the input/load current of the matrix converter and track the reference well well with an input power factor of one.

The following steps are the real-time fault detection and diagnostic steps for the matrix converter system:

S7: Establish a state space model of the equivalent circuit under the open circuit fault condition, and obtain the change law of the fault phase load current after the open circuit fault of the switch, which specifically includes the following steps:

S7011: The sampling period is T _s ;

S7012: As shown in FIG. 5, the input voltage period is divided into six sectors in which I, II, III, IV, V, and VI are sequentially arranged, and the input voltage phase angle is in the sector V and is about to enter the sector VI; The reference current value i ^* _oA and the actual value i _{oA of the} phase are both positive values;

S7013: S _Aa has an open circuit fault. During the k ^th sampling period T _s ^k , the predictive controller always turns on the S _Aa turn-on signal in order to reduce the deviation between the actual value of the A output phase current and the reference value.

Then, the A output phase is open, and the A phase load current i _oA flows through the clamp circuit, as shown in Figure 6(a). Its circuit state space model is described as follows:

Where u _N is the voltage between the load neutral point and the power supply ground.

The voltage of the output phase is expressed as follows:

Where U _cp is the capacitance voltage of the clamp circuit, and u _max is the maximum value of the input three-phase voltage.

Where u _min represents the minimum value of the input three-phase voltage.

Available from (31):

Substituting (33) into (32) gives:

When i _oA > 0 (34).

S703: Construct energy function v:

The rate of change is expressed as:

So, get:

When i _oA <0 (37)

It can be known from equations (34) and (37) that when the fault switch satisfies S _Aa =1, the amplitude of the phase A load current will become smaller and smaller, so that the deviation between the actual value and the reference value becomes more and more Large; because S _{Aa is} in the sector VI and I connected to the A corresponding output of the maximum input voltage, the predictive controller always turns on the switch S _Aa , however, S _Aa cannot be turned on because of the fault, the result is the A phase load current amplitude The value decreases and S _Aa =1 for several consecutive sampling periods; the amplitude of the phase A load current gradually decreases to zero for a certain period of time.

S704: When the input voltage phase is in the sectors III and IV, the reference current value i ^* _oA and the actual value _{ioA of the} A output phase are both negative values, and the equivalent circuit is as shown in FIG. 6(b). The rate of change of the energy function is expressed as:

Step S8 includes the following steps:

T _zero =T _fault -T _discharge (39).

In order to distinguish the fault phase from the situation where the current remains at zero when the fault occurs and the current crosses zero at normal, the fault diagnosis time needs to be satisfied:

k _f T _s >T _ε (40)

Comprehensive (38) ~ (40), the time required for fault diagnosis should meet:

S802: Perform fault diagnosis according to the finite set model prediction control switch state, identify the fault switch position, and assume that the open fault of the _{SA Aa} , the logic of the fault switch positioning may be expressed as:

Figure 7 is a flow chart showing the real-time fault detection and diagnosis steps of the above matrix converter system. The fault diagnosis time is 1.4ms.

The following are the steps to perform a fault-tolerant operation after identifying a fault:

S9: After identifying the position of the fault switch, the switch states related to the action of the fault switch in all switch combination states are removed, and the number of remaining switch combination states is determined. For example, when S _{Aa is} open, the constraints of the operation of the matrix converter are:

S10: In all the remaining switch combination states, the evaluation function related to the input current of the matrix converter is not considered, that is, the value of λ in the total evaluation function of (28) is zero, and the evaluation function under the specific fault is re-determined as:

g ^F = g ₂ (44)

S11: In the course of operation with fault matrix converter using a finite set of model predictive control methods, see FIG. 8, i.e., the calculated value of ^F g of all the remaining switches in the combined state of each sampling period, in which a selected minimum value g ^F The switch combination state is used as the switch state of the next sampling period; the fault-tolerant operation that satisfies the dynamic and steady-state performance indexes of the system after the fault is realized.

Figure 9 shows a comparison of the load current waveforms before and after the diagnostic and fault-tolerant measures are added to the matrix converter in the event of an open-circuit fault. It can be seen from Fig. 9(a) that when the matrix converter system is in normal operation in the early stage, the load current is basically sinusoidal, the harmonic distortion rate is very low, and the reference value is well tracked. After the S _Aa switch open circuit fault is introduced, the load current i _oA occurs. Distortion and rapid decay to zero, i _oB and i _oC are also relatively large. As shown in Fig. 9(b), after the fault diagnosis is added, after the introduction of the S _Aa switch open circuit fault, the predictive controller always turns on the S _Aa signal for several consecutive sampling periods. However, the _{SA Aa} cannot be guided due to the fault. Therefore, the current of the A output phase will remain at zero for several consecutive sampling periods, and an open circuit fault of S _Aa can be diagnosed at this time, and the fault diagnosis process can be more clearly seen in FIG. 9(c). After the fault switch is located, fault tolerance measures are added, and the switch combination state in which the evaluation function value can be minimized is selected in the remaining switch combination states by the predictive control. As can be seen from the figure, the three-phase load current of the matrix converter can be substantially balanced and better track the reference sinusoidal current during fault-tolerant operation.

Table 2 shows the THD comparison table for normal, fault, and fault tolerant operation of the matrix converter.

Table 2

It can be seen from Table 2 that after the open circuit fault of the switch S _Aa , the fundamental component of the phase load A of the matrix converter is significantly reduced, the harmonic distortion rate is significantly increased, and the waveform quality of the load current of the two phases B and C is also affected. A relatively large impact. In the fault-tolerant operation state of the embodiment of the invention, the fundamental wave components of the load currents of the three phases A, B and C are increased, the harmonic distortion rate is decreased, and the waveform quality is improved, which further proves the invention. Effectiveness.

The above description is only the preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes can be made to the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and scope of the present invention are intended to be included within the scope of the present invention.

Claims

A matrix converter switch open circuit fault diagnosis method based on predictive control, characterized in that it comprises the following steps:

S1: in the case that the matrix converter works normally, the output voltage, the input voltage, the input current and the load current of the matrix converter are collected according to the sampling period, and a relationship model between the output voltage and the input voltage, the input current and the load current is established;

S2: determining the number of all switch combination states of the matrix converter if the constraint is satisfied;

S3: establishing a state space model of the load transformer load end, and obtaining a predicted value of the load current in the next sampling period;

S4: establishing a state space model of the input filter of the matrix converter, and discretizing, obtaining a predicted value of the input current and the input voltage of the next sampling period;

S5: determining an evaluation function;

S6: using a finite set model predictive control strategy, in each sampling period, selecting a switch state that minimizes the evaluation function value in all switch combination states as a switch state of the next sampling period;

S7: Establish a state space model of the equivalent circuit under the open circuit fault condition, and obtain the change law of the fault phase load current after the open circuit fault of the switch;

S8: On-line monitoring of the load current. When the load current continues to be zero, the fault state is determined according to the finite set model prediction control switch state, and the position of the fault switch is identified.
The fault diagnosis method according to claim 1, wherein after the step S8 is completed, the method further comprises:

S9: after identifying the location of the fault switch, removing the switch states related to the action of the fault switch in all switch combination states, and determining the number of remaining switch combination states;

S10: re-determining the evaluation function under the specific fault in all the remaining switch combination states;

S11: Using a finite set model predictive control strategy, in each sampling period, a switch combination state that minimizes an evaluation function value is selected among all remaining switch combination states as a switch state of a next sampling period.
The fault diagnosis method according to claim 1 or 2, wherein the relationship between the output voltage and the input voltage, the input current, and the load current is as follows:

Where u oA , u oB and u oC are the output voltages of the three phases A, B and C respectively; u ea , u eb and u ec are the input voltages of the three phases A, B and C respectively; i ea , i eb and i ec is the input current of three phases A, B and C respectively; i oA , i oB and i oC are the load currents of three phases A, B and C respectively; S is the switching state matrix; the element S XY in S (X ∈{A, B, C}, Y∈{a, b, c}) indicates the switching state of the switch, S XY =1 indicates that the switch S XY is closed, and S XY =0 indicates that the switch S XY is off. .
The fault diagnosis method according to claim 3, wherein the constraint condition comprises:

S201: the input end is not short-circuited, and the output end is continuous;

S202: at any time, among the three bidirectional switches connected to the same phase output, one and only one switch is turned on, and the other two switches are turned off;

S203: satisfying non-negativeity of duty ratios of the respective switches;

Under the above constraints, each element in the switch state matrix S satisfies:

The state of all switch combinations in the normal operation of the matrix converter is 3*3*3=27.
The fault diagnosis method according to claim 4, wherein the step S3 comprises the following steps:

S301: Establish a state space model of the matrix converter load end as follows:

Where i o and u o are the load current and the output voltage, respectively; L and R represent the inductance and resistance of the load, respectively;

S302: Define a switching period T s , and obtain a predicted value i o k+1 of the load current in the next sampling period according to the forward Euler's formula:

Where i o k and u o k are the load current value and the output voltage value at the current sampling time, respectively; the coefficients F 1 and F 2 are:
The fault diagnosis method according to claim 5, wherein the step S4 comprises the following steps:

S401: Establish a filter state space model at the input end of the matrix converter as follows:

Where u s , u e , i s and i e represent the supply voltage, the filter capacitor voltage, the supply current and the input current, respectively; R i , L i and C i represent the filter resistor, the power supply and the filter inductor, and the filter capacitor, respectively;

S402: Combining (9) and (10) to obtain:

among them:

S403: discretizing (11) to obtain:

In the middle
with
Representing the filter capacitor voltage and the supply current value in the next sampling period;
with
The power supply voltage, filter capacitor voltage, power supply current, and input current value at the current sampling time are respectively indicated; and:

S404: Substituting (15) into equation (14) to obtain an input current and a filter capacitor voltage for the next sampling period:

Where E 2 , E 3 , D 2 and D 3 are elements of the matrix G, and E 1 , E 4 , D 1 and D 4 are elements of the matrix H, by R i , L i , C i and the sampling period T s decided.
The fault diagnosis method according to claim 6, wherein the step S5 comprises the following steps:

S501: determining an evaluation function related to an input current of the matrix converter;

The reference value of the matrix converter input current is:

Where φ is the phase angle of the input voltage,
Reference input current
The magnitude and have:

Where U sm is the magnitude of the supply voltage u s ; η is the efficiency of the matrix converter, obtained by the following formula:

P o and P in represent the output power and input power, respectively, which are obtained by the following formula:

Where I * om represents the magnitude of the reference load current i * o ; the expected input current i s has a power factor of 1, so θ is 0;

S502: determining the predictive control on the input current i s reaches unity input power factor is desirable to evaluate the performance as a function of g 1:

Wherein, Δi sα and Δi sβ represent the input current two-phase stationary coordinate system in the next sampling period of a predicted value with a reference value i s p i α and β components s * deviation, obtained by the following equation:

In the formula, T abctoαβ is a transformation matrix of a three-phase stationary coordinate system converted into a two-phase rotating coordinate system, which is obtained by:

S503: Determine an evaluation function related to the matrix converter load current in the predictive control:

The evaluation function g 2 defining the expected value of the load current predicted to achieve the tracking reference load current is:

Wherein, Δi oα and Δi oβ represent load current two-phase stationary coordinate system in the next sampling period of a predicted value with a reference value i o p i o * deviation α and β components, obtained by the following formula:

S504: Define the total evaluation function as:

g=λ*g 1 +g 2 (28)

Where λ is a weighting factor, indicating the importance of the performance index of the input current in the control of the entire matrix converter system;

Then, the step S6 is specifically: in the operation of the matrix converter, the finite set model is used to predict the control strategy, that is, the g value of the total evaluation function of all the switch combination states is calculated in each sampling period, and the selection can be performed. The switch state with the smallest value, as the switch state of the next sampling period, achieves the operation of achieving the desired dynamic performance and steady state performance of the system.
The fault diagnosis method according to claim 7, wherein the step S7 comprises the following steps:

S701: It is assumed that the following conditions S7011 to S7013 are established:

S7011: The sampling period is T s ;

S7012: The input voltage period is divided into six sectors in which I, II, III, IV, V and VI are sequentially arranged, the input voltage phase angle is in the sector V and is about to enter the sector VI; the reference current value i of the A output phase * oA and the actual value i oA are positive values;

S7013: S Aa has an open circuit fault. During the k th sampling period T s k , the predictive controller always turns on the S Aa turn-on signal in order to reduce the deviation between the actual value of the A output phase current and the reference value;

Then, the A output phase is open, and the A phase load current i oA flows through the clamp circuit. The circuit state space model is described as follows:

Where u N is the voltage between the load neutral point and the power ground point;

The voltage of the output phase is expressed as follows:

Where U cp is the capacitance voltage of the clamp circuit, and u max is the maximum value of the input three-phase voltage;

Since the phase A load current flows through the clamp circuit, U cp is much larger than u max , which is expressed as:

Where u min represents the minimum value of the input three-phase voltage;

According to the equations (29) and (30), the rate of change of the phase A load current i oA is obtained as follows:

Available from (31):

Substituting (33) into (32) gives:

When i oA >0 (34)

S703: Construct energy function v:

The rate of change is expressed as:

So, get:

When i oA <0 (37)

It can be known from equations (34) and (37) that when the fault switch satisfies S Aa =1, the amplitude of the phase A load current will become smaller and smaller, so that the deviation between the actual value and the reference value becomes more and more Large; because S Aa is in the sector VI and I connected to the A corresponding output of the maximum input voltage, the predictive controller always turns on the switch S Aa , however, S Aa cannot be turned on because of the fault, the result is the A phase load current amplitude The value decreases and S Aa =1 for several consecutive sampling periods; the amplitude of the phase A load current gradually decreases to zero within a certain period of time;

S704: When the input voltage phase is in the sectors III and IV, the reference current value i * oA and the actual value ioA of the A output phase are both negative values, and the energy function change rate at this time is expressed as:

The amplitude of the phase A load current will also decrease to zero for a certain period of time.
The fault diagnosis method according to claim 8, wherein the step S8 comprises the following steps:

S801: when the duration T zero of the load current remains zero reaches the time required for fault diagnosis, and then according to the finite set model predictive control selected switch state for fault diagnosis, identifying the position of the fault switch, the fault condition and the input voltage period It is related to the load current fundamental period T o and represents the fault duration as:

Wherein T in and T o represent the input voltage fundamental period and the load current fundamental period, respectively;

The time T discharge required to define the load current of the fault phase to zero by the clamp circuit, and the time T zero at which the fault phase current remains zero is expressed as:

T zero =T fault -T discharge (39)

The troubleshooting time needs to be met:

k f T s >T ε (40)

Where k f is the sensitivity coefficient of the fault diagnosis; T s is the zero-crossing time of the reference current of the output phase under normal operation; the threshold ε is a small positive value set to reduce the false alarm rate of the fault diagnosis;

Comprehensive (38) ~ (40), the time required for fault diagnosis should meet:

S802: Perform fault diagnosis according to the finite set model prediction control switch state, identify the fault switch position, and assume that the open fault of the SA Aa , the logic of the fault switch positioning may be expressed as:

If |i oA | ≤ ε and S Aa =1 are satisfied in several consecutive sampling periods, then S Aa has an open circuit fault, namely:
A fault diagnosis method according to claim 9, wherein the step S9, S Aa when open, the constraint matrix converter is running:

Therefore, the remaining switch combination states are 2*3*3=18; when the other switches are open, and so on;

In the step S10, in all the remaining switch combination states, the evaluation function related to the input current of the matrix converter is not considered, that is, the value of λ in the total evaluation function of (28) is zero, and the evaluation under the specific fault is re-determined. The function is:

g F = g 2 (44)

In the step S11, the process operation with fault matrix converter using a finite set of model predictive control, i.e. at each sampling period is calculated F value g for all remaining combinations of state switches, in which a selected minimum value g F The switch combination state is used as the switch state of the next sampling period; the fault-tolerant operation that satisfies the dynamic and steady-state performance indexes of the system after the fault is realized.