CN101900595A - Intelligent material level detection method of double inlet and outlet coal mill - Google Patents
Intelligent material level detection method of double inlet and outlet coal mill Download PDFInfo
- Publication number
- CN101900595A CN101900595A CN2009100117442A CN200910011744A CN101900595A CN 101900595 A CN101900595 A CN 101900595A CN 2009100117442 A CN2009100117442 A CN 2009100117442A CN 200910011744 A CN200910011744 A CN 200910011744A CN 101900595 A CN101900595 A CN 101900595A
- Authority
- CN
- China
- Prior art keywords
- material level
- coal
- signal
- noise
- coal mill
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Disintegrating Or Milling (AREA)
Abstract
The invention belongs to the field of precise detection of coal mill material level and particularly relates to an intelligent material level detection method of a double inlet and outlet coal mill and adopted equipment thereof. Firstly, a differential head is used for primary material level detection; when the detected material level is lower than the defined low coal level or higher than the defined high coal level, differential pressure in the barrel of the coal mill is measured, and the material level is detected according to the relationship between the differential pressure and a coal storage material level; when the detected material level is positioned between the defined low coal level and the high coal level, the differential head is switched into a noise method to detect. The noise method comprises: (A) collecting the noise signal of the barrel of the coal mill on site; (B) carrying out wavelet packet transform to the noise signal in the step (A); (C) reconstructing a signal with the characteristic frequency band disclosed in step (B); (D) utilizing the input and output relationship of a neural network to realize the corresponding relationship of the energy signal of each frequency band of the noise and the coal mill material level. The invention has high detection precision, good stability, low cost and wide applicable range.
Description
Technical field
The invention belongs to the accurate detection range of coal pulverizer material level, relate in particular to a kind of intelligent material level detection method of double inlet and outlet coal mill.
Background technology
That double inlet and outlet coal mill equipment has is rational in infrastructure, the reserve capabillity of function admirable, reliability height, equipment is big, equipment operation flexibly, exert oneself and fineness is stable, the consumable accessory life-span is long, the rate of working continuously advantages of higher, the application of succeeding in industries such as thermal power generation.And in fact most of coal pulverizer system does not all have to move under optimum condition, makes the pulverized coal preparation system consumption rate quite big, and its one of the main reasons is also not have level gauging means in a kind of mill tube accurately and reliably up to now.
At present, coal pulverizer material level detection method commonly used both at home and abroad mainly contains pressure differential method, bear vibration method, current method, Strain Method, sonic method etc.Wherein differential pressure method is a kind of traditional indirect measurement method, also is maximum a kind of method of using at present.Its principle of work is: if coal is many in the coal pulverizer, material level is just high, and pressure reduction is just big; On the contrary, if coal is few in the coal pulverizer, material level is just low, and pressure reduction is just little, and pressure reduction and material level exist certain relation.But differential pressure signal is not the single-valued function of coal load quantity, but the multivariate function of coal load quantity, ventilation, coal pulverizer structural parameters.Therefore differential pressure signal can not reflect coal load quantity exactly.Studied of Anhui electric power experimental study carried out the material level detection method of coal pulverizer by current method, but because the complicacy of double inlet and outlet coal mill operating condition, tends to occur the situation of electric current shakiness, causes the out of true of measurement.Kolacz etc. have proposed the material level of strain method for measuring coal pulverizer, and the weight effect by steel ball in the cylindrical shell and coal cinder causes that the deflection of cylindrical shell carries out direct level gauging.Yet these methods or be to measure inaccurately or are that the refitting engineer amount is big, and investment is big, so fail to promote the use of always.Ultrasonic type and nuclear radiation formula level sensor mostly are external introducing product, ultrasonic type level-sensing device as triumphant auspicious (Kay-Ray) company of the U.S. and mug(unit of measure) nit (Magnitrol) company, but examined scope of this kind equipment and sound power restriction, in the scope of minimum blind area (0.7m), can not use, because maximum distance is limited by acoustical power, can only survey the material level in the 40m simultaneously.The QG type of Germany E+H company is a nuclear radiation formula level sensor, but nuclear radiation has harm to surrounding environment, and be subjected to instrument fixedly range ability limit.In addition, the import checkout equipment costs an arm and a leg, and causes the complete machine cost too high, makes that the user is difficult to receive.
Summary of the invention
The present invention is intended to overcome the deficiencies in the prior art part and a kind of accuracy of detection height is provided, and good stability is with low cost, the intelligent material level detection method of the double inlet and outlet coal mill that the scope of application is extensive.
For achieving the above object, the present invention is achieved in that
A kind of intelligent material level detection method of double inlet and outlet coal mill, at first using pressure differential method to carry out preliminary material level detects, when the material level that detects is lower than the low coal position of definition or is higher than the high coal position of definition, measure the pressure reduction in the coal pulverizer cylindrical shell, material level is detected according to described pressure reduction and the relation of depositing the coal charge position; When the material level that detects is between the low coal position of definition and high coal position, switches to the noise method material level is detected.
As a kind of preferred version, noise method of the present invention comprises:
(A) collection in worksite coal pulverizer cylindrical shell noise signal;
(B) the described noise signal of step (A) is carried out wavelet package transforms, with the characteristic frequency section after obtaining decomposing;
(C) signal to the described characteristic frequency section of step (B) is reconstructed, and obtains eigenwert;
(D), utilize the input of neural network, output relation to realize the characteristic signal of each frequency range of noise and the corresponding relation of coal pulverizer material level with the input signal of the described eigenwert of step (C) as neural network.
As another kind of preferred version, after the collection in worksite coal pulverizer cylindrical shell noise signal of the present invention, transfer it to discrete digital signal, and then it is carried out sampling analysis.
Further, when step of the present invention (B) is carried out wavelet package transforms, carry out three layers of WAVELET PACKET DECOMPOSITION earlier, utilize the wavelet package reconstruction algorithm that it is reconstructed again.
Further, use the BP neural network to realize the characteristic signal of each frequency range of noise and the corresponding relation of coal pulverizer material level in the step of the present invention (D).
The present invention compared with prior art has following characteristics:
1, the present invention uses the material level that detects double inlet and outlet coal mill based on the two-factor method of noise method, by audio sensor or microphone audio frequency acquiring signal, the noise difference that its cylinder sends when working down according to different coal load quantity states is judged coal load quantity, it is safe in utilization, the automaticity height can realize effectively that double inlet and outlet coal mill detects at the accurate material level of whole service operating mode.
2, the related pick-up unit of the present invention is simple in structure, is easy to go into operation, and simple installation in use, is easy to maintenance and safeguards.
3, the present invention has the degree of accuracy height, and power consumption is low, long service life, and characteristics such as cost performance height can fully satisfy the urgent need of power station industry.It is about 2/3 that the present invention can reduce cost, and has good economic benefits.
Description of drawings
The invention will be further described below in conjunction with the drawings and specific embodiments.Protection scope of the present invention not only is confined to the statement of following content.
Fig. 1 is a dual factors coal pulverizer material level testing process block diagram of the present invention;
Fig. 2 is a noise method material level testing process FB(flow block) of the present invention;
Fig. 3 is a noise method material level checkout equipment schematic block circuit diagram of the present invention.
Embodiment
As shown in Figure 1, the present invention adopts dual factors material level detection method (pressure differential method+noise method) to detect, to improve precision and the stability that the coal pulverizer material level detects.The dual factors material level detection method is exactly to utilize pressure differential method to carry out material level earlier to detect, carry out the detection of the coal position of coal pulverizer roughly, in view of there is certain inaccuracy in pressure differential method, but under the condition of chute blockage and coal cinder, can receive effect preferably, because it is the measuring method that directly contacts with coal.So the design only utilizes pressure differential method to judge one high (chute blockage), low (coal cinder) coal position shelves, between high and low coal position, use the noise method to carry out the detection of coal-grinding machine-made egg-shaped or honey-comb coal briquets position.
When the coal pulverizer normal output was worked, the coal position was about 50%, and it is the most desirable that this moment, the noise method was carried out the detection of coal position.Interval at this because the percussive action of steel ball is stronger, so noise radiation is the strongest relatively, is that the noise method is measured interval more accurately.But traditional noise method is at coal pulverizer cylindrical shell outer sensor, and with the noise gathered is approximate cylindrical shell is produced when thinking the coal pulverizer coal-grinding noise, this point is not very accurate.Because the equipment beyond the coal pulverizer cylindrical shell all can produce certain pollution to the noise of cylindrical shell, the noise that make to detect coal pulverizer is the fusion of multiple noise informations such as the cylindrical shell noise of coal pulverizer, noise of motor, Powder discharging fan noise, gear tooth noise, and what can accurately reflect coal-grinding machine-made egg-shaped or honey-comb coal briquets position is the noise that the steel ball percussive action is produced in the coal pulverizer cylindrical shell.
Therefore the present invention adopts the wavelet packet technology that the survey noise is carried out the frequency band decomposition, obtains the information of different frequency section, and wherein the information of some frequency band is exactly the information of reflection coal-grinding machine-made egg-shaped or honey-comb coal briquets position.After obtaining reflecting the proper vector of coal position, utilize neural network to set up the corresponding relation of noise and coal position.
As shown in Figure 2, noise method of the present invention comprises:
(A) collection in worksite coal pulverizer cylindrical shell noise signal; After the collection in worksite coal pulverizer cylindrical shell noise signal of the present invention, transfer it to discrete digital signal, and then it is carried out sampling analysis.
(B) the described noise signal of step (A) is carried out wavelet package transforms, with the characteristic frequency section after obtaining decomposing; When carrying out wavelet package transforms, carry out three layers of WAVELET PACKET DECOMPOSITION earlier, utilize the wavelet package reconstruction algorithm that it is reconstructed again.
(C) signal to the described characteristic frequency section of step (B) is reconstructed, and obtains eigenwert;
(D), utilize the input of BP neural network, output relation to realize the energy signal of each frequency range of noise and the corresponding relation of coal pulverizer material level with the input signal of the described eigenwert of step (C) as neural network.
The dual factors coal pulverizer material level detection method that the present invention adopts pressure differential method and noise method to combine.At first use pressure differential method to carry out preliminary material level and detect, when the material level that detects is lower than the low coal position of definition or is higher than the high coal position of definition, use pressure differential method to detect; But the material level that detects uses the noise method of invention to carry out the material level detection between the low coal position and high coal position of definition.
It at first is that signal is carried out wavelet package transforms that noise method material level detects, purpose is for the characteristic frequency section after obtaining decomposing, signal to the characteristic frequency section is reconstructed then, obtains eigenwert it is set up model as the input signal of neural network and the coal position of coal pulverizer.After obtaining coal pulverizer characteristics of noise frequency band, problem is how relation one to one to be set up in the coal position of proper vector and coal pulverizer.The present invention adopts the neural net model establishing method, the calculated signals of each frequency band is gone out energy value, and then with the signal of each frequency range input signal as neural network, the study by neural network is to set up the corresponding relation of noise signal feature and coal pulverizer material level.
Noise method material level testing process of the present invention:
1, on-the-spot coal pulverizer cylindrical shell noise signal of being gathered at first transfers signal to discrete digital signal, then it is carried out sampling analysis.
2, the data after will changing are carried out the wavelet packet deployment analysis, carry out three layers of WAVELET PACKET DECOMPOSITION, utilize the wavelet package reconstruction algorithm that it is reconstructed, and can obtain 8 independently signals after the reconstruct.Each node is represented a signal characteristic of coal pulverizer noise signal respectively.
3, after obtaining the coefficient of each frequency band, it is reconstructed, to extract the signal of each frequency band range.Obtain wavelet packet after the reconstruct and decompose each sub-frequency bands range signal for three layers
(i=0,1, L 7).Original noise S is decomposed into
By the analysis of acquired signal frequency being determined 8 of the signal that extracts the independently frequency ranges of frequency range.
4, behind decomposition and the reconstruction signal, preserve, analyze the feature of each frequency range the signal of each frequency band.In the design, to obtain the energy of each band signal, and with it as proper vector.The design adopts three layers of WAVELET PACKET DECOMPOSITION, band signal
(i=0,1,2 ..., 7) and pairing energy tries to achieve by following formula:
X wherein
Ik(i=0,1 ..., 7, k=1,2 ..., n) expression reconstruction signal
The amplitude of discrete point.So just obtained the proper vector of the signal after the reconstruct.Because the coal position during system works in the coal pulverizer cylindrical shell changes, and has so just caused the energy of the signal of the several frequency ranges of system to change, and utilizes this to concern the relation of setting up coal pulverizer noise and coal position.Existing selected characteristic signal is:
When energy was big, the element of this proper vector was a bigger number, can bring some inconvenience in numerical analysis, therefore to carry out normalization and handle this proper vector,
5, with the energy of each frequency range of noise signal
(i=0,1, L, 7) and its corresponding material level are as the training sample of neural network.Here use the BP neural network to realize the energy of frequency range and the corresponding relation of coal pulverizer material level.
6, setting network layer, the neuron number R of the input layer of this network is all 8 mutually with the number of input vector, and the neuron number of output layer is 1, and rule of thumb can to obtain the neuron number of hidden layer be 2 * R+1=17 to formula.The transport function of the hidden neuron of network is tan sig, and the neuronic transport function of output layer is log sig, and this is because the element of object vector is positioned at the output requirement of just in time satisfying function in the interval [1,1].
7, utilize training sample that neural network is trained, after the neural network convergence, can utilize input, the output relation of neural network to realize the energy signal of each frequency range of noise and the corresponding relation of coal pulverizer material level, thereby the accurate material level of realizing coal pulverizer detect.
The equipment that intelligent material level detection method adopted of above-mentioned double inlet and outlet coal mill, it comprises noise collecting part 1, signal condition part 2, adapter 3, Signal Spacing part 4, A/D conversion portion 5, microprocessor 6 and output transform part 7; The input end of the output termination signal condition part 2 of described noise collecting part 1; The output terminal of described signal condition part 2 joins through the input end of adapter 3 with Signal Spacing part 4; The output terminal of described Signal Spacing part 4 joins through the input end of A/D conversion portion 5 with microprocessor 6; The input end of the output termination output transform part 7 of described microprocessor 6; Described signal condition part 2 comprises: audio frequency amplifier section and low-pass filtering part; The input end of the output termination low-pass filtering part of described audio frequency amplifier section.
(1) noise collecting part.Checkout equipment is converted to electric signal by microphone with the voice signal that collects.
(2) audio frequency amplifies.System is by the microphone collected sound signal, and the built-in automatic gain of process is controlled (AGC) and can be carried out audio frequency and amplify for most of electret microphones provide special-purpose low-cost, the high-quality amplifier of microphone MAX9814 of bias voltage.
(3) low-pass filtering.Other frequency contents of the non-noise of filtering reduce the interference of unnecessary frequency content to noise analysis.Concentrate on frequency domain characteristics in the 0-5KHz scope according to useful sound signal, the signal beyond the 5K is carried out low-pass filtering.System has adopted universal switch electric capacity programmable filter MAX260.Just can improve filtering characteristic by MAX260 is carried out simple programming, reach filtering performance preferably.
(4) adapter.Adopt the circuit adapter to carry out the coupling of input signal, make signal after the Filtering Processing meet the requirement of isolator input signal.In order to protect circuit-under-test, to reduce the influence of neighbourhood noise, and make the system linearity degree reach best, need carry out linear isolation processing at this to test circuit.Do not meet the requirement of isolator input signal through the signal after the Filtering Processing, so before isolating, need to carry out the adaptation processing of signal, so that signal meets the input requirement of optocoupler.
(5) Signal Spacing part.Adopting high-precision linear optical coupling TIL300 to carry out photoelectricity isolates, it is that an isolation feedback light diode and the output optical diode by infrared light LED irradiation divergent configuration is formed, this device adopts special manufacturing technology to come the non-linear of compensation LED time and temperature characterisitic, thus the linear ratio of four width of cloth luminous fluxes that output signal and LED are sent.
(6) carry out linearization process, make between the output signal of isolator and the noise signal to be collected linear, to reduce to measure the error of noise signal.
(7) A/D conversion portion.With analogue noise conversion of signals to be measured is digital signal.After a series of processing such as sound signal process to be measured amplification, filtering, isolation, for it is further handled, needing analogue noise conversion of signals to be measured be digital signal (that is A/D conversion).System selects for use four and half double integrator D/A ICL7135 as ADC, and this ADC is exportable ± 20000 sign indicating numbers, provide clock by the crystal oscillator frequency division for ADC, making sample time is the integral multiple of power frequency, can effectively suppress power frequency and disturb.
(8) microprocessor.Utilize MPU to carry out the noise material level detection method of invention.After noise signal to be measured changes digital signal into, can utilize MPU (intelligent object) to handle to measured signal, not only realized utilizing software further to handle the purpose of (small echo and neural network) to measured signal, and remedied the function that can't realize on the hardware, thereby make the performance of the software and hardware aspect of system more become rationally also perfect.
(9) output transform part.Output signal is converted to current signal, to be used for on-the-spot system's control.In order to have simplified the display unit of signal conditioning unit, and make things convenient for the field personnel to read, so the normalized current of the output 4-20mA of system.Need before this output signal of MPU is carried out the DAC conversion, system selects for use MAX504 to be DAC again.This sheet is a SPI serial line interface form, includes DAC, reference source and operational amplifier.MAX504 is converted to analog output voltage with the digital signal of input, just the data of input can be converted to the electric current output of standard 4-20mA again through electric current series connection negative-feedback circuit.
Claims (5)
1. the intelligent material level detection method of a double inlet and outlet coal mill, it is characterized in that: at first use pressure differential method to carry out preliminary material level and detect, when the material level that detects is lower than the low coal position of definition or is higher than the high coal position of definition, measure the pressure reduction in the coal pulverizer cylindrical shell, material level is detected according to described pressure reduction and the relation of depositing the coal charge position; When the material level that detects is between the low coal position of definition and high coal position, switches to the noise method material level is detected.
2. the intelligent material level detection method of double inlet and outlet coal mill according to claim 1, it is characterized in that: described noise method comprises:
(A) collection in worksite coal pulverizer cylindrical shell noise signal;
(B) the described noise signal of step (A) is carried out wavelet package transforms, with the characteristic frequency section after obtaining decomposing;
(C) signal to the described characteristic frequency section of step (B) is reconstructed, and obtains eigenwert;
(D), utilize the input of neural network, output relation to realize the characteristic signal of each frequency range of noise and the corresponding relation of coal pulverizer material level with the input signal of the described eigenwert of step (C) as neural network.
3. the intelligent material level detection method of double inlet and outlet coal mill according to claim 2 is characterized in that: after the collection in worksite coal pulverizer cylindrical shell noise signal, transfer it to discrete digital signal, and then it is carried out sampling analysis.
4. the intelligent material level detection method of double inlet and outlet coal mill according to claim 3 is characterized in that: when described step (B) is carried out wavelet package transforms, carry out three layers of WAVELET PACKET DECOMPOSITION earlier, utilize the wavelet package reconstruction algorithm that it is reconstructed again.
5. the intelligent material level detection method of double inlet and outlet coal mill according to claim 4 is characterized in that: use the BP neural network to realize the characteristic signal of each frequency range of noise and the corresponding relation of coal pulverizer material level in the described step (D).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009100117442A CN101900595A (en) | 2009-05-27 | 2009-05-27 | Intelligent material level detection method of double inlet and outlet coal mill |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009100117442A CN101900595A (en) | 2009-05-27 | 2009-05-27 | Intelligent material level detection method of double inlet and outlet coal mill |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101900595A true CN101900595A (en) | 2010-12-01 |
Family
ID=43226331
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2009100117442A Pending CN101900595A (en) | 2009-05-27 | 2009-05-27 | Intelligent material level detection method of double inlet and outlet coal mill |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101900595A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013189126A1 (en) * | 2012-06-19 | 2013-12-27 | Guo Yunchang | Method related to improving signal-to-noise ratio in passive nucleonic level gauge |
| CN106582961A (en) * | 2016-12-16 | 2017-04-26 | 大唐东北电力试验研究所有限公司 | Device for monitoring coal position of coal mill through multiquadrant noise method |
| CN111366388A (en) * | 2020-03-16 | 2020-07-03 | 重庆邮电大学 | Grinder load detection method based on wavelet packet energy spectrum |
-
2009
- 2009-05-27 CN CN2009100117442A patent/CN101900595A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013189126A1 (en) * | 2012-06-19 | 2013-12-27 | Guo Yunchang | Method related to improving signal-to-noise ratio in passive nucleonic level gauge |
| CN106582961A (en) * | 2016-12-16 | 2017-04-26 | 大唐东北电力试验研究所有限公司 | Device for monitoring coal position of coal mill through multiquadrant noise method |
| CN106582961B (en) * | 2016-12-16 | 2019-04-12 | 大唐东北电力试验研究院有限公司 | Using the device of more quadrant Noise Methods monitoring coal-grinding machine-made egg-shaped or honey-comb coal briquets position |
| CN111366388A (en) * | 2020-03-16 | 2020-07-03 | 重庆邮电大学 | Grinder load detection method based on wavelet packet energy spectrum |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105606933A (en) | Vibration-noise-based on-line monitoring system of converter transformer | |
| CN102608452A (en) | High speed train equipment state and electric energy quality monitoring system and method | |
| JP2006275845A (en) | Lightning current measurement apparatus | |
| CN101900595A (en) | Intelligent material level detection method of double inlet and outlet coal mill | |
| CN101187693A (en) | Motor performance automatic test system based on dummy instrument and its working method | |
| CN105067492A (en) | Dust concentration meter | |
| CN103049984B (en) | Combustible gas alarm instrument based on Wireless HART communication technology | |
| CN110470963A (en) | A kind of high voltage capacitive apparatus insulated performance on-line monitoring device based on DSP | |
| CN201671805U (en) | Tangential fan performance testing device | |
| CN109470389A (en) | Strain-type torch measuring system and method | |
| CN204142613U (en) | A kind of high precision small dust sensor | |
| CN103728501B (en) | A kind of continuous iron loss measurement mechanism of electrical sheet and method | |
| CN203687927U (en) | High-stability and high-accuracy biaxial inclination remote intelligent monitoring system | |
| CN205898852U (en) | System takes place for base mark signal | |
| CN206161147U (en) | Bridge vibration monitoring devices | |
| CN205982612U (en) | On --spot parameter record appearance of electric energy meter | |
| CN202383241U (en) | Ultrahigh-frequency sensor for detecting GIS partial discharge | |
| CN209157099U (en) | Vacuum pressure casting machine casting temperature measuring device | |
| CN219161436U (en) | Voiceprint monitoring system | |
| CN113702897A (en) | Anti-harmonic interference three-phase electric energy meter on-site calibration instrument | |
| CN202350904U (en) | Integrated vibration transmitter | |
| CN209876531U (en) | Pipeline leakage acoustic emission signal detection system | |
| CN202141543U (en) | Temperature-pressure complementation device of pressure sensor based on two-dimensional orthogonal function | |
| CN203719760U (en) | Intelligent grinding sound frequency spectrum analyzer | |
| CN101509928A (en) | Apparatus for detecting engine speed by exhaust pressure |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20101201 |