CN101493354A - Material level detecting method for tube mill based on multi-sensor fusing technology - Google Patents

Abstract

The invention relates to a drum ball mill level detection method, in particular to a drum ball mill level detection method based on a multi-sensor technique, and solves the problem of low detection precision in the existing drum ball mill level detection methods. The method comprises the following steps: (1) respectively acquiring noise signals s1, s2 and s3 at a drum of a drum ball mill, the position at which self-background noise is higher and the environment in which the drum ball mill is located; (2) performing conditioning, analog to digital conversion and pre-whitening processing on the noise signals s1, s2 and s3 to obtain signals w1, w2 and w3; (3) respectively performing blind source separation processing on the signals w1, w2 and w3 to obtain signals x1, x2 and x3; (4) screening information representing drum ball mill level information from the signals x1, x2 and x3, and performing discrete Fourier transform, module taking and normalization on the information to obtain an output signal z(n); and (5) computing the value of a drum ball mill level L according to the output signal z(n). The detection method is reasonable, can effectively reduce interference of environmental noise and the self-background noise, and can perform accurate and reliable detection on the drum ball mill level.

Description

Ball mill level detection method based on multi-sensor fusion technology

Technical field

The present invention relates to ball mill level detection method, specifically is a kind of ball mill level detection method based on multi-sensor fusion technology.

Background technology

The roller milling machine is a crucial powder manufacturing apparatus in a kind of commercial production, and it is widely used in industries such as electric power, mine, metallurgy, chemical industry and building materials, and main effect is that raw material or semi-manufacture are ground into powder, uses for subsequent handling.For example, in thermal power generation, raw coal must grind to reach certain fineness fully through bowl mill, just can enter the burner hearth burning; In cement production process, the roller milling machine still is the main component devices of raw grinding and cement grinding system.

The roller milling machine belongs to the equipment of a kind of high energy consumption, poor efficiency, is consuming a large amount of electric energy every day.The roller milling machine utilizes the rotating cylinder of low speed to take certain altitude to as a large amount of steel balls of abrasive body, and a large amount of then steel ball impellings are fallen, and material is smashed.The steel ball that the roller milling machine is loaded and the weight ratio of material often reach ten to one and even higher, and therefore a large amount of energy are by the bowl mill autophage.According to statistics, in the cement mill, the power consumption more than 70% is to consume on grinding machine, and in the electric energy that grinding machine consumed only the energy less than 10% really be used for grinding of raw material, and the energy more than 90% is converted into vibration, heating and noise.And in the fuel-burning power plant, the energy resource consumption of roller milling machine accounts for 1/3rd of station service.Because the roller milling machine is operated under the rotation status, the equipment that all the time lacks direct its inner material level of detection, phenomenons such as full mill, empty mill and stifled mill often take place, have to often stop material or stop mill, cause production efficiency to reduce, also influence the raising of output and quality, the life-span to equipment also can produce adverse consequences simultaneously.According to incompletely statistics, existing nearly 30,000 large-size ball mills in the whole nation are applied in various industries such as electric power, petrochemical industry, metallurgy, chemical industry, mine, cement.According to experimental study, bowl mill all has the energy-saving potential more than at least 10% and saves the raw-material potentiality of 9% above steel, need can accurately measure the material level of roller milling machine to the excavation of these potentiality, and control effectively.Therefore, the reliable measurements ball mill level, and the stable operation of index drum formula bowl mill is in optimum Working, for the stability that improves its operation and economy and energy-saving and cost-reducing having great significance.

At present, ball mill level detection method is mainly contained differential pressure method, power method, vibration measuring method and acoustic method.1, differential pressure method utilizes the differential pressure of bowl mill gateway to represent the material level of bowl mill inside.Though can represent the material stock amount with differential pressure in theory,, also be subjected to the influence of factors such as the steel ball loading capacity of bowl mill and ventilation because bowl mill gateway differential pressure is not only relevant with the material stock amount of bowl mill.Therefore, differential pressure method can not reflect the material level information of bowl mill exactly in the use of reality.The variation of current of electric reflected a kind of method of material stock amount indirectly when 2, power method moved by detecting bowl mill, because the current of electric of bowl mill mainly is subjected to the influence of steel ball loading capacity, unloaded variation with full load mill electric current is little, and be not single valued relation, so the precision of this method is not high.3, vibration measuring method is determined the material stock amount by frequency and the amplitude vibrated in the low speed coal mill operational process, it is bigger that but this method lags behind, be subjected to the vibration effect of miscellaneous equipment in the environment easily, because sensor directly contacts with the bowl mill body of vibration sensor is damaged easily in addition, the maintenance cost height is unfavorable for the continuous operation of system.4, the acoustic method is commonly called as the electric ear method, and it utilizes collision produces between the liner plate of steel ball and bowl mill inwall in the bowl mill operational process and the steel ball voice signal to reflect the material level of bowl mill.When coal load quantity after a little while, between the steel ball and the bump between steel ball and the liner plate more violent, sound is bigger, sound reduces with the increase of coal load quantity.Owing to be non-cpntact measurement, this method has obtained to use comparatively widely.But also there is defective in this method, mainly contains following several respects: the first, and the noise of adjacent bowl mill can produce bigger influence to measuring; The second, when high charge level moves, because bowl mill materials inside compactedness height, the noise magnitude that steel-ball collision produces is lower, and the ground unrest of equipment such as the machinery of bowl mill and motor transmission mechanism is relative higher, and it is low to cause high charge level to measure sensitivity, can not accurately reflect material level.

Summary of the invention

Problems such as the present invention can not accurately reflect level of material for ball mill information for the existing detection method that solves ball mill level exists, accuracy of detection is low, cost is high, measurement sensitivity is low provide a kind of ball mill level detection method based on multi-sensor fusion technology.

The present invention adopts following technical scheme to realize: based on the ball mill level detection method of multi-sensor fusion technology, step is as follows:

(1), uses three audio sensors simultaneously and gather the noise signal s at roller milling machine cylinder place respectively ₁, the relative higher position of the own ground unrest of roller milling machine (referring generally to the position of body machinery and motor transmission mechanism) noise signal s ₂, and the noise signal s of roller milling machine environment of living in ₃

(2), the noise signal s that step (1) is collected ₁, s ₂, s ₃Nurse one's health signal u ₁, u ₂, u ₃, with signal u ₁, u ₂, u ₃Analog to digital conversion gets numerical sequence signal v ₁, v ₂, v ₃, then with numerical sequence signal v ₁, v ₂, v ₃Prewhitening handle signal w ₁, w ₂, w ₃, wherein, to noise signal s ₁, s ₂, s ₃The signal u that obtains after the conditioning ₁, u ₂, u ₃Be suitable for carrying out the input range scope of the A/D converter that analog to digital conversion adopts;

(3), the signal w that respectively step (2) is obtained ₁, w ₂, w ₃Carry out blind source separating treatment and get three signal x independently on statistics ₁, x ₂, x ₃

(4), because the solving result of blind source separating treatment has uncertainty, mainly show as blind source and separate uncertain, the amplitude of signal of arrangement position of back signal phasor and the variation of initial phase; Therefore need from signal x ₁, x ₂, x ₃In screen out the signal of representing ball mill level information; Discriminating method is as follows: a, to the noise signal s of roller milling machine cylinder place ₁The corresponding signal w that after step 2 is handled, obtains ₁, and the signal x that obtains through step (3) ₁, x ₂, x ₃Carry out spectrum analysis, press following formula signal w ₁, x ₁, x ₂, x ₃Carry out the conversion of N point discrete Fourier,

$F_w\left(n\right)=\underset{k=0}{\overset{N-1}{\mathrm{\&Sigma;}}}{w}_1\left(k\right)e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20091007386200121.png,A20091007386200122.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;nk}/N},n=\mathrm{0,1,2},...,N-1---\left(1\right)$

$F_i\left(n\right)=\underset{k=0}{\overset{N-1}{\mathrm{\&Sigma;}}}{x}_i\left(k\right)e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20091007386200121.png,A20091007386200122.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;nk}/N},n=\mathrm{0,1,2},...,N-1,i=\mathrm{1,2,3}---\left(2\right)$

B, press following formula to signal w ₁, x ₁, x ₂, x ₃Discrete Fourier transform (DFT) delivery as a result, and normalized,

$A_w\left(n\right)=\frac{\left|F_w\left(n\right)\right|}{\max \left(\right|F_w\left(n\right)\left|\right)}---\left(3\right)$

$A_i\left(n\right)=\frac{\left|F_i\left(n\right)\right|}{\max \left({|F}_i\left(n\right)\right|)},i=\mathrm{1,2,3}---\left(4\right)$

C, obtain A respectively by following formula _i(n), i=1,2,3 and A _w(n) Euclidean distance D _i,

$D_i=\mathrm{dis}\tan \mathrm{ce}(A_i,A_w)=\underset{k=0}{\overset{N-1}{\mathrm{\&Sigma;}}}\sqrt{{\left|\right|A_{i,k}-A_{w,k}\left|\right|}^2},i=\mathrm{1,2,3}---\left(5\right)$

The Euclidean distance of two groups of signals has been represented the similarity of two groups of signals, and Euclidean distance is more for a short time to mean that two groups of signals are similar more, therefore with minimum D _iCorresponding signal x _iBe the signal of representing ball mill level information, get and minimum D _iCorresponding A _i(n) first half in the sequence as sequence length is

Output signal z (n), promptly

z(n)＝A _i(n)， $n=\mathrm{0,1,2},...,\frac{N}{2}-1---\left(6\right);$

(5), according to the output signal z (n) of step (4), be calculated as follows the value of material level L in the roller milling machine,

$L=\frac{\underset{n=0}{\overset{N_c}{\mathrm{\&Sigma;}}}z\left(n\right)}{\underset{n=0}{\overset{\frac{N}{2}-1}{\mathrm{\&Sigma;}}}z\left(n\right)}---\left(7\right),$

Wherein, N in the formula (7) _cDefinite method of value is as follows:

Under the ball mill level condition with higher, select two kinds of material level operating modes, (1) set by step-(4) are obtained and two kinds of output signal z that the material level operating mode is corresponding _h(n), z _l(n); Make N then _cFrom 1 to

Value successively, according to two kinds of output signal z that the material level operating mode is corresponding _h(n), z _l(n), respectively two kinds of material level operating modes are carried out material level according to the formula (7) in the step (5) and calculate, obtain and each N _cCorresponding material level value L _h(k), L _l(k), wherein, $k=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,\frac{N}{2}-1;$ L _h(k)-L _lPairing k value was the N that will determine when (k) result's absolute value was maximum _cValue.

Compared with prior art, the present invention adopts a plurality of audio sensors to obtain the noise signal of roller milling machine cylinder, the ambient noise signal of roller milling machine itself and the noise signal of roller milling machine site environment of living in respectively; And utilize blind source separate technology that the noise signal of being gathered is handled, using Spectral Analysis Method screens signal, separate, determine from noise signal representing the sound signal of level of material for ball mill, adopt the frequency spectrum ratioing technigue that the material level in the roller milling machine is calculated then, the accuracy height, the precision height, highly sensitive.Wherein, blind source separate technology is a kind of information processing method that just grew up in recent years, by specificity analysis to the mixed signal of gathering, source signal in the mixed signal can be separated, and do not need the too much information of source signal and transmission channel, be widely used in fields such as Audio Signal Processing and picture signal processing.At present, veteran operations staff can judge the material level of bowl mill inside by on-the-spot sound in the site environment of complexity, chief reason is the characteristic of people's ear and the human auditory system processing power to voice signal, and the method for the invention just is to have imitated the characteristic of people's ear and human auditory system the processing power of voice signal is realized judgement to ball mill level.

The method of the invention is reasonable, can reduce neighbourhood noise, ground unrest own and adjacent bowl mill interference of noise effectively, and ball mill level is detected accurately and reliably.

Description of drawings

Fig. 1 is that synoptic diagram is arranged in the sensing of the method for the invention step 1 sound intermediate frequency sensor;

Fig. 2 is the process flow diagram of the method for the invention;

1,2, the 3-audio sensor among the figure:; 4-roller milling machine; The 5-motor transmission mechanism; The 6-blanking port.

Embodiment

As shown in Figure 1, 2, based on the ball mill level detection method of multi-sensor fusion technology, step is as follows:

(1), uses three audio sensors simultaneously and gather the noise signal s at roller milling machine cylinder place respectively ₁, the relative higher position of the own ground unrest of roller milling machine (referring generally to the position of body machinery and motor transmission mechanism) noise signal s ₂, and the noise signal s of roller milling machine environment of living in ₃

(2), the noise signal s that step (1) is collected ₁, s ₂, s ₃Nurse one's health signal u ₁, u ₂, u ₃, with signal u ₁, u ₂, u ₃Analog to digital conversion gets numerical sequence signal v ₁, v ₂, v ₃, then with numerical sequence signal v ₁, v ₂, v ₃Prewhitening handle signal w ₁, w ₂, w ₃, wherein, to noise signal s ₁, s ₂, s ₃The signal u that obtains after the conditioning ₁, u ₂, u ₃Be suitable for carrying out the input range scope of the A/D converter that analog to digital conversion adopts;

(3), the signal w that respectively step (2) is obtained ₁, w ₂, w ₃Carry out blind source separating treatment and get three signal x independently on statistics ₁, x ₂, x ₃

(4), from signal x ₁, x ₂, x ₃In screen out the signal of representing ball mill level information; Discriminating method is as follows: a, to the noise signal s of roller milling machine cylinder place ₁The corresponding signal w that after step 2 is handled, obtains ₁, and the signal x that obtains through step (3) ₁, x ₂, x ₃Carry out spectrum analysis, press following formula signal w ₁, x ₁, x ₂, x ₃Carry out the conversion of N point discrete Fourier,

$F_w\left(n\right)=\underset{k=0}{\overset{N-1}{\mathrm{\&Sigma;}}}{w}_1\left(k\right)e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20091007386200121.png,A20091007386200122.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;nk}/N},n=\mathrm{0,1,2},...,N-1---\left(1\right)$

$F_i\left(n\right)=\underset{k=0}{\overset{N-1}{\mathrm{\&Sigma;}}}{x}_i\left(k\right)e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20091007386200121.png,A20091007386200122.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;nk}/N},n=\mathrm{0,1,2},...,N-1,i=\mathrm{1,2,3}---\left(2\right)$

B, press following formula to signal w ₁, x ₁, x ₂, x ₃Discrete Fourier transform (DFT) delivery as a result, and normalized,

$A_w\left(n\right)=\frac{\left|F_w\left(n\right)\right|}{\max \left(\right|F_w\left(n\right)\left|\right)}---\left(3\right)$

$A_i\left(n\right)=\frac{\left|F_i\left(n\right)\right|}{\max \left({|F}_i\left(n\right)\right|)},i=\mathrm{1,2,3}---\left(4\right)$

C, obtain A respectively by following formula _i(n), i=1,2,3 and A _w(n) Euclidean distance D _i,

$D_i=\mathrm{dis}\tan \mathrm{ce}(A_i,A_w)=\underset{k=0}{\overset{N-1}{\mathrm{\&Sigma;}}}\sqrt{{\left|\right|A_{i,k}-A_{w,k}\left|\right|}^2},i=\mathrm{1,2,3}---\left(5\right)$

With minimum D _iCorresponding signal x _iBe the signal of representing ball mill level information, get and minimum D _iCorresponding A _i(n) first half in the sequence as sequence length is

Output signal z (n), promptly

z(n)＝A _i(n)， $n=\mathrm{0,1,2},...,\frac{N}{2}-1---\left(6\right);$

(5), according to the output signal z (n) of step (4), be calculated as follows the value of material level L in the roller milling machine,

$L=\frac{\underset{n=0}{\overset{N_c}{\mathrm{\&Sigma;}}}z\left(n\right)}{\underset{n=0}{\overset{\frac{N}{2}-1}{\mathrm{\&Sigma;}}}z\left(n\right)}---\left(7\right),$

Wherein, N in the formula (7) _cDefinite method of value is as follows:

Under the ball mill level condition with higher, select two kinds of material level operating modes, (1) set by step-(4) are obtained and two kinds of output signal z that the material level operating mode is corresponding _h(n), z _l(n); Make N then _cFrom 1 to

Value successively, according to two kinds of output signal z that the material level operating mode is corresponding _h(n), z _l(n), respectively two kinds of material level operating modes are carried out material level according to the formula (7) in the step (5) and calculate, obtain and each N _cCorresponding material level value L _h(k), L _l(k), wherein, $k=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,\frac{N}{2}-1;$ L _h(k)-L _lPairing k value was the N that will determine when (k) result's absolute value was maximum _cValue.

During concrete enforcement, the roller milling machine 4 noise signal s of cylinder place ₁Collection generally realize by the audio sensor 1 that points to roller milling machine 4 blanking ports 6 positions; The noise signal s of the relative higher position of the ground unrest of roller milling machine own ₂Collection realize by the audio sensor 2 that points to the higher relatively motor transmission mechanism 5 of roller milling body ground unrest; Roller milling machine environmental audio signal s of living in ₃Collection by pointing to audio sensor 3 realizations opposite with audio sensor 1.

Claims (1)

1, a kind of ball mill level detection method based on multi-sensor fusion technology is characterized in that step is as follows:

(1), uses three audio sensors simultaneously and gather the noise signal s at roller milling machine cylinder place respectively ₁, the relative higher position of the ground unrest of roller milling machine own noise signal s ₂, and the noise signal s of roller milling machine environment of living in ₃

(2), the noise signal s that step (1) is collected ₁, s ₂, s ₃Nurse one's health signal u ₁, u ₂, u ₃, with signal u ₁, u ₂, u ₃Analog to digital conversion gets numerical sequence signal v ₁, v ₂, v ₃, then with numerical sequence signal v ₁, v ₂, v ₃Prewhitening handle signal w ₁, w ₂, w ₃, wherein, to noise signal s ₁, s ₂, s ₃The signal u that obtains after the conditioning ₁, u ₂, u ₃Be suitable for carrying out the input range scope of the A/D converter that analog to digital conversion adopts;

(3), the signal w that respectively step (2) is obtained ₁, w ₂, w ₃Carry out blind source separating treatment and get three signal x independently on statistics ₁, x ₂, x ₃

(4), from signal x ₁, x ₂, x ₃In screen out the signal of representing ball mill level information; Discriminating method is as follows: a, to the noise signal s of roller milling machine cylinder place ₁The corresponding signal w that after step 2 is handled, obtains ₁, and the signal x that obtains through step (3) ₁, x ₂, x ₃Carry out spectrum analysis, press following formula signal w ₁, w ₁, w ₂, w ₃Carry out the conversion of N point discrete Fourier,

$F_w\left(n\right)=\underset{k=0}{\overset{N-1}{\mathrm{\&Sigma;}}}{w}_1\left(k\right)e^{-j2\mathrm{\&pi;nk}/N},$ n＝0，1，2，...，N-1 (1)

$F_i\left(n\right)=\underset{k=0}{\overset{N-1}{\mathrm{\&Sigma;}}}{x}_i\left(k\right)e^{-j2\mathrm{\&pi;nk}/N},$ n＝0，1，2，...，N-1，i＝1，2，3 (2)

B, press following formula to signal w ₁, w ₁, w ₂, w ₃Discrete Fourier transform (DFT) delivery as a result, and normalized,

$A_w\left(n\right)=\frac{\left|F_w\left(n\right)\right|}{\max \left(\right|F_w\left(n\right)\left|\right)}---\left(3\right)$

$A_i\left(n\right)=\frac{\left|F_i\left(n\right)\right|}{\max \left(\right|F_i\left(n\right)\left|\right)},i=\mathrm{1,2,3}---\left(4\right)$

C, obtain A respectively by following formula _i(n), i=1,2,3 and A _w(n) Euclidean distance D _i,

$D_i=\mathrm{dis}\tan \mathrm{ce}(A_i,A_w)=\underset{k=0}{\overset{N-1}{\mathrm{\&Sigma;}}}\sqrt{{\left|\right|A_{i,k}-A_{w,k}\left|\right|}^2},i=\mathrm{1,2,3}---\left(5\right)$

With minimum D _iCorresponding signal x _iBe the signal of representing ball mill level information, get and minimum D _iCorresponding A _i(n) first half in the sequence as sequence length is Output signal z (n), promptly

z(n)＝A _i(n)， $n=\mathrm{0,1,2},...,\frac{N}{2}-1---\left(6\right);$

(5), according to the output signal z (n) of step (4), be calculated as follows the value of material level L in the roller milling machine,

$L=\frac{\underset{n=0}{\overset{N_c}{\mathrm{\&Sigma;}}}z\left(n\right)}{\underset{n=0}{\overset{\frac{N}{2}-1}{\mathrm{\&Sigma;}}}z\left(n\right)}---\left(7\right),$

Wherein, N in the formula (7) _cDefinite method of value is as follows:

Under the ball mill level condition with higher, select two kinds of material level operating modes, (1) set by step-(4) are obtained and two kinds of output signal z that the material level operating mode is corresponding _h(n), z _l(n); Make N then _cFrom 1 to

Value successively, according to two kinds of output signal z that the material level operating mode is corresponding _h(n), z _l(n), respectively two kinds of material level operating modes are carried out material level according to the formula (7) in the step (5) and calculate, obtain and each N _cCorresponding material level value L _h(k), L _l(k), wherein, k=1,2 ..., L _h(k)-L _lPairing k value was the N that will determine when (k) result's absolute value was maximum _cValue.

Images