CN102128656B - Slightly bent Koch mass flow meter digital signal processing method and system - Google Patents

Abstract

Translated fromChinese

本发明涉及一种微弯型科氏质量流量计数字信号处理方法和系统，包括放大滤波电路、ADC1、ADC2、电压基准源、电压跟随器、电流源、差分放大器、ADC3、模拟驱动模块、DSP、外扩SRAM、外扩EEPROM、4～20mA输出及脉冲输出、电源模块及软件。模拟驱动模块激励科氏质量流量传感器内部的激振器，使流量管以固有频率振动。两路传感器信号经过两路参数相同的调理电路放大、滤波后，分别由ADC1和ADC2采样转换为数字量，通过DSP的多通道缓冲串口传送至DSP。DSP对采集的信号进行数字滤波消除噪声，采用基于拉格朗日插值的过零检测方法计算信号的频率、相位差，得到流体的密度、质量流量等参量，并结合温度，进行补偿。

The invention relates to a digital signal processing method and system of a micro-curved Coriolis mass flowmeter, including an amplification filter circuit, ADC1, ADC2, a voltage reference source, a voltage follower, a current source, a differential amplifier, ADC3, an analog drive module, and a DSP , External expansion SRAM, external expansion EEPROM, 4 ~ 20mA output and pulse output, power supply module and software. The analog drive module excites the exciter inside the Coriolis mass flow sensor, causing the flow tube to vibrate at its natural frequency. After the two sensor signals are amplified and filtered by two conditioning circuits with the same parameters, they are sampled and converted into digital quantities by ADC1 and ADC2 respectively, and then transmitted to the DSP through the multi-channel buffer serial port of the DSP. DSP performs digital filtering on the collected signal to eliminate noise, uses the zero-crossing detection method based on Lagrangian interpolation to calculate the frequency and phase difference of the signal, and obtains parameters such as fluid density and mass flow rate, and compensates it in combination with temperature.

Description

A kind of little curved Ke's mass flowmeter digital signal processing method and system

Technical field

The present invention relates to field of flow detection, be a kind of Ke's mass flowmeter digital signal processing method and system, particularly higher at little curved coriolis mass flow sensor natural frequency, phase differential is less and the IIR bandpass filter of a kind of passband width, the depth-adjustment of design, based on signal processing method and the system of the zero passage detection of Lagrange's interpolation.

Background technology

Coriolis mass flowmeters (abbreviation Coriolis mass flowmeter) can directly be measured mass rate, is one of current flowmeter with the fastest developing speed, has broad application prospects.Coriolis mass flowmeter is made up of primary instrument and secondary instrument, and primary instrument comprises measuring tube (or claiming flowtube), magnetoelectric velocity transducer (perhaps optical displacement sensor), vibrator and temperature sensor; Secondary instrument is called transmitter again, mainly comprises signal condition, processing and driving circuit.Measuring tube has U-shaped cast (being designated hereinafter simply as U-shaped), straight pipe type and little bent tube type (being designated hereinafter simply as little curved) etc., wherein, U-shaped is highly sensitive, and namely the torsional deformation that produces of the coriolis force that is caused by flow is big, and natural frequency (resonance frequency) is generally between 70Hz to 150Hz; The sensitivity of straight pipe type is low, and natural frequency is generally more than 1kHz; U-shaped low of little curved remolding sensitivity, natural frequency is generally between 300Hz to 900Hz.The effect of transmitter is mistiming or frequency and the phase differential of measuring between the two-way magnetoelectric velocity transducer output signal.For computational accuracy and the antijamming capability that improves Coriolis mass flowmeter, Digital Signal Processing is used for the processing of coriolis mass flow sensor output signal both at home and abroad, development is based on the Coriolis mass flow transmitter of DSP (digital signal processor), for example, HeFei University of Technology combines lattice type adaptive notch filter and the DTFT algorithm of taking into account negative frequency, handle the output signal of coriolis mass flow sensor, and adopt the TMS320F28335 dsp chip to develop Coriolis mass flow transmitter (Xu Kejun, Zhu Yong is strong etc., a kind of Coriolis mass flow transmitter based on DSP, patent of invention, application number: 200910185560.8, application publication number: CN 101706299 A, Shen Qing Publication day: 2010.05.12).Sensor output signal carries out pre-service through digital filter earlier, adopts lattice type adaptive notch filter tracking signal frequency afterwards, adopts the DTFT algorithm of taking into account negative frequency to calculate the phase differential of signal.A whole set of algorithm has stronger antijamming capability, speed of convergence, higher computational accuracy faster, the transmitter of developing matches with U-shaped Coriolis mass flowmeter primary instrument, carries out discharge and demarcates, and measuring accuracy reaches 0.1 grade, more than the range ratio 20:1, effect is fine.But this transmitter is mainly at U-shaped.When for the treatment of little curved coriolis mass flow sensor output signal, because the natural frequency of little bend pipe is higher, in order to obtain computational accuracy preferably, just must improve the sample frequency of transmitter; In order to reflect the variation of flow in real time, must between the sampling interval of adjacent two data, finish the processor active task of algorithm; And present dsp chip is because the restriction of resource can't be finished this task in real time.If simplify present algorithm, reduce calculated amount, certainly will reduce computational accuracy.And the phase differential of little curved coriolis mass flow sensor output is littler than U-shaped, and for example, the phase differential of the U-shaped Coriolis mass flowmeter maximum flow correspondence of domestic certain Coriolis flowmeter manufacturer production is about 2⁰, and the phase differential of little curved Coriolis mass flowmeter maximum flow correspondence is about 0.5⁰So, must propose new, not only satisfy computational accuracy require, but also can be on DSP the algorithm of real-time implementation, the handling problem that solves little curved coriolis mass flow sensor output signal.

Summary of the invention

The present invention is for addressing the above problem, by the following technical solutions: little curved Ke's mass flowmeter digital signal processing method and system comprise amplification filtering circuit, analog to digital converter (ADC1), analog to digital converter (ADC2), voltage-reference, voltage follower, current source, differential amplifier, analog to digital converter (ADC3), analog-driven module, digital signal processor DSP, extend out SRAM, extend out EEPROM, man-machine interface, the output of 4～20mA electric current, pulse output, power module and software.The vibrator of analog-driven module excitation coriolis mass flow sensor inside makes flowtube with natural frequency vibration; Be positioned at two magnetoelectric sensor output two-way sinusoidal signals of flowtube both sides, two paths of signals is after the identical modulate circuit amplification of two-way parameter, filtering, be digital quantity by two identical ADC1 and analog to digital converter ADC2 sample conversion of model, the multichannel buffered serial port McBSP by digital signal processor DSP sends into digital signal processor DSP; Digital signal processor DSP is eliminated noise effect to the filtering of the advanced line number word of sampled signal, adopts based on the zero passage detection method of Lagrange (Lagrange) interpolation again and calculates frequency, the mistiming of signal, and then obtain mass rate and the fluid density of fluid; Steady current is through coriolis mass flow sensor temperature inside sensor (RTD), the RTD output voltage is behind differential amplifier, be that digital quantity is sent into digital signal processor DSP by analog to digital converter ADC3 sample conversion, digital signal processor DSP compensates according to the temperature convection amount; The measurement result of digital signal processor DSP shows at liquid crystal, and exports measurement result by the output of 4～20mA electric current and pulse output module.

The advantage of patent of the present invention is: adopt a kind of simply and effectively signal processing algorithm to handle the Ke Shi sensor signal, reduce the operand of algorithm greatly, be easy to realize, be fit to very much the higher Ke Shi sensor signal of the little curved equifrequent of processing; Take corresponding measure to overcome that single-chip microcomputer computing number of significant digit is limited, there is biasing problem in hardware circuit, guaranteed the realization precision of algorithm; Based on DSP in real time, implementation algorithm accurately, develop the system that industry member can be accepted and use.

Description of drawings

Fig. 1 is the hardware The general frame of system of the present invention.

Fig. 2 is the master monitor process flow diagram of system of the present invention.

Fig. 3 is the signal acquisition process synoptic diagram of system of the present invention.

Fig. 4 is the ADC break in service process flow diagram of system of the present invention.

Fig. 5 is the DMA break in service process flow diagram of system of the present invention.

Fig. 6 is the zero passage detection schematic diagram of system of the present invention.

Fig. 7 is the zero passage detection algorithm block diagram of system of the present invention.

Fig. 8 finds the solution the process flow diagram at zero point for system of the present invention.

Fig. 9 is that the circuit bias of system of the present invention is to the influence of algorithm.

Figure 10 is the pulse output synoptic diagram of system of the present invention.

Figure 11 is the PWM divide ratio allocation scheme of system of the present invention.

Embodiment

Hardware block diagram of the present invention as shown in Figure 1, by amplification filtering circuit, analog to digital converter (ADC1), analog to digital converter (ADC2), voltage-reference, voltage follower, current source, differential amplifier, analog to digital converter (ADC3), analog drive circuit, digital signal processor (DSP), extend out SRAM, extend out EEPROM, man-machine interface, the output of 4～20mA electric current and pulse output, power module form.

The course of work of the present invention as shown in Figure 2.After system powered on, on the one hand, the vibrator of analog drive system excitation Coriolis mass flowmeter primary instrument made flowtube with natural frequency vibration; On the other hand, digital signal processor DSP is finished the initialization of each module of system and the initialization of algorithmic variable.Be positioned at two magnetoelectric sensor output two-way sinusoidal signals of flowtube both sides, after the modulate circuit amplification that this two paths of signals process two-way parameter is identical, the filtering, by two analog to digital converter ADC1, analog to digital converter ADC2 samplings that model is identical, sampled result is sent in the digital signal processor DSP by the multichannel buffered serial port McBSP of digital signal processor DSP; Digital signal processor DSP adopts the frequency and the phase differential that calculate signal based on the zero passage detection method of Lagrange's interpolation again to the filtering of the advanced line number word of sampled data, obtains mass rate; Digital signal processor DSP reads the numerical value of analog to digital converter ADC3, is converted into temperature, and flow is carried out temperature compensation; Finally, flow measurement is the result be presented on the LCD, and by 4～20mA and pulse output module, exports corresponding electric current and pulse signal.

After system powers on, mimic channel produces the vibration that drives signal excitation Coriolis flowmeter on the one hand, digital signal processor DSP is finished the initialization of system's various piece on the other hand, starts also the sample conversion of two-way analog to digital converter ADC1 and analog to digital converter ADC2 synchronously then.After analog to digital converter ADC1 data-switching is good, trigger interruption by GPIO to digital signal processor DSP.Consider that analog to digital converter ADC1, analog to digital converter ADC2 open synchronously, data-switching is finished in theory simultaneously, therefore, does not open analog to digital converter ADC2 and interrupts, by the state of inquiry analog to digital converter ADC2 in ADC1 interrupts, judge whether analog to digital converter ADC2 converts.After converting, digital signal processor DSP provides clock by McBSP-A, McBSP-B to analog to digital converter ADC1, analog to digital converter ADC2, transformation result is moved to the receiving register (DRR) of McBSP; It then is to finish by direct memory access (DMA) (DMA) module that data in the McBSP receiving register are specified the transfer in random-access memory (ram) space to the user.During initialization DMA, thepassage 1 of DMA, the source address pointer ofpassage 2 are respectively the receiving register of McBSP-A and McBSP-B, the purpose pointer is interim array bufferL, the bufferR of directed towards user definition then, and then DMA automatically stores the ADC transformation result into bufferL and bufferR from DRR.When interim array deposit full after, produce DMA and interrupt, in the interrupt service routine, DMA changes the purpose pointer earlier, prepares for transmitting next time, the content in the ephemeral data is copied in two circulation arrays in the external SRAM again, waits for calling of algorithm.Like this, leaving alone under the situation of CPU, finishing reading of ADC sampled result.Whole signals collecting, transport process as shown in Figure 3, ADC break in service flow process as shown in Figure 4, DMA break in service flow process is as shown in Figure 5.

As shown in Figure 2, after having produced 500 new point data, namely begin to call algorithm.Detection signal amplitude size at first if signal amplitude, represents then that sensor does not open fully less than setting value and shake, continues to wait for; After signal amplitude is greater than setting value, allow 500 point data pass through the IIR bandpass filter of a passband width and depth-adjustment earlier, carry out pre-service, adopt the frequency, the mistiming that allow the zero passage detection method of day interpolation calculate signal based on glug afterwards, draw instantaneous mass flow in conjunction with instrument coefficient; Read temperature, flow is carried out temperature compensation.The whether set of inquiry keyboard zone bit is if set is then called keyboard and handled subroutine.By regularly 1s interruption of cputimer0, in interrupt service routine, the instantaneous delivery that adds up obtains integrated flow; In addition, according to the instantaneous delivery value, by DA or outwardsoutput 4～20mA electric current and the pulse of ePWM module.Refresh LCD, show each measurement result.

There is multiple noise in the actual industrial scene, disturb as a certain fixed frequency that random noise, power frequency interference, motor and pipe vibration etc. cause, the frequency band of these noise distributes very wide, for this reason, the present invention adopts a kind of IIR bandpass filter with trapper structure that sensor signal is carried out filtering.Filter transfer function is:

(1)

In the formula,

,

Be the trap frequency,

,

WillWith

Substitution formula (1) can get its gain at trap frequency place and is:

(2)

When

,

Be in close proximity to 1, and

During not 0, near π, 2 π, formula (2) can be reduced to:

(3)

As seen, when

The time, the trap place is decay; When

The time, the trap place is for amplifying.Trap depth by

Determine, and be subjected to

Influence very little.During designing filter, can fix earlier

, by regulatingChange the trap width of trapper, its value is more close to 1, and the trap width is more narrow; Regulate again, change trap depth.This bandpass filter passband is narrower, and in working in the reality, when fluid media (medium) was full of flowtube, the natural frequency basic fixed of Coriolis mass flowmeter was feasible so with this wave filter the Coriolis mass flowmeter output signal is carried out pre-service.

Zero passage detection obtained the time interval between zero crossing, and then asks for signal frequency, phase differential, as shown in Figure 6 by the moment of tracer signal zero crossing.As seen, signal frequency is

, the two paths of signals mistiming is

, phase differential is

In the actual operation, DSP handles is discrete signal after the sampling.Because ADC can not sample cycle signal zero-cross point just, this just need carry out curve fitting to sampled data, obtains signal zero crossing, as shown in Figure 7.The present invention adopts Lagrangian quadratic interpolation match.If sample sequence

, work as appearance

The time, namely show at [n-1, n] to have a zero point constantly.Generate the quadratic interpolation polynomial expression

, by Lagrangian computing formula:

Simplify:

(4)

(5)

(6)

Therefore, can obtain by solving an equation

The time moment correspondingt, i.e. zero crossing.To give up that root outside [n-1, n] in the equation in actual applications.

Finding in implementation procedure, is that situations between [n-1, n] not of two roots that the match quadratic equation obtains all can appear in 3 matches of Lagrange's interpolation or 5 matches of least square method.When this situation occurring, find the solution null position again with approach based on linear interpolation, find the solution flow process as shown in Figure 8.

After 500 signal datas find zero point, just signal parameter can have been calculated.Ask phase differential to need the zero point of two paths of signals, this takes place to misplace to subtract each other easily, as shown in Figure 6, if 500 do not comprise just

, will produce

Therefore such dislocation, need be judged.Suppose that two paths of signals respectively calculates,

Individual zero point is because the phase range of little curved Coriolis mass flowmeter is no more than10 °, so do as judging:

(1) if, L road signal lag then is described and dislocation can take place and subtract each other modification method: should beTo zero point, the mistiming by

Try to achieve;

(2) if

, R road signal lag then is described and dislocation can take place and subtract each other modification method: should be

To zero point, the mistiming by

Try to achieve;

(3) ifNo matter illustrate to beLStillRDislocation can not take place road signal lag subtracts each other, and hasTo zero point, the mistiming directly by

Try to achieve.

The zero passage detection method is simple, and calculated amount is little, but when DSP realizes, is subjected to the influence that DSP computing number of significant digit is limited, there is biasing in hardware circuit easily.So, must take corresponding measure to guarantee the realization precision of algorithm.

Use 32 float categorical variable when carrying out interpolation arithmetic, the coefficient number of significant digit that is obtained by formula (4)～(6) is few, and the error of calculation is big.For this reason, intermediate variable must use 64 double types of variables when interpolation arithmetic.

Algorithm is transplanted to when DSP goes up test and is found, when the used sampled value of Lagrange's interpolation

Very near 0 o'clock, the resultant error of calculating can be very big, and when separating the quadratic polynomial equation, along withnIncrease, increasing by the equation root error that formula (4)～(6) solve, these all are the limited and truncation errors that cause of DSP computing number of significant digit.The method that solves is that the sampled value that Lagrange's interpolation is used is amplified 100 times earlier, and subscript unification is constantly carried out interpolation calculation by (n-2), (n-1), n normalizing to 0,1,2 again, and formula (4)～(6) become:

(7)

(8)

(9)

Through after the above-mentioned processing, the number of significant digit of result of calculation is improved, and has guaranteed the realization precision of algorithm.

Because the hardware circuit parameter can not accomplish to mate fully, the modulate circuit of two-way sensor signal is symmetry fully, and this just causes the signal after the two-way conditioning to have biasing, is not strict with 0 symmetry, and the biasing difference of two paths of signals.Be cycle signal zero-cross point and algorithm detects, therefore, must take measures on customs clearance overcomes circuit bias to the influence of algorithm.

Suppose that left road sensor signal just is biased to, the right wing sensor signal is biased to negative, the zero crossing of algorithm detection signal, as shown in Figure 9.

During calculated rate, be example with left road, [L₁, L₂] between the time interval less than the half period of signal, and [L₂, L₃] between the half period interval greater than signal, so, if only with L₁, L₂Or L₂, L₃Calculating signal frequency must introduce than the computation error.For the influence that reduces to setover to frequency computation part, should use L₁, L₃Ask for signal frequency, the zero point that namely guarantees to ask for frequency and be employed zero point a signal period is right.

When calculating phase differential, because the two paths of signals biasing is different, make from L₁, R₁, L road signal lag is from L₂, R₂, L road signal is leading.In the reality, biasing may not can so obviously (allows one road signal in negative edge hysteresis at zero point, leading zero point at rising edge) to the influence of algorithm, allows one road signal in negative edge hysteresis at zero point, in rising edge leading trend at zero point, if only with L but can produce one₁, R₁Or L₂, R₂For primitive calculates phase differential, calculating the result will inevitably fluctuate.For this reason, should be with L₁, R₁And L₂, R₂Be primitive, namely comprise two groups of zero points of rising edge and negative edge to calculating phase differential.When adopting said method calculated rate and phase differential, the fluctuation of result of calculation obviously reduces.

After calculating signal frequency, phase differential, result of calculation is sorted, and abandon maximum, smallest point, get rid of singular point, after intermediate point is averaged, again the result is sent into second level running mean.The numerical values recited of the average array length MeanN of macro definition in the software, and open up a TimeDiffAarry[MeanN] array, then each running mean is calculated as:

TimeDiffMean=TimeDiffMean+(TimeDiff-TimeDiffAarry[mean_cur])/MeanN;

TimeDiffAarry[mean_cur]=TimeDiff;

mean_cur++;

if(mean_cur>= MeanN)

{mean_cur=0;}

As seen, only by the size of macro definition MeanN, can change the length of running mean array, take into account algorithm response speed and response accuracy, and do not need to change the internal algorithm program.

In cputimer0 interrupts, flow need be converted to electric current output.Send 12 bit serial data with GPIO simulation SPI sequential to DA7513, finish digital quantity to the conversion of analog voltage amount by D/A, again by a V/I translation circuit, be 4～20mA with voltage transitions, be that DSP needs earlier instantaneous delivery to be converted to electric current, be voltage with current conversion again, write the corresponding digital quantity of voltage therewith to D/A at last.

It then is to finish by the ePWM module that instantaneous delivery is converted to pulse, and the present invention has designed a kind of software and realized the quantitatively method of output pulse.In the 1s that cputimer0 produces interrupts, the flow rate conversion that is accumulated in this 1s is become will export the number of pulse, and stipulate these pulses output when cputimer0 interrupts next time finish (in 1s, export and finish).The flow value out_flow that exports gets the integral part of integrated flow in this 1s, and fraction part left_flow is added to next time and calculates.If pulse equivalency is the 1g/ pulse, then in 1s, export N=out_flow pulse, the frequency of pulse is f_PwmSo=N is the periodic quantity PRD=f of general purpose timer_TBCLK/ f_Pwm, base when TBCLK is the input of timer, its frequency is f_TBCLK=HSPCLK/DIV, HSPCLK=75MHz is the high-frequency clock of DSP, and DIV is divide ratio, and value is 1,2,4,8,16,32,64,128.Because of (f_TBCLK/ f_Pwm) the possibility of result be decimal, and PRD only gets result's integral part, maximum error is-1.Therefore, pwm pulse of every output, timer can be counted a TBCLK pulse at most more, so in 1s, timer at most can many meter f_PwmIndividual TBCLK pulse, but as long as guarantee CMP〉fpwm, the part of many meters just can not produce new pwm pulse.In cputimer0 interrupts next time, can be earlier with timer count value CNT zero clearing, the part of many meters can not add up, as shown in figure 10.

As seen, a certain number of pulses switch of output becomes the PWM ripple of output certain frequency in 1s.The main task of upgrading the PWM output valve namely is according to different f_Pwm, select corresponding divide ratio DIV, guaranteeing PRD between 0～65535, and CMP f_PwmAt HSPCLK=75MHz, under the CMP=PRD/2 situation, the frequency range that DSP can export is 8.94～6123Hz, at the sensor of different bores, can revise pulse equivalency, cooperates pulse output.At this, Figure 11 provides the concrete allocation scheme of a kind of DIV.

It is unusual to occur some in the measuring process, the situation that for example external device initialization error, transmitter do not have that input signal, measurement result go beyond the scope, Array Bound, data cover etc., for the ease of monitoring these mistakes, also debugging for convenience simultaneously, in system software, add error module, be used for the abnormal conditions that produce in mark and the logging program implementation.Definition ERROR structure and global variable cmf_error:

typedef struct

{ int16 ad_error;

int16 no_signal_error;

int16 array_overflow_error;

int16 zero_cross_error;

int16 eeprom_error;

int16 ad_temp_error;

int16 error_flag; }

ERROR;

ERROR cmf_error;

When the configuration peripheral hardware, to peripheral hardware operate, when algorithm process is made mistakes, the number of times that corresponding module will be made mistakes at corresponding program segment adds 1, and the error_flag of set simultaneously, inquiry cmf_error variable state, and take corresponding error handling processing measure.

Claims (1)

1. little curved Ke's mass flowmeter digital signal processing method, adopt little curved Ke's mass flowmeter digital signal processing system, system comprises the amplification filtering circuit, analog to digital converter ADC1, analog to digital converter ADC2, voltage-reference, voltage follower, differential amplifier, analog to digital converter ADC3, the analog-driven module, digital signal processor DSP, extend out SRAM, extend out EEPROM, man-machine interface, the output of 4～20mA electric current and pulse output module, power module and software, it is characterized in that: the vibrator of analog-driven module excitation Coriolis mass flowmeter inside makes flowtube with natural frequency vibration; Be positioned at two magnetoelectric sensor output two paths of signals of flowtube both sides, two paths of signals is after the identical modulate circuit amplification of two-way parameter, filtering, be digital quantity by model identical analog to digital converter ADC1 and analog to digital converter ADC2 sample conversion respectively, the multichannel buffered serial port McBSP by digital signal processor DSP sends into digital signal processor DSP; Digital signal processor DSP is eliminated noise effect to the filtering of the advanced line number word of sampled signal, adopts frequency, the mistiming of calculating signal based on the zero passage detection method of Lagrange's interpolation again, obtains mass rate and the fluid density of fluid again; Steady current is through Coriolis mass flowmeter temperature inside sensor RTD, temperature sensor RTD output voltage is behind differential amplifier, be that digital quantity is sent into digital signal processor DSP by analog to digital converter ADC3 sample conversion, digital signal processor DSP compensates according to the temperature convection amount; The measurement result of digital signal processor DSP shows at liquid crystal, and by the output of 4～20mA electric current and pulse output module, exports corresponding electric current and pulse signal; After system powers on, analog-driven produces and drives the vibration that signal is kept Coriolis mass flowmeter on the one hand, on the other hand, the initialization of the initialization of system's various piece and algorithm, variable is finished in digital signal processor DSP inside, opens then and the sample conversion of synchronized AD converter ADC1 and analog to digital converter ADC2 two-way AD; After data-switching is good, produce interruption by GPIO; Interrupt service routine is sent to the receiving register of McBSP with the AD transformation result, by the direct memory access (DMA) dma module, the data of McBSP receiving register is sent in the interim array in user-defined random access memory ram space; After interim array is deposited and expired, produce DMA and interrupt, in the interrupt service routine, the new data in the interim array is stored in the user-defined SRAM of the extending out circulation array, be used for algorithm and transfer data; Whenever producing 500 new point data, begin to call algoritic module data are handled;

After producing 500 new point data, algorithm is transferred 500 point data from the circulation array, handle; At first judge the signal amplitude size, if signal amplitude, show that Coriolis mass flowmeter does not also open fully less than setting value and shake, continue to wait for; If signal amplitude is greater than setting value, showing that Coriolis mass flowmeter has opened shakes, earlier two paths of signals is carried out pre-service by the IIR bandpass filter of a passband width, depth-adjustment, eliminate noise, call the zero passage detection method based on Lagrange's interpolation then, calculate frequency, the mistiming of two paths of signals; Result of calculation is sorted, get rid of singular value, and through after the running mean, obtain measuring average, in conjunction with instrument coefficient, calculate mass rate; Read temperature value, flow is carried out temperature compensation;

The sampled signal that digital signal processor DSP is handled is discrete signal, and not lucky zero crossing carries out curve fitting to sampled data, obtains signal zero crossing; Adopt Lagrangian quadratic interpolation match; If sample sequence x (n) when x (n-1) x (n)＜0 occurring, namely shows at [n-1, n] to have a zero point constantly; Generate quadratic interpolation polynomial expression x=at²+ bt+c, wherein

a＝0.5x(n-2)-x(n-1)+0.5x(n)

b＝-0.5x(n-2)(n-1+n)+x(n-1)(n-2+n)-0.5x(n)(n-2+n-1)

c＝0.5x(n-2)(n-1)n-x(n-1)(n-2)n+0.5x(n)(n-2)(n-1)

Moment corresponding t, i.e. zero crossing when obtaining x=0 by solving an equation; Give up that root outside [n-1, n] in actual applications;

After 500 point data find zero point, calculate signal parameter; When asking phase differential, need the zero point of two paths of signals, dislocation takes place easily subtract each other; If two paths of signals is respectively L and R, and 3 pairs of zero points of establishing L road and R road are to being respectively L₁And R₁, L₂And R₂, L₃And R₃, if 500 do not comprise L just₁, will produce L₂-R₁Such dislocation supposes that two paths of signals respectively calculates nL, nR zero point, the phase range of little curved Coriolis mass flowmeter is no more than ± and 10 °, do as judging:

(1) as if ZeroL (0)-ZeroR (0)＜-10 °, then modification method is subtracted each other in L road signal lag and generation dislocation: (nL-1) to zero point, the mistiming is tried to achieve by ZeroL (n+1)-ZeroR (n);

(2) as if ZeroL (0)-ZeroR (0)＞-10 °, then modification method is subtracted each other in R road signal lag and generation dislocation: (nR-1) to zero point, the mistiming is tried to achieve by ZeroL (n)-ZeroR (n+1);

(3) if-10 °＜ZeroL (0)-ZeroR (0)＜10 °, no matter be that L or R road signal lag dislocation can not take place subtract each other, have min (nL, nR) to zero point, the mistiming is directly tried to achieve by ZeroL (n)-ZeroR (n);

Be the influence that reduces to setover to frequency computation part, use L₁, L₃Ask for signal frequency, the zero point that namely guarantees to ask for frequency and be employed zero point a signal period is right; When calculating phase differential, with L₁, R₁And L₂, R₂Be primitive, namely comprise two groups of zero points of rising edge and negative edge to calculating phase differential.

Images