CN112250161A - Self-adaptive potential controller of microbial electrolysis battery for azo wastewater treatment and control method thereof - Google Patents

Self-adaptive potential controller of microbial electrolysis battery for azo wastewater treatment and control method thereof Download PDF

Info

Publication number
CN112250161A
CN112250161A CN202010890924.9A CN202010890924A CN112250161A CN 112250161 A CN112250161 A CN 112250161A CN 202010890924 A CN202010890924 A CN 202010890924A CN 112250161 A CN112250161 A CN 112250161A
Authority
CN
China
Prior art keywords
output
neuron
hidden layer
module
azo
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
CN202010890924.9A
Other languages
Chinese (zh)
Other versions
CN112250161B (en
Inventor
远野
殷万欣
丁成
陈天明
乔椋
李朝霞
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.)
Yancheng Institute of Technology
Yancheng Institute of Technology Technology Transfer Center Co Ltd
Original Assignee
Yancheng Institute 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 Yancheng Institute of Technology filed Critical Yancheng Institute of Technology
Priority to CN202010890924.9A priority Critical patent/CN112250161B/en
Publication of CN112250161A publication Critical patent/CN112250161A/en
Application granted granted Critical
Publication of CN112250161B publication Critical patent/CN112250161B/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
    • C02F3/00—Biological treatment of water, waste water, or sewage
    • C02F3/005—Combined electrochemical biological processes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • C—CHEMISTRY; METALLURGY
    • C02—TREATMENT OF WATER, WASTE WATER, SEWAGE, OR SLUDGE
    • C02F—TREATMENT OF WATER, WASTE WATER, SEWAGE, OR SLUDGE
    • C02F2101/00—Nature of the contaminant
    • C02F2101/30—Organic compounds
    • C02F2101/308—Dyes; Colorants; Fluorescent agents
    • C—CHEMISTRY; METALLURGY
    • C02—TREATMENT OF WATER, WASTE WATER, SEWAGE, OR SLUDGE
    • C02F—TREATMENT OF WATER, WASTE WATER, SEWAGE, OR SLUDGE
    • C02F2101/00—Nature of the contaminant
    • C02F2101/30—Organic compounds
    • C02F2101/38—Organic compounds containing nitrogen

Landscapes

  • Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Molecular Biology (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Chemical & Material Sciences (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Electrochemistry (AREA)
  • Organic Chemistry (AREA)
  • Chemical Kinetics & Catalysis (AREA)
  • Biodiversity & Conservation Biology (AREA)
  • Water Supply & Treatment (AREA)
  • Environmental & Geological Engineering (AREA)
  • General Chemical & Material Sciences (AREA)
  • Hydrology & Water Resources (AREA)
  • Microbiology (AREA)
  • Water Treatment By Electricity Or Magnetism (AREA)

Abstract

The invention discloses a self-adaptive potential controller of a microbial electrolysis battery for azo wastewater treatment and a control method thereof. The artificial intelligence algorithm is adopted to carry out soft measurement on the concentration of the pollutants in the azo wastewater, so that the real-time monitoring on the concentration of the pollutants in the azo wastewater is realized, the optimal potential applied is determined and controlled according to the influence of the azo dye on the activity of the microorganisms on the electrodes, the azo dye processing capacity of the microorganisms on the electrodes can be stimulated to the maximum extent, and the processing efficiency of the azo wastewater is improved.

Description

Self-adaptive potential controller of microbial electrolysis battery for azo wastewater treatment and control method thereof
Technical Field
The invention relates to a self-adaptive potential controller of a microbial electrolysis battery for azo wastewater treatment and a control method thereof.
Background
Azo dyes are the largest chemical classification in dyes, and are widely applied to industrial production such as printing, dyeing and papermaking and the like due to the characteristics of low price, stability, various colors and the like. However, azo dyes have the characteristics of high toxicity, difficult biodegradation and high chromaticity, and if the azo dyes are directly discharged into natural water, the survival of aquatic organisms can be seriously influenced, so that the search for a proper treatment method of azo dye wastewater is particularly important.
The prior method for pretreating azo dye wastewater can be basically divided into a chemical oxidation method and a biological method. The chemical oxidation method has an advantage of high reaction speed by adding a strong oxidizing agent to wastewater to degrade azo, but cannot be widely used because of its high treatment cost and the possibility of causing secondary pollution. Biological processes are processes that convert certain refractory materials into easily degradable materials by coupling hydrolytic acidification with microorganisms. Compared with the chemical oxidation method, the method has the advantages of low treatment cost and no secondary pollution, and is widely concerned by researchers. However, the biological treatment method has low treatment efficiency, and if the tolerance and the degradation capability of microorganisms to toxic pollutants in the hydrolysis acidification process can be improved, the dependence on chemical agents can be reduced certainly, and the treatment cost of wastewater can be reduced.
Microbial Electrolysis Cells (MEC) are a method of enhancing the degradation and transformation of microorganisms by using the extracellular electron transfer of microorganisms. The method is characterized in that the transfer direction of microorganism electrons is directionally controlled by the aid of the potential, the capability of microorganisms to metabolize pollutants is stimulated, the complexity of the pollutants is reduced, the breakage of azo bonds is strengthened, and the biodegradability of the azo wastewater is greatly improved.
In the process of removing the pollutants from the azo wastewater, the azo concentration in the wastewater has fluctuation, and the constant potential applied to the cathode is obtained by analyzing the fluctuation range of the azo content in the azo wastewater, so that in the treatment process, the azo reducing capability of microorganisms under different azo concentrations is lost to a certain extent under the current constant potential, and the treatment efficiency of the azo wastewater of the MEC cannot be fully exerted.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a self-adaptive potential controller of a microbial electrolysis cell for azo wastewater treatment and a control method thereof.
In order to solve the problems of the prior art, the invention adopts the technical scheme that:
a self-adaptive potential controller of a microbial electrolysis battery for azo wastewater treatment comprises a data acquisition module, a data preprocessing module, a RBF neural network prediction model, a logic analysis module and a PWM conversion voltage output module; the data acquisition module transmits acquired information to the data preprocessing module for preprocessing, and then the information enters the RBF neural network prediction model for training; inputting the trained data into a PWM-voltage output module, wherein the positive and negative electrodes of the voltage output of the PWM-voltage output module are connected with the positive and negative electrodes of the biological electrolytic cell through electric wires; the RBF neural network module consists of an input layer, a hidden layer and an output layer; the data preprocessing module, the RBF neural network prediction model and the logic analysis module are all embedded into the raspberry pie; the data acquisition module and the PWM voltage conversion module are respectively connected with GPIO ports of the raspberry group, so that data receiving and voltage output are realized.
And trapping the RBF neural network prediction model into a raspberry pie, and judging the magnitude of output voltage according to the coupling relation among the concentration of azo wastewater pollutants, the activity of electrode microorganisms (the activity of the electrode microorganisms is reflected by the current in the current collector) and the applied potential of the working electrode.
The data acquisition module is responsible for receiving various data in the azo wastewater, preprocessing the various data, and transmitting the processed data into the RBF neural network prediction model.
The input layer in the RBF neural network prediction model is used for receiving all the parameters which are preprocessed in the data acquisition module; the hidden layer in the RBF neural network module is the position of the center of the radial basis function and the size of the width of the radial basis function; and an output layer in the RBF neural network module is used for outputting the concentration of the pollutants in the azo wastewater, and the concentration is a single value.
And the logic analysis module receives the concentration of the pollutants in the azo wastewater output by the output layer in the RBF neural network prediction module, and the current and the potential acquired by the current collector, judges the magnitude of the potential to be applied under the current condition according to the coupling relation among the concentration of the pollutants in the azo wastewater, the magnitude of the current and the applied potential, and converts the magnitude of the potential into PWM frequency to be output.
The improvement is that the data acquisition module comprises one or more of a total nitrogen sensor, a biological oxygen demand sensor, a chemical oxygen demand sensor, a pH value sensor, a current collector and an analog-to-digital converter; the total nitrogen sensor, the biological oxygen demand sensor, the chemical oxygen demand sensor, the pH value sensor and the current collector are connected with a GPIO port of the raspberry pie through an analog-to-digital converter for signal input.
As an improvement, the data preprocessing module performs normalization processing on the monitoring data in the data acquisition module, the processed data is used as an input parameter of the RBF neural network, and the formula for the normalization processing is as follows:
Figure RE-GDA0002778084510000031
as an improvement, the construction method of the RBF neural network prediction model comprises the following steps:
step 1, constructing a RBF neural network model
Step 1.1, determining an input parameter as X according to the data processed by the data preprocessing module,
X＝[x1,x2,L,xn]Tn is the number of input layer units;
step 1.2, determine the output vector Y as the output vector, i.e. the value of the predicted output voltage, Y ═ Y1,y2,L,yq]Q is the number of cells of the output layer;
step 1.3, initializing hidden layer values
Connection weight W of output layerk
Wk＝[wk1,wk2,L,wkp]T,(k＝1,2,L,q)，
Where p is the number of hidden layer units, q is the number of output layer units, k is denoted as the kth output in the output layer;
the weight initialization method from the hidden layer to the output layer comprises the following steps:
Figure RE-GDA0002778084510000032
where mink is the minimum of all expected outputs in the kth output neuron in the training set; maxk is the maximum value of the expected output in the kth output neuron in the training set, and j represents the jth unit in the hidden layer unit number;
initializing the center parameter C of neurons in the hidden layerj＝[cj1,cj2,L,cjn]T，CjiCalculation formula of initial value:
Figure RE-GDA0002778084510000033
j is 1,2, L, p, where p is the number of hidden layer neurons, j represents the jth of the number of hidden layer units, and i represents the ith of the number of hidden layer units;
the calculation method of the initialized width vector comprises the following steps:
Figure RE-GDA0002778084510000041
dfthe value of the width adjustment coefficient is 1, and i represents the ith in the number of hidden layer units; the effect is that each hidden layer neuron easily realizes the sensing ability to local information, the width vector influences the action range of the neuron to input information, and the smaller the width is, the narrower the shape of the hidden layer neuron action function is, the smaller the response of the information near the neuron center is; n is the number of groups of all output parameters;
step 2: training RBF neural network model
Step 2.1, computing the output of the jth neuron of the hidden layer
Figure RE-GDA0002778084510000042
cjIs the central vector of the jth neuron of the hidden layer, which is composed of the central components of the jth neuron of the hidden layer corresponding to all the neurons of the input layer, Cj＝[cj1,cj2,L,cjn]T，DjWidth vector for the jth neuron in the hidden layer, and CjCorresponding to, Dj＝[dj1,dj2,L,djn]T,DjThe larger the hidden layer is, the larger the influence range of the hidden layer on the input vector is, and the smoothness of the neural element is better; P.P is the Euclidean norm;
step 2.2, computing output of neurons in the output layer
Y＝[y1,y2,L,yq]T,
Figure RE-GDA0002778084510000043
Wherein wkjThe adjustment weight of the kth neuron of the output layer and the jth neural element of the hidden layer is obtained, and the k value is the kth output in the output layer;
step 2.3, iterative calculation of weight parameters;
and 2.4, calculating the root mean square error (RMS) of the RBF neural network:
Figure RE-GDA0002778084510000044
if RMS is less than or equal to ε, training is ended, otherwise go to step 2.3.
It is further improved that the training method of the weighting parameters of the Rbf neural network in step 2.3 is selected as a gradient descent method. The center, width and interceding weights are optimized by the following formula, the iterative calculation formula is as follows:
Figure RE-GDA0002778084510000051
Figure RE-GDA0002778084510000052
Figure RE-GDA0002778084510000053
wkj(t) is the adjustment weight calculated between the kth output neuron and the jth hidden layer neuron at the t-th iteration, cji(t) is the intermediate component of the jth hidden layer neuron to the ith output neuron at the time of the t iteration, dji(t) is the sum of center cji(t), η is a learning factor, E is an RBF neural network evaluation function:
Figure RE-GDA0002778084510000054
wherein O islkFor the expected output value, y, of the kth output neuron at the l-th input samplelkThe net output value of the kth output neuron at the ith input sample is obtained.
The control method of the self-adaptive potential controller of the microbial battery for azo wastewater treatment comprises the following steps:
firstly, arranging a data acquisition module in a polluted water body, wherein the data acquisition module is connected with a GPIO port of a raspberry pie through a line to realize data acquisition from the water body;
secondly, data collected from a water body are transmitted to a data preprocessing module of the raspberry pie for preprocessing, the processed data are input into an RBF neural network prediction model for training, and a proper voltage value is predicted after training;
thirdly, predicting a proper voltage value, judging a PWM signal to be output by the current GPIO according to the output voltage through a logic analysis module, and outputting the PWM signal from the GPIO;
and fourthly, transmitting the PWM signal output from the raspberry to a PWM voltage conversion module, realizing the output of voltage through the module, and connecting the positive electrode and the negative electrode of the module with the positive electrode and the negative electrode of the biological fuel cell through leads respectively to finally realize the self-adaptive control of the applied voltage of the biological electrolytic cell.
Has the advantages that:
compared with the conventional device for treating azo wastewater by applying MEC with constant potential, the self-adaptive potential controller for the microbial electrolysis cell for treating the azo wastewater and the control method thereof provided by the invention have the advantages that the concentration of the pollutants in the azo wastewater is subjected to soft measurement by adopting an artificial intelligence algorithm, the real-time monitoring on the concentration of the pollutants in the azo wastewater is realized, the applied optimal potential is determined and controlled according to the influence of the azo dye on the activity of the microorganisms on the electrode, the azo dye treatment capacity of the microorganisms on the electrode can be maximally excited, and the azo treatment efficiency is improved.
Drawings
FIG. 1 is a system framework diagram;
FIG. 2 is a flow chart of RBF training;
FIG. 3 is a diagram of a raspberry pi GPIO interface;
FIG. 4 is a logic diagram of an ADC DAC;
fig. 5 is a wiring diagram of the PWM voltage converter.
Detailed Description
As shown in FIG. 1, a self-adaptive electric potential controller for a microbial electrolysis cell for azo wastewater treatment comprises a data acquisition module, a data preprocessing module, an RBF neural network prediction model, a logic analysis module and a PWM voltage conversion output module; the data acquisition module comprises a Total Nitrogen (TN) sensor and a Biological Oxygen Demand (BOD)5) The system comprises a sensor, a Chemical Oxygen Demand (COD) sensor, a pH value (pH) sensor and a current collector, namely a data acquisition module, wherein the data acquisition module is responsible for receiving indexes of various parameters in the azo wastewater;
the data preprocessing module is used for carrying out normalization processing on the monitoring data in the data acquisition module, the processed data are used as input parameters of the RBF neural network, and the formula for the normalization processing is as follows:
Figure RE-GDA0002778084510000061
the RBF neural network prediction model is responsible for receiving all data of the data preprocessing module and predicting the concentration of azo wastewater pollutants by the trained RBF neural network;
the logic analysis module judges the optimal potential to be applied under the current condition according to the concentration index of the pollutants in the azo wastewater output by the RBF neural network prediction model and the index of the current monitor, calculates the voltage to be output according to the current potential, and outputs PWM under the corresponding voltage through a GPIO port of the raspberry group;
the PWM voltage conversion output module receives the voltage output from the raspberry pi GPIO port.
As shown in fig. 2, the construction of the RBF neural network prediction model of the pollutants in the azo wastewater includes the following steps:
step 1, constructing a RBF neural network model
Step 1.1, according to the data processed by the data preprocessing module, determining that the input parameter is X, where X is ═ X1,x2,L,xn]TN is the number of input layer units;
step 1.2, determining the output vector Y as the output vector, i.e. the predicted value of the output voltage, Y＝[y1,y2,L,yq]Q is the number of cells of the output layer;
step 1.3, initializing hidden layer values
Connection weight W of output layerk，
Wk＝[wk1,wk2,L,wkp]T,(k＝1,2,L,q),
Where p is the number of hidden layer units, q is the number of output layer units, k is denoted as the kth output in the output layer;
the weight initialization method from the hidden layer to the output layer comprises the following steps:
Figure RE-GDA0002778084510000071
where mink is the minimum of all expected outputs in the kth output neuron in the training set; maxk is the maximum value of the expected output in the kth output neuron in the training set, and j represents the jth unit in the hidden layer unit number;
initializing the center parameter C of neurons in the hidden layerj＝[cj1,cj2,L,cjn]T，CjiCalculation formula of initial value:
Figure RE-GDA0002778084510000072
j is 1,2, L, p, where p is the number of hidden layer neurons, and j represents the jth of the number of hidden layer cells;
the calculation method of the initialized width vector comprises the following steps:
Figure RE-GDA0002778084510000073
dfthe value of the width adjustment coefficient is 1, and i represents the ith in the number of input layer units; the effect is that each hidden layer neuron easily realizes the sensing ability to local information, and the width vector influences the action range of the neuron to input informationThe smaller the width, the narrower the shape of the hidden layer neuron action function, and the smaller the response of the information near the neuron center.
Step 2: training RBF neural network model
Step 2.1, calculating the output of the jth neuron of the hidden layer:
Figure RE-GDA0002778084510000081
cjis the central vector of the jth neuron of the hidden layer, which is composed of the central components of the jth neuron of the hidden layer corresponding to all the neurons of the input layer, Cj＝[cj1,cj2,L,cjn]T，DjWidth vector for the jth neuron in the hidden layer, and CjCorresponding to, Dj＝[dj1,dj2,L,djn]T,DjThe larger the hidden layer is, the larger the influence range of the hidden layer on the input vector is, and the smoothness of the neural element is better; P.P is the Euclidean norm.
Step 2.2, computing output of neurons in the output layer
Y＝[y1,y2,L,yq]T，
Figure RE-GDA0002778084510000082
Wherein wkjOutputting the k-th output in the layer for the k-th neuron of the output layer and the adjustment weight of the j-th neural element of the hidden layer;
step 2.3, iterative calculation of weight parameters;
the training method of the Rbf neural network weight parameter is selected to be a gradient descent method. The center, width and interceding weights are optimized by the following formula, the iterative calculation formula is as follows:
Figure RE-GDA0002778084510000083
Figure RE-GDA0002778084510000084
Figure RE-GDA0002778084510000085
wkj(t) is the adjustment weight calculated between the kth output neuron and the jth hidden layer neuron at the t-th iteration, cji(t) is the intermediate component of the jth hidden layer neuron to the ith output neuron at the time of the t iteration, dji(t) is the sum of center cji(t), η is a learning factor, E is an RBF neural network evaluation function:
Figure RE-GDA0002778084510000091
wherein O islkFor the expected output value, y, of the kth output neuron at the l-th input samplelkThe net output value of the kth output neuron at the ith input sample is obtained.
And 2.4, calculating the root mean square error (RMS) of the RBF neural network:
Figure RE-GDA0002778084510000092
if RMS is less than or equal to ε, training is ended, otherwise go to step 2.3.
The control method of the self-adaptive potential controller of the microbial battery for azo wastewater treatment comprises the following steps:
firstly, arranging a data acquisition module in a polluted water body, wherein the data acquisition module is connected with a GPIO port of a raspberry pie through a line to realize data acquisition from the water body;
secondly, data collected from a water body are transmitted to a data preprocessing module of the raspberry pie for preprocessing, the processed data are input into an RBF neural network prediction model for training, and a proper voltage value is predicted after training;
thirdly, predicting a proper voltage value, judging a PWM signal to be output by the current GPIO according to the output voltage through a logic analysis module, and outputting the PWM signal from the GPIO;
and fourthly, transmitting the PWM signal output from the raspberry to a PWM voltage conversion module, realizing the output of voltage through the module, and connecting the positive electrode and the negative electrode of the module with the positive electrode and the negative electrode of the biological fuel cell through leads respectively to finally realize the self-adaptive control of the applied voltage of the biological electrolytic cell.
As shown in fig. 3, each sensor and the PWM-to-voltage output module are connected to the raspberry pi through a GPIO interface of the raspberry pi;
as shown in fig. 4, the input of the ADC digital-to-analog converter is a data acquisition module, is connected to each sensor in the data acquisition module, and converts an input analog signal into a digital signal.
As shown in fig. 5, the PWM voltage output module is externally connected to an external power supply, the input terminal of the PWM voltage output module is connected to the PWM output terminal of the GPIO of the raspberry pi, the PWM input is converted into a voltage output, and the output terminal is connected to the positive and negative electrodes of the bioelectrolysis cell.
The above description is only a preferred embodiment of the present invention, and the scope of the present invention is not limited thereto, and any simple modifications or equivalent substitutions of the technical solutions that can be obviously obtained by those skilled in the art within the technical scope of the present invention are within the scope of the present invention.

Claims (6)

1. A self-adaptive electric potential controller of a microbial electrolysis battery for azo wastewater treatment is characterized by comprising a data acquisition module, a data preprocessing module, an RBF neural network prediction model, a logic analysis module and a PWM voltage conversion output module; the data acquisition module transmits acquired information to the data preprocessing module for preprocessing, and then the information enters the RBF neural network prediction model for training; inputting the trained data into a PWM-voltage output module, wherein the positive and negative electrodes of the voltage output of the PWM-voltage output module are connected with the positive and negative electrodes of the biological electrolytic cell through electric wires; the RBF neural network module consists of an input layer, a hidden layer and an output layer; the data preprocessing module, the RBF neural network prediction model and the logic analysis module are all embedded into the raspberry pie; the data acquisition module and the PWM voltage conversion module are respectively connected with GPIO ports of the raspberry group, so that data receiving and voltage output are realized.
2. The adaptive voltage controller for the microbial electrolysis cell for azo wastewater treatment according to claim 1, wherein the data acquisition module comprises one or more combinations of a total nitrogen sensor, a biological oxygen demand sensor, a chemical oxygen demand sensor, a pH value sensor, a current collector and an analog-to-digital converter; the total nitrogen sensor, the biological oxygen demand sensor, the chemical oxygen demand sensor, the pH value sensor and the current collector are connected with a GPIO port of the raspberry pie through an analog-to-digital converter for signal input.
3. The adaptive voltage controller for the microbial electrolysis cell for azo wastewater treatment according to claim 1, wherein the data preprocessing module normalizes the monitoring data in the data acquisition module, and the processed data is used as an input parameter of the RBF neural network, and the formula for the normalization process is as follows:
Figure RE-FDA0002778084500000011
4. the adaptive voltage controller for the microbial electrolysis cell for azo wastewater treatment, according to claim 1, wherein the RBF neural network prediction model is constructed by the following steps:
step 1, constructing a RBF neural network model
Step 1.1, after the data is processed by a data preprocessing moduleDetermining the input parameter as X, X ═ X1,x2,L,xn]TN is the number of input layer units;
step 1.2, determine the output vector Y as the output vector, i.e. the value of the predicted output voltage, Y ═ Y1,y2,L,yq]Q is the number of cells of the output layer;
step 1.3, initializing hidden layer values
Connection weight W of output layerk，
Wk＝[wk1,wk2,L,wkp]T,(k＝1,2,L,q)，
Where p is the number of hidden layer units, q is the number of output layer units, k is denoted as the kth output in the output layer;
the weight initialization method from the hidden layer to the output layer comprises the following steps:
Figure RE-FDA0002778084500000021
wherein min k is the minimum of all expected outputs in the kth output neuron in the training set; max k is the maximum value of the expected output in the kth output neuron in the training set, and j represents the jth unit in the hidden layer unit number;
initializing the center parameter C of neurons in the hidden layerj＝[cj1,cj2,L,cjn]T，CjiCalculation formula of initial value:
Figure RE-FDA0002778084500000022
j is 1,2, L, p, where p is the number of hidden layer neurons, j represents the jth of the number of hidden layer cells, and i represents the ith input feature in the input layer; min i is the minimum value of a certain characteristic parameter in all the groups of input samples, and max i is the maximum value of a certain characteristic parameter in all the groups of input samples;
the calculation method of the initialized width vector comprises the following steps:
Figure RE-FDA0002778084500000023
dfthe width adjustment coefficient is 1, and i is expressed as the ith input characteristic in the input; the effect is that each hidden layer neuron easily realizes the sensing ability to local information, the width vector influences the action range of the neuron to input information, and the smaller the width is, the narrower the shape of the hidden layer neuron action function is, the smaller the response of the information near the neuron center is; n is the number of groups of all output parameters;
step 2: training RBF neural network model
Step 2.1, computing the output of the jth neuron of the hidden layer
Figure RE-FDA0002778084500000031
cjIs the central vector of the jth neuron of the hidden layer, which is composed of the central components of the jth neuron of the hidden layer corresponding to all the neurons of the input layer, Cj＝[cj1,cj2,L,cjn]T，DjWidth vector for the jth neuron in the hidden layer, and CjCorresponding to, Dj＝[dj1,dj2,L,djn]T,DjThe larger the hidden layer is, the larger the influence range of the hidden layer on the input vector is, and the smoothness of the neural element is better; P.P is the Euclidean norm;
step 2.2, computing output of neurons in the output layer
Y＝[y1,y2,L,yq]T,
Figure RE-FDA0002778084500000032
Wherein wkjThe adjustment weight of the kth neuron of the output layer and the jth neural element of the hidden layer is obtained, and the k value is the kth output in the output layer;
step 2.3, iterative calculation of weight parameters;
and 2.4, calculating the root mean square error (RMS) of the RBF neural network:
Figure RE-FDA0002778084500000033
if RMS is less than or equal to ε, training is ended, otherwise go to step 2.3.
5. The adaptive voltage controller for the microbial electrolysis cell for azo wastewater treatment according to claim 4, wherein the training method of the Rbf neural network weight parameter in step 2.3 is selected to be a gradient descent method. The center, width and interceding weights are optimized by the following formula, the iterative calculation formula is as follows:
Figure RE-FDA0002778084500000034
Figure RE-FDA0002778084500000035
Figure RE-FDA0002778084500000036
wkj(t) is the tuning weight between the kth output neuron and the jth hidden layer neuron calculated at the t iteration, cji(t) is the intermediate component of the jth hidden layer neuron to the ith output neuron at the time of the t iteration, dji(t) is the sum of center cji(t), η is a learning factor, E is an RBF neural network evaluation function:
Figure RE-FDA0002778084500000041
wherein O islkFor the expected output value, y, of the kth output neuron at the l-th input samplelkThe net output value of the kth output neuron at the ith input sample is obtained.
6. The method for controlling the adaptive potential controller of the microbial electrolysis cell for azo wastewater treatment based on claim 1 is characterized by comprising the following steps:
firstly, arranging a data acquisition module in a polluted water body, wherein the data acquisition module is connected with a GPIO port of a raspberry pie through a line to realize data acquisition from the water body;
secondly, data collected from a water body are transmitted to a data preprocessing module of the raspberry pie for preprocessing, the processed data are input into an RBF neural network prediction model for training, and a proper voltage value is predicted after training;
thirdly, predicting a proper voltage value, judging a PWM signal to be output by the current GPIO according to the output voltage through a logic analysis module, and outputting the PWM signal from the GPIO;
and fourthly, transmitting the PWM signal output from the raspberry to a PWM voltage conversion module, realizing the output of voltage through the module, and connecting the positive electrode and the negative electrode of the module with the positive electrode and the negative electrode of the biological fuel cell through leads respectively to finally realize the self-adaptive control of the applied voltage of the biological electrolytic cell.
CN202010890924.9A 2020-08-29 2020-08-29 Self-adaptive potential controller of microbial electrolysis cell for azo wastewater treatment and control method thereof Active CN112250161B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010890924.9A CN112250161B (en) 2020-08-29 2020-08-29 Self-adaptive potential controller of microbial electrolysis cell for azo wastewater treatment and control method thereof

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010890924.9A CN112250161B (en) 2020-08-29 2020-08-29 Self-adaptive potential controller of microbial electrolysis cell for azo wastewater treatment and control method thereof

Publications (2)

Publication Number Publication Date
CN112250161A true CN112250161A (en) 2021-01-22
CN112250161B CN112250161B (en) 2023-05-30

Family

ID=74224149

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010890924.9A Active CN112250161B (en) 2020-08-29 2020-08-29 Self-adaptive potential controller of microbial electrolysis cell for azo wastewater treatment and control method thereof

Country Status (1)

Country Link
CN (1) CN112250161B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113860496A (en) * 2021-10-12 2021-12-31 山东建筑大学 AAO (anaerobic-anoxic-oxic) external carbon source intelligent adding system and method based on BP (Back propagation) neural network
CN117303510A (en) * 2023-09-07 2023-12-29 兰州理工大学 Self-adaptive control method for waveform characteristic analysis of wastewater treatment power supply
CN117476125A (en) * 2023-12-27 2024-01-30 豆黄金食品有限公司 Dried beancurd stick raffinate recovery data processing system based on data analysis

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103197544A (en) * 2013-02-25 2013-07-10 北京工业大学 Sewage disposal process multi-purpose control method based on nonlinear model prediction
CN104360035A (en) * 2014-11-02 2015-02-18 北京工业大学 Sewage TP soft measurement method based on self-organized particle swarm and radial basis function neural network
CN106295800A (en) * 2016-07-28 2017-01-04 北京工业大学 A kind of water outlet total nitrogen TN intelligent detecting method based on recurrence Self organizing RBF Neural Network
CN107728479A (en) * 2017-09-29 2018-02-23 北京城市排水集团有限责任公司 A kind of biological phosphate-eliminating accuracy control method based on RBF neural

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103197544A (en) * 2013-02-25 2013-07-10 北京工业大学 Sewage disposal process multi-purpose control method based on nonlinear model prediction
CN104360035A (en) * 2014-11-02 2015-02-18 北京工业大学 Sewage TP soft measurement method based on self-organized particle swarm and radial basis function neural network
CN106295800A (en) * 2016-07-28 2017-01-04 北京工业大学 A kind of water outlet total nitrogen TN intelligent detecting method based on recurrence Self organizing RBF Neural Network
CN107728479A (en) * 2017-09-29 2018-02-23 北京城市排水集团有限责任公司 A kind of biological phosphate-eliminating accuracy control method based on RBF neural

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
HABIB-ALLAH TAYEBI等: "Modeling of reactive orange 16 dye removal from aqueous media by", 《JOURNAL OF MOLECULAR STRUCTURE》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113860496A (en) * 2021-10-12 2021-12-31 山东建筑大学 AAO (anaerobic-anoxic-oxic) external carbon source intelligent adding system and method based on BP (Back propagation) neural network
CN117303510A (en) * 2023-09-07 2023-12-29 兰州理工大学 Self-adaptive control method for waveform characteristic analysis of wastewater treatment power supply
CN117476125A (en) * 2023-12-27 2024-01-30 豆黄金食品有限公司 Dried beancurd stick raffinate recovery data processing system based on data analysis
CN117476125B (en) * 2023-12-27 2024-04-05 豆黄金食品有限公司 Dried beancurd stick raffinate recovery data processing system based on data analysis

Also Published As

Publication number Publication date
CN112250161B (en) 2023-05-30

Similar Documents

Publication Publication Date Title
CN112250161B (en) Self-adaptive potential controller of microbial electrolysis cell for azo wastewater treatment and control method thereof
CN108898215B (en) Intelligent sludge bulking identification method based on two-type fuzzy neural network
KR100456413B1 (en) System and method for AI controlling waste-water treatment by neural network and back-propagation algorithm
WO2012109900A1 (en) Method and system for wastewater treatment based on dissolved oxygen control by fuzzy neural network
CN103744293B (en) Wastewater treatment monitoring method and system based on fuzzy neural network
CN114740361B (en) Fuel cell voltage prediction method based on long-short-term memory neural network model
CN111245313A (en) Brushless direct current motor PID speed regulation method based on genetic neural network
CN109978024B (en) Effluent BOD prediction method based on interconnected modular neural network
CN113448245A (en) Deep learning-based dissolved oxygen control method and system in sewage treatment process
CN112250246B (en) Intelligent azo dye wastewater detoxification and advanced treatment method
CN112250166B (en) Printing and dyeing wastewater efficient detoxification gradient treatment method based on intelligent SBR electrode process
CN111898799B (en) BFA-Elman-based power load prediction method
CN112819087A (en) Effluent BOD sensor abnormity detection method based on modular neural network
CN112836432A (en) Indoor particle suspended matter concentration prediction method based on transfer learning
CN117113250A (en) Abnormality detection method for fine chemical production process
CN110837886A (en) Effluent NH4-N soft measurement method based on ELM-SL0 neural network
CN116306803A (en) Method for predicting BOD concentration of outlet water of ILSTM (biological information collection flow) neural network based on WSFA-AFE
Deng et al. Stdp and competition learning in spiking neural networks and its application to image classification
CN114462208A (en) Effluent BOD online soft measurement method based on self-organizing RBFNN
Petrova et al. Neural network modelling of fermentation processes. Microorganisms cultivation model
CN114091351A (en) Mechanical bearing service life prediction method based on exponential smoothing and bidirectional long-and-short-term memory network
CN114149076A (en) Intelligent debugging system of anaerobic ammonia oxidation sewage treatment system
CN116822570A (en) Soft measurement method for total phosphorus concentration of effluent based on fuzzy random configuration network
Liu et al. A PSO-RBF neural network for BOD multi-step prediction in wastewater treatment process
CN115058343B (en) Selenium-enriched yeast cell wall breaking method, controller and system based on ultrasonic waves

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
TA01 Transfer of patent application right
TA01 Transfer of patent application right

Effective date of registration: 20211228

Address after: 224051 middle road of hope Avenue, Yancheng City, Jiangsu Province, No. 1

Applicant after: YANCHENG INSTITUTE OF TECHNOLOGY

Applicant after: Yancheng Institute of Technology Technology Transfer Center Co.,Ltd.

Address before: 224051 middle road of hope Avenue, Yancheng City, Jiangsu Province, No. 1

Applicant before: YANCHENG INSTITUTE OF TECHNOLOGY

GR01 Patent grant
GR01 Patent grant