CN113104945A - Intelligent coal slime water dosing method and system based on floc characteristic feedback - Google Patents

Intelligent coal slime water dosing method and system based on floc characteristic feedback Download PDF

Info

Publication number
CN113104945A
CN113104945A CN202110472310.3A CN202110472310A CN113104945A CN 113104945 A CN113104945 A CN 113104945A CN 202110472310 A CN202110472310 A CN 202110472310A CN 113104945 A CN113104945 A CN 113104945A
Authority
CN
China
Prior art keywords
floc
image
fractal dimension
target
intelligent
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.)
Granted
Application number
CN202110472310.3A
Other languages
Chinese (zh)
Other versions
CN113104945B (en
Inventor
董宪姝
马晓敏
樊玉萍
韩宜纯
姚素玲
孙冬
巩固
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Taiyuan University of Technology
Original Assignee
Taiyuan University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Taiyuan University of Technology filed Critical Taiyuan University of Technology
Priority to CN202110472310.3A priority Critical patent/CN113104945B/en
Publication of CN113104945A publication Critical patent/CN113104945A/en
Application granted granted Critical
Publication of CN113104945B publication Critical patent/CN113104945B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • C—CHEMISTRY; METALLURGY
    • C02—TREATMENT OF WATER, WASTE WATER, SEWAGE, OR SLUDGE
    • C02F—TREATMENT OF WATER, WASTE WATER, SEWAGE, OR SLUDGE
    • C02F1/00—Treatment of water, waste water, or sewage
    • C02F1/52—Treatment of water, waste water, or sewage by flocculation or precipitation of suspended impurities
    • C—CHEMISTRY; METALLURGY
    • C02—TREATMENT OF WATER, WASTE WATER, SEWAGE, OR SLUDGE
    • C02F—TREATMENT OF WATER, WASTE WATER, SEWAGE, OR SLUDGE
    • C02F2209/00—Controlling or monitoring parameters in water treatment
    • C02F2209/001—Upstream control, i.e. monitoring for predictive control

Abstract

The invention relates to a coal slime water intelligent dosing method and system based on floc characteristic feedback, belonging to the technical field of coal slime water intelligent dosing; the technical problem to be solved is as follows: the improvement of the intelligent chemical adding method for the slime water based on floc characteristic feedback is provided; the technical scheme for solving the technical problems is as follows: the method comprises the following steps: collecting the feeding flow, feeding concentration and overflow water turbidity data on a feeding pipe and sending the data to an intelligent controller, and collecting floc image data of the slime water and sending the floc image data to the intelligent controller; the intelligent controller carries out image processing, BP neural network prediction and fuzzy recognition on the acquired data; processing the acquired floc image of the coal slime water to obtain a fractal dimension of a target floc at the current moment; predicting a fractal dimension target value of a target floc group at the next moment through a BP neural network; comparing the fractal dimension of the floc obtained through image processing with a fractal dimension target value of the floc to control the drug adding amount; the invention is applied to coal mines.

Description

Intelligent coal slime water dosing method and system based on floc characteristic feedback
Technical Field
The invention discloses a coal slime water intelligent dosing method and system based on floc characteristic feedback, and belongs to the technical field of coal slime water intelligent dosing methods and systems.
Background
Coal occupies an important position in the energy structure of China, clean production and utilization of coal are vital to ecological environment, with continuous optimization and adjustment of the energy structure in recent years, countries and society put forward higher requirements on clean use of energy and environmental protection, and for the coal preparation industry, the core goal of a coal preparation plant is to improve economic benefit, and the mathematical expression of the maximum economic benefit of the coal preparation plant is as follows: max [ f (x)]＝∑RiPi－c1－c2(i ═ l, …, n), in the above formula: riYield of the i product, PiIs the selling price of the product in the ith, n is the total number of the products in the scheme, c1For processing cost, c2F (x) is the raw coal cost and is the benefit function.
The intelligent construction of the coal preparation plant starts with the two aspects of reducing manpower, reducing processing cost and improving yield, and can effectively improve the economic benefit of the coal preparation plant, thereby realizing the aims of accurate separation, fine management and increment efficiency improvement.
The solid-liquid separation is very important in the construction of an intelligent coal preparation plant, and the concentration and sedimentation process occupies a core position in the solid-liquid separation intelligence. The traditional concentration link mainly depends on manual experience and partial technical parameters to select the addition amounts of the flocculating agent and the flocculating agent, so that the medicament is wasted, and the medicament cannot play the maximum role. At present, a concentration and sedimentation link is rough and lacks of an on-line detection device, and many key variables related to the concentration link of a concentrator cannot be acquired in real time, so that the concentration and sedimentation process cannot be dynamically known, and the phenomenon of lag exists in medicament addition and adjustment. Therefore, the coal slime water intelligent dosing method and system based on floc characteristic feedback are provided.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention aims to solve the technical problems that: the improvement of the intelligent coal slime water dosing method based on floc characteristic feedback and the hardware improvement of the system are provided.
In order to solve the technical problems, the invention adopts the technical scheme that: a coal slime water intelligent dosing method based on floc characteristic feedback comprises the following steps:
the method comprises the following steps: data acquisition: the detection system is used for acquiring the feeding flow, feeding concentration and overflow water turbidity data on the feeding pipe and sending the data to the intelligent controller, and the image acquisition device is used for acquiring floc image data of the coal slime water and sending the floc image data to the intelligent controller;
step two: data processing: the intelligent controller carries out image processing, BP neural network prediction and fuzzy recognition on the data acquired in the step one;
preprocessing, denoising and target edge extracting the acquired floc image of the coal slime water through image processing to obtain a fractal dimension of a target floc at the current moment;
predicting a fractal dimension target value of a target floc group at the next moment through a BP neural network;
step three: and (3) dosing control: and comparing the fractal dimension of the floc obtained through image processing with the fractal dimension target value of the floc to control the medicine adding amount.
The specific steps of obtaining the fractal dimension of the floc through image processing in the step two are as follows:
step 2.1.1: image preprocessing: performing space domain convolution on the image and the Laplacian in the four fields;
step 2.1.2: denoising an image: filtering out image noise by adopting three-window stop filtering;
step 2.1.3: extracting target edges: extracting the edge contour of the floc cluster by adopting a Canny operator, performing morphological expansion treatment on the image of the target floc cluster detected by the Canny operator, filling up the missing part and refining;
step 2.1.4: image segmentation: determining the minimum peak gray value of the image gray histogram as a threshold value w, setting pixel points smaller than w as 0, and setting pixel points larger than w as 255;
step 2.1.5: characteristic parameter extraction: (1) fractal dimension of floc: determining the long axis L of the flocs and the area A of the flocs in the image by adopting a traversal method, and determining lnA-D according to an area-long axis methodflnL + ln alpha, and linear fitting lnA and lnL to obtain slope of straight line, i.e. fractal dimension D of flocfWherein α is a proportionality constant; (2) particle size of flocs: deleting flocs with the area smaller than four neighborhoods in the binary image, approximating the number larger than four connected domains as the granularity of the flocs, and calculating the number of the flocs; (3) particle size of flocs: the maximum diameter of the floc equivalent circle is approximately regarded as the floc particle diameter.
The second step adopts a BP neural network to predict the fractal dimension of the target floc group, and comprises the following specific steps:
step 2.2.1: adopting a three-layer BP network structure, wherein the nodes of an input layer comprise a feeding flow, a feeding concentration, detected overflow water turbidity and overflow water target turbidity;
step 2.2.2: the intelligent controller requires to output the fractal dimension of the target floc group, and the output layer node of the BP network structure is the fractal dimension of the target floc group;
the three-layer BP network structure is as follows:
the input layer is: xi＝[y(k-1)，u(k-1)，yr(k)]，
The hidden layer is:
Figure BDA0003045957820000021
the output layer is:
Figure BDA0003045957820000022
U(k)＝Ock(k)，
in the above formula: oc ofj(k) Outputting for the jth node of the hidden layer; oc ofk(k) Outputting for the network; g is an excitation functionUsing logsig (n) function: logsig (n) ═ 1/(1+ exp (-n));
Figure BDA0003045957820000023
the weight value of the hidden output layer is obtained;
Figure BDA0003045957820000024
outputting a layer threshold value for the hidden layer; y (k-1) is the last overflow water turbidity, u (k-1) is the fractal dimension of the last floc, yr(k) Characteristic parameters of the flock are carved; u (k) is the fractal dimension of the target floc.
And the step two of judging whether the fractal dimension target value of the floc is correct through fuzzy recognition comprises the following specific steps:
step 2.3.1: the input variable of the fuzzy control is the deviation e of the fractal dimension of the floc and the target value and the change rate ec thereof, and the output variable is the output frequency u of the frequency converter;
selecting a linguistic variable with input variable deviation E as E, and selecting a linguistic variable with input variable deviation change rate Ec as Ec;
selecting a language variable of the output frequency U as U;
step 2.3.2: calculating the deviation and the change rate between the target prediction value of the fractal dimension of the floc and the given turbidity, wherein the calculation formula is as follows: e (k + p) ═ yg-y(k+p)，ec(k+p)＝e(k+p)-e(k+p-1)；
In the above formula: e is the deviation of the fractal dimension of the floc from the target value, and ec is the deviation change rate of the fractal dimension of the floc from the target value;
step 2.3.2: three assigned value tables are established according to the basic fuzzy control theory and respectively correspond to different fuzzy subsets on a fuzzy domain.
A coal slime water intelligent dosing system based on floc characteristic feedback comprises a detection system, an image recognition system, an intelligent controller and a dosing execution device, wherein the detection system and the image recognition system send acquired feeding flow, feeding concentration and overflow water turbidity data on a feeding pipe and floc image data of coal slime water to the intelligent controller, the intelligent controller constructs a BP neural network to pre-judge the state of the next moment according to the detected data at the moment, and carries out fuzzy control according to the deviation from a target value and the deviation change rate, so that a medicament regulating quantity is obtained to control the dosing device to adjust;
the detection system comprises a feeding detection device and an overflow water detection device, the feeding detection device comprises a flowmeter and a first concentration meter which are arranged on a feeding pipe, and the overflow water detection device comprises a second concentration meter which is arranged on an output pipe of the concentration tank;
the image identification system comprises an image acquisition device and an image analysis system, wherein the image acquisition device is arranged on a rake of the concentration tank, slowly moves along with the rake and performs floc detection in real time;
the medicine adding execution device comprises a medicine preparing device, a medicine storage pool, a controllable valve and a medicine adding pump, the medicine preparing device is connected with the medicine storage pool, the medicine storage pool is connected with an inlet end of the medicine adding pump, an outlet end of the medicine adding pump is connected with a concentration pool, and a control end of the medicine adding pump is connected with the medicine adding measurement and control system after being connected with a frequency converter through a wire.
The image acquisition device comprises a square shell made of steel plates, rubber is arranged at the joint of the steel plates for sealing, an industrial camera, a light source, an observation column, a water tank, a first small pump and a second small pump are arranged in the square shell, and shading cloth is arranged at the top end of the square shell;
the inlet end of the second small pump is connected with the feeding pipe, the outlet end of the second small pump is connected with the observation column, the inlet end of the first small pump is connected with the water tank, and the outlet end of the first small pump is connected with the observation column.
The control ends of the flowmeter, the first concentration meter and the second concentration meter are respectively connected with the intelligent controller through leads, and the dosing control system is connected with the intelligent controller through leads; the intelligent controller is connected with the control host through wired or wireless communication.
The flowmeter specifically adopts the electromagnetic flowmeter, first concentration meter specifically adopts the differential concentration meter, the second concentration meter specifically adopts the ultrasonic wave concentration meter, the light source specifically adopts laser light source.
The dosing pump specifically adopts a metering pump to dose.
Compared with the prior art, the invention has the beneficial effects that: the intelligent coal slime water dosing method based on floc characteristic feedback provided by the invention judges and adjusts the dosing amount by adopting an artificial intelligent image recognition mode, can replace the current workers to adjust the dosage by experience according to the sedimentation condition of the coal slime water, can enable a coal preparation plant to more stably, accurately and quickly dose the chemicals, and saves the labor cost and time.
Drawings
The invention is further described below with reference to the accompanying drawings:
FIG. 1 is a schematic diagram of the system of the present invention;
FIG. 2 is a schematic structural diagram of an image capturing device according to the present invention;
FIG. 3 is a diagram of a floc after image processing according to the present invention;
in the figure: 1 is a water tank, 2 is a light source, 3 is a first small pump, 4 is an observation column, 5 is shading cloth, 6 is a second small pump, and 7 is an industrial camera.
Detailed Description
As shown in FIGS. 1 to 3, the intelligent chemical adding method for slime water based on floc characteristic feedback comprises the following steps:
the method comprises the following steps: data acquisition: the detection system is used for acquiring the feeding flow, feeding concentration and overflow water turbidity data on the feeding pipe and sending the data to the intelligent controller, and the image acquisition device is used for acquiring floc image data of the coal slime water and sending the floc image data to the intelligent controller;
step two: data processing: the intelligent controller carries out image processing, BP neural network prediction and fuzzy recognition on the data acquired in the step one;
preprocessing, denoising and target edge extracting the acquired floc image of the coal slime water through image processing to obtain a fractal dimension of a target floc at the current moment;
predicting a fractal dimension target value of a target floc group at the next moment through a BP neural network;
judging whether the fractal dimension target value of the floc is correct or not through fuzzy recognition, and when the fractal dimension target value of the floc is incorrect, trimming the fractal dimension target value of the floc by comparing the effluent turbidity of the concentration tank with the turbidity target value;
step three: and (3) dosing control: and comparing the fractal dimension of the floc obtained through image processing with the fractal dimension target value of the floc to control the medicine adding amount.
Step three: and (3) dosing control: and comparing the fractal dimension of the floc obtained through image processing with the fractal dimension target value of the floc to control the medicine adding amount.
The specific steps of obtaining the fractal dimension of the floc through image processing in the step two are as follows:
step 2.1.1: image preprocessing: performing space domain convolution on the image and the Laplacian in the four fields;
step 2.1.2: denoising an image: filtering out image noise by adopting three-window stop filtering;
step 2.1.3: extracting target edges: extracting the edge contour of the floc cluster by adopting a Canny operator, performing morphological expansion treatment on the image of the target floc cluster detected by the Canny operator, filling up the missing part and refining;
step 2.1.4: image segmentation: determining the minimum peak gray value of the image gray histogram as a threshold value w, setting pixel points smaller than w as 0, and setting pixel points larger than w as 255;
step 2.1.5: characteristic parameter extraction: (1) fractal dimension of floc: determining the long axis L of the flocs and the area A of the flocs in the image by adopting a traversal method, and determining lnA-D according to an area-long axis methodflnL + ln alpha, and linear fitting lnA and lnL to obtain slope of straight line, i.e. fractal dimension D of flocfWherein α is a proportionality constant; (2) particle size of flocs: deleting flocs with the area smaller than four neighborhoods in the binary image, approximating the number larger than four connected domains as the granularity of the flocs, and calculating the number of the flocs; (3) particle size of flocs: the maximum diameter of the floc equivalent circle is approximately regarded as the floc particle diameter.
The second step adopts a BP neural network to predict the fractal dimension of the target floc group, and comprises the following specific steps:
step 2.2.1: adopting a three-layer BP network structure, wherein the nodes of an input layer comprise a feeding flow, a feeding concentration, detected overflow water turbidity and overflow water target turbidity;
step 2.2.2: the intelligent controller requires to output the fractal dimension of the target floc group, and the output layer node of the BP network structure is the fractal dimension of the target floc group;
the three-layer BP network structure is as follows:
the input layer is: xi＝[y(k-1)，u(k-1)，yr(k)]，
The hidden layer is:
Figure BDA0003045957820000051
the output layer is:
Figure BDA0003045957820000052
U(k)＝Ock(k)，
in the above formula: oc ofj(k) Outputting for the jth node of the hidden layer; oc ofk(k) Outputting for the network; g is the excitation function, which adopts logsig (n) function: logsig (n) ═ 1/(1+ exp (-n));
Figure BDA0003045957820000053
the weight value of the hidden output layer is obtained;
Figure BDA0003045957820000054
outputting a layer threshold value for the hidden layer; y (k-1) is the last overflow water turbidity, u (k-1) is the fractal dimension of the last floc, yr(k) Characteristic parameters of the flock are carved; u (k) is the fractal dimension of the target floc.
And the step two of judging whether the fractal dimension target value of the floc is correct through fuzzy recognition comprises the following specific steps:
step 2.3.1: the input variable of the fuzzy control is the deviation e of the fractal dimension of the floc and the target value and the change rate ec thereof, and the output variable is the output frequency u of the frequency converter;
selecting a linguistic variable with input variable deviation E as E, and selecting a linguistic variable with input variable deviation change rate Ec as Ec;
selecting a language variable of the output frequency U as U;
step 2.3.2: calculating the deviation and the change rate between the target prediction value of the fractal dimension of the floc and the given turbidity, wherein the calculation formula is as follows: e (k + p) ═ yg-y(k+p)，ec(k+p)＝e(k+p)-e(k+p-1)；
In the above formula: e is the deviation of the fractal dimension of the floc from the target value, and ec is the deviation change rate of the fractal dimension of the floc from the target value;
step 2.3.2: three assigned value tables are established according to the basic fuzzy control theory and respectively correspond to different fuzzy subsets on a fuzzy domain.
A coal slime water intelligent dosing system based on floc characteristic feedback comprises a detection system, an image recognition system, an intelligent controller and a dosing execution device, wherein the detection system and the image recognition system send acquired feeding flow, feeding concentration and overflow water turbidity data on a feeding pipe and floc image data of coal slime water to the intelligent controller, the intelligent controller constructs a BP neural network to pre-judge the state of the next moment according to the detected data at the moment, and carries out fuzzy control according to the deviation from a target value and the deviation change rate, so that a medicament regulating quantity is obtained to control the dosing device to adjust;
the detection system comprises a feeding detection device and an overflow water detection device, the feeding detection device comprises a flowmeter and a first concentration meter which are arranged on a feeding pipe, and the overflow water detection device comprises a second concentration meter which is arranged on an output pipe of the concentration tank;
the image identification system comprises an image acquisition device and an image analysis system, wherein the image acquisition device is arranged on a rake of the concentration tank, slowly moves along with the rake and performs floc detection in real time;
the medicine adding execution device comprises a medicine preparing device, a medicine storage pool, a controllable valve and a medicine adding pump, the medicine preparing device is connected with the medicine storage pool, the medicine storage pool is connected with an inlet end of the medicine adding pump, an outlet end of the medicine adding pump is connected with a concentration pool, and a control end of the medicine adding pump is connected with the medicine adding measurement and control system after being connected with a frequency converter through a wire.
The image acquisition device comprises a square shell made of steel plates, rubber is arranged at the joint of the steel plates for sealing, an industrial camera 7, a light source 2, an observation column 4, a water tank 1, a first small pump 3 and a second small pump 6 are arranged in the square shell, and shading cloth 5 is arranged at the top end of the square shell;
the inlet end of the second small pump 6 is connected with the feeding pipe, the outlet end of the second small pump 6 is connected with the observation column 4, the inlet end of the first small pump 3 is connected with the water tank 1, and the outlet end of the first small pump 3 is connected with the observation column 4;
the control ends of the flowmeter, the first concentration meter and the second concentration meter are respectively connected with the intelligent controller through leads, and the dosing control system is connected with the intelligent controller through leads; the intelligent controller is connected with the control host through wired or wireless communication.
The flowmeter specifically adopts the electromagnetic flowmeter, first concentration meter specifically adopts the differential concentration meter, the second concentration meter specifically adopts the ultrasonic wave concentration meter, 2 specifically adopt laser light source in the light source.
The dosing pump specifically adopts a metering pump to dose.
The invention provides a coal slime water intelligent dosing method and system based on floc characteristic feedback, the system mainly comprises: detection system, image acquisition system, intelligent control ware, medicine implementation device.
The detection system comprises a feeding detection device and an overflow water detection device, wherein in the feeding detection device, a differential pressure type concentration meter and an electromagnetic flow meter are arranged on a feeding pipe, the concentration and the flow of the coal slime water feeding are detected in real time, the overflow detection device mainly comprises an ultrasonic concentration meter, the concentration of the overflow water is detected, and the data of the detection system need to be transmitted to an intelligent controller.
The image acquisition system comprises an image acquisition device and an image processing system, wherein the image acquisition device and an image recognition device are arranged on the rake of the concentration tank and can slowly move along with the rake to perform floc detection in real time; the image analysis system pre-judges the fractal dimension of the target floc group by adopting a three-layer BP network structure.
The image acquisition device is a steel container which is provided with an industrial camera 7, a light source 2, an observation column 4, a water tank 1, a first small pump 3 and a second small pump 6, the whole image recognition device is a square container made of steel plates, the requirement on the sealing performance of the square container is extremely high, and rubber sealing is needed at the joint of the steel plates. Collection system is connected with the external world through two pipes, and the diameter of pipe is 6 ~ 12mm, and 6 entry ends of second minipump link to each other with the pan feeding pipe of device, and 6 exit ends of second minipump link to each other with observation post 4, with in the pump extraction observation post of wadding group. One end of the first small pump 3 is connected with a water tank, and the volume of the water tank is 10L; the other end of the first small pump 3 is connected with the observation column 4, the volume of the observation column 4 is 400mL, water is pumped into the observation column 4 from the water tank 1, and laser is used as the light source 2, so that the floccule can be conveniently and clearly shot.
The intelligent controller mainly carries out three major parts of image processing, BP neural network prediction and fuzzy recognition, an image processing system firstly carries out preprocessing on an image obtained by an image acquisition device, a four-neighborhood Laplacian operator and the image are adopted to carry out space domain convolution, the transformation can enhance the contrast of a gray mutation position in the image, and the image outline of a floc cluster and the details of the floc cluster edge become clear; the acquired floc images are influenced by factors such as illumination and the like, various noises inevitably exist, and the noises are removed by adopting three-window median filtering; and because the target edge detection and extraction effect of the Canny operator is good, the Canny operator is adopted to extract the edge contour of the floc, and the morphological expansion treatment is carried out on the image of the Canny operator detection target floc, so that the missing part is filled and refined. Determining a threshold value w according to the wave crest of the image gray histogram, setting the pixel points smaller than w as 0, setting the pixel points larger than w as 255, determining the long axis L of the flocs and the area A of the flocs in the image by adopting a traversal method, and determining lnA-D according to an area-long axis methodflnL+lnαfMeasuring different L and A, and performing linear fitting on lnA and lnL to obtain the slope of a straight line, namely the fractal dimension D of the flocf。
The invention adopts a three-layer BP network structure to process images. And the nodes of the input layer comprise the feeding flow, the feeding concentration, the detected overflow water turbidity and the overflow water target turbidity. The controller requires to output the fractal dimension of the target floc, so that the fractal dimension of the target floc is determined as the output layer node. The layers were as follows:
the input layer is: xi＝[y(k-1)，u(k-1)，yr(k)]，
The hidden layer is:
Figure BDA0003045957820000081
the output layer is:
Figure BDA0003045957820000082
U(k)＝Ock(k)，
in the above formula: oc ofj(k) Outputting for the jth node of the hidden layer; oc ofk(k) Outputting for the network; g is the excitation function, which adopts logsig (n) function: logsig (n) ═ 1/(1+ exp (-n));
Figure BDA0003045957820000083
the weight value of the hidden output layer is obtained;
Figure BDA0003045957820000084
outputting a layer threshold value for the hidden layer; y (k-1) is the last overflow water turbidity, u (k-1) is the fractal dimension of the last floc, yr(k) Characteristic parameters of the flock are carved; u (k) is the fractal dimension of the target floc. . And comparing the fractal dimension of the flocs obtained by the image recognition system with a target value to control the dosage. And if the fractal dimension target value of the flocs is correct, trimming by comparing the effluent turbidity of the concentration tank with the turbidity target value.
The invention can ensure the stability of the turbidity of the circulating water in the slime water system by applying fuzzy control, reduce the influence of detection lag on the response process, reduce the overshoot and weaken the oscillation phenomenon. The input variable of the fuzzy control is the deviation e of the fractal dimension of the floc and the target value and the change rate ec thereof, and the output variable of the fuzzy control is the output frequency u of the frequency converter. Selecting the linguistic variable with input variable deviation E as E and the linguistic variable with change rate Ec as Ec; the output U linguistic variable is selected as U.
Calculating the deviation between the target prediction value of the fractal dimension of the floc and a given turbidity and the change rate E (k + p) ═ yg-Y (k + p) and ec (k + p) -E (k + p-1), in the control system, selecting the argument Y { -6, -5, -4, -3, -2, -1, -0, +0, +1, +2, +3, +4, +5, +6} of the deviation change E, and using a triangular function as the language variable value { NM, NS, NO, PO, PS, PM, PB }, wherein the membership function adopts a triangular function. The domain of variation Ec is { -6, -5, -4, -3, -2, -1, -0, +0, +1, +2, +3, +4, +5, +6}, and the language variable value is { NB, NM, NS, NO, PO, PS, PM, PB } membership functions are triangular functions. The domain Z of the linguistic variable U of the output frequency is { -6, -5, -4, -3, -2, -1, -0, +0, +1, +2, +3, +4, +5, +6}, the linguistic variable value is selected as { NB, NM, NS, NO, PO, PS, PM, PB }, and the membership function adopts a triangular function. Wherein NB in the linguistic variables represents the negative maximum, NM represents the negative medium, NS represents the negative minimum, NO represents the negative 0, PO represents the positive 0, PS represents the positive minimum, PM represents the positive medium, and PB represents the positive maximum.
Three assigned value tables are established according to the basic fuzzy control theory and respectively correspond to different fuzzy subsets on a fuzzy domain. The quantization factor and the value thereof have great influence on the dynamic performance of the control system. The concrete embodiment is as follows: when the quantitative factor Ke of the deviation e of the fractal dimension of the floc from the target value is larger, the response is accelerated, the oscillation is accelerated, the overshoot is larger, and the transition process is longer. When the quantization factor Kec of the deviation change rate ec of the fractal dimension of the floc and the target value is larger, the rapidity is good, and the overshoot is reduced; however, when Kec is too large, overshoot is reduced, but the system response speed is affected.
The medicine adding executing device mainly comprises a medicine preparing device, a controllable valve, a medicine storage tank and a screw pump, and the screw pump is controlled to accurately add medicine according to the medicine adding amount output by the intelligent controller.
Examples
In this embodiment, detecting system includes that the ultrasonic wave concentration meter of installation electromagnetic flowmeter, differential pressure formula concentration meter and overflow water department on the pan feeding pipe.
In this embodiment, the image acquisition device on the rake firstly adds 300mL of water in the water tank into the observation column 4 through the peristaltic pump, then extracts 100mL of coal slurry through the peristaltic pump, when the peristaltic pump stops working, the industrial camera takes pictures at a speed of 10s per sheet, and then transmits the pictures to the intelligent controller for image data processing.
In this embodiment, a three-layer BP network structure is adopted. The feeding flow, feeding concentration, detected overflow water turbidity and overflow water target turbidity are input layers. And outputting the fractal dimension of the target floc group, so that the fractal dimension of the target floc group is determined as the output layer node.
In this embodiment, the difference between the fractal dimension of the floc and the target value and the deviation change rate are fuzzified to obtain the adjustment amount of the medicament.
In the embodiment, the control module of the dosing execution device controls the metering pump to accurately dose according to the compensation value, so that the dosing amount is adjusted, and the dosing amount is optimized.
The following further describes an embodiment of the image capturing device according to the present invention with reference to the accompanying drawings. The image acquisition system shown in fig. 1 is a steel container provided with an industrial camera 7, a light source 2, an observation column 4, a water tank 1, a first small pump and a second small pump, and the square container has extremely high sealing requirements and needs to be sealed by rubber at the joint of steel plates. The collecting device is provided with two pipes connected with the outside, one is a feeding pipe, the other is a discharging pipe, the observing column is connected with a pump room, a water tank is connected with the pump room by a hose, the diameter of the pipes is 6-12 mm, the two small pumps adopt peristaltic pumps, the light source adopts laser as a light source, a Mide Vervier MV-GE502GC-T-CL model camera and an MV-LD-12-10M-J model lens are selected, the resolution of the camera can reach 2592 multiplied by 2048 to the maximum, the collecting frame number per second is 22fps, the shutter is a global shutter, the exposure time is 0.005ms-327ms, the lens is a 1000 ten thousand pixel fixed focus lens, the focal length is 12mm, the camera is connected with an industrial personal computer by a GIgE bus mode, a data interface is an RJ45 kilomega network interface, a 100M network interface is compatible downwards, the observing column is 400mL, light-transmitting organic glass is adopted, the water needs to be added, as shown in fig. 1, a second small pump 6 is connected to the inlet of the feed pipe and the outlet of the observation column, and the flock is pumped into the observation column. 3 one end of first micropump links to each other with the water tank, and the other end links to each other with the observation post, takes out the observation post with water from the water tank in, dilutes and is convenient for clearly shoot wadding group to install image recognition device on the rake of concentrated pond, can follow the rake and slowly move, carry out wadding group in real time and detect.
Firstly, the water pump pumps water in the water tank, when the water pump reaches the position of 300mL, the controller stops the pumping of the water pump, meanwhile, the pumping of the coal slime water pump is started, and similarly, the pump stops the pumping when the water pump reaches the position of 400 mL. The camera shoots at a speed of 10s while the coal slime water pump extracts, the image is transmitted to the intelligent controller, the shooting is stopped after 5min, the electromagnetic valve is opened, the coal slime water is discharged, and 100mL of water is rapidly extracted to wash the observation column. And repeating the processes until the water level of the water tank rises to 100mL, giving an alarm, and adding water into the water tank.
It should be noted that, regarding the specific structure of the present invention, the connection relationship between the modules adopted in the present invention is determined and can be realized, except for the specific description in the embodiment, the specific connection relationship can bring the corresponding technical effect, and the technical problem proposed by the present invention is solved on the premise of not depending on the execution of the corresponding software program.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.

Claims (9)

1. A coal slime water intelligent dosing method based on floc characteristic feedback is characterized by comprising the following steps: the method comprises the following steps:
the method comprises the following steps: data acquisition: the detection system is used for acquiring the feeding flow, feeding concentration and overflow water turbidity data on the feeding pipe and sending the data to the intelligent controller, and the image acquisition device is used for acquiring floc image data of the coal slime water and sending the floc image data to the intelligent controller;
step two: data processing: the intelligent controller carries out image processing, BP neural network prediction and fuzzy recognition on the data acquired in the step one;
preprocessing, denoising and target edge extracting the acquired floc image of the coal slime water through image processing to obtain a fractal dimension of a target floc at the current moment;
predicting a fractal dimension target value of a target floc group at the next moment through a BP neural network;
step three: and (3) dosing control: and comparing the fractal dimension of the floc obtained through image processing with the fractal dimension target value of the floc to control the medicine adding amount.
2. The intelligent chemical adding method for the slime water based on the floc characteristic feedback as claimed in claim 1, which is characterized in that: the specific steps of obtaining the fractal dimension of the floc through image processing in the step two are as follows:
step 2.1.1: image preprocessing: performing space domain convolution on the image and the Laplacian in the four fields;
step 2.1.2: denoising an image: filtering out image noise by adopting three-window median filtering;
step 2.1.3: extracting target edges: extracting the edge contour of the floc cluster by adopting a Canny operator, performing morphological expansion treatment on the image of the target floc cluster detected by the Canny operator, filling up the missing part and refining;
step 2.1.4: image segmentation: determining the minimum peak of the image gray level histogram as a threshold value w, setting pixel points smaller than w as 0, and setting pixel points larger than w as 255;
step 2.1.5: characteristic parameter extraction: (1) fractal dimension of floc: determining the long axis L of the flocs and the area A of the flocs in the image by adopting a traversal method, and determining lnA-D according to an area-long axis methodflnL + ln alpha, and linear fitting lnA and lnL to obtain slope of straight line, i.e. fractal dimension D of flocfWherein α is a proportionality constant; (2) particle size of flocs: deleting the floc images with the areas smaller than the four neighborhoods in the binary image, approximating the number of the floc images larger than the four connected domains as the granularity of the floc, and calculating the number of the floc images; (3) particle size of flocs: the maximum diameter of the floc equivalent circle is approximately regarded as the floc particle diameter.
3. The intelligent chemical adding method for the slime water based on the floc characteristic feedback as claimed in claim 1, which is characterized in that: the second step adopts a BP neural network to predict the fractal dimension of the target floc group, and comprises the following specific steps:
step 2.2.1: adopting a three-layer BP network structure, wherein the nodes of an input layer comprise the feed concentration, the detected turbidity of the overflow water and the target turbidity of the overflow water;
step 2.2.2: the intelligent controller requires to output the fractal dimension of the target floc group, and the output layer node of the BP network structure is the fractal dimension of the target floc group;
the three-layer BP network structure is as follows:
the input layer is: xi＝[y(k-1)，u(k-1)，yr(k)]，
The hidden layer is:
Figure FDA0003045957810000021
the output layer is:
Figure FDA0003045957810000022
U(k)＝Ock(k)，
in the above formula: oc ofj(k) Outputting for the jth node of the hidden layer; oc ofk(k) Outputting for the network; g is the excitation function, which adopts logsig (n) function: logsig (n) ═ 1/(1+ exp (-n));
Figure FDA0003045957810000023
the weight value of the hidden output layer is obtained;
Figure FDA0003045957810000024
outputting a layer threshold value for the hidden layer; y (k-1) is the turbidity of the overflow water at the last time, u (k-1) is the fractal dimension of the flocs at the last time, yr(k) Characteristic parameters of the flock are carved; u (k) is the fractal dimension of the target floc.
4. The intelligent chemical adding method for the slime water based on the floc characteristic feedback as claimed in claim 1, which is characterized in that: and the step two of judging whether the fractal dimension target value of the floc is correct through fuzzy recognition comprises the following specific steps:
step 2.3.1: the input variable of the fuzzy control is the deviation e of the fractal dimension of the floc and the target value and the change rate ec thereof, and the output variable is the output frequency u of the frequency converter;
selecting a linguistic variable with input variable deviation E as E, and selecting a linguistic variable with input variable deviation change rate Ec as Ec;
selecting a language variable of the output frequency U as U;
step 2.3.2: calculating the deviation and the change rate between the target prediction value of the fractal dimension of the floc and the given turbidity, wherein the calculation formula is as follows: e (k + p) ═ yg-y(k+p)，ec(k+p)＝e(k+p)-e(k+p-1)；
In the above formula: e is the deviation of the fractal dimension of the floc from the target value, and ec is the deviation change rate of the fractal dimension of the floc from the target value;
step 2.3.2: three assigned value tables are established according to the basic fuzzy control theory and respectively correspond to different fuzzy subsets on a fuzzy domain.
5. The utility model provides a coal slime water intelligence medicine system based on wadding characteristic feedback, includes detecting system, image recognition system, intelligent control ware, adds medicine executive device, its characterized in that: the detection system and the image recognition system send the acquired feeding flow, feeding concentration and overflow water turbidity data on the feeding pipe and the acquired floc image data of the coal slime water to the intelligent controller, the intelligent controller constructs a BP neural network to pre-judge the state of the next moment according to the detected data at the moment, and carries out fuzzy control according to the deviation from a target value and the deviation change rate, so that the obtained medicament regulating quantity control dosing device is adjusted;
the detection system comprises a feeding detection device and an overflow water detection device, the feeding detection device comprises a flowmeter and a first concentration meter which are arranged on a feeding pipe, and the overflow water detection device comprises a second concentration meter which is arranged on an output pipe of the concentration tank;
the image identification system comprises an image acquisition device and an image analysis system, wherein the image acquisition device is arranged on a rake of the concentration tank, slowly moves along with the rake and performs floc detection in real time;
the medicine adding execution device comprises a medicine preparing device, a medicine storage pool, a controllable valve and a medicine adding pump, the medicine preparing device is connected with the medicine storage pool, the medicine storage pool is connected with an inlet end of the medicine adding pump, an outlet end of the medicine adding pump is connected with a concentration pool, and a control end of the medicine adding pump is connected with the medicine adding measurement and control system after being connected with a frequency converter through a wire.
6. The intelligent chemical adding system for slime water based on floc characteristic feedback as claimed in claim 5, wherein: the image acquisition device comprises a square shell made of a steel plate, rubber is arranged at the joint of the steel plate for sealing, an industrial camera (7), a light source (2), an observation column (4), a water tank (1), a first small pump (3) and a second small pump (6) are arranged in the square shell, and shading cloth (5) is arranged at the top end of the square shell;
the entry end of second small-size pump (6) links to each other with the pan feeding pipe, the exit end of second small-size pump (6) links to each other with observation post (4), the entry end of first small-size pump (3) links to each other with water tank (1), the exit end of first small-size pump (3) links to each other with observation post (4).
7. The intelligent chemical adding system for slime water based on floc characteristic feedback as claimed in claim 6, wherein: the control ends of the flowmeter, the first concentration meter and the second concentration meter are respectively connected with the intelligent controller through leads, and the dosing control system is connected with the intelligent controller through leads; the intelligent controller is connected with the control host through wired or wireless communication.
8. The intelligent chemical adding system for slime water based on floc characteristic feedback as claimed in claim 7, wherein: the flowmeter specifically adopts the electromagnetic flowmeter, first concentration meter specifically adopts the differential concentration meter, the second concentration meter specifically adopts the ultrasonic wave concentration meter, light source (2) specifically adopts laser light source.
9. The intelligent chemical adding system for slime water based on floc characteristic feedback as claimed in claim 8, wherein: the dosing pump specifically adopts a metering pump to dose.
CN202110472310.3A 2021-04-29 2021-04-29 Intelligent coal slime water dosing method and system based on floc characteristic feedback Active CN113104945B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110472310.3A CN113104945B (en) 2021-04-29 2021-04-29 Intelligent coal slime water dosing method and system based on floc characteristic feedback

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110472310.3A CN113104945B (en) 2021-04-29 2021-04-29 Intelligent coal slime water dosing method and system based on floc characteristic feedback

Publications (2)

Publication Number Publication Date
CN113104945A true CN113104945A (en) 2021-07-13
CN113104945B CN113104945B (en) 2023-04-14

Family

ID=76720398

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110472310.3A Active CN113104945B (en) 2021-04-29 2021-04-29 Intelligent coal slime water dosing method and system based on floc characteristic feedback

Country Status (1)

Country Link
CN (1) CN113104945B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113916732A (en) * 2021-11-02 2022-01-11 哈尔滨工业大学（深圳） Active sludge microscopic image real-time observation recording pulse flow cell
CN115108617A (en) * 2022-07-06 2022-09-27 中冶南方城市建设工程技术有限公司 Coagulation dosing method and system

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000218263A (en) * 1999-02-01 2000-08-08 Meidensha Corp Water quality controlling method and device therefor
CN1538177A (en) * 2003-05-30 2004-10-20 哈尔滨工业大学 Coagulating process flocculate detection method based on image processing technology and optimization control system
CN103011356A (en) * 2012-08-15 2013-04-03 重庆水务集团股份有限公司 Method for controlling automatic chemical dosing of high-turbidity water system
CN106442526A (en) * 2016-08-29 2017-02-22 青岛理工大学 Activated sludge floc analysis method based on MATLAB
CN111233118A (en) * 2020-03-19 2020-06-05 中冶赛迪工程技术股份有限公司 Intelligent control system and control method for high-density sedimentation tank

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000218263A (en) * 1999-02-01 2000-08-08 Meidensha Corp Water quality controlling method and device therefor
CN1538177A (en) * 2003-05-30 2004-10-20 哈尔滨工业大学 Coagulating process flocculate detection method based on image processing technology and optimization control system
CN103011356A (en) * 2012-08-15 2013-04-03 重庆水务集团股份有限公司 Method for controlling automatic chemical dosing of high-turbidity water system
CN106442526A (en) * 2016-08-29 2017-02-22 青岛理工大学 Activated sludge floc analysis method based on MATLAB
CN111233118A (en) * 2020-03-19 2020-06-05 中冶赛迪工程技术股份有限公司 Intelligent control system and control method for high-density sedimentation tank

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
常颖等: "絮凝体分形维数投药控制研究", 《环境污染治理技术与设备》 *
赵洋等: "紫金洗煤厂絮凝剂自动添加系统的研究", 《煤炭技术》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113916732A (en) * 2021-11-02 2022-01-11 哈尔滨工业大学（深圳） Active sludge microscopic image real-time observation recording pulse flow cell
CN113916732B (en) * 2021-11-02 2024-02-27 哈尔滨工业大学（深圳） Activated sludge microscopic image real-time observation recording pulse flow cell
CN115108617A (en) * 2022-07-06 2022-09-27 中冶南方城市建设工程技术有限公司 Coagulation dosing method and system
CN115108617B (en) * 2022-07-06 2023-06-20 中冶南方城市建设工程技术有限公司 Method and system for adding medicine by coagulation

Also Published As

Publication number Publication date
CN113104945B (en) 2023-04-14

Similar Documents

Publication Publication Date Title
CN113104945B (en) Intelligent coal slime water dosing method and system based on floc characteristic feedback
CN103011356B (en) Method for controlling automatic chemical dosing of high-turbidity water system
CN109325403A (en) A kind of water pollution identification administering method and system based on image recognition
CN103708592A (en) Novel water treatment method based on machine vision and device thereof
CN103630473A (en) Active sludge online computer image analysis early warning system and an active sludge online computer image analyzing and early warning method thereof
CN205216301U (en) Sedimentation tank automatic sludge discharging machine construct
CN115108617B (en) Method and system for adding medicine by coagulation
CN111210152A (en) Drainage system scheduling method and device
CN205933447U (en) Seawater desalination pretreatment systems
CN109139442A (en) Elevator pump priority control method, device and storage medium based on genetic algorithm
CN103995547B (en) Oil/gas Well Large-scale Acid Fracturing returns and is drained through residual sour Online Processing System and disposal route in journey
CN116540538A (en) Water consumption adjusting method and system
CN105712542A (en) Hydrofluoric acid wastewater recycling treatment device for photovoltaic and semiconductor industries and technology thereof
CN208626714U (en) A kind of plate and frame filter press on-line cleaning device
CN114693493B (en) IoT-based polluted river water ecological restoration system
CN108009307A (en) Sewage treatment plant's managerial experiences succession method
CN209052517U (en) Modularization movable reverse osmosis concentrated water processing equipment
CN209386273U (en) Hydrophobic recycling system in a kind of production of yeast extract
CN206417940U (en) Chemical circulation water zero discharge processing system
CN207727723U (en) A kind of Initial Runoff collection and the separate system of water quality and liquid level coordinated signals
CN208684598U (en) A kind of sewage disposal system of automatic water supplement
CN112112631A (en) Intelligent identification method for indicator diagram of oil pumping unit
CN205953767U (en) Dirty water sedimentation equipment of biodegradable
CN206109125U (en) Spent acid concentration and recycling system that contain heavy metal and particulate matter
CN207130080U (en) A kind of high aluminum content boils film Waste Water Treatment

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
GR01 Patent grant
GR01 Patent grant