CN108362654A - Solid waste prevents monitoring system - Google Patents
Solid waste prevents monitoring system Download PDFInfo
- Publication number
- CN108362654A CN108362654A CN201810130556.0A CN201810130556A CN108362654A CN 108362654 A CN108362654 A CN 108362654A CN 201810130556 A CN201810130556 A CN 201810130556A CN 108362654 A CN108362654 A CN 108362654A
- Authority
- CN
- China
- Prior art keywords
- spectrum
- vector
- signal
- sampling
- sample
- 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
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2201/00—Features of devices classified in G01N21/00
- G01N2201/12—Circuits of general importance; Signal processing
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The present invention provides a kind of solid waste to prevent monitoring system comprising sequentially connected such as lower unit：Solid waste sampling unit, pretreatment unit, spectral scan unit, spectral signal processing unit and monitoring unit.Confidence level of the present invention is higher, overcome it is not good to soil metal contaminant content prediction effect in the prior art, cannot achieve soil metal contaminant content monitoring defect.
Description
Technical field
The invention belongs to field of spectral analysis technology, and in particular to a kind of solid waste prevention monitoring system.
Background technology
It is especially main to two kinds of soil with organic matter and full nitrogen at present at home and abroad about in soil constituent study on monitoring
Component content research in, it is most of to have obtained preferable achievement in research.On this basis, Many researchers also use light
Some metal element contents in spectral analysis technology quantitative analysis soil, it is pre- although also obtaining certain achievement for monitoring
Effect is surveyed but not as good as apparent to the prediction effect of organic matter and total nitrogen content.
That attempt spectral technique being introduced into Soil K+adsorption for the first time is American scientist Browsers and Hanks, 1965
Find that the organic matter in soil has relevant feature line mark near infrared spectrum region and organic matter year.Then in 1970, with
Afterwards in 1970, Salisbury and Hunt are equally found that certain minerals in soil near infrared spectrum there is also phases
The feature line mark of pass.The achievement of the above several researchers has started the new method detected for soil constituent, while in theory
On demonstrate using spectral analysis technique in soil Cucumber or ingredient carry out qualitative or quantitative analysis possibility.
At home and abroad in the result of study of scholar it is recognised that at present for soil constituent detection in, organic matter and
The prediction effect of full nitrogen is generally forged a good relationship, and, that has not consistent in each document report for the prediction effect of metallic element
The prediction effect of other scholar is preferable, and some is then not fully up to expectations, and at every in report, the processing for soil sampling object
It is not quite similar, some researchers the operations such as can be dried, be ground up, sieved to collected sample early period in experiment, experiment
As a result show that the prediction effect of spectrum analysis can be improved by carrying out processing appropriate to soil sampling object.Report proves sample
The physical properties such as water content, particle diameter precision of prediction can all be had an impact, and testing conditions such as sample whether rotate,
The factors such as detection height, angle also can all have an impact prediction effect.But the experiment conducted in having been reported that is adopted close red
External spectrum instrument carries out, and tests and carry out in laboratory environments mostly, rarely has the scene of field progress in the wild is practical to measure
Report.
When analyzing soil constituent using spectral analysis technique, the performance of spectral instrument has a fixing for prediction effect
It rings, and different prediction model method for building up has also embodied extremely important effect during prediction.Even if adopting sometimes
With the spectral instrument of same model, the method that prediction model is established is different, and obtained detection result is also different, wherein mainly adopting
With the methods of principle component regression, offset minimum binary and artificial neural network.Also have many scholars for a variety of data at
Reason method is compared, and as a result shows that each method has the characteristics that itself, but which kind of specifically used data processing method into
The testing result that row modeling can most be had, does not obtain an exact conclusion in current achievement in research.
Li Jie of China Agricultural Machinery Science Research Academy in 2012 et al. is complete to the soil sampling object near 72 Beijing
Nitrogen, full potassium, the content of organic matter and pH value are analyzed, and are modeled by Partial Least Squares, prediction result and actual value
With good consistency, the coefficient of determination reaches as high as 0.9554.What of Zhejiang University brave et al. has studied variable grain diameter
With different modeling methods for the image of soil organic matter content near-infrared spectral analytical method, sample is divided into particle in research
A diameter of 0.169-2mm and be less than two sample sets of 0.169mm, establish three kinds of prediction models, the results showed that work as particle diameter
When between 0.169-2mm, prediction related coefficient is all higher than 0.84, and predicted root mean square error is respectively less than 0.20；And when particle is straight
When diameter is less than 0.169mm, prediction related coefficient is respectively less than 0.71, and predicted root mean square error is all higher than 0.23.
From the point of view of development present situation, spectra methods is generally defined as a kind of completely new soil constituent inspection at present
Quick, the loseless method surveyed, and new development trend from now on, for solving the traditional chemical in soil constituent detection at present
The efficiency of detection method is low, it is of high cost, can not the drawbacks such as Fields detection.According to current achievement in research, spectral analysis technique for
The prediction effect of the soil organism and total nitrogen content is generally preferable, and the related coefficient of prediction result and chemical detection can generally reach
0.9 or more is even higher.The size of diameter of soil pellet has a certain impact for prediction effect, but less, soil moisture content
Also prediction result can be produced bigger effect, influence of the data processing method used by prediction mode to prediction result is little.
And generally it is not so good as the prediction effect of organic matter and total nitrogen content for the prediction effect of the metal components content such as phosphorus, potassium, calcium, magnesium
It is good, but can also reach 0.8 in some studies, what effect was relatively good in individual researchs can reach 0.9 or more, this just needs building
Suitable data processing method is selected when vertical prediction model.
Invention content
In view of the above analysis, the main purpose of the present invention is to provide a kind of solid waste to prevent monitoring system.The system
Spectral analysis technique is used, and a kind of data processing mechanism of synthesis is proposed according to the characteristics of spectrum analysis so that no
It is only preferable for the prediction effect of the soil organism and total nitrogen content, but also suitable for soil metal pollutant component content
Monitoring, it is also basicly stable good for the prediction effect of the metal pollutants component content such as phosphorus, potassium, calcium, magnesium.
The purpose of the present invention is what is be achieved through the following technical solutions.
A kind of solid waste prevention monitoring system comprising sequentially connected such as lower unit：
Solid waste sampling unit, for carrying out soil sampling to selection area；
Pretreatment unit, for handling soil sampling object；
Spectral scan unit, for carrying out spectral scan to soil sampling object；
Spectral signal processing unit, for handling spectral signal；
Monitoring unit, for will treated that signal is compared with given threshold, and to the signal value more than threshold value into
Row alarm.
Further, the solid waste sampling unit for carrying out soil sampling to selection area includes：
Number of samples determination subelement is by variance yields and average value and true value for determining number of samples P before sampling
Mathematic interpolation number of samples, and P be >=19 natural number；
Sampling processing subelement, for the size according to sampling area, using snakelike sampling, sampling depth is selected as
30CM is sampled according to calculated number of samples, when sampling, is first produced a sampling section, is carried out being parallel to section
Sampling, adopts the earthwork to perpendicular to ground, the soil after sampling is mixed, then mixing thickness soil sampling object is crushed,
Stone, root system of plant are removed, mixes and transform into quadrangle, diagonal line is divided and is divided into four parts, takes diagonal two parts therein, repeatedly
Carry out repeatedly after, by the Weight control of soil sampling object needs weight, then to by sample contents through hydrochloric acid, nitric acid, high chlorine
Acid is handled；
The pretreatment unit includes：
Cloth subelement is spread, for soil sampling object to be positioned in sample container, is shaken with making its upper surface substantially flat
Sample, is then positioned on quartz window by distribution；
Subelement is rotated, in measurement process, making soil sampling object be rotated with sample container instrument；
Spectral scan unit includes following subelement：
Averaged spectrum obtains subelement, for being carried out spectrum point in the rotary course by means of the rotation subelement
Analysis obtains the averaged spectrum obtained after multiple rotary scanning analysis, and the rotary speed of sample container is 4cm/s, and soil sampling object exists
It can be scanned in 40s 64 times, the spectrum after arithmetic mean is as a sampling spectrum；Each sample is positioned over different sample containers
In, it measures twice, the spectroscopic data after being averaged again is used for signal processing；
The signal processing is based on the spectral signal processing unit, and the spectral signal processing unit includes：
Subelement is corrected, for acquiring reference signal, measured signal is acquired, obtains original absorbance spectrum, determine correction ginseng
Reference wavelength is examined, correction reference data wavelength calibration absorption spectrum is utilized；The correction reference data wavelength by calculating as follows
It obtains：
Spectrum matrix is denoted as X (M*K), wherein i-th of wavelength variable represents i-th of column vector in spectrum matrix X, remembers N=
Min (m-1, k), the n=1 before carrying out first time interative computation, the arbitrary a certain row for selecting spectrum matrix, are denoted as variable xk(0)；
The column vector location sets not being selected into are denoted as T,
Calculate separately xk(n-1)To remaining the projection of following variables, it is denoted as：
Note k (n-1)=arg (max (| | Pj||),j∈T)；N=n+1 enables x if n=Nk(n-1)As the row being selected into
Variable, returns to above-mentioned column vector position and calculates step and calculated again, using multiple linear regression method to the variables set that newly selects
Close { xk(0),xk(1),…,xk(n-1)Assessed, obtain correction reference data wavelength value；
After the spectral scan output valve for obtaining soil sampling object, in order to remove the soil that background interference obtains more high s/n ratio
Spectroscopic data is needed with improving the robustness and predictive ability of the soil constituent prediction model that the later stage is established to P sample
Data are handled as follows：
Calculate the spectrum average of sample contents：
To unknown sample spectrum x (1 × m), pass through the following formula spectrum x that obtains that treatedcentered：
Then spectrum standard deviation is calculated:
Spectrum sample value is handled as follows：
Entire spectrum sample signal value is handled as follows, value X after being handled：
Wherein, n is sample number, is counted for wavelength, and the spectrum that Xn corresponds to the sample contents as n-th sample is equal
Value；
After obtaining spectral signal, the spectral signal processing further includes that spectrum sample, carries out 2-d wavelet change at equal intervals
It changes, then carries out matrix of wavelet coefficients stationary window transversal scanning, obtain nonstationary noise variance evaluation vector sum signal variance
Estimate vector adjusts window width, then carries out rescan point by point, and it is accurate to obtain accurate noise variance evaluation vector sum
Signal variance estimate vector, then by Bayes's threshold value shrink denoising, finally carry out 2-d wavelet contravariant exchange for central row to
Amount is used as signal value output；
It is described to include to the point-by-point adjustment of window width progress：The Noise Variance Estimation of window is fixed first, definition is solid
Determine window Wm×n, width m is set as empirical value, and height n is coefficient matrix height, by window width diagonally high frequency coefficient matrix
Center row vectorOn slide laterally point by point, whereinUsing in wavelet coefficient estimation in window
The noise variance vector of heart row, according to the change rate of noise variance vector, dynamic adjusts window width, and specially basis is made an uproar
The linear phase property of the sound property that approaches uniformity is distributed in all wavelet coefficients and wavelet basis, high-rise wavelet coefficient
Noise variance vector can be by being averaging to obtain, to signal variance vector in the noise variance vector of last layer by two neighboring variance
It is estimated, evaluation method is similar with above-mentioned noise variance vector evaluation method, but since to concentrate on part small for signal variance
In wave system number, there is larger difference on each wavelet decomposition layer, need to carry out layering estimation, in all directions coefficient square of the layer calculated
It is slid laterally respectively in the center row vector of battle array, using all wavelet coefficients respectively decomposed in this layer of window on direction, estimation should
The signal variance vector of layer, is set dynamically according to the change rate of signal variance vector, resets serial ports width, setting signal window
Mouth adjustment threshold value, every layer of signal variance vector is estimated using new window again, then acquires the threshold value of every layer of matrix of wavelet coefficients
Vector uses threshold for every a line of high frequency transverse direction coefficient matrix, longitudinal high frequency coefficient matrix, diagonal high frequency coefficient matrix
Value vector carries out contraction noise reduction, and spectrum matrix is changed in the wavelet coefficient contravariant after noise reduction, takes the center row vector of spectrum matrix
As the spectral signal after noise reduction；
The monitoring unit is compared after the spectral signal that obtains that treated with the spectral signal threshold value of setting, when
It when beyond threshold value, alarms, reaches Metal Pollution Study In Soils object content detection purpose.
Further, further include being removed background interference and noise reduction process to spectral signal.
Further, it after the spectral signal that obtains that treated, is compared with the spectral signal threshold value of setting, when beyond threshold
It when value, alarms, reaches Metal Pollution Study In Soils object content detection purpose.
Technical scheme of the present invention has the following advantages：
A kind of solid waste of the present invention prevents monitoring system, and synthesis proposes soil sampling and subsequent spectral signal
Treatment mechanism, multi task process especially has been carried out to spectral signal, and interfere it and noise has carried out fusion treatment, made
The spectral signal value that must be obtained disclosure satisfy that prediction effect, and confidence level is higher, overcome in the prior art to soil metallic pollution
Object content prediction effect is not good, cannot achieve the defect of soil metal contaminant content monitoring.
Description of the drawings
Fig. 1 shows the composition frame chart of the monitoring system of the present invention.
Specific implementation mode
The mechanism of the solid waste prevention monitoring system of the present invention determines number of samples before specifically including sampling, i.e., by variance
The mathematic interpolation number of samples P of value and average value and true value, and the natural number that P is >=19, according to the size of sampling area,
Using snakelike sampling, sampling depth is selected as 30CM, is sampled according to calculated number of samples, when sampling, first produces one
A sampling section, is sampled being parallel to section, adopts the earthwork to perpendicular to ground, the soil after sampling is mixed,
Again will mixing thickness soil sampling object break into pieces, removal stone, plant big root system, be sufficiently mixed and transform into quadrangle, divide
Diagonal line is divided into four parts, takes diagonal two parts therein, after being repeated a number of times, the Weight control of soil sampling object is being needed
Weight.
Then to handling sample contents through hydrochloric acid, nitric acid, perchloric acid.
Preferably, soil sampling object is positioned in sample container, gentle agitation makes its upper surface flat distribution, then will
Sample is positioned on quartz window, and in measurement process, soil sampling object can be rotated with sample container instrument to obtain average light
Spectrum, the rotary speed of sample container are 4cm/s, soil sampling object can be scanned in 40s 64 this, the spectrum after arithmetic mean is made
For a sampling spectrum.Each sample is positioned in different sample containers, is measured twice, the spectroscopic data after being averaged again is used for
Signal processing.
Preferably, further include absorption spectrum aligning step, including：Reference signal is acquired, measured signal is acquired, is obtained original
Absorption spectrum determines correction reference data wavelength, utilizes correction reference data wavelength calibration absorption spectrum.
The correction reference data wavelength by being calculated as follows：
Spectrum matrix is denoted as X (M*K), wherein i-th of wavelength variable represents i-th of column vector in spectrum matrix X, remembers N=
Min (m-1, k), the n=1 before carrying out first time interative computation, the arbitrary a certain row for selecting spectrum matrix, are denoted as variable xk(0)；
The column vector location sets not being selected into are denoted as T,
Calculate separately xk(n-1)To remaining the projection of following variables, it is denoted as：
Note k (n-1)=arg (max (| | Pj||),j∈T)；N=n+1 enables x if n=Nk(n-1)As the row being selected into
Variable, returns to above-mentioned column vector position and calculates step and calculated again, using multiple linear regression method to the variables set that newly selects
Close { xk(0),xk(1),…,xk(n-1)Assessed, obtain correction reference data wavelength value.
Preferably, after obtaining spectral signal, described to carry out processing to spectral signal include spectrum sample at equal intervals, is carried out
Two-dimensional wavelet transformation then carries out matrix of wavelet coefficients stationary window transversal scanning, obtains nonstationary noise variance evaluation vector
With signal variance estimate vector, window width is adjusted point by point, then carry out rescan, obtains accurate noise variance evaluation
Vector sum precise signal variance evaluation vector, then shrinks denoising by Bayes's threshold value, finally carries out 2-d wavelet inverse transformation
Take center row vector as signal value output.
After the spectral scan output valve for obtaining soil sampling object, have compared with high s/n ratio to remove background interference and obtain
Soil spectrum data are needed with improving the robustness and predictive ability of the soil constituent prediction model that the later stage is established to data
It is handled as follows：
Calculate the spectrum average of sample contents：
To unknown sample spectrum x (1 × m), pass through the following formula spectrum x that obtains that treatedcentered：
Then spectrum standard deviation is calculated:
Spectrum sample value is handled as follows：
Entire spectrum sample signal value is handled as follows, value X after being handled：
Wherein, n is sample number, is counted for wavelength, and the spectrum that Xn corresponds to the sample contents as n-th sample is equal
Value；
The spectral signal of sampling has a degree of inhomogeneities, and it is that significant non-stationary property is made an uproar to cause spectral noise
Sound, spectral noise matrix are the algebraical sums of the notable nonstationary noise on short-term stationarity noise and spectrum axis on a timeline.
Generally use wavelet transformation carries out denoising, however the Wavelet-denoising Method of single threshold value be difficult to cope with it is notable on spectrum axis
Nonstationary noise is easily lost the signal detail of low-noise area.To containing nonstationary noise image B, M layers points of small echo is carried out
Solution.The window W that an area is m*n is arranged to slide point by point in matrix of wavelet coefficients, the denoising of window center point wavelet coefficient
Threshold value only has the wavelet coefficient in window and is calculated, and the Spatial distributions estimation of noise-removed threshold value may be implemented.The area of W, by passing through
Value setting is tested, it is too small, then it can not utilize statistical law Accurate Prediction noise variance；It is too big, then it is unable to the fast of response noises variance
Speed variation.Therefore, it when dealing with objects different, needs to continuously attempt to its area, noise variance changes small image, selection
Larger window, conversely, then selection is compared with wicket.
But in this present embodiment, as a result of the soil collection and processing mode of the present embodiment, and being adopted to soil
For sample object, spectral noise concentrates on specific region, and noise variance is big in region and variation is strong, and rest part noise variance
Relatively small and steady, fixed window area is easy that threshold value estimation is made to generate error.Therefore, applicant proposed a kind of windows
The point-by-point adjustment mode of width dynamic carries out noise reduction process to spectral signal matrix.
The Noise Variance Estimation of window is fixed first, defines stationary window Wm×n, width m is set as empirical value, height
N is coefficient matrix height, by window width diagonally high frequency coefficient matrix center row vectorOn slide laterally point by point,
WhereinThe noise variance vector that central row is estimated using wavelet coefficient in window, according to the noise
The change rate of variance vectors, dynamic adjust window width, and specially according to noise, approaches uniformity is distributed in all wavelet coefficients
Property and wavelet basis linear phase property, the noise variance vector of high-rise wavelet coefficient can be by the noise side of last layer
Two neighboring variance is averaging to obtain in difference vector, signal variance vector is estimated, evaluation method and above-mentioned noise
Variance vectors evaluation method is similar, but since signal variance concentrates in the wavelet coefficient of part, has on each wavelet decomposition layer larger
Difference needs to carry out layering estimation, is slid laterally respectively in the center row vector of all directions coefficient matrix of the layer calculated,
Using all wavelet coefficients respectively decomposed in this layer of window on direction, the signal variance vector of this layer is estimated, according to signal variance
The change rate dynamic setting of vector, resets serial ports width, and setting signal window is adjusted threshold value, estimated again using new window
Every layer of signal variance vector, then the threshold vector of every layer of matrix of wavelet coefficients is acquired, for high frequency transverse direction coefficient matrix, longitudinal direction
Every a line of high frequency coefficient matrix, diagonal high frequency coefficient matrix, carries out contraction noise reduction using threshold vector, will be small after noise reduction
Spectrum matrix is changed in wave system number contravariant, takes the center row vector of spectrum matrix as the spectral signal after noise reduction.
Accordingly, as shown in Figure 1, the present invention also provides the detecting systems for implementing the above method comprising be sequentially connected
Such as lower unit：
Solid waste sampling unit, for carrying out soil sampling to selection area；
Pretreatment unit, for handling soil sampling object；
Spectral scan unit, for carrying out spectral scan to soil sampling object；
Spectral signal processing unit, for handling spectral signal；
Monitoring unit, for will treated that signal is compared with given threshold, and to the signal value more than threshold value into
Row alarm.
According to a preferred embodiment of the invention, the solid waste for carrying out soil sampling to selection area samples single
Member includes：
Number of samples determination subelement is by variance yields and average value and true value for determining number of samples P before sampling
Mathematic interpolation number of samples, and P be >=19 natural number；
Sampling processing subelement, for the size according to sampling area, using snakelike sampling, sampling depth is selected as
30CM is sampled according to calculated number of samples, when sampling, is first produced a sampling section, is carried out being parallel to section
Sampling, adopts the earthwork to perpendicular to ground, the soil after sampling is mixed, then mixing thickness soil sampling object is crushed,
Stone, root system of plant are removed, mixes and transform into quadrangle, diagonal line is divided and is divided into four parts, takes diagonal two parts therein, repeatedly
Carry out repeatedly after, by the Weight control of soil sampling object needs weight, then to by sample contents through hydrochloric acid, nitric acid, high chlorine
Acid is handled；
The pretreatment unit includes：
Cloth subelement is spread, for soil sampling object to be positioned in sample container, is shaken with making its upper surface substantially flat
Sample, is then positioned on quartz window by distribution；
Subelement is rotated, in measurement process, making soil sampling object be rotated with sample container instrument；
Spectral scan unit includes following subelement：
Averaged spectrum obtains subelement, for being carried out spectrum point in the rotary course by means of the rotation subelement
Analysis obtains the averaged spectrum obtained after multiple rotary scanning analysis, and the rotary speed of sample container is 4cm/s, and soil sampling object exists
It can be scanned in 40s 64 times, the spectrum after arithmetic mean is as a sampling spectrum；Each sample is positioned over different sample containers
In, it measures twice, the spectroscopic data after being averaged again is used for signal processing；
The signal processing is based on the spectral signal processing unit, and the spectral signal processing unit includes：
Subelement is corrected, for acquiring reference signal, measured signal is acquired, obtains original absorbance spectrum, determine correction ginseng
Reference wavelength is examined, correction reference data wavelength calibration absorption spectrum is utilized；The correction reference data wavelength by calculating as follows
It obtains：
Spectrum matrix is denoted as X (M*K), wherein i-th of wavelength variable represents i-th of column vector in spectrum matrix X, remembers N=
Min (m-1, k), the n=1 before carrying out first time interative computation, the arbitrary a certain row for selecting spectrum matrix, are denoted as variable xk(0)；
The column vector location sets not being selected into are denoted as T,
Calculate separately xk(n-1)To remaining the projection of following variables, it is denoted as：
Note k (n-1)=arg (max (| | Pj||),j∈T)；N=n+1 enables x if n=Nk(n-1)As the row being selected into
Variable, returns to above-mentioned column vector position and calculates step and calculated again, using multiple linear regression method to the variables set that newly selects
Close { xk(0),xk(1),…,xk(n-1)Assessed, obtain correction reference data wavelength value；
After the spectral scan output valve for obtaining soil sampling object, in order to remove the soil that background interference obtains more high s/n ratio
Spectroscopic data is needed with improving the robustness and predictive ability of the soil constituent prediction model that the later stage is established to P sample
Data are handled as follows：
Calculate the spectrum average of sample contents：
To unknown sample spectrum x (1 × m), pass through the following formula spectrum x that obtains that treatedcentered：
Then spectrum standard deviation is calculated:
Spectrum sample value is handled as follows：
Entire spectrum sample signal value is handled as follows, value X after being handled：
Wherein, n is sample number, is counted for wavelength, and the spectrum that Xn corresponds to the sample contents as n-th sample is equal
Value；
After obtaining spectral signal, the spectral signal processing further includes that spectrum sample, carries out 2-d wavelet change at equal intervals
It changes, then carries out matrix of wavelet coefficients stationary window transversal scanning, obtain nonstationary noise variance evaluation vector sum signal variance
Estimate vector adjusts window width, then carries out rescan point by point, and it is accurate to obtain accurate noise variance evaluation vector sum
Signal variance estimate vector, then by Bayes's threshold value shrink denoising, finally carry out 2-d wavelet contravariant exchange for central row to
Amount is used as signal value output；
It is described to include to the point-by-point adjustment of window width progress：The Noise Variance Estimation of window is fixed first, definition is solid
Determine window Wm×n, width m is set as empirical value, and height n is coefficient matrix height, by window width diagonally high frequency coefficient matrix
Center row vectorOn slide laterally point by point, whereinUsing in wavelet coefficient estimation in window
The noise variance vector of heart row, according to the change rate of noise variance vector, dynamic adjusts window width, and specially basis is made an uproar
The linear phase property of the sound property that approaches uniformity is distributed in all wavelet coefficients and wavelet basis, high-rise wavelet coefficient
Noise variance vector can be by being averaging to obtain, to signal variance vector in the noise variance vector of last layer by two neighboring variance
It is estimated, evaluation method is similar with above-mentioned noise variance vector evaluation method, but since to concentrate on part small for signal variance
In wave system number, there is larger difference on each wavelet decomposition layer, need to carry out layering estimation, in all directions coefficient square of the layer calculated
It is slid laterally respectively in the center row vector of battle array, using all wavelet coefficients respectively decomposed in this layer of window on direction, estimation should
The signal variance vector of layer, is set dynamically according to the change rate of signal variance vector, resets serial ports width, setting signal window
Mouth adjustment threshold value, every layer of signal variance vector is estimated using new window again, then acquires the threshold value of every layer of matrix of wavelet coefficients
Vector uses threshold for every a line of high frequency transverse direction coefficient matrix, longitudinal high frequency coefficient matrix, diagonal high frequency coefficient matrix
Value vector carries out contraction noise reduction, and spectrum matrix is changed in the wavelet coefficient contravariant after noise reduction, takes the center row vector of spectrum matrix
As the spectral signal after noise reduction；
The monitoring unit is compared after the spectral signal that obtains that treated with the spectral signal threshold value of setting, when
It when beyond threshold value, alarms, reaches Metal Pollution Study In Soils object content detection purpose.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
All any modification, equivalent and improvement etc., should all be included in the protection scope of the present invention made by within refreshing and principle.
Claims (2)
1. a kind of solid waste prevents monitoring system comprising sequentially connected such as lower unit：
Solid waste sampling unit, for carrying out soil sampling to selection area；
Pretreatment unit, for handling soil sampling object；
Spectral scan unit, for carrying out spectral scan to soil sampling object；
Spectral signal processing unit, for handling spectral signal；
Monitoring unit, for will treated that signal is compared with given threshold, and the signal value more than threshold value is reported
It is alert.
2. solid waste as claimed in claim 1 prevents monitoring system, the solid for carrying out soil sampling to selection area
Waste sampling unit includes：
Number of samples determination subelement is the difference by variance yields and average value and true value for determining number of samples P before sampling
Value calculates number of samples, and the natural number that P is >=19；
Sampling processing subelement, for the size according to sampling area, using snakelike sampling, sampling depth is selected as
30CM is sampled according to calculated number of samples, when sampling, is first produced a sampling section, is carried out being parallel to section
Sampling, adopts the earthwork to perpendicular to ground, the soil after sampling is mixed, then mixing thickness soil sampling object is crushed,
Stone, root system of plant are removed, mixes and transform into quadrangle, diagonal line is divided and is divided into four parts, takes diagonal two parts therein, repeatedly
Carry out repeatedly after, by the Weight control of soil sampling object needs weight, then to by sample contents through hydrochloric acid, nitric acid, high chlorine
Acid is handled；
The pretreatment unit includes：
Cloth subelement is spread, for soil sampling object to be positioned in sample container, shake is distributed with making its upper surface substantially flat,
Then sample is positioned on quartz window；
Subelement is rotated, in measurement process, making soil sampling object be rotated with sample container instrument；
Spectral scan unit includes following subelement：
Averaged spectrum obtains subelement, is obtained for carrying out spectrum analysis in the rotary course by means of the rotation subelement
The averaged spectrum obtained after multiple rotary scanning analysis is taken, the rotary speed of sample container is 4cm/s, and soil sampling object is in 40s
Interior to be scanned 64 times, the spectrum after arithmetic mean is as a sampling spectrum；Each sample is positioned in different sample containers,
It measures twice, the spectroscopic data after being averaged again is used for signal processing；
The signal processing is based on the spectral signal processing unit, and the spectral signal processing unit includes：
Subelement is corrected, for acquiring reference signal, measured signal is acquired, obtains original absorbance spectrum, determine that correction refers to base
Quasi wave is long, utilizes correction reference data wavelength calibration absorption spectrum；The correction reference data wavelength by being calculated as follows：
Spectrum matrix is denoted as X (M*K), wherein i-th of wavelength variable represents i-th of column vector in spectrum matrix X, remembers N=min
(m-1, k), the n=1 before carrying out first time interative computation, the arbitrary a certain row for selecting spectrum matrix are denoted as variable xk(0)；Not having
There are the column vector location sets being selected into be denoted as T,
Calculate separately xk(n-1)To remaining the projection of following variables, it is denoted as：
Note k (n-1)=arg (max (| | Pj||),j∈T)；N=n+1 enables x if n=Nk(n-1)As the row variable being selected into,
It returns to above-mentioned column vector position and calculates step and calculated again, using multiple linear regression method to the variables collection that newly selects
{xk(0),xk(1),…,xk(n-1)Assessed, obtain correction reference data wavelength value；
After the spectral scan output valve for obtaining soil sampling object, in order to remove the soil spectrum that background interference obtains more high s/n ratio
Data need the data to P sample to improve the robustness and predictive ability of the soil constituent prediction model that the later stage is established
It is handled as follows：
Calculate the spectrum average of sample contents：
To unknown sample spectrum x (1 × m), pass through the following formula spectrum x that obtains that treatedcentered：
Then spectrum standard deviation is calculated:
Spectrum sample value is handled as follows：
Entire spectrum sample signal value is handled as follows, value X after being handled：
Wherein, n is sample number, is counted for wavelength, and Xn corresponds to the spectrum average of the sample contents as n-th sample；
After obtaining spectral signal, the spectral signal processing further includes that spectrum sample, carries out two-dimensional wavelet transformation, connect at equal intervals
Carry out matrix of wavelet coefficients stationary window transversal scanning, obtain nonstationary noise variance evaluation vector sum signal variance estimate to
Amount, adjusts window width, then carry out rescan point by point, obtains accurate noise variance evaluation vector sum precise signal side
Then poor estimate vector shrinks denoising by Bayes's threshold value, finally carry out 2-d wavelet contravariant and exchange central row vector conduct for
Signal value output；
It is described to include to the point-by-point adjustment of window width progress：The Noise Variance Estimation of window is fixed first, defines fixed window
Mouth Wm×n, width m is set as empirical value, and height n is coefficient matrix height, by window width diagonally high frequency coefficient matrix center
Row vectorOn slide laterally point by point, whereinCentral row is estimated using wavelet coefficient in window
Noise variance vector, according to the change rate of noise variance vector, dynamic adjustment window width specially exists according to noise
The linear phase property of approaches uniformity is distributed in all wavelet coefficients property and wavelet basis, the noise of high-rise wavelet coefficient
Variance vectors can carry out signal variance vector by being averaging to obtain by two neighboring variance in the noise variance vector of last layer
Estimation, evaluation method is similar with above-mentioned noise variance vector evaluation method, but since signal variance concentrates on part wavelet systems
In number, there is larger difference on each wavelet decomposition layer, need to carry out layering estimation, in all directions coefficient matrix of the layer calculated
It is slid laterally respectively in the row vector of center, using all wavelet coefficients respectively decomposed in this layer of window on direction, estimates this layer
Signal variance vector, is set dynamically according to the change rate of signal variance vector, resets serial ports width, setting signal window tune
Whole threshold value, every layer of signal variance vector is estimated using new window again, then acquires the threshold vector of every layer of matrix of wavelet coefficients,
For every a line of high frequency transverse direction coefficient matrix, longitudinal high frequency coefficient matrix, diagonal high frequency coefficient matrix, threshold vector is used
Contraction noise reduction is carried out, spectrum matrix is changed into the wavelet coefficient contravariant after noise reduction, takes the center row vector of spectrum matrix as drop
Spectral signal after making an uproar；
The monitoring unit is compared after the spectral signal that obtains that treated with the spectral signal threshold value of setting, when beyond
It when threshold value, alarms, reaches Metal Pollution Study In Soils object content detection purpose.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810130556.0A CN108362654A (en) | 2018-02-08 | 2018-02-08 | Solid waste prevents monitoring system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810130556.0A CN108362654A (en) | 2018-02-08 | 2018-02-08 | Solid waste prevents monitoring system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108362654A true CN108362654A (en) | 2018-08-03 |
Family
ID=63005392
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810130556.0A Pending CN108362654A (en) | 2018-02-08 | 2018-02-08 | Solid waste prevents monitoring system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108362654A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114558881A (en) * | 2022-03-10 | 2022-05-31 | 广东省车汇莱再生物资回收有限公司 | New energy scraped car resource recovery system based on it is intelligent |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107462542A (en) * | 2017-09-19 | 2017-12-12 | 黑龙江省能源环境研究院 | A kind of heavy metal-polluted soil detection warning device and its application method |
-
2018
- 2018-02-08 CN CN201810130556.0A patent/CN108362654A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107462542A (en) * | 2017-09-19 | 2017-12-12 | 黑龙江省能源环境研究院 | A kind of heavy metal-polluted soil detection warning device and its application method |
Non-Patent Citations (2)
| Title |
|---|
| 吴德操: "面向地表水环境的分布式紫外-可见光谱水质在线检测监测系统关键技术研究", 《中国博士学位论文全文数据库 工程科技Ⅰ辑》 * |
| 尚廷义 等: "《基于光谱技术的寒地水稻稻瘟病害分析及机理研究》", 30 June 2016, 哈尔滨工程大学出版社 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114558881A (en) * | 2022-03-10 | 2022-05-31 | 广东省车汇莱再生物资回收有限公司 | New energy scraped car resource recovery system based on it is intelligent |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Modeling salinity effects on soil reflectance under various moisture conditions and its inverse application: A laboratory experiment | |
| Whiting et al. | Predicting water content using Gaussian model on soil spectra | |
| Gomez et al. | Regional predictions of eight common soil properties and their spatial structures from hyperspectral Vis–NIR data | |
| CN106645037A (en) | Method for detecting heavy metal content of coal gangue filling reclamation reconstruction soil based on high spectrum technology | |
| Jahn et al. | Mid-infrared spectroscopic determination of soil nitrate content | |
| AU2017377078A1 (en) | Classification of soil texture and content by near-infrared spectroscopy | |
| CN110531054B (en) | Soil organic carbon prediction uncertainty estimation method based on Bootstrap sampling | |
| CN109738380B (en) | Hyperspectral remote sensing judgment method for soil salinization degree | |
| CN108344691A (en) | Soil contamination grade detecting system | |
| Mura et al. | Vibrational spectroscopy and Aquaphotomics holistic approach to determine chemical compounds related to sustainability in soil profiles | |
| Xiong et al. | The total P estimation with hyper-spectrum–a novel insight into different P fractions | |
| CN108362654A (en) | Solid waste prevents monitoring system | |
| Šestak et al. | Visible and near infrared reflectance spectroscopy for field-scale assessment of Stagnosols properties. | |
| CN108375553A (en) | A kind of soil tenor detection method | |
| CN108387539A (en) | Oilfield drilling waste disposes early warning equipment | |
| CN108344704A (en) | Consumer waste incineration handles monitoring device | |
| CN108362655A (en) | A kind of monitoring water quality on line system of more administrative grade linkages | |
| CN108344705A (en) | Rubbish from cooking disposes monitoring method | |
| Zhuo et al. | Quantitative retrieving of soil organic matter using field spectrometer and hyperspectral remote sensing | |
| Arrouays et al. | Short‐range spatial variability of metal contents in soil on a one hectare agricultural plot | |
| CN108362656A (en) | A kind of PH on-line computing models of networking | |
| CN108344692A (en) | Using the COD online automatic monitors of technology of Internet of things | |
| Jahn et al. | Wavelet-based spectral analysis for soil nitrate content measurement | |
| 李硕 et al. | Prediction of vertical distribution of soil nitrogen content in soil profile using spectral imaging technique | |
| Okunev et al. | Application of ATR-FTIR spectrometry for express prediction of the organic matter properties of arable leached chernozem |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180803 |