CN104767433B - For identifying the time varying signal method of sampling of induction less brush-less motor initial position - Google Patents

Abstract

The invention discloses a kind of time varying signal method of sampling for identifying induction less brush-less motor initial position, step comprises: 1) select to power up mode; Setting cycle-index and sampling time, arranges cumulative sum and clear 0; 2) short duration voltage pulse of designated magnitude is powered up induction less brush-less motor according to the mode that powers up selected, after power High frequency filter and amplification are carried out to the changing currents with time sampled signal exported, carry out AD sampling in the sampling time, and sampled result is added to cumulative sum; 3) cycle-index is subtracted 1, judge whether cycle-index is 0, as performed step 4) for 0 redirect; If not being 0 redirect execution step 2); 4) cumulative sum is extracted, result will be extracted and export.The present invention does not need to increase hardware circuit, does not need the resolution improving AD converter, can obtain anti-interference good, the sampled result that resolution figure place is high, can not produce obvious noise of motor, rotor can not be made to rotate or shake during identification.

Description

For identifying the time varying signal method of sampling of induction less brush-less motor initial position

Technical field

The present invention relates to motor-drive technique field, be specifically related to a kind of time varying signal method of sampling for identifying induction less brush-less motor initial position, wherein induction less brush-less motor comprises brshless DC motor and permagnetic synchronous motor.

Background technology

Along with permanent magnetic material, power MOS pipe and control chip function update and perfect, the induction less brush-less motor comprising brshless DC motor and permagnetic synchronous motor obtains at industrial control field, field of household appliances and new energy field to be applied more and more widely.The brushless electric machine of relative position sensor, the induction less brush-less motor of position-sensor-free has that volume is little, cost is low, control system is not easily disturbed the remarkable advantage such as high with reliability.

As everyone knows, motor be made to obtain effective detent torque and to start smoothly, need accurately to know initial position of rotor, if the initial position of rotor obtained and actual value deviation are comparatively large, the problems such as carrying load ability decline, even reversion during electric motor starting, will be there will be.The original position of electric motor's rotator of usual position sensor obtains by transducer, and the initial position of rotor of position-sensor-free induction less brush-less motor can be determined by pre-determined bit method or location recognition usually.

Determine the method mainly pre-determined bit method of induction less brush-less original position of electric motor's rotator at present.But pre-determined bit method is higher to actuator power capacity requirement, there is the shortcomings such as motor reversal probability during starting slow, pre-determined bit is large, concussion, this is unallowed in some occasion.

Initial position of rotor identification is generally the theoretical research launched based on inductance method at present.Inductance ratio juris is: the inductance value of motor stator winding is relevant with rotor-position, by apply to any two-phase of motor short duration voltage pulse (namely adopt that the positive V of U is negative, the positive W of U is negative, the positive W of V is negative, the positive U of V is negative, the positive U of W is negative successively for the UVW three-phase of induction less brush-less motor, the positive V of W negative totally six kinds power up mode), to measure and the size of more consequent 6 current sampling signals identifies initial position of rotor.Relative to pre-determined bit method, inductance method does not need pre-aligned, but directly starts after the initial position identifying rotor.Under can accurately identifying the prerequisite of initial position of rotor, have during electric motor starting detent torque large, start fast, nonreversible when starting, the remarkable advantage such as not shake when starting.

But in actual application, there is following unfavorable factor in the initial position of rotor identification based on inductance method: A) added voltage pulse time is short, and thus current sampling signal is little; B) there is the interference such as power supply disturbance, CPU core interference and I/O mouth; C) diverse location motor inductances amount difference is small; D) ADC limited resolution.Just due to the objective reality of above-mentioned unfavorable factor, make to be difficult to obtain sampled value accurately when adopting traditional inductance method to carry out time varying signal AD sampling technically, so that accurately can not identify initial position of rotor, cause the method for inductance method identification initial position of rotor not in fact to be widely used.

Summary of the invention

The technical problem to be solved in the present invention is: for the above-mentioned technical problem of prior art, one is provided not need to increase hardware circuit, do not need the resolution improving AD converter, anti-interference is good, resolution figure place is high, obvious noise of motor can not being produced, the time varying signal method of sampling for identifying induction less brush-less motor initial position that rotor rotates or shakes during identification, can not being made.

In order to solve the problems of the technologies described above, the technical solution used in the present invention is:

For identifying a time varying signal method of sampling for induction less brush-less motor initial position, step comprises:

1) from six kinds of the UVW three-phase of induction less brush-less motor power up that the positive V of mode U is negative, the positive W of U is negative, the positive W of V is negative, the positive U of V is negative, the positive U of W is negative, the positive V of W negative selection one power up mode; Setting cycle-index N and sampling time Δ T, arranges cumulative sum SUM _iand clear 0;

2) short duration voltage pulse of designated magnitude is powered up induction less brush-less motor according to the mode that powers up selected, the pulse duration Δ T of described short duration voltage pulse when not using sampling hold circuit ₀be more than or equal to sampling time Δ T, the pulse duration Δ T of described short duration voltage pulse when using sampling hold circuit ₀be less than sampling time Δ T, after power to the current sampling resistor R of induction less brush-less motor _sNSthe changing currents with time sampled signal exported carries out High frequency filter and amplification, and carries out AD sampling in the Δ T moment in sampling time to High frequency filter and the changing currents with time sampled signal after amplifying, and sampled result is added to cumulative sum SUM _i;

3) cycle-index N is subtracted 1, judge whether cycle-index N is 0, if be 0, redirect performs step 4); If be not 0, redirect performs step 2);

4) to cumulative sum SUM _iextract, result will be extracted and export.

Preferably, described step 2) in High frequency filter specifically refer to and carry out High frequency filter with low pass filter, by adjusting the cut-off frequency f of low pass filter ₀, make the white noise amplitude comprised through the filtered signal of low pass filter be greater than 1 least significant bit LSB of AD sampler.

Preferably, the extraction in described step 4) specifically refers to 1. following ~ one 3. in three kinds of methods: 1., by cumulative sum SUM _imove to right as extracting result behind lgN/lg4 position, wherein N is the positive integer power of 4; 2., by cumulative sum SUM _idivided by cumulative sum SUM _ias extracting result after the mean value that the quantity of middle comprised sampled result obtains also retains the precision of specifying; 3., by cumulative sum SUM _idirect as extracting result.

Preferably, in described step 4) to cumulative sum SUM _ithe step of all sampled result obtained being carried out to data screening is also comprised before extracting, described data screening specifically refers to: first pass through in step 2) middle record sampled result, then step 2 is performed for cycle-index N time) the N number of sampled result of cycle-index that obtains, select maximum wherein and from cumulative sum SUM _ithis maximum of middle removal, or the minimum value selected wherein from cumulative sum SUM _ithis minimum value of middle removal, or the maximum simultaneously selected wherein and minimum value from cumulative sum SUM _ithis maximum of middle removal and minimum value.

Preferably, also comprise the step of demarcating cycle-index N before described step 1) setting cycle-index N, detailed step comprises:

1.1) receptible minimum accuracy P during setting identification initial position of rotor; Setting sampling time Δ T; The initial value of setting cyclic variable M, this initial value is more than or equal to 4; Setting is used for the testing time CNT of statistical correction rate, desired location discrimination model MOD, arranges cumulative sum SUM1 ~ SUM6 and clear 0, and the corresponding six kinds of one powered up in mode of each cumulative sum power up mode;

1.2) the rotor physical location POS of induction less brush-less motor is navigated to 1;

1.3) short duration voltage pulse of designated magnitude is powered up mode according to six kinds successively to power up induction less brush-less motor, the pulse duration Δ T of described short duration voltage pulse when not using sampling hold circuit ₀be more than or equal to sampling time Δ T, the pulse duration Δ T of described short duration voltage pulse when using sampling hold circuit ₀be less than sampling time Δ T, after power to the current sampling resistor R of induction less brush-less motor _sNSthe changing currents with time sampled signal exported carries out High frequency filter and amplification, and carry out AD sampling in the Δ T moment in sampling time to through High frequency filter and the changing currents with time sampled signal after amplifying, and sampled result is added to corresponding cumulative sum respectively by the mode that powers up, and mode powered up to each repeats step 1.3) M time altogether;

1.4) six cumulative sum SUM1 ~ SUM6 are extracted respectively, obtain six altogether and extract result; Described extraction specifically refers to 1. following ~ one 3. in three kinds of methods: as extraction result S1 ~ S6 behind the lgN/lg4 position that 1., using cumulative sum SUM1 ~ SUM6 moves to right respectively, wherein N is the positive integer power of 4; 2. the mean value (retaining enough precision), obtained divided by M respectively by cumulative sum SUM1 ~ SUM6 is afterwards as extraction result S1 ~ S6; 3., using cumulative sum SUM1 ~ SUM6 directly as extracting result S1 ~ S6.

1.5) judge whether position discrimination model MOD is 1, if so, then redirect performs step 1.6); If not, then jump to step 1.11);

1.6) arranging the intermediate variable i being used for location recognition is 0;

1.7) judge that extraction result S1 is greater than extraction result S4 and whether sets up, if set up, intermediate variable i is added 1;

1.8) judge that extraction result S2 is greater than extraction result S5 and whether sets up, if set up, intermediate variable i is added 2;

1.9) judge that extraction result S3 is greater than extraction result S6 and whether sets up, if set up, intermediate variable i is added 4;

1.10) search default location lookup table according to the value of intermediate variable i, obtain rotor recognizing site pos, jump to 1.14);

1.11) whether maximumly S1 in result S1, S2, S3 is judged to extract, if it is maximum for extracting result S1, judge that extraction result S1 is greater than S4 and whether sets up again, set up if S1 is greater than S4, make the value of rotor recognizing site pos be 1 and jump to step 1.14), be false if S1 is greater than S4, make the value of rotor recognizing site pos be 4 and jump to step 1.14); If it is not maximum for extracting result S1, perform step 1.12);

1.12) whether maximumly S2 in result S1, S2, S3 is judged to extract, if it is maximum for extracting result S2, judge that extraction result S2 is greater than S5 and whether sets up again, if extract result S2 to be greater than S5 establishment, the value of rotor recognizing site pos is made to be 2 and to jump to step 1.14), be false if extraction result S2 is greater than S5, make the value of rotor recognizing site pos be 5 and jump to step 1.14); If it is not maximum for extracting result S2, perform step 1.13);

1.13) whether maximumly S3 in result S1, S2, S3 is judged to extract, if it is maximum for extracting result S3, judge that extraction result S3 is greater than S6 and whether sets up again, if set up, the value of rotor recognizing site pos is made to be 3 and to jump to step 1.14), if be false, make the value of rotor recognizing site pos be 6 and jump to step 1.14); If it is not maximum for extracting result S3, then increases the value of cyclic variable M, reset the testing time CNT for statistical correction rate, redirect performs step 1.2);

1.14) rotor recognizing site pos and the current rotor physical location POS of induction less brush-less motor is recorded;

1.15) whether the rotor current actual positions judging induction less brush-less motor is 6, if be not 6, in advance the rotor physical location POS of induction less brush-less motor is navigated to the next position, and redirect performs step 1.3); If be 6, then perform next step;

1.16) testing time CNT is subtracted 1, judge whether new testing time CNT is 0, if new testing time CNT is not 0, then redirect performs step 1.2); If new testing time CNT is 0, then redirect performs step 1.17);

1.17) each record rotor recognizing site pos and rotor physical location POS is compared, in the test of CNT time, count the accuracy C that six each location recognition are all correct; Whether judgment accuracy C is less than receptible minimum accuracy P when identifying initial position of rotor and sets up, if set up, then increase the value of cyclic variable M, reset the testing time CNT for statistical correction rate, redirect performs step 1.2), if be false, then the value of current cyclic variable M is powered up the calibration result of the common cycle-index N of mode as six kinds.

The present invention is for identifying that the time varying signal method of sampling of induction less brush-less motor initial position has following advantage: first, the present invention uses the short duration voltage pulse of designated magnitude, the pulse duration Δ T of short duration voltage pulse when not using sampling hold circuit ₀be more than or equal to sampling time Δ T, the pulse duration Δ T of short duration voltage pulse when using sampling hold circuit ₀be less than sampling time Δ T, after power the Δ T moment in sampling time to High frequency filter and amplify after changing currents with time sampled signal carry out AD sampling obtain sampled result, by aforesaid operations cycle criterion N time, make the amplitude of the changing currents with time sampled signal gathered for N time substantially constant, meet input signal in oversampler method be the condition of a constant direct voltage 1.; Secondly, the current sampling resistor R of induction less brush-less motor after power _sNSthe interference signal of the approximate white noises such as power supply disturbance, CPU core interference, I/O mouth is contained, by adjusting the cut-off frequency f of low pass filter in the changing currents with time sampled signal exported ₀the white noise amplitude comprised by signal after low pass filter is made to be greater than 1 least significant bit LSB of AD sampler, thus can to meet the noise in oversampler method, input signal superposed be white noise or approximate white noise, and 2. white noise has the condition of enough amplitudes.Based on above-mentioned two conditions 1., condition 2. meet, make the present invention for identify induction less brush-less motor initial position the time varying signal method of sampling on the one hand do not increase hardware circuit, do not use more high-resolution AD converter instead prerequisite under, resolution and the interference free performance of AD conversion can be improved, even if apply relatively short potential pulse, also accurately can differentiate the size of six kinds of current sampling signals, and then accurately identify the initial position of rotor; On the other hand can apply relatively short potential pulse due to during initial position because identifying rotor, therefore in identifying, both obvious noise of motor can not be produced, also rotor can not be made to rotate or shake, have and do not need to increase hardware circuit, require low to AD converter, anti-interference is good, and resolution is high, obvious noise of motor can not be produced, the advantage that rotor rotates or shakes during identification, can not be made.

Accompanying drawing explanation

Fig. 1 is the basic procedure schematic diagram of embodiment of the present invention method.

Fig. 2 is the waveform schematic diagram of time varying signal sampling in embodiment of the present invention method.

Front 6 groups of measured datas of Fig. 3 is N when being 1 initial position of rotor of inductance method identification brshless DC motor.

Rear 6 groups of measured datas of Fig. 4 is N when being 1 initial position of rotor of inductance method identification brshless DC motor.

Front 6 groups of measured datas of Fig. 5 is N when being 2 initial position of rotor of inductance method identification brshless DC motor.

Rear 6 groups of measured datas of Fig. 6 is N when being 2 initial position of rotor of inductance method identification brshless DC motor.

Front 6 groups of measured datas of Fig. 7 is N when being 4 initial position of rotor of inductance method identification brshless DC motor.

Rear 6 groups of measured datas of Fig. 8 is N when being 4 initial position of rotor of inductance method identification brshless DC motor.

Front 6 groups of measured datas of Fig. 9 is N when being 20 initial position of rotor of inductance method identification brshless DC motor.

Rear 6 groups of measured datas of Figure 10 is N when being 20 initial position of rotor of inductance method identification brshless DC motor.

Front 6 groups of measured datas of Figure 11 is N when being 40 initial position of rotor of inductance method identification brshless DC motor.

Rear 6 groups of measured datas of Figure 12 is N when being 40 initial position of rotor of inductance method identification brshless DC motor.

Front 6 groups of measured datas of Figure 13 is N when being 1 initial position of rotor of inductance method identification permagnetic synchronous motor.

Rear 6 groups of measured datas of Figure 14 is N when being 1 initial position of rotor of inductance method identification permagnetic synchronous motor.

Front 6 groups of measured datas of Figure 15 is N when being 2 initial position of rotor of inductance method identification permagnetic synchronous motor.

Rear 6 groups of measured datas of Figure 16 is N when being 2 initial position of rotor of inductance method identification permagnetic synchronous motor.

Front 6 groups of measured datas of Figure 17 is N when being 4 initial position of rotor of inductance method identification permagnetic synchronous motor.

Rear 6 groups of measured datas of Figure 18 is N when being 4 initial position of rotor of inductance method identification permagnetic synchronous motor.

Front 6 groups of measured datas of Figure 19 is N when being 20 initial position of rotor of inductance method identification permagnetic synchronous motor.

Rear 6 groups of measured datas of Figure 20 is N when being 20 initial position of rotor of inductance method identification permagnetic synchronous motor.

Front 6 groups of measured datas of Figure 21 is N when being 40 initial position of rotor of inductance method identification permagnetic synchronous motor.

Rear 6 groups of measured datas of Figure 22 is N when being 40 initial position of rotor of inductance method identification permagnetic synchronous motor.

The result that Figure 23 obtains when being identical measured data the second recognition mode with Figure 17.

The result that Figure 24 obtains when being identical measured data the second recognition mode with Figure 18.

Figure 25 is the schematic flow sheet demarcating cycle-index N in the embodiment of the present invention one.

Embodiment

As shown in Figure 1, the present embodiment comprises for the step of the time varying signal method of sampling identifying induction less brush-less motor initial position:

1) from six kinds of the UVW three-phase of induction less brush-less motor power up that the positive V of mode U is negative, the positive W of U is negative, the positive W of V is negative, the positive U of V is negative, the positive U of W is negative, the positive V of W negative selection one power up mode; Setting cycle-index N and sampling time Δ T, arranges cumulative sum SUM _iand clear 0;

2) short duration voltage pulse of designated magnitude is powered up induction less brush-less motor according to the mode that powers up selected, the pulse duration Δ T of described short duration voltage pulse when not using sampling hold circuit ₀be more than or equal to sampling time Δ T, the pulse duration Δ T of described short duration voltage pulse when using sampling hold circuit ₀be less than sampling time Δ T, after power to the current sampling resistor R of induction less brush-less motor _sNSthe changing currents with time sampled signal exported carries out High frequency filter and amplification, and carries out AD sampling in the Δ T moment in sampling time to High frequency filter and the changing currents with time sampled signal after amplifying, and sampled result is added to cumulative sum SUM _i;

3) cycle-index N is subtracted 1, judge whether cycle-index N is 0, if be 0, redirect performs step 4); If be not 0, redirect performs step 2);

4) to cumulative sum SUM _iextract, result will be extracted and export.

As everyone knows, adopt oversampler method need meet following two conditions, condition 1.: input signal is a constant direct voltage; Condition is 2.: the noise that input signal superposes is white noise or approximate white noise, and white noise has enough amplitudes.Meeting under over-sampling condition, sample frequency often doubles, and in-band noise will reduce 3dB, and Measurement Resolution is by increase by 1/2.When Voltage Pulse Width Δ T is less, current sampling signal is the time varying signal of approximate triangular wave, time varying signal obviously can not meet first condition required by oversampler method, because signal is being the marked change that near linear rises in the process of multiple repairing weld always.The time varying signal method of sampling of the present embodiment is in fact a kind of discrete oversampler method, its principle is: repeated to produce N time by same current sampling signal (current sampling signal produced when such as the positive V of making alive pulse U bears), if disregard the impact of various interference, each is stable in the amplitude of the current sampling signal in identical Δ T moment in sampling time, as the formula (1), thus the condition that can meet over-sampling 1..Simultaneously, the present embodiment considers that the real input signal of low pass filter includes the interference signals such as power supply disturbance, CPU core interference, I/O mouth, by carrying out High frequency filter to the input signal of low pass filter, the output signal of low pass filter can be made to have superposed the interference of the approximate white noise of certain amplitude, thus the condition that can meet over-sampling 2..Therefore, the present embodiment repeats N time (N is enough large) the selected mode that powers up, and carries out High frequency filter and amplification, and all carry out AD sampling in the Δ T moment to each current sampling signal.Mode is powered up to selected, the amplitude of this N sub-sampling signal is substantially constant in the Δ T moment, and signal has superposed the interference of approximate white noise, therefore over-sampling condition is met, carry out adding up and extracting to this N time sampled value, just can obtain the discrete over sampled values that anti-interference is good, resolution is higher, adopt the shorter potential pulse of the present embodiment accurately can identify the initial position of rotor.

V _SNS=R _SNS·(V _MOTOR/R)(1-e ^-ΔT·R/L)（1）

In formula (1), V _sNSrepresent the amplitude of the current sampling signal in Δ T moment in sampling time, R _sNSrepresent the resistance value of the current sampling resistor of induction less brush-less motor, V _mOTORrepresent the amplitude of short duration voltage pulse, R represents that the current sampling resistor of induction less brush-less motor is connected the all-in resistance after two switching tube conducting resistance, and L represents the inductance value of the corresponding two-phase winding coil of the mode that powers up that induction less brush-less motor is current.For specific motor, certain rotor position, specificly power up mode and specific busbar voltage, L, R, V _mOTOR, R _sNSall constant, V _sNSbe variable quantity, Δ T time is longer, the amplitude V of the current sampling signal in Δ T moment in sampling time _sNSlarger.The amplitude V of the current sampling signal in Δ T moment in sampling time _sNSallowed band need consider two factors: one is the amplitude V of the current sampling signal in Δ T moment in sampling time _sNSincrease can make the gate source voltage V of field effect transistor switch pipe _gSreduce, but be as the criterion with the conducting resistance of not obvious increase switching tube; Two is amplitude V of the current sampling signal in Δ T moment in sampling time _sNScarry out High frequency filter, be amplified into the voltage of AD converter input and do not exceed the scope that its electrical characteristic bears.At the amplitude V of the current sampling signal in Δ T moment in sampling time _sNSallowed band in, Δ T time is longer, and the various otherness powering up the current amplitude that mode produces is more obvious, but sampling time Δ T time is longer, and the noise of motor of generation is larger, and even rotor rotates.Adopt the method for the present embodiment, accurately can differentiate the various otherness powering up the current amplitude that mode produces with less sampling time Δ T.

See the T on Fig. 2, t axle ₁, T ₂, T ₃..., T _nrespectively represent the 1st, 2,3 ..., N number of potential pulse initial time, the T1 moment can be 0, Δ T is the pulse duration of each short duration voltage pulse, curve L be by the 1st, 2,3 ..., N number of potential pulse at the curve of the amplitude point smooth connection of respective Δ T moment current sampling signal, by T ₁~ T _nn primary current sampled signal couple together smoothly at the amplitude point in the Δ T moment of each time, just define the curve L in Fig. 2, so repeat artificially to produce N time and the process of sampling and extracting to same current sampling signal, just be equivalent to N the over-sampling to curve L, substantially constant see the known each sampled signal of Fig. 2 in the amplitude in Δ T moment, and also can there is certain fluctuation up and down by the impact of white noise interference signal, thus the condition that can meet oversampler method 1. with condition 2., and according to over-sampling principle, N over-sampling had both added the resolution of AD conversion, turn improve interference free performance.Even if therefore apply relatively little short duration voltage pulse, also accurately can differentiate the size of six kinds of current sampling signals, and then accurately identify the initial position of rotor.

In the present embodiment, step 2) in High frequency filter specifically refer to and carry out High frequency filter with low pass filter, by adjusting the cut-off frequency f of low pass filter ₀, make the white noise amplitude comprised through the filtered signal of low pass filter be greater than 1 least significant bit LSB of AD sampler.

In the present embodiment, the extraction in step 4) specifically refers to 1. following ~ one 3. in three kinds of methods: 1., by cumulative sum SUM _imove to right as extracting result behind lgN/lg4 position, wherein N is the positive integer power of 4; 2., by cumulative sum SUM _idivided by cumulative sum SUM _ias extracting result after the mean value that the quantity of middle comprised sampled result obtains also retains the precision of specifying; 3., by cumulative sum SUM _idirect as extracting result.It should be noted that, 1. above-mentioned ~ 3. three kinds of methods are not for data acquisition is the exhaustive of adopted extraction, certainly, those skilled in the art also can adopt the abstracting method of other types as required, it equally also can complete the extraction of all sampled result, equally also can reach the resolution increasing AD conversion, the object improving interference free performance.In the present embodiment, 3. extraction in step 4) specifically refers to method, the all sampled result being about to obtain add up, using the cumulative cumulative sum obtained as extraction result, method advantage 3. carry out cumulative after make difference between the discrete extraction result (cumulative sum) that finally obtains larger, to guarantee between discrete extraction result that upper and lower disturbance is larger, can guarantee to meet the condition of oversampler method 2..

In the present embodiment, to cumulative sum SUM in step 4) _ithe step of all sampled result obtained being carried out to data screening is also comprised before extracting, described data screening specifically refers to: first pass through in step 2) middle record sampled result, then step 2 is performed for cycle-index N time) the N number of sampled result of cycle-index that obtains, select maximum wherein and from cumulative sum SUM _ithis maximum of middle removal, or the minimum value selected wherein from cumulative sum SUM _ithis minimum value of middle removal, or the maximum simultaneously selected wherein and minimum value from cumulative sum SUM _ithis maximum of middle removal and minimum value.Certainly, by above-mentioned data screening, the noise in sampled result can be removed, promote the accuracy of sampled result.

See the measured data of carrying out for certain permagnetic synchronous motor shown in the measured data of carrying out for certain brshless DC motor shown in Fig. 3 ~ Figure 12, Figure 13 ~ Figure 22.In Fig. 3 ~ Figure 12, S1 ~ S6 represents that employing the present embodiment time varying signal method of sampling extracts six extractions result (cumulative sum) obtained, POS represents rotor physical location, pos represents the rotor recognizing site adopting inductance method identification to obtain, i represents the intermediate variable adopting inductance method identification initial position of rotor to obtain, and the transformational relation between intermediate variable i and rotor recognizing site pos refers to table 1.

For in the measured data that certain brshless DC motor carries out: Fig. 3 and Fig. 4 is cycle-index N value adopts when being 1 12 of inductance method identification initial position of rotor to take turns measured data, for six the rotor physical locations being numbered 1 ~ 6, its accuracy adds up to 50%, and each whole accuracy of taking turns measured data is 0%.Fig. 5 and Fig. 6 is cycle-index N value adopts when being 2 12 of inductance method identification initial position of rotor to take turns measured data, and for six the rotor physical locations being numbered 1 ~ 6, its accuracy adds up to 64%, and each whole accuracy of taking turns measured data is 0%.Fig. 7 and Fig. 8 is cycle-index N value adopts when being 4 12 of inductance method identification initial position of rotor to take turns measured data, and for six the rotor physical locations being numbered 1 ~ 6, its accuracy adds up to 89%, and each whole accuracy of taking turns measured data is 33%.Fig. 9 and Figure 10 is cycle-index N value adopts when being 8 12 of inductance method identification initial position of rotor to take turns measured data, and for six the rotor physical locations being numbered 1 ~ 6, its accuracy adds up to 93%, and each whole accuracy of taking turns measured data is 58%.Figure 11 and Figure 12 is cycle-index N value adopts when being 40 12 of inductance method identification initial position of rotor to take turns measured data, for six the rotor physical locations being numbered 1 ~ 6, its accuracy adds up to 100%, and each whole accuracy of taking turns measured data is 100%.

For in the measured data that certain permagnetic synchronous motor carries out: Figure 13 and Figure 14 is cycle-index N value adopts when being 1 12 of inductance method identification initial position of rotor to take turns measured data, for six the rotor physical locations being numbered 1 ~ 6, its accuracy adds up to 36%, and each whole accuracy of taking turns measured data is 0%.Figure 15 and Figure 16 is cycle-index N value adopts when being 2 12 of inductance method identification initial position of rotor to take turns measured data, and for six the rotor physical locations being numbered 1 ~ 6, its accuracy adds up to 46%, and each whole accuracy of taking turns measured data is 0%.Figure 17 and Figure 18 is cycle-index N value adopts when being 4 12 of inductance method identification initial position of rotor to take turns measured data, and for six the rotor physical locations being numbered 1 ~ 6, its accuracy adds up to 57%, and each whole accuracy of taking turns measured data is 0%.Figure 19 and Figure 20 is cycle-index N value adopts when being 20 12 of inductance method identification initial position of rotor to take turns measured data, and for six the rotor physical locations being numbered 1 ~ 6, its accuracy adds up to 93%, and each whole accuracy of taking turns measured data is 58%.Figure 21 and Figure 22 is cycle-index N value adopts when being 40 12 of inductance method identification initial position of rotor to take turns measured data, for six the rotor physical locations being numbered 1 ~ 6, its accuracy adds up to 100%, and each whole accuracy of taking turns measured data is 100%.

Show through above-mentioned measured data, after given suitable pulse duration Δ T, along with the increase of cycle-index N, the accuracy rate of location recognition is also along with increase, after cycle-index N is increased to certain value, the rate of accuracy reached of initial position of rotor identification is to acceptable degree.Therefore, how the setting of cycle-index N completes the time varying signal method of sampling for the present embodiment accurately and fast and has very important effect.Certainly, a general enough large cycle-index N(such as N=40 can be set for induction less brush-less motor) rate of accuracy reached of initial position of rotor identification can be made to acceptable degree; But as the time varying signal method of sampling for a certain induction less brush-less motor will be made more quick and accurate, the cycle-index N for this induction less brush-less motor is then needed to demarcate, thus make cycle-index N meet make the rate of accuracy reached of initial position of rotor identification to the prerequisite of acceptable degree under little as much as possible, thus save the time of time varying signal sampling, make the initial position of rotor identification of induction less brush-less motor more rapidly quick.

It should be noted that, negative for the positive V of U, the positive W of U negative, the positive W of V negative, the positive U of V is negative, the positive U of W is negative, negative six kinds of the positive V of W powers up mode, both can adopt unified cycle-index N, and also can power up mode according to difference and adopt different cycle-index N.In the present embodiment, the positive V of U is born, the positive W of U bears, the positive W of V bears, the positive U of V bears, the positive U of W bears, the positive V of W bears six kinds and power up mode and adopt unified cycle-index N.As shown in figure 25, also comprise the step of demarcating cycle-index N before step 1) setting cycle-index N in the present embodiment, detailed step comprises:

1.1) receptible minimum accuracy P during setting identification initial position of rotor; Setting sampling time Δ T; The initial value of setting cyclic variable M, this initial value is more than or equal to 4; Setting is used for the testing time CNT of statistical correction rate, desired location discrimination model MOD, arranges cumulative sum SUM1 ~ SUM6 and clear 0, and the corresponding six kinds of one powered up in mode of each cumulative sum power up mode;

1.2) the rotor physical location POS of induction less brush-less motor is navigated to 1;

1.3) short duration voltage pulse of designated magnitude is powered up mode according to six kinds successively to power up induction less brush-less motor, the pulse duration Δ T of described short duration voltage pulse when not using sampling hold circuit ₀be more than or equal to sampling time Δ T, the pulse duration Δ T of described short duration voltage pulse when using sampling hold circuit ₀be less than sampling time Δ T, after power to the current sampling resistor R of induction less brush-less motor _sNSthe changing currents with time sampled signal exported carries out High frequency filter and amplification, and carry out AD sampling in the Δ T moment in sampling time to through High frequency filter and the changing currents with time sampled signal after amplifying, and sampled result is added to corresponding cumulative sum respectively by the mode that powers up, and mode powered up to each repeats steps A 1.3) M time altogether;

1.4) six cumulative sum SUM1 ~ SUM6 are extracted respectively, obtain six altogether and extract result; Described extraction specifically refers to 1. following ~ one 3. in three kinds of methods: as extraction result S1 ~ S6 behind the lgN/lg4 position that 1., using cumulative sum SUM1 ~ SUM6 moves to right respectively, wherein N is the positive integer power of 4; 2. the mean value (retaining enough precision), obtained divided by M respectively by cumulative sum SUM1 ~ SUM6 is afterwards as extraction result S1 ~ S6; 3., using cumulative sum SUM1 ~ SUM6 directly as extracting result S1 ~ S6.

1.5) judge whether position discrimination model MOD is 1, if so, then redirect performs step 1.6); If not, then jump to step 1.11);

1.6) arranging the intermediate variable i being used for location recognition is 0;

1.7) judge that extraction result S1 is greater than extraction result S4 and whether sets up, if set up, intermediate variable i is added 1;

1.8) judge that extraction result S2 is greater than extraction result S5 and whether sets up, if set up, intermediate variable i is added 2;

1.9) judge that extraction result S3 is greater than extraction result S6 and whether sets up, if set up, intermediate variable i is added 4;

1.10) search default location lookup table according to the value of intermediate variable i, obtain rotor recognizing site pos, jump to 1.14);

1.11) whether maximumly S1 in result S1, S2, S3 is judged to extract, if it is maximum for extracting result S1, judge that extraction result S1 is greater than S4 and whether sets up again, set up if S1 is greater than S4, make the value of rotor recognizing site pos be 1 and jump to step 1.14), be false if S1 is greater than S4, make the value of rotor recognizing site pos be 4 and jump to step 1.14); If it is not maximum for extracting result S1, perform step 1.12);

1.12) whether maximumly S2 in result S1, S2, S3 is judged to extract, if it is maximum for extracting result S2, judge that extraction result S2 is greater than S5 and whether sets up again, if extract result S2 to be greater than S5 establishment, the value of rotor recognizing site pos is made to be 2 and to jump to step 1.14), be false if extraction result S2 is greater than S5, make the value of rotor recognizing site pos be 5 and jump to step 1.14); If it is not maximum for extracting result S2, perform step 1.13);

1.13) whether maximumly S3 in result S1, S2, S3 is judged to extract, if it is maximum for extracting result S3, judge that extraction result S3 is greater than S6 and whether sets up again, if set up, the value of rotor recognizing site pos is made to be 3 and to jump to step 1.14), if be false, make the value of rotor recognizing site pos be 6 and jump to step 1.14); If it is not maximum for extracting result S3, then increases the value of cyclic variable M, reset the testing time CNT for statistical correction rate, redirect performs step 1.2);

1.14) rotor recognizing site pos and the current rotor physical location POS of induction less brush-less motor is recorded;

1.15) whether the rotor current actual positions judging induction less brush-less motor is 6, if be not 6, in advance the rotor physical location POS of induction less brush-less motor is navigated to the next position, and redirect performs step 1.3); If be 6, then perform next step;

1.16) testing time CNT is subtracted 1, judge whether new testing time CNT is 0, if new testing time CNT is not 0, then redirect performs step 1.2); If new testing time CNT is 0, then redirect performs step 1.17);

1.17) each record rotor recognizing site pos and rotor physical location POS is compared, in the test of CNT time, count the accuracy C that six each location recognition are all correct; Whether judgment accuracy C is less than receptible minimum accuracy P when identifying initial position of rotor and sets up, if set up, then increase the value of cyclic variable M, reset the testing time CNT for statistical correction rate, redirect performs step 1.2), if be false, then the value of current cyclic variable M is powered up the calibration result of the common cycle-index N of mode as six kinds.

In the location lookup table preset in the present embodiment, when middle variable i is 0, rotor recognizing site pos is 5; When middle variable i is 1, rotor recognizing site pos is 6; When middle variable i is 0, rotor recognizing site pos is 5; When middle variable i is 3, rotor recognizing site pos is 1; By that analogy, table 1 is referred to.

Table 1: location lookup table.

i	0	1	2	3	4	5	6	7
									pos	5	6	Invalid	1	4	Invalid	3	2

Induction less brush-less motor in the present embodiment comprises UVW three-phase, rotor pole logarithm is 4, and the stator of therefore induction less brush-less motor is marked with 4 group echos, and comprise 1 ~ 6 totally 6 positions in each group echo, therefore a circumference is divided into a position.Stator can exist the position that four are all identified as 1, abovementioned steps 1.2) in when the rotor physical location POS of induction less brush-less motor is navigated to 1, specifically refer to the rotor rotating induction less brush-less motor, the positioning pointer of rotor is turned on stator, and any one is identified as the position of 1.

In the present embodiment, according to initial value setting employing two kinds of optional position discrimination models of position discrimination model MOD, the first position discrimination model is step 1.6) ~ 1.10), the first position discrimination model is accurate diagnostic method; The second position discrimination model is step 1.11) ~ 1.13), the second position discrimination model is Fuzzy evaluation.But two kinds of position discrimination models can both realize the accurate calibration to cycle-index N.Compare with the first position discrimination model, the requirement of discrimination model to cycle-index N of the second position is smaller, is described as follows: cycle-index N=4 for Figure 17,18, when adopting the first position discrimination model, the accuracy that position differentiates is only 57%, and the while of 6 positions, accuracy is 0%; With Figure 17,18 identical test datas, adopt the second position discrimination model to obtain Figure 23,24, the while that the accuracy that now position differentiates being 100%, 6 positions, accuracy is 100%.Clearly adopt the second position discrimination model to reduce cycle-index, also namely shorten recognition time, but specifically select which kind of recognition mode, need to decide according to concrete motor characteristic.

The above is only the preferred embodiment of the present invention, protection scope of the present invention be not only confined to above-described embodiment, and all technical schemes belonged under thinking of the present invention all belong to protection scope of the present invention.It should be pointed out that for those skilled in the art, some improvements and modifications without departing from the principles of the present invention, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (4)

1., for identifying a time varying signal method of sampling for induction less brush-less motor initial position, it is characterized in that step comprises:

1) from six kinds of the UVW three-phase of induction less brush-less motor power up that the positive V of mode U is negative, the positive W of U is negative, the positive W of V is negative, the positive U of V is negative, the positive U of W is negative, the positive V of W negative selection one power up mode; Setting cycle-index N and sampling time Δ T, arranges cumulative sum SUM _iand clear 0;

2) short duration voltage pulse of designated magnitude is powered up induction less brush-less motor according to the mode that powers up selected, the pulse duration Δ T of described short duration voltage pulse when not using sampling hold circuit ₀be more than or equal to sampling time Δ T, the pulse duration Δ T of described short duration voltage pulse when using sampling hold circuit ₀be less than sampling time Δ T, after power to the current sampling resistor R of induction less brush-less motor _sNSthe changing currents with time sampled signal exported carries out High frequency filter and amplification, and carries out AD sampling in the Δ T moment in sampling time to High frequency filter and the changing currents with time sampled signal after amplifying, and sampled result is added to cumulative sum SUM _i;

3) cycle-index N is subtracted 1, judge whether cycle-index N is 0, if be 0, redirect performs step 4); If be not 0, redirect performs step 2);

4) to cumulative sum SUM _iextract, result will be extracted and export;

Described step 2) in High frequency filter specifically refer to and carry out High frequency filter with low pass filter, by adjusting the cut-off frequency f of low pass filter ₀, make the white noise amplitude comprised through the filtered signal of low pass filter be greater than 1 least significant bit LSB of AD sampler.

2. the time varying signal method of sampling for identifying induction less brush-less motor initial position according to claim 1, is characterized in that, the extraction in described step 4) specifically refers to 1. following ~ one 3. in three kinds of methods: 1., by cumulative sum SUM _imove to right as extracting result behind lgN/lg4 position, wherein N is the positive integer power of 4; 2., by cumulative sum SUM _idivided by cumulative sum SUM _ias extracting result after the mean value that the quantity of middle comprised sampled result obtains also retains the precision of specifying; 3., by cumulative sum SUM _idirect as extracting result.

3. the time varying signal method of sampling for identifying induction less brush-less motor initial position according to claim 2, is characterized in that: to cumulative sum SUM in described step 4) _ithe step of all sampled result obtained being carried out to data screening is also comprised before extracting, described data screening specifically refers to: first pass through in step 2) middle record sampled result, then step 2 is performed for cycle-index N time) the N number of sampled result of cycle-index that obtains, select maximum wherein and from cumulative sum SUM _ithis maximum of middle removal, or the minimum value selected wherein from cumulative sum SUM _ithis minimum value of middle removal, or the maximum simultaneously selected wherein and minimum value from cumulative sum SUM _ithis maximum of middle removal and minimum value.

4. according to the time varying signal method of sampling for identifying induction less brush-less motor initial position in claims 1 to 3 described in any one, it is characterized in that: also comprise the step of demarcating cycle-index N before described step 1) setting cycle-index N, detailed step comprises:

1.1) receptible minimum accuracy P during setting identification initial position of rotor; Setting sampling time Δ T; The initial value of setting cyclic variable M, this initial value is more than or equal to 4; Setting is used for the testing time CNT of statistical correction rate, desired location discrimination model MOD, arranges cumulative sum SUM1 ~ SUM6 and clear 0, and the corresponding six kinds of one powered up in mode of each cumulative sum power up mode;

1.2) the rotor physical location POS of induction less brush-less motor is navigated to 1;

1.3) short duration voltage pulse of designated magnitude is powered up mode according to six kinds successively to power up induction less brush-less motor, the pulse duration Δ T of described short duration voltage pulse when not using sampling hold circuit ₀be more than or equal to sampling time Δ T, the pulse duration Δ T of described short duration voltage pulse when using sampling hold circuit ₀be less than sampling time Δ T, after power to the current sampling resistor R of induction less brush-less motor _sNSthe changing currents with time sampled signal exported carries out High frequency filter and amplification, and carry out AD sampling in the Δ T moment in sampling time to through High frequency filter and the changing currents with time sampled signal after amplifying, and sampled result is added to corresponding cumulative sum respectively by the mode that powers up, and mode powered up to each repeats step 1.3) M time altogether;

1.4) six cumulative sum SUM1 ~ SUM6 are extracted respectively, obtain six altogether and extract result; Described extraction specifically refers to 1. following ~ one 3. in three kinds of methods: as extraction result S1 ~ S6 behind the lgN/lg4 position that 1., using cumulative sum SUM1 ~ SUM6 moves to right respectively, wherein N is the positive integer power of 4; 2. the mean value (retaining enough precision), obtained divided by M respectively by cumulative sum SUM1 ~ SUM6 is afterwards as extraction result S1 ~ S6; 3., using cumulative sum SUM1 ~ SUM6 directly as extracting result S1 ~ S6;

1.5) judge whether position discrimination model MOD is 1, if so, then redirect performs step 1.6); If not, then jump to step 1.11);

1.6) arranging the intermediate variable i being used for location recognition is 0;

1.7) judge that extraction result S1 is greater than extraction result S4 and whether sets up, if set up, intermediate variable i is added 1;

1.8) judge that extraction result S2 is greater than extraction result S5 and whether sets up, if set up, intermediate variable i is added 2;

1.9) judge that extraction result S3 is greater than extraction result S6 and whether sets up, if set up, intermediate variable i is added 4;

1.10) search default location lookup table according to the value of intermediate variable i, obtain rotor recognizing site pos, jump to 1.14);

1.11) whether maximumly S1 in result S1, S2, S3 is judged to extract, if it is maximum for extracting result S1, judge that extraction result S1 is greater than S4 and whether sets up again, set up if S1 is greater than S4, make the value of rotor recognizing site pos be 1 and jump to step 1.14), be false if S1 is greater than S4, make the value of rotor recognizing site pos be 4 and jump to step 1.14); If it is not maximum for extracting result S1, perform step 1.12);

1.12) whether maximumly S2 in result S1, S2, S3 is judged to extract, if it is maximum for extracting result S2, judge that extraction result S2 is greater than S5 and whether sets up again, if extract result S2 to be greater than S5 establishment, the value of rotor recognizing site pos is made to be 2 and to jump to step 1.14), be false if extraction result S2 is greater than S5, make the value of rotor recognizing site pos be 5 and jump to step 1.14); If it is not maximum for extracting result S2, perform step 1.13);

1.13) whether maximumly S3 in result S1, S2, S3 is judged to extract, if it is maximum for extracting result S3, judge that extraction result S3 is greater than S6 and whether sets up again, if set up, the value of rotor recognizing site pos is made to be 3 and to jump to step 1.14), if be false, make the value of rotor recognizing site pos be 6 and jump to step 1.14); If it is not maximum for extracting result S3, then increases the value of cyclic variable M, reset the testing time CNT for statistical correction rate, redirect performs step 1.2);

1.14) rotor recognizing site pos and the current rotor physical location POS of induction less brush-less motor is recorded;

1.15) whether the rotor current actual positions judging induction less brush-less motor is 6, if be not 6, in advance the rotor physical location POS of induction less brush-less motor is navigated to the next position, and redirect performs step 1.3); If be 6, then perform next step;

1.16) testing time CNT is subtracted 1, judge whether new testing time CNT is 0, if new testing time CNT is not 0, then redirect performs step 1.2); If new testing time CNT is 0, then redirect performs step 1.17);

1.17) each record rotor recognizing site pos and rotor physical location POS is compared, in the test of CNT time, count the accuracy C that six each location recognition are all correct; Whether judgment accuracy C is less than receptible minimum accuracy P when identifying initial position of rotor and sets up, if set up, then increase the value of cyclic variable M, reset the testing time CNT for statistical correction rate, redirect performs step 1.2), if be false, then the value of current cyclic variable M is powered up the calibration result of the common cycle-index N of mode as six kinds.