CN109019717B - Intelligent treatment method and system for thermal power plant desulfurization wastewater - Google Patents

Intelligent treatment method and system for thermal power plant desulfurization wastewater Download PDF

Info

Publication number
CN109019717B
CN109019717B CN201810987345.9A CN201810987345A CN109019717B CN 109019717 B CN109019717 B CN 109019717B CN 201810987345 A CN201810987345 A CN 201810987345A CN 109019717 B CN109019717 B CN 109019717B
Authority
CN
China
Prior art keywords
svr
desulfurization wastewater
historical
kernel
kernel function
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.)
Active
Application number
CN201810987345.9A
Other languages
Chinese (zh)
Other versions
CN109019717A (en
Inventor
刘宁
崔焕民
杨建慧
牟伟腾
袁照威
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.)
Datang Beijing Water Engineering Technology Co ltd
Original Assignee
Datang Beijing Water Engineering Technology Co ltd
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 Datang Beijing Water Engineering Technology Co ltd filed Critical Datang Beijing Water Engineering Technology Co ltd
Priority to CN201810987345.9A priority Critical patent/CN109019717B/en
Publication of CN109019717A publication Critical patent/CN109019717A/en
Application granted granted Critical
Publication of CN109019717B publication Critical patent/CN109019717B/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
    • 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
    • C02F1/5236—Treatment of water, waste water, or sewage by flocculation or precipitation of suspended impurities using inorganic agents
    • C—CHEMISTRY; METALLURGY
    • C02—TREATMENT OF WATER, WASTE WATER, SEWAGE, OR SLUDGE
    • C02F—TREATMENT OF WATER, WASTE WATER, SEWAGE, OR SLUDGE
    • C02F2103/00—Nature of the water, waste water, sewage or sludge to be treated
    • C02F2103/18—Nature of the water, waste water, sewage or sludge to be treated from the purification of gaseous effluents
    • 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

Landscapes

  • Chemical & Material Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Hydrology & Water Resources (AREA)
  • Engineering & Computer Science (AREA)
  • Environmental & Geological Engineering (AREA)
  • Water Supply & Treatment (AREA)
  • Organic Chemistry (AREA)
  • Inorganic Chemistry (AREA)
  • Separation Of Suspended Particles By Flocculating Agents (AREA)
  • Treating Waste Gases (AREA)

Abstract

The invention discloses an intelligent pretreatment method and system for desulfurization wastewater of a thermal power plant. The intelligent preprocessing method comprises the following steps: acquiring a training sample, wherein the training sample comprises historical monitoring index data and historical dosing amount data of the water quality of the desulfurization wastewater; constructing an intelligent processing model according to the historical water quality monitoring index data and the historical medicine adding amount data; acquiring water quality monitoring index data of the desulfurization wastewater at the current moment; and inputting the water quality monitoring index data of the desulfurization wastewater at the current moment into the intelligent pretreatment model to obtain the optimal dosage of the current desulfurization wastewater. By adopting the method or the system, the problem of accurate calculation of the dosage of the traditional desulfurization wastewater can be solved, and the rapid, convenient and effective dosing process is realized.

Description

Intelligent treatment method and system for thermal power plant desulfurization wastewater
Technical Field
The invention relates to the technical field of thermal power plant wastewater treatment, in particular to an intelligent treatment method and system for thermal power plant desulfurization wastewater.
Background
At present, the requirements of the national desulfurization wastewater discharge standard are increasingly strict, and particularly, after a State administration releases a Water pollution prevention action plan (Ten Water for short) in 2015, the state puts forward more strict requirements on the treatment of various water body pollutions. Meanwhile, the national 'thirteen-five' plan further strictly controls the use of water resources to improve the water environment protection to the national strategic level. In order to strictly execute national laws and regulations and industrial specifications, the thermal power plant is taken as a large user for water utilization and drainage, and zero emission of the coal-fired power plant is urgent.
At present, most of water treatment systems of thermal power plants are being or are planned to be transformed to realize the recycling of wastewater, but the desulfurization wastewater is used as a strand of water treated at the tail end of the power plant, and the water quality has the characteristics of high calcium and magnesium ion concentration, high suspended matter content, high heavy metal content, high chloride ion content and the like. At present, the technology of pretreatment, concentration and decrement and end solidification is commonly adopted, and the pretreatment operation is the most basicThe working aims at meeting the requirements of anti-scaling, concentration and decrement and the like of a subsequent device, and the removal degree of impurity ions in the incoming water in the pretreatment process determines the process requirements and the cost of subsequent membrane concentration and flue evaporation. Therefore, the pretreatment is a more critical process in the whole process flow, and the addition of lime Na is usually adopted2SO4+Na2CO3And removing impurities in the wastewater by using the agent. The dosage of the medicament added is determined by the effluent quality of the operating personnel or experiments in the prior art, the accurate dosage is difficult to achieve, a large amount of time and labor are wasted, and the effluent index can not meet the effluent requirement.
Therefore, based on the defect in the traditional desulfurization wastewater dosing process, the realization of intelligent dosing of desulfurization wastewater is a problem to be solved urgently in the water treatment industry.
Disclosure of Invention
The invention aims to provide an intelligent treatment method and system for thermal power plant desulfurization wastewater, which overcome the defect that the dosage of a dosing agent is difficult to accurately determine in the traditional desulfurization wastewater pretreatment process and realize an accurate and effective dosing process.
In order to achieve the purpose, the invention provides the following scheme:
an intelligent treatment method for desulfurization wastewater of a thermal power plant comprises the following steps:
acquiring a training sample, wherein the training sample comprises historical monitoring index data and historical dosing amount data of the water quality of the desulfurization wastewater;
constructing an intelligent processing model according to the historical water quality monitoring index data and the historical medicine adding amount data;
acquiring water quality monitoring index data of the desulfurization wastewater at the current moment;
and inputting the water quality monitoring index data of the desulfurization wastewater at the current moment into the intelligent pretreatment model to obtain the optimal dosage of the current desulfurization wastewater.
Optionally, an intelligent processing model is constructed according to the historical water quality monitoring index data and the historical dosing amount data, and the intelligent processing model specifically comprises the following steps:
obtaining a training sample according to the historical water quality monitoring index data and the historical dosing amount data;
based on a plurality of single kernel functions, constructing a weighted kernel function comprising the single kernel functions in a weighted mode, wherein the weighted kernel function comprises kernel function parameters;
establishing an SVR mathematical prediction model according to the training sample, the kernel function parameters and set SVR parameters, wherein the SVR parameters comprise a penalty coefficient and insensitive parameters;
traversing the parameter range by adopting a grid search method according to the SVR mathematical prediction model;
obtaining an optimal SVR parameter by adopting a five-fold cross verification method according to the parameter range;
judging whether the optimal SVR parameter reaches a set precision threshold range;
if the optimal SVR parameter reaches a set precision threshold range, establishing an intelligent processing model according to the optimal SVR parameter range;
and if the optimal SVR parameter does not reach the set precision threshold range, returning to traverse the parameter range by adopting a grid search method according to the SVR mathematical prediction model.
Optionally, the training sample is a time series set (X, Y) composed of an input time series signal and an output time series signal, wherein the input time series signal is historical monitoring index data of the water quality of the desulfurization wastewater, and the output time series signal is the historical dosing data;
the input time series signal is X ═ Xij]nⅹp＝[x1,x2,…,xi,…,xn]，i＝1,2,……n，j＝1,2,……p，xi＝[xi1,xi2,…,xip]N is the sample number of the historical monitoring index data of the quality of the desulfurization wastewater in the training sample, and p is the historical monitoring index data of the quality of the desulfurization wastewater; the desulfurization wastewater water quality monitoring index data comprises Mg2+、Ca2+、Cl-、SO4 2-One or more of water quality indicators;
the said transfusionGiving a time-series signal of Y ═ Yi]nⅹ1I is 1,2, … … n, and n is the sample number of the desulfurization wastewater historical dosage data in the training sample; the historical dosing data of the desulfurization wastewater comprises lime and Na2SO4、Na2CO3One or more of a coagulant aid or flocculant agent.
Optionally, the constructing a weighted kernel function including a plurality of single kernel functions in a weighted manner based on a plurality of single kernel functions specifically includes:
different mononuclear functions k are weighted in a set modetAnd combining to obtain a multi-core weighted core function kappa, wherein the multi-core weighted core function kappa is represented by the following formula:
Figure BDA0001779996980000031
wherein, mutWeight coefficients for different kernel functions are obtained by a genetic algorithm, an
Figure BDA0001779996980000032
κtAre different single kernel functions, which are often linear kernel functions, polynomial kernel functions, gaussian radial basis kernel functions; and t is 1,2 … …, m and m are the number of single kernel functions, and a global polynomial kernel function and a local Gaussian radial basis kernel function are adopted in the calculation process.
Optionally, the establishing an SVR mathematical prediction model according to the multi-kernel weighted kernel function specifically includes:
substituting the multi-kernel weighted kernel function into the SVR basic mathematical model f (x) ═ ωTx + b to obtain the expression form f (x) ═ omega of the SVR model in the high-dimensional spaceTκ(xi,xj) + b, where ω represents a weight vector matrix; b represents an offset, which is constant;
the SVR mathematical model is converted into:
Figure BDA0001779996980000041
the specific conversion process is as follows:
the mathematical model f (x) corresponds to an objective function of:
Figure BDA0001779996980000042
the corresponding constraint conditions are as follows:
yi-[ωTκ(xi)+b]≤ε+ξi
s.t.[ωTκ(xi)+b]-yi≤ε+ξi'
ξi,ξi'≥0,i＝1,2,…,n
the solution of the problem needs to introduce a Lagrange multiplier
Figure BDA0001779996980000043
And defines the lagrangian function L:
Figure BDA0001779996980000044
for parameters omega, b, xi respectivelyi,
Figure BDA0001779996980000045
Calculating a partial derivative:
Figure BDA0001779996980000046
Figure BDA0001779996980000047
Figure BDA0001779996980000048
Figure BDA0001779996980000049
the objective function is converted to its dual problem according to the above equation:
Figure BDA0001779996980000051
Figure BDA0001779996980000052
the SVR mathematical model is converted into:
Figure BDA0001779996980000053
the meaning of the symbols is explained as follows:
ω represents a weight vector matrix; b represents an offset, which is constant; c is a penalty coefficient; n is the sample number of the historical monitoring index data of the quality of the desulfurization wastewater in the training sample; xiiAnd xii' is a relaxation factor; epsilon is an insensitive parameter; alpha is alphai、
Figure BDA0001779996980000054
ηiAnd
Figure BDA0001779996980000055
is a lagrange multiplier; l is a Lagrangian function.
In order to achieve the purpose, the invention provides the following scheme:
an intelligent thermal power plant desulfurization wastewater treatment system, the system comprising:
the first acquisition module is used for acquiring a training sample, wherein the training sample comprises historical monitoring index data and historical dosing amount data of the water quality of the desulfurization wastewater;
the intelligent processing model building module is used for building an intelligent processing model according to the historical water quality monitoring index data and the historical medicine adding amount data;
the second acquisition module is used for acquiring the water quality monitoring index data of the desulfurization wastewater at the current moment;
and the optimal dosage determining module is used for inputting the water quality monitoring index data of the desulfurization wastewater at the current moment into the intelligent pretreatment model to obtain the optimal dosage of the current desulfurization wastewater.
Optionally, the intelligent processing model building module specifically includes:
the training sample determining unit is used for obtaining a training sample according to the water quality historical monitoring index data and the historical medicine adding amount data;
the multi-kernel weighted kernel function establishing unit is used for establishing a weighted kernel function comprising a plurality of single kernel functions in a weighted mode based on the single kernel functions, and the weighted kernel function comprises kernel function parameters;
the SVR mathematical prediction model establishing unit is used for establishing an SVR mathematical prediction model according to the training sample, the kernel function parameters and set SVR parameters, and the SVR parameters comprise punishment coefficients and insensitive parameters;
the parameter optimization range determining unit is used for traversing the parameter range by adopting a grid search method according to the SVR mathematical prediction model;
the optimal SVR parameter range determining unit is used for obtaining optimal SVR parameters by adopting a five-fold cross verification method according to the parameter range; (ii) a
The judging unit is used for judging whether the optimal SVR parameter range reaches a set precision threshold range;
the first judgment result unit is used for establishing an intelligent processing model according to the optimal SVR parameter range if the optimal SVR parameter range reaches a set precision threshold range;
and the second judgment result unit is used for returning to traverse the parameter range by adopting a grid search method according to the SVR mathematical prediction model if the optimal SVR parameter range does not reach the set precision threshold range.
Optionally, the training sample is a time series set (X, Y) composed of an input time series signal and an output time series signal, wherein the input time series signal is historical monitoring index data of the water quality of the desulfurization wastewater, and the output time series signal is the historical dosing data;
the input time series signal is X ═ Xij]nⅹp＝[x1,x2,…,xi,…,xn]，i＝1,2,……n，j＝1,2,……p，xi＝[xi1,xi2,…,xip]N is the sample number of the historical monitoring index data of the quality of the desulfurization wastewater in the training sample, and p is the historical monitoring index data of the quality of the desulfurization wastewater; the desulfurization wastewater water quality monitoring index data comprises Mg2+、Ca2+、Cl-、SO4 2-One or more of water quality indicators;
the output time series signal is Y ═ Yi]nⅹ1I is 1,2, … … n, and n is the sample number of the desulfurization wastewater historical dosage data in the training sample; the historical dosing data of the desulfurization wastewater comprises lime and Na2SO4、Na2CO3One or more of a coagulant aid or flocculant agent.
Optionally, the multi-kernel weighted kernel function establishing unit specifically includes:
the mononuclear function constructing subunit is used for constructing different mononuclear functions according to the water quality historical monitoring index data and the historical dosing amount data;
a multi-core weighted kernel function establishing subunit for establishing different single kernel functions kappa in a set weighting modetAnd combining to obtain a multi-core weighted core function kappa, wherein the multi-core weighted core function kappa is represented by the following formula:
Figure BDA0001779996980000071
wherein, mutAre weight coefficients of different kernel functions, an
Figure BDA0001779996980000072
κtIs a different sheetThe single kernel function is a linear kernel function, a polynomial kernel function and a Gaussian radial basis kernel function; t is 1,2 … …, m is the number of single kernel functions.
Optionally, the SVR mathematical prediction model establishing unit specifically includes:
an SVR mathematical prediction model establishing subunit for establishing the multi-core weighted kernel function
Figure BDA0001779996980000073
Substituting the SVR basic mathematical model f (x) ═ ωTx + b to obtain the expression form f (x) ═ omega of the SVR model in the high-dimensional spaceTκ(xi,xj) + b, where ω represents a weight vector matrix; b represents an offset, which is constant;
the SVR mathematical model is converted into:
Figure BDA0001779996980000074
the specific conversion process is as follows:
the mathematical model f (x) corresponds to an objective function of:
Figure BDA0001779996980000075
the corresponding constraint conditions are as follows:
yi-[ωTκ(xi)+b]≤ε+ξi
s.t.[ωTκ(xi)+b]-yi≤ε+ξi'
ξi,ξi'≥0,i＝1,2,…,n
the solution of the problem needs to introduce a Lagrange multiplier
Figure BDA0001779996980000081
And defines the lagrangian function L:
Figure BDA0001779996980000082
for parameters omega, b, xi respectivelyi,
Figure BDA0001779996980000083
Calculating a partial derivative:
Figure BDA0001779996980000084
Figure BDA0001779996980000085
Figure BDA0001779996980000086
Figure BDA0001779996980000087
the objective function is converted to its dual problem according to the above equation:
Figure BDA0001779996980000088
Figure BDA0001779996980000089
the SVR mathematical model is converted into:
Figure BDA00017799969800000810
the meaning of the symbols is explained as follows:
ω represents a weight vector matrix; b represents an offset, which is constant; c is a penalty coefficient; n is the sample of the historical monitoring index data of the quality of the desulfurization wastewater in the training sampleCounting; xiiAnd xii' is a relaxation factor; epsilon is an insensitive parameter; alpha is alphai、
Figure BDA0001779996980000091
ηiAnd
Figure BDA0001779996980000092
is a lagrange multiplier; k () is a kernel function; l is a Lagrangian function.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects: the invention provides an intelligent pretreatment method for desulfurization wastewater of a thermal power plant, which comprises the following steps: acquiring a training sample, wherein the training sample comprises historical monitoring index data and historical dosing amount data of the water quality of the desulfurization wastewater; constructing an intelligent processing model according to the historical water quality monitoring index data and the historical medicine adding amount data; acquiring water quality monitoring index data of the desulfurization wastewater at the current moment; and inputting the water quality monitoring index data of the desulfurization wastewater at the current moment into the intelligent pretreatment model to obtain the optimal dosage of the current desulfurization wastewater. By establishing an intelligent treatment model, the problem of accurate calculation of the dosage of the traditional desulfurization wastewater is solved, and a rapid, convenient and effective dosing process is realized.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a flow chart of an intelligent treatment method for desulfurization wastewater of a thermal power plant according to an embodiment of the invention;
FIG. 2 is a flow chart of a method for constructing an intelligent processing model according to an embodiment of the present invention;
FIG. 3 is a structural diagram of an intelligent desulfurization wastewater treatment system of a thermal power plant according to an embodiment of the invention;
FIG. 4 is a block diagram of an intelligent process model building block according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
SVR (Support vector Regression) is an important application branch of Support Vector Machines (SVMs), the former is mainly used for Regression, and the latter is mainly used for classification. The SVM method was first proposed by the Vanpik team in 1963, and subsequently Platt and Chih-Jen Lin et al further pushed the SVM method to application. In the field of machine learning, the SVM/SVR method is a supervised learning model, which is commonly used for pattern recognition, classification, and regression analysis. The method can convert the nonlinear problem into a high-dimensional space through a kernel function, so that the nonlinear problem can be linearly divided. Meanwhile, the method is based on the structure risk minimization theory, the learner can obtain global optimization, and the expected risk of the whole sample space meets a certain upper bound with a certain probability.
The invention provides a thermal power plant desulfurization wastewater pretreatment method and a thermal power plant desulfurization wastewater pretreatment system. The method avoids the nonlinear optimization problems of blindness, local optimum and the like of kernel function design in the SVR model through the multi-kernel weighted kernel function, solves the nonlinear problem in the intelligent prediction model through the SVR method, optimizes parameters such as punishment coefficient C, insensitive parameter epsilon, kernel function parameters and the like of the SVR model through a grid search method, finally obtains the SVR intelligent prediction model based on the multi-kernel weighting, and realizes the intelligent pretreatment of the desulfurization wastewater through the model.
FIG. 1 is a flow chart of an intelligent treatment method for desulfurization wastewater of a thermal power plant in an embodiment of the invention. As shown in fig. 1, an intelligent treatment method for desulfurization wastewater of a thermal power plant comprises the following steps:
step 101: acquiring a training sample, wherein the training sample comprises historical monitoring index data and historical dosing amount data of the water quality of the desulfurization wastewater;
step 102: constructing an intelligent processing model according to the historical water quality monitoring index data and the historical medicine adding amount data;
step 103: acquiring water quality monitoring index data of the desulfurization wastewater at the current moment;
step 104: and inputting the water quality monitoring index data of the desulfurization wastewater at the current moment into the intelligent pretreatment model to obtain the optimal dosage of the current desulfurization wastewater.
FIG. 2 is a flowchart of a method for constructing an intelligent processing model according to an embodiment of the present invention. As shown in fig. 2, the method for constructing an intelligent processing model specifically includes:
step 1021: obtaining a training sample according to the historical water quality monitoring index data and the historical dosing amount data;
step 1022: based on a plurality of single kernel functions, constructing a weighted kernel function comprising the single kernel functions in a weighted mode, wherein the weighted kernel function comprises kernel function parameters;
step 1023: establishing an SVR mathematical prediction model according to the training sample, the kernel function parameters and set SVR parameters, wherein the SVR parameters comprise a penalty coefficient and insensitive parameters;
step 1024: traversing the parameter range by adopting a grid search method according to the SVR mathematical prediction model;
step 1025: obtaining an optimal SVR parameter by adopting a five-fold cross verification method according to the parameter range;
step 1026: judging whether the optimal SVR parameter reaches a set precision threshold range;
step 1027: if the optimal SVR parameter reaches a set precision threshold range, establishing an intelligent processing model according to the optimal SVR parameter range;
if the optimal SVR parameter does not reach the set precision threshold range, the process returns to step 1022.
The training sample is a time sequence set (X, Y) consisting of an input time sequence signal and an output time sequence signal, wherein the input time sequence signal is historical monitoring index data of the water quality of the desulfurization wastewater, and the output time sequence signal is historical dosing data;
the input time series signal is X ═ Xij]nⅹp＝[x1,x2,…,xi,…,xn]，i＝1,2,……n，j＝1,2,……p，xi＝[xi1,xi2,…,xip]N is the sample number of the historical monitoring index data of the quality of the desulfurization wastewater in the training sample, and p is the historical monitoring index data of the quality of the desulfurization wastewater; the desulfurization wastewater water quality monitoring index data comprises Mg2+、Ca2+、Cl-、SO4 2-One or more of water quality indicators;
the output time series signal is Y ═ Yi]nⅹ1I is 1,2, … … n, and n is the sample number of the desulfurization wastewater historical dosage data in the training sample; the historical dosing data of the desulfurization wastewater comprises lime and Na2SO4、Na2CO3One or more of a coagulant aid or flocculant agent.
Step 1021 specifically includes:
constructing different mononuclear functions according to the historical water quality monitoring index data and the historical dosing amount data;
different mononuclear functions k are weighted in a set modetAnd combining to obtain a multi-core weighted core function kappa, wherein the multi-core weighted core function kappa is represented by the following formula:
Figure BDA0001779996980000121
wherein, mutFor different kernel functionsA weight coefficient, and
Figure BDA0001779996980000122
κtthe single kernel functions are different single kernel functions, and the single kernel functions are linear kernel functions, polynomial kernel functions and Gaussian radial basis kernel functions; t is 1,2 … …, m is the number of single kernel functions.
Step 1022 specifically includes:
weighting the multi-core kernel function
Figure BDA0001779996980000123
Substituting the SVR basic mathematical model f (x) ═ ωTx + b to obtain the expression form f (x) ═ omega of the SVR model in the high-dimensional spaceTκ(xi,xj) + b, where ω represents a weight vector matrix; b represents an offset, which is constant;
the SVR mathematical model is converted into:
Figure BDA0001779996980000124
the specific conversion process is as follows:
the mathematical model f (x) corresponds to an objective function of:
Figure BDA0001779996980000125
the corresponding constraint conditions are as follows:
yi-[ωTκ(xi)+b]≤ε+ξi
s.t.[ωTκ(xi)+b]-yi≤ε+ξi'
ξi,ξi'≥0,i＝1,2,…,n
the solution of the problem needs to introduce a Lagrange multiplier
Figure BDA0001779996980000126
And defines the lagrangian function L:
Figure BDA0001779996980000131
for parameters omega, b, xi respectivelyi,
Figure BDA0001779996980000132
Calculating a partial derivative:
Figure BDA0001779996980000133
Figure BDA0001779996980000134
Figure BDA0001779996980000135
Figure BDA0001779996980000136
the objective function is converted to its dual problem according to the above equation:
Figure BDA0001779996980000137
Figure BDA0001779996980000138
the SVR mathematical model is converted into:
Figure BDA0001779996980000139
the meaning of the symbols is explained as follows:
ω represents a weight vector matrix; b represents an offset, which is constant; c isA penalty coefficient; n is the sample number of the historical monitoring index data of the quality of the desulfurization wastewater in the training sample; xiiAnd xii' is a relaxation factor; epsilon is an insensitive parameter; alpha is alphai、
Figure BDA00017799969800001311
ηiAnd
Figure BDA00017799969800001310
is a lagrange multiplier; l is a Lagrangian function.
The SVR parameters comprise a penalty coefficient C, an insensitive parameter epsilon and a kernel function parameter; the kernel function parameters are for a polynomial kernel function k (x, x)i)＝((xi,xj)+r)dThe parameters are r and d, and the default value of r is 0; for the Gaussian kernel function kappa (x, x)i)＝exp(-||xi-xj||/2σ2) The parameter is σ.
The grid searching method is an exhaustive searching method for specified parameter values, and mainly traverses the value ranges of SVR parameters such as penalty coefficient C, insensitive parameter epsilon and kernel function parameter. The value range of the penalty coefficient C is generally [2 ]-5,225](ii) a The value range of the insensitive parameter epsilon is generally [0.002,0.1 ]](ii) a When the kernel function is a gaussian kernel function, the value range of the parameter sigma is generally [2 ]-15,215]When the kernel function selects the polynomial kernel function, the value range of the parameter d is generally [1,10 ]]。
The 5-fold cross validation is to divide the training sample (X, Y) into 5 parts, one part is selected as a test set each time, and four parts are selected as a training set. And (5) calculating the mean square error according to the SVR mathematical model in the step (5), repeating for 5 times, and solving the mean square error for 5 times as the final calculation precision.
And further reducing the range of the punishment coefficient C, the insensitive parameter epsilon, the kernel function parameter and other SVR parameters according to the calculation precision, and repeating the steps until the calculation precision reaches the optimal precision to obtain the optimal intelligent prediction model.
Desulfurization wastewater monitoring index dataPredominantly Mg2+、Ca2+、Cl-And SO4 2-The water quality indexes are equal; the type of the medicine added is mainly lime and Na2SO4、Na2CO3Coagulant aids or flocculants, and the like; the intelligent preprocessing model is built by adopting a multi-core weighted SVR (support vector regression) model. The invention solves the problem that the optimal dosage is difficult to achieve by a single kernel function in the SVR dosage prediction process under the condition of complex water quality. Meanwhile, the multi-core weighted kernel function avoids the problems of blindness of kernel function selection and local optimum nonlinear optimization, and is beneficial to improving the performance of the learning machine. Therefore, the intelligent pretreatment model provided by the invention overcomes the defect that the dosage of the medicament is difficult to accurately determine in the desulfurization wastewater treatment process, and realizes quick and accurate dosage.
FIG. 3 is a structural diagram of an intelligent desulfurization wastewater treatment system of a thermal power plant in an embodiment of the invention. As shown in fig. 3, an intelligent treatment system for desulfurization wastewater of a thermal power plant comprises:
the first acquisition module 201 is used for acquiring a training sample, wherein the training sample comprises historical monitoring index data and historical dosing amount data of the water quality of the desulfurization wastewater;
the intelligent processing model building module 202 is used for building an intelligent processing model according to the historical water quality monitoring index data and the historical medicine adding amount data;
the second obtaining module 203 is used for obtaining the water quality monitoring index data of the desulfurization wastewater at the current moment;
and the optimal dosage determining module 204 is used for inputting the data of the water quality monitoring index of the desulfurization wastewater at the current moment into the intelligent pretreatment model to obtain the optimal dosage of the current desulfurization wastewater.
FIG. 4 is a block diagram of an intelligent process model building block according to an embodiment of the present invention. As shown in fig. 4, the intelligent processing model building module 202 specifically includes:
the training sample determining unit 2021 is configured to obtain a training sample according to the historical water quality monitoring index data and the historical chemical dosing amount data;
the multi-kernel weighted kernel function establishing unit 2022 is configured to establish, based on the plurality of single kernel functions, a weighted kernel function including the plurality of single kernel functions in a weighted manner, where the weighted kernel function includes kernel function parameters;
the SVR mathematical prediction model establishing unit 2023 is configured to establish an SVR mathematical prediction model according to the training samples, the kernel function parameters, and set SVR parameters, where the SVR parameters include a penalty coefficient and an insensitive parameter;
the parameter optimization range determining unit 2024 is configured to traverse a parameter range by using a grid search method according to the SVR mathematical prediction model;
the optimal SVR parameter range determining unit 2025 is configured to obtain an optimal SVR parameter by using a five-fold cross validation method according to the parameter range;
the judging unit 2026 is configured to judge whether the optimal SVR parameter range reaches a set accuracy threshold range;
the first judgment result unit 2027 is configured to establish an intelligent processing model according to the optimal SVR parameter range if the optimal SVR parameter range reaches a set precision threshold range;
and a second result judgment unit 2028, configured to return to establishing the SVR mathematical prediction model according to the multi-core weighted kernel function if the optimal SVR parameter range does not reach the set precision threshold range.
The training sample is a time sequence set (X, Y) consisting of an input time sequence signal and an output time sequence signal, wherein the input time sequence signal is historical monitoring index data of the water quality of the desulfurization wastewater, and the output time sequence signal is historical dosing data;
the input time series signal is X ═ Xij]nⅹp＝[x1,x2,…,xi,…,xn]，i＝1,2,……n，j＝1,2,……p，xi＝[xi1,xi2,…,xip]N is the sample number of the historical monitoring index data of the quality of the desulfurization wastewater in the training sample, and p is the historical monitoring index data of the quality of the desulfurization wastewater; the desulfurization wastewater water quality monitoring indexData include Mg2+、Ca2+、Cl-、SO4 2-One or more of water quality indicators;
the output time series signal is Y ═ Yi]nⅹ1I is 1,2, … … n, and n is the sample number of the desulfurization wastewater historical dosage data in the training sample; the historical dosing data of the desulfurization wastewater comprises lime and Na2SO4、Na2CO3One or more of a coagulant aid or flocculant agent.
The multi-kernel weighted kernel function establishing unit 2021 specifically includes:
the mononuclear function constructing subunit is used for constructing different mononuclear functions according to the water quality historical monitoring index data and the historical dosing amount data;
a multi-core weighted kernel function establishing subunit for establishing different single kernel functions kappa in a set weighting modetAnd combining to obtain a multi-core weighted core function kappa, wherein the multi-core weighted core function kappa is represented by the following formula:
Figure BDA0001779996980000161
wherein, mutAre weight coefficients of different kernel functions, an
Figure BDA0001779996980000162
κtThe single kernel functions are different single kernel functions, and the single kernel functions are linear kernel functions, polynomial kernel functions and Gaussian radial basis kernel functions; t is 1,2 … …, m is the number of single kernel functions.
The SVR mathematical prediction model establishing unit 2022 specifically includes:
an SVR mathematical prediction model establishing subunit for establishing the multi-core weighted kernel function
Figure BDA0001779996980000163
Substituting the SVR basic mathematical model f (x) ═ ωTx + b to obtain the expression form f (x) ═ omega of the SVR model in the high-dimensional spaceTκ(xi,xj) + b, where ω represents a weight vector matrix; b represents an offset, which is constant;
the mathematical model f (x) corresponds to an objective function of:
Figure BDA0001779996980000171
the corresponding constraint conditions are as follows:
yi-[ωTκ(xi)+b]≤ε+ξi
s.t.[ωTκ(xi)+b]-yi≤ε+ξi'
ξi,ξi'≥0,i＝1,2,…,n
the dual problem is as follows:
Figure BDA0001779996980000172
Figure BDA0001779996980000173
the SVR mathematical model is converted into:
Figure BDA0001779996980000174
the meaning of the symbols is explained as follows:
ω represents a weight vector matrix; b represents an offset, which is constant; c is a penalty coefficient; n is the sample number of the historical monitoring index data of the quality of the desulfurization wastewater in the training sample; xiiAnd xii' is a relaxation factor; epsilon is an insensitive parameter; alpha is alphai、
Figure BDA0001779996980000175
ηiAnd
Figure BDA0001779996980000176
is a lagrange multiplier; l is a Lagrangian function.
The SVR parameters comprise a penalty coefficient C, an insensitive parameter epsilon and a kernel function parameter; the kernel function parameters are for a polynomial kernel function k (x, x)i)＝((xi,xj)+r)dThe parameters are r and d, and the default value of r is 0; for the Gaussian kernel function kappa (x, x)i)＝exp(-||xi-xj||/2σ2) The parameter is σ.
The grid searching method is an exhaustive searching method for specified parameter values, and mainly traverses the value ranges of SVR parameters such as penalty coefficient C, insensitive parameter epsilon and kernel function parameter. The value range of the penalty coefficient C is generally [2 ]-5,225](ii) a The value range of the insensitive parameter epsilon is generally [0.002,0.1 ]](ii) a When the kernel function is a gaussian kernel function, the value range of the parameter sigma is generally [2 ]-15,215]When the kernel function selects the polynomial kernel function, the value range of the parameter d is generally [1,10 ]]。
The 5-fold cross validation is to divide the training sample (X, Y) into 5 parts, one part is selected as a test set each time, and four parts are selected as a training set. And (5) calculating the mean square error according to the SVR mathematical model in the step (5), repeating for 5 times, and solving the mean square error for 5 times as the final calculation precision.
And further reducing the range of the punishment coefficient C, the insensitive parameter epsilon, the kernel function parameter and other SVR parameters according to the calculation precision, and repeating the steps until the calculation precision reaches the optimal precision to obtain the optimal intelligent prediction model.
Compared with the prior art, the invention has the beneficial effects that:
1. the problem of non-linear optimization such as blindness and local optimum of kernel function design in an SVR model is solved by adopting a weighted kernel function formed by a global polynomial kernel function and a local Gaussian radial basis kernel function.
2. And optimizing a penalty coefficient C, an insensitive parameter epsilon and a kernel function parameter of the SVR model through a grid search algorithm, and endowing the obtained SVR optimal parameter to the SVR mathematical model, so that the prediction precision of the SVR model is improved.
3. The intelligent prediction model solves the adverse effect of the time-varying characteristic of the incoming water quality on the operation process in the traditional desulfurization wastewater pretreatment process, reduces the waste of reagents, reduces the labor cost and the operation cost, and meets the water quality requirement of the outgoing water in real time.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.

Claims (6)

1. An intelligent treatment method for desulfurization wastewater of a thermal power plant is characterized by comprising the following steps:
acquiring a training sample, wherein the training sample comprises historical monitoring index data and historical dosing amount data of the water quality of the desulfurization wastewater; the method specifically comprises the following steps: obtaining a training sample according to the historical water quality monitoring index data and the historical dosing amount data; based on a plurality of single kernel functions, constructing a weighted kernel function comprising the single kernel functions in a weighted mode, wherein the weighted kernel function comprises kernel function parameters; establishing an SVR mathematical prediction model according to the training sample, the kernel function parameters and set SVR parameters, wherein the SVR parameters comprise a penalty coefficient and insensitive parameters; traversing the parameter range by adopting a grid search method according to the SVR mathematical prediction model; obtaining an optimal SVR parameter by adopting a five-fold cross verification method according to the parameter range; judging whether the optimal SVR parameter reaches a set precision threshold range; if the optimal SVR parameter reaches a set precision threshold range, establishing an intelligent processing model according to the optimal SVR parameter range; if the optimal SVR parameter does not reach the set precision threshold range, returning to traverse the parameter range by adopting a grid search method according to the SVR mathematical prediction model;
constructing an intelligent processing model according to the historical water quality monitoring index data and the historical medicine adding amount data;
acquiring water quality monitoring index data of the desulfurization wastewater at the current moment;
inputting the water quality monitoring index data of the desulfurization wastewater at the current moment into the intelligent pretreatment model to obtain the optimal dosage of the current desulfurization wastewater;
traversing the value range of the SVR parameters by a grid searching method; the SVR parameters include: penalty factorCInsensitive parameter
Figure DEST_PATH_IMAGE002
And kernel function parameters; the penalty coefficientCHas a value range of [2 ]-5,225](ii) a The insensitive parameter
Figure DEST_PATH_IMAGE002A
Has a value range of [0.002,0.1 ]](ii) a When the kernel function selects the Gaussian kernel function, the value range of the parameter sigma is [2 ]-15,215]When the kernel function selects a polynomial kernel function, its parametersdThe value range is [1,10 ]]；
Establishing an SVR mathematical prediction model according to the multi-core weighted kernel function, which specifically comprises the following steps:
substituting the multi-kernel weighted kernel function into the SVR mathematical model
Figure DEST_PATH_IMAGE004
Obtaining the expression form of the SVR model in the high-dimensional space
Figure DEST_PATH_IMAGE006
In the formula (I), wherein,
Figure DEST_PATH_IMAGE008
representing a weight vector matrix;
Figure DEST_PATH_IMAGE010
represents a bias, which is constant;
the SVR mathematical model is converted into:
Figure DEST_PATH_IMAGE012
the specific conversion process is as follows:
the mathematical model
Figure DEST_PATH_IMAGE014
The corresponding objective function is:
Figure DEST_PATH_IMAGE016
the corresponding constraint conditions are as follows:
Figure DEST_PATH_IMAGE018
introducing lagrange multipliers
Figure DEST_PATH_IMAGE020
And defining a Lagrangian functionL：
Figure DEST_PATH_IMAGE022
Respectively to the parameters
Figure DEST_PATH_IMAGE024
Calculating a partial derivative:
Figure DEST_PATH_IMAGE026
the objective function is converted to its dual problem according to the above equation:
Figure DEST_PATH_IMAGE028
Figure DEST_PATH_IMAGE030
the SVR mathematical model is converted into:
Figure DEST_PATH_IMAGE012A
Figure DEST_PATH_IMAGE008A
representing a weight vector matrix;
Figure DEST_PATH_IMAGE010A
represents a bias, which is constant;
Figure DEST_PATH_IMAGE032
is a penalty coefficient;nthe number of samples of the historical monitoring index data of the quality of the desulfurization wastewater in the training sample is;
Figure DEST_PATH_IMAGE034
and
Figure DEST_PATH_IMAGE036
is a relaxation factor;
Figure DEST_PATH_IMAGE002AA
is an insensitive parameter;
Figure DEST_PATH_IMAGE038
、
Figure DEST_PATH_IMAGE040
、
Figure DEST_PATH_IMAGE042
and
Figure DEST_PATH_IMAGE044
is a lagrange multiplier;Lis the lagrange function.
2. The intelligent treatment method for desulfurization wastewater of thermal power plant according to claim 1, wherein the training sample is a time series set consisting of an input time series signal and an output time series signal(s) ((X,Y) Wherein the input time series signal is historical monitoring index data of the water quality of the desulfurization wastewater, and the output time series signal is historical dosing data;
the input time series signal isX=[x ij ] nⅹp =[x 1,x 2,…,x i ,…,x n ]，i=1,2,……n，j=1,2,……p，x i =[x i1,x i2,…,x ip ]，nThe number of samples of the historical monitoring index data of the quality of the desulfurization wastewater in the training sample,phistorical monitoring index data of the water quality of the desulfurization wastewater; the desulfurization wastewater water quality monitoring index data comprises Mg2+、Ca2+、Cl-、SO4 2-One or more of water quality indicators;
the output time series signal isY=[y i ] nⅹ1，i=1,2,……n，nThe number of samples of the historical dosage data of the desulfurization wastewater in the training samples is calculated; the historical dosing data of the desulfurization wastewater comprises lime and Na2SO4、Na2CO3One or more of a coagulant aid or flocculant agent.
3. The intelligent thermal power plant desulfurization wastewater treatment method according to claim 1, wherein the constructing a weighted kernel function including a plurality of single kernel functions in a weighted manner based on the plurality of single kernel functions specifically comprises:
different single-kernel functions are combined in a set weighting mode
Figure DEST_PATH_IMAGE046
Combining to obtain multi-core weighted kernel function
Figure DEST_PATH_IMAGE048
Said multi-core weighted kernel function
Figure DEST_PATH_IMAGE048A
Expressed by the following formula:
Figure DEST_PATH_IMAGE050
；
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE052
weight coefficients for different kernel functions are obtained by a genetic algorithm, an
Figure DEST_PATH_IMAGE054
；
Figure DEST_PATH_IMAGE046A
Are different single kernel functions, which are often linear kernel functions, polynomial kernel functions, gaussian radial basis kernel functions;t=1, 2……, m，mthe number of the single kernel functions is the number of the single kernel functions, and a global polynomial kernel function and a local Gaussian radial basis kernel function are adopted in the calculation process.
4. The utility model provides a desulfurization waste water intelligence processing system of thermal power plant which characterized in that, the system includes:
the first acquisition module is used for acquiring a training sample, wherein the training sample comprises historical monitoring index data and historical dosing amount data of the water quality of the desulfurization wastewater;
the intelligent processing model building module is used for building an intelligent processing model according to the historical water quality monitoring index data and the historical medicine adding amount data;
the second acquisition module is used for acquiring the water quality monitoring index data of the desulfurization wastewater at the current moment;
the optimal dosage determining module is used for inputting the water quality monitoring index data of the desulfurization wastewater at the current moment into the intelligent pretreatment model to obtain the optimal dosage of the current desulfurization wastewater;
the intelligent processing model building module specifically comprises:
the training sample determining unit is used for obtaining a training sample according to the water quality historical monitoring index data and the historical medicine adding amount data;
the multi-kernel weighted kernel function establishing unit is used for establishing a weighted kernel function comprising a plurality of single kernel functions in a weighted mode based on the single kernel functions, and the weighted kernel function comprises kernel function parameters;
the SVR mathematical prediction model establishing unit is used for establishing an SVR mathematical prediction model according to the training sample, the kernel function parameters and set SVR parameters, and the SVR parameters comprise punishment coefficients and insensitive parameters;
the parameter optimization range determining unit is used for traversing the parameter range by adopting a grid search method according to the SVR mathematical prediction model;
the optimal SVR parameter range determining unit is used for obtaining optimal SVR parameters by adopting a five-fold cross verification method according to the parameter range;
the judging unit is used for judging whether the optimal SVR parameter range reaches a set precision threshold range;
the first judgment result unit is used for establishing an intelligent processing model according to the optimal SVR parameter range if the optimal SVR parameter range reaches a set precision threshold range;
the second judgment result unit is used for returning to traverse the parameter range by adopting a grid search method according to the SVR mathematical prediction model if the optimal SVR parameter range does not reach the set precision threshold range;
traversing the value range of the SVR parameters by a grid searching method; the SVR parameters include: penalty factorCInsensitive parameter
Figure DEST_PATH_IMAGE002AAA
And kernel function parameters; the penalty coefficientCHas a value range of [2 ]-5,225](ii) a The insensitive parameter
Figure DEST_PATH_IMAGE002AAAA
Has a value range of [0.002,0.1 ]](ii) a When the kernel function selects the Gaussian kernel function, the value range of the parameter sigma is [2 ]-15,215]When the kernel function selects a polynomial kernel function, its parametersdThe value range is [1,10 ]]；
The SVR mathematical prediction model establishing unit specifically comprises:
an SVR mathematical prediction model establishing subunit for establishing the multi-core weighted kernel function
Figure DEST_PATH_IMAGE050A
Substituting into the SVR mathematical model
Figure DEST_PATH_IMAGE004A
Obtaining the expression form of the SVR model in the high-dimensional space
Figure DEST_PATH_IMAGE006A
In the formula (I), wherein,
Figure DEST_PATH_IMAGE008AA
representing a weight vector matrix;
Figure DEST_PATH_IMAGE010AA
represents a bias, which is constant;
the SVR mathematical model is converted into:
Figure DEST_PATH_IMAGE012AA
；
the specific conversion process is as follows:
the mathematical model
Figure DEST_PATH_IMAGE014A
The corresponding objective function is:
Figure DEST_PATH_IMAGE016A
the corresponding constraint conditions are as follows:
Figure DEST_PATH_IMAGE018A
introducing lagrange multipliers
Figure DEST_PATH_IMAGE020A
And defining a Lagrangian functionL：
Figure DEST_PATH_IMAGE022A
Respectively to the parameters
Figure DEST_PATH_IMAGE024A
Calculating a partial derivative:
Figure DEST_PATH_IMAGE026A
the objective function is converted to its dual problem according to the above equation:
Figure DEST_PATH_IMAGE028A
Figure DEST_PATH_IMAGE030A
the SVR mathematical model is converted into:
Figure DEST_PATH_IMAGE012AAA
Figure DEST_PATH_IMAGE008AAA
representing a weight vector matrix;
Figure DEST_PATH_IMAGE010AAA
represents a bias, which is constant;
Figure DEST_PATH_IMAGE032A
is a penalty coefficient;nthe number of samples of the historical monitoring index data of the quality of the desulfurization wastewater in the training sample is;
Figure DEST_PATH_IMAGE034A
and
Figure DEST_PATH_IMAGE036A
is a relaxation factor;
Figure DEST_PATH_IMAGE002AAAAA
is an insensitive parameter;
Figure DEST_PATH_IMAGE038A
、
Figure DEST_PATH_IMAGE040A
、
Figure DEST_PATH_IMAGE042A
and
Figure DEST_PATH_IMAGE044A
is a lagrange multiplier;Lis the lagrange function.
5. The intelligent thermal power plant desulfurization wastewater treatment system according to claim 4, wherein the training samples are a time series set consisting of an input time series signal and an output time series signal(s) ((X,Y) Wherein the input time series signal is historical monitoring index data of the water quality of the desulfurization wastewater, and the output time series signal is historical dosing data;
the input time series signal isX=[x ij ] nⅹp =[x 1,x 2,…,x i ,…,x n ]，i=1,2,……n，j=1,2,……p，x i =[x i1,x i2,…,x ip ]，nThe number of samples of the historical monitoring index data of the quality of the desulfurization wastewater in the training sample is p, and p is the historical monitoring index data of the quality of the desulfurization wastewater; the desulfurization wastewater water quality monitoring index data comprises Mg2+、Ca2+、Cl-、SO4 2-One or more of water quality indicators;
the output time series signal isY=[y i ] nⅹ1，i=1,2,……n，nThe number of samples of the historical dosage data of the desulfurization wastewater in the training samples is calculated; the historical dosing data of the desulfurization wastewater comprises lime and Na2SO4、Na2CO3One or more of a coagulant aid or flocculant agent.
6. The intelligent thermal power plant desulfurization wastewater treatment system according to claim 4, wherein the multi-core weighted kernel function establishing unit specifically comprises:
the mononuclear function constructing subunit is used for constructing different mononuclear functions according to the water quality historical monitoring index data and the historical dosing amount data;
a multi-kernel weighted kernel function establishing subunit for establishing different single kernel functions in a set weighting mode
Figure DEST_PATH_IMAGE046AA
Combining to obtain multi-core weighted kernel function
Figure DEST_PATH_IMAGE048AA
Said multi-core weighted kernel function
Figure DEST_PATH_IMAGE048AAA
Expressed by the following formula:
Figure DEST_PATH_IMAGE050AA
；
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE052A
are weight coefficients of different kernel functions, an
Figure DEST_PATH_IMAGE054A
；
Figure DEST_PATH_IMAGE046AAA
The single kernel functions are different single kernel functions, and the single kernel functions are linear kernel functions, polynomial kernel functions and Gaussian radial basis kernel functions;t=1, 2……, m，mis the number of single kernel functions.
CN201810987345.9A 2018-08-28 2018-08-28 Intelligent treatment method and system for thermal power plant desulfurization wastewater Active CN109019717B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810987345.9A CN109019717B (en) 2018-08-28 2018-08-28 Intelligent treatment method and system for thermal power plant desulfurization wastewater

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810987345.9A CN109019717B (en) 2018-08-28 2018-08-28 Intelligent treatment method and system for thermal power plant desulfurization wastewater

Publications (2)

Publication Number Publication Date
CN109019717A CN109019717A (en) 2018-12-18
CN109019717B true CN109019717B (en) 2021-08-24

Family

ID=64625548

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810987345.9A Active CN109019717B (en) 2018-08-28 2018-08-28 Intelligent treatment method and system for thermal power plant desulfurization wastewater

Country Status (1)

Country Link
CN (1) CN109019717B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112850996B (en) * 2021-01-20 2021-12-07 福建省福能龙安热电有限公司 Desulfurization waste reduction optimizing device
CN115231653A (en) * 2022-09-22 2022-10-25 江苏晨逸京泽生物科技有限公司 Mixed treatment method and system for hemodialysis waste liquid
CN117238389B (en) * 2023-11-14 2024-01-30 江苏海峡环保科技发展有限公司 Fluorine-containing wastewater treatment system and method based on intelligent dosing

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100830518B1 (en) * 2006-12-19 2008-05-21 영남대학교 산학협력단 Wind speed estimating method of wind generation system using svr algorithm
CN105938116A (en) * 2016-06-20 2016-09-14 吉林大学 Gas sensor array concentration detection method based on fuzzy division and model integration
CN106502093A (en) * 2016-10-25 2017-03-15 大唐（北京）水务工程技术有限公司 Water island dosing On-Line Control Method and device based on GA SVR
CN107703480A (en) * 2017-08-28 2018-02-16 南京邮电大学 Mixed kernel function indoor orientation method based on machine learning

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100830518B1 (en) * 2006-12-19 2008-05-21 영남대학교 산학협력단 Wind speed estimating method of wind generation system using svr algorithm
CN105938116A (en) * 2016-06-20 2016-09-14 吉林大学 Gas sensor array concentration detection method based on fuzzy division and model integration
CN106502093A (en) * 2016-10-25 2017-03-15 大唐（北京）水务工程技术有限公司 Water island dosing On-Line Control Method and device based on GA SVR
CN107703480A (en) * 2017-08-28 2018-02-16 南京邮电大学 Mixed kernel function indoor orientation method based on machine learning

Also Published As

Publication number Publication date
CN109019717A (en) 2018-12-18

Similar Documents

Publication Publication Date Title
CN109019717B (en) Intelligent treatment method and system for thermal power plant desulfurization wastewater
US10570024B2 (en) Method for effluent total nitrogen-based on a recurrent self-organizing RBF neural network
CN101763085B (en) System and method for optimizing temperature of pesticide production waste liquid incinerator
CN104376380A (en) Ammonia concentration predicting method based on recursion self-organization neural network
CN102262147A (en) Soft measurement method and system for effluent chemical oxygen demand (COD) of waste water treatment system
Liu et al. Comprehensive evaluation of water resources security in the Yellow River basin based on a fuzzy multi-attribute decision analysis approach
CN111428914B (en) Method and device for determining water environment treatment engineering scheme and storage medium
CN101759236B (en) Regulating and controlling method of point source pollutant of industrial park
CN103454390A (en) Method and device for measuring concentration of dissolved oxygen
CN115758125A (en) Industrial sewage treatment soft measurement method based on feature structure optimization and deep learning
CN107944205B (en) Water area characteristic model establishing method based on Gaussian smoke plume model
Naser et al. Simulation of low TDS and biological units of Fajr industrial wastewater Treatment plant using artificial neural network and principal component analysis hybrid method
CN105372995A (en) Measurement and control method for sewage disposal system
CN115775042B (en) Crop heavy metal enrichment risk prediction method and system based on Bayesian theory
CN110109904B (en) Environment-friendly big data oriented water quality soft measurement method
CN103942600B (en) Sludge bulking Forecasting Methodology based on spike self-organizing radial base neural net
CN116119877A (en) Sewage automatic treatment method and system based on Internet of things technology
Chang Soft measurement modeling of turbidity in flocculation process of drinking water treatment using gaussian process regression
CN114693493A (en) IoT-based polluted river water ecological restoration system
Liu et al. Soft-sensors based on Gaussian process regression for wastewater treatment plants
CN114971164A (en) Sludge treatment equipment abnormity detection method and system based on artificial intelligence
Rahbari et al. Evaluating the process efficiency of industrial wastewater treatment plants using data envelopment analysis approach case study: Khuzestan steel company treatment plant
Prihasto et al. Fuzzy madm method for decision support system based on artificial neural network to water quality assessment in surabaya river
Wang et al. Successful prediction for coagulant dosage and effluent turbidity of a coagulation process in a drinking water treatment plant based on the Elman neural network and random forest models
Wang et al. A holistic assessment of spatio-temporal pattern and water quality in the typical basin of northeast China using multivariate statistical methods

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