CN113313230A - Intelligent membrane pollution decision-making method based on knowledge fuzzy width learning - Google Patents
Intelligent membrane pollution decision-making method based on knowledge fuzzy width learning Download PDFInfo
- Publication number
- CN113313230A CN113313230A CN202110421204.2A CN202110421204A CN113313230A CN 113313230 A CN113313230 A CN 113313230A CN 202110421204 A CN202110421204 A CN 202110421204A CN 113313230 A CN113313230 A CN 113313230A
- Authority
- CN
- China
- Prior art keywords
- layer
- fuzzy
- iteration
- network
- output
- 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
Links
- 239000012528 membrane Substances 0.000 title claims abstract description 134
- 238000000034 method Methods 0.000 title claims abstract description 54
- 210000002569 neuron Anatomy 0.000 claims abstract description 122
- 230000008569 process Effects 0.000 claims abstract description 32
- 239000010865 sewage Substances 0.000 claims abstract description 30
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 83
- 238000013528 artificial neural network Methods 0.000 claims description 71
- 230000035699 permeability Effects 0.000 claims description 35
- 239000010802 sludge Substances 0.000 claims description 30
- 238000010606 normalization Methods 0.000 claims description 27
- 238000004519 manufacturing process Methods 0.000 claims description 24
- 238000005201 scrubbing Methods 0.000 claims description 19
- 238000012549 training Methods 0.000 claims description 17
- 230000006870 function Effects 0.000 claims description 15
- 238000011084 recovery Methods 0.000 claims description 13
- 238000005273 aeration Methods 0.000 claims description 12
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 claims description 12
- 238000004140 cleaning Methods 0.000 claims description 12
- 238000009285 membrane fouling Methods 0.000 claims description 12
- 239000001301 oxygen Substances 0.000 claims description 12
- 229910052760 oxygen Inorganic materials 0.000 claims description 12
- 238000012360 testing method Methods 0.000 claims description 11
- 239000007789 gas Substances 0.000 claims description 9
- 230000002159 abnormal effect Effects 0.000 claims description 7
- 238000010992 reflux Methods 0.000 claims description 6
- 239000000126 substance Substances 0.000 claims description 6
- 238000011160 research Methods 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 claims description 3
- 230000009191 jumping Effects 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 230000000694 effects Effects 0.000 description 7
- 238000011161 development Methods 0.000 description 4
- 230000002401 inhibitory effect Effects 0.000 description 4
- 238000005265 energy consumption Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000000926 separation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/02—Computing arrangements based on specific mathematical models using fuzzy logic
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Fuzzy Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Automation & Control Theory (AREA)
- Computational Linguistics (AREA)
- Algebra (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Activated Sludge Processes (AREA)
Abstract
The invention provides a knowledge fuzzy width learning-based membrane pollution intelligent decision-making method, which aims at the problems of frequent membrane pollution events, great harm and the like in a membrane bioreactor-sewage treatment process. The invention utilizes the fuzzy width learning network to establish the membrane pollution decision model, adopts the linear fuzzy neuron to replace the characteristic layer neuron of the width learning network, and utilizes the experience knowledge to set the parameters of the fuzzy width learning network decision model, thereby solving the problem of insufficient data in the actual membrane pollution decision process, realizing the accurate decision of the membrane pollution and improving the processing capacity of the membrane pollution. Experimental results show that the method can provide accurate decision suggestions, reduce the harm caused by membrane pollution and ensure the safe, stable and efficient operation of the sewage treatment process.
Description
Technical Field
The invention constructs a membrane pollution intelligent decision model by utilizing a fuzzy width learning network based on the operation characteristics of a membrane bioreactor-sewage treatment process, provides decision suggestions for inhibiting membrane pollution, adopts a fuzzy neuron to replace a characteristic layer neuron of the width learning network, and sets parameters of the fuzzy width learning network decision model by utilizing empirical knowledge, solves the problem of insufficient data in the actual membrane pollution decision process, realizes accurate decision of membrane pollution and improves the treatment capacity of membrane pollution. The intelligent decision method is applied to the sewage treatment process, has important influence on the safe, stable and efficient operation of sewage treatment, is an important branch of the advanced manufacturing technical field, and belongs to the field of water treatment and intelligent control. Therefore, the intelligent decision of membrane pollution has important significance in the sewage treatment process.
Background
The membrane bioreactor sewage treatment process combines a membrane separation technology and a biological treatment technology, and has the advantages of good solid-liquid separation effect, low sludge load, small occupied area and the like. At present, China popularizes the technology in the field of urban sewage treatment and obtains better sewage treatment effect. However, the problem of membrane pollution becomes a bottleneck problem which restricts the safe and stable operation of the membrane bioreactor sewage treatment process, which not only increases the operation energy consumption, but also reduces the effluent quality, and even leads to the collapse of the whole sewage treatment process. Therefore, the method for deeply analyzing the membrane pollution phenomenon and researching the membrane pollution decision-making method has important research significance for ensuring the safe and stable operation of the sewage treatment process and improving the sewage treatment efficiency.
Currently, the research aiming at the membrane pollution decision has attracted the attention of experts and scholars at home and abroad, but the realization effect is not optimistic. On one hand, the membrane pollution mechanism is complex, and the reaction process is variable, so that the membrane pollution decision method based on the mechanism model can hardly meet the requirement of safe and stable operation in the sewage treatment process. On the other hand, although a decision method based on operation data obtains a certain effect, the sewage treatment process is a dynamic time-varying system, and a plurality of parameters are involved, so that the traditional membrane pollution decision model is difficult to adapt to the dynamic change of the working condition, and the effect of accurate decision cannot be realized. Therefore, by combining the operation data and expert experience knowledge, a real-time accurate membrane pollution decision-making method is designed, and the method has important theoretical significance and application value for realizing safe, stable and efficient operation of the sewage treatment process, inhibiting membrane pollution and improving effluent quality.
The invention designs a membrane pollution intelligent decision method based on knowledge fuzzy width learning, which is characterized in that a decision model based on a fuzzy width learning network is constructed, wherein a linear fuzzy neuron is adopted to replace a characteristic layer neuron of the width learning network, and parameters of the fuzzy width learning network decision model are set by using empirical knowledge, so that the problem of insufficient data in the actual membrane pollution decision process is solved, and accurate decision of membrane pollution is realized. The invention not only solves the problem that the sewage treatment plant inhibits membrane pollution, but also can improve the effluent quality and improve the safety and stability of the operation of the sewage treatment plant.
Disclosure of Invention
The invention obtains a membrane pollution intelligent decision method based on knowledge fuzzy width learning, which constructs a decision model based on a fuzzy width learning network, wherein a model fuzzy neuron is adopted to replace a characteristic layer neuron of the width learning network, and parameters of the fuzzy width learning network decision model are set by using empirical knowledge, so that the problem of insufficient data in the actual membrane pollution decision process is solved, and the accurate decision of membrane pollution is realized.
The invention adopts the following technical scheme and implementation steps:
1. a membrane pollution intelligent decision method based on knowledge fuzzy width learning is characterized in that a decision model based on a fuzzy width learning network is established to provide decision support for inhibiting membrane pollution, wherein a fuzzy neuron is adopted to replace a characteristic layer neuron of the width learning network, and parameters of the fuzzy width learning network decision model are set by using empirical knowledge, so that the problem of insufficient data in the actual membrane pollution decision process is solved, and accurate decision of membrane pollution is realized, and the method comprises the following steps:
(1) determining input and output variables of a membrane fouling decision model: one bottleneck problem of the membrane bioreactor sewage treatment process is the phenomenon of membrane pollution, which not only causes the reduction of effluent quality and the improvement of production energy consumption, but also can cause the collapse of the sewage treatment process, and seriously restricts the popularization and application of the membrane bioreactor. It is therefore necessary to provide decision support to inhibit membrane fouling. Taking a sewage treatment process of a membrane bioreactor as a research object, carrying out characteristic analysis on a sewage treatment process variable, and selecting a process variable related to membrane pollution as an input variable of an intelligent decision model: water permeability, water permeability attenuation speed, water production flow, membrane scrubbing air flow, sludge concentration, transmembrane pressure difference, water production turbidity and water permeability recovery rate; normalizing the acquired input variables to [0,1 ]; the operation suggestion is used as an output variable of the membrane pollution intelligent decision model; dividing all sample data into two groups, wherein one group comprises P training samples, the other group comprises M test samples, and P is generally required to be more than M;
(2) establishing an intelligent decision model based on a fuzzy width learning network: establishing a membrane pollution decision model by using a fuzzy width learning network, wherein the fuzzy width learning network has four layers including an input layer, a fuzzy neural network layer, an enhancement layer and an output layer; the structure is a connection mode of 8-I-J-8, the number of neuron of model input layer for determining membrane pollution is determined to be 8, the number of sub-network of fuzzy neural network layer is determined to be I, and I is 1,10]Any positive integer in between; the number of the neuron groups in the enhancement layer is J, and the J is taken as [2,20]]Any positive integer in between; the number of neurons in an output layer is 8, a fuzzy width learning network is trained by using P training samples, and the input quantity of a membrane pollution decision model is x (t) ═ x1(t),x2(t),…,xP(t)]，xn(t)＝[x1n(t),x2n(t),x3n(t),x4n(t),x5n(t),x6n(t),x7n(t),x8n(t)]For the nth sample in the t iteration after normalization, n is 1,2, …, P, x1n(t) is the water permeability, x, in the nth sample at the t iteration after normalization2n(t) is the water permeability attenuation speed x in the nth sample at the t iteration after normalization3n(t) is the water flow rate, x, in the nth sample at the t iteration after normalization4n(t) is the amount of membrane scrubbing gas in the nth sample at the tth iteration after normalization, x5n(t) is the sludge concentration in the nth sample at the t iteration after normalization, x6n(t) is the transmembrane pressure difference in the nth sample at the tth iteration after normalization, x7n(t) is the turbidity of water produced in the nth sample at the t iteration after normalization, x8n(t) is the water permeability recovery rate in the nth sample at the t iteration after normalization, and the output quantity of the membrane pollution decision model is the operation suggestion Yd(t)＝[Yd1(t),Yd2(t),…,YdP(t)]，Ydn(t)＝[Y1dn(t),Y2dn(t),Y3dn(t),Y4dn(t),Y5dn(t),Y6dn(t),Y7dn(t),Y8dn(t)]For the operation suggestion of the nth sample at the t-th iteration, Y1dn(t) is an operation suggestion 1, namely that the operation is not suitable for more than 4 hours and the water yield is reduced by 5 to 10 percent, and Y is2dn(t) is an operation suggestion 2, namely, the transmembrane pressure difference is controlled to 10-40kPa, Y3dn(t) is an operation suggestion 3, namely the dissolved oxygen concentration is controlled to be 1.5-3.5mg/L and Y is controlled to increase the aeration quantity4dn(t) is an operation suggestion 4, namely, the water quality, the water production flow, the membrane scrubbing air quantity, whether the concentration of the sludge in the membrane tank is abnormal or not are checked, and Y is5dn(t) is an operation suggestion 5, namely the sludge discharge of the membrane pool is increased, the reflux ratio of the membrane pool is increased, and the sludge concentration of the membrane pool is controlled to be 8000-12000mg/L and Y6dn(t) is an operation suggestion 6, namely adjusting the operation parameters, reducing the water yield by 5-10 percent, or increasing the aeration rate, controlling the concentration of dissolved oxygen by 2-4mg/L, and Y7dn(t) is an operating recommendation 7, namely the recent development of an on-line physical cleaning, Y8dn(t) is an operation suggestion 8, that is, for implementing online chemical cleaning within 24 hours, each layer in the decision model based on the fuzzy width learning network is represented as follows:
fuzzy width learning network input layer: this layer consists of 8 neurons, each outputting as follows:
us(t)＝xs(t),s＝1,2,...,8, (1)
wherein u iss(t) is the output value of the s-th neuron of the input layer of the fuzzy width learning network at the t-th iteration;
fuzzy neural network layer of fuzzy width learning network: the fuzzy neural network layer consists of I sub-networks, and each fuzzy neural network sub-network structure comprises four layers including an input layer, a hidden layer, a rule layer and an output layer; the structure is a connection mode of 8-L-L-V, the number of input layer neurons of each fuzzy neural network is determined to be 8, the number of hidden layer neurons is determined to be L, the number of regular layer neurons is determined to be L, and L is any positive integer between [2 and 20 ]; the number of neurons in the output layer is V, and the V is any positive integer between [2 and 10 ]; the individual layer outputs can be expressed as:
fuzzy neural network sub-network input layer: this layer consists of 8 neurons, each outputting as follows:
rs(t)＝us(t), (2)
wherein r iss(t) is the output value of the s-th neuron of the input layer of the fuzzy neural network sub-network at the t-th iteration;
fuzzy neural network sub-network hidden layer: the layer consists of L neurons, and the layer output can be expressed as:
wherein,
the output value of the ith neuron of the hidden layer of the fuzzy neural network sub-network at the t iteration, cls(t) is the center of the s-th membership function of the l-th neuron of the hidden layer at the t-th iteration; sigmals(t) is the width of the s-th membership function of the l-th neuron of the hidden layer at the t-th iteration;
fuzzy neural network sub-network rule layer: the layer consists of L neurons, and the layer output can be expressed as:
wherein f isl(t) is the output value of the ith neuron of the fuzzy neural network sub-network rule layer at the tth iteration;
fuzzy neural network sub-network output layer: the layer is composed of V neurons, and the layer output can be expressed as:
wherein o isv(t) is the output value of the v-th neuron of the output layer of the fuzzy neural network sub-network at the t-th iteration, wvl(t) is the weight between the output layer vth neuron and the rule layer lth neuron in the tth iteration;
fuzzy width learning network enhancement layer: the layer consists of J groups of neurons, and the layer output can be expressed as:
H(t)＝[h1(t),h2(t),...,hJ(t)], (6)
hj(t)＝ζj(O(t)whj(t)+βhj(t)),j＝1,2,...,J, (7)
wherein, H (t) is the output value of the fuzzy width learning network enhancement layer at the t iteration, hj(t) is the output value of the j group of neurons in the enhancement layer at the t iteration, ζj() Activation function for the jth group of neurons in the enhancement layer, o (t) ═ o1(t),o2(t),…,oI(t)]Is the output value of the fuzzy neural network layer at the t iteration, oi(t)＝[o1(t),o2(t),…,oV(t)]For the output value of ith sub-network of fuzzy neural network layer at the t iteration, I is 1,2, …, I, whj(t) and betahj(t) weight and deviation values between the fuzzy neural network layer and the jth group of neurons in the enhancement layer during the tth iteration;
fuzzy width learning network output layer: the layer consists of 8 neurons, and the layer output can be expressed as:
wherein Y (t) is the output value of the fuzzy width learning network output layer at the t iteration, wi(t) is the weight between the output layer of the fuzzy width learning network and the ith sub-network of the fuzzy neural network layer at the t iteration, wj(t) is the weight between the output layer of the fuzzy width learning network and the jth group of neurons in the enhancement layer during the tth iteration;
(3) setting an initial structure and parameters of the fuzzy width learning network intelligent decision model by using empirical knowledge: converting empirical knowledge into the form of fuzzy rules:
wherein,
and
as the water permeability x in rule q1nThe lower and upper bound values of (a),
and
for a regular q middle membrane scrubbing gas volume x4nThe lower and upper bound values of (a),
and
for the water permeability recovery x in rule q8nThe lower and upper bound values of (a),
for the output value of the v-th neuron of the output layer of the fuzzy neural network sub-network in the rule Q, Q is 1, …, and Q is the number of rules. Setting the number L of neurons of a hidden layer and a regular layer of each fuzzy neural network layer sub-network to be Q;
setting initial parameters of the fuzzy width learning network intelligent decision model by using formula (9):
wherein, cqsn(t) is the center of the s-th membership function of the q-th neuron of the hidden layer in the nth sample at the t-th iteration,
and
the lower bound and the upper bound of the input layer s neuron corresponding to the q neuron of the hidden layer during the t iteration; sigmaqsn(t) is the width of the s-th membership function of the q-th neuron of the hidden layer in the nth sample at the tth iteration, rho is a normal number, and the value is randomly selected in an interval (0, 1);
(4) training a decision model based on a fuzzy width learning network, specifically:
setting the current iteration time t as 1, and setting the maximum iteration time K of a decision model based on a fuzzy width learning network as 30;
giving the number of initial fuzzy neural network layer sub-networks of the fuzzy width learning network as I, the number of enhancement layer neuron groups as J, the number of hidden layer and regular layer neurons of each fuzzy neural network layer sub-network as Q, and the number of output layer neurons as V;
thirdly, calculating the output Y (t) of the fuzzy width learning network according to the formulas (1) to (10), wherein the accuracy of the output of the fuzzy width learning network is represented as:
A(t)＝TF(t)/P, (12)
where a (t) is the output accuracy of the fuzzy width learning network at the t-th iteration, and tf (t) is the number of samples that the fuzzy width learning network outputs the same as the actual value at the t-th iteration. And (3) adjusting parameters of the two-type fuzzy width learning network by applying a pseudo-inverse algorithm:
W(t)＝D(t)+Yd(t), (13)
wherein W (t) [ < W >i(t)，Wj(t)]For the weight, W, between the output layer of the fuzzy width learning network and the fuzzy neural network layer and the enhancement layer at the t-th iterationi(t) is the weight between the fuzzy width learning network output layer and the fuzzy neural network layer at the t-th iteration, Wj(t) is the weight between the fuzzy width learning network output layer and the enhancement layer at the tth iteration; d (t) ([ O (t), H (t))]λ is a normal number in the interval (0,1)]Taking a value at random, and E is an identity matrix;
fourthly, if t is not more than K or A (t) is more than 0.05, returning to the third step; if t is more than K and A (t) is less than or equal to 0.05, stopping calculating and jumping out of the cycle to finish training;
(5) membrane fouling intelligent decision making
Learning a network membrane pollution decision model by using the trained fuzzy width, and obtaining a membrane pollution class output value of the model by using the water permeability, the water permeability attenuation speed, the water production flow, the membrane scrubbing air quantity, the sludge concentration, the transmembrane pressure difference, the water production turbidity and the water permeability recovery rate of M test samples as input variables of the model, namely an operation suggestion corresponding to the class to which the sample to be tested belongs, wherein the operation suggestion comprises the following steps: operation advice 1: the operation is not suitable for more than 4 hours, the water yield is reduced by 5-10%, and the operation suggestion is 2: reducing the water yield of the membrane pool, controlling the transmembrane pressure difference to 10-40kPa, and recommending 3: increasing aeration quantity, controlling the dissolved oxygen concentration to be 1.5-3.5mg/L, and recommending 4: and (3) checking whether the water quality, the water production flow, the membrane scrubbing gas quantity and the membrane pool sludge concentration are abnormal or not, and recommending the operation 5: increasing the sludge discharge amount of the membrane pool, increasing the reflux ratio of the membrane pool, controlling the sludge concentration of the membrane pool to be 8000-one 12000mg/L, and recommending the operation to be 6: adjusting operation parameters, reducing the water yield by 5-10%, or increasing the aeration rate, controlling the dissolved oxygen concentration to be 2-4mg/L, and recommending the operation to be 7: recent development of online physical cleaning, operation recommendation 8: and the online chemical cleaning is carried out within 24 hours, and decision support is provided for the production process.
The invention is mainly characterized in that:
(1) the invention provides a membrane pollution intelligent decision-making method based on knowledge fuzzy width learning, aiming at the problem of low precision of membrane pollution types obtained by means of operation data in the current membrane bioreactor sewage treatment, and solving the problem of accurate decision-making of membrane pollution;
(2) aiming at the problem of insufficient robustness of the traditional width learning network, the invention provides a fuzzy width learning network structure, and a characteristic layer neuron of the width learning network is replaced by a type of fuzzy neuron so as to improve the robustness of the network;
(3) aiming at the problem of initial setting of the structure and parameters of the fuzzy width learning network, the invention sets the parameters of the fuzzy width learning network decision model by using empirical knowledge, so that the constructed model has proper precision and network structure.
Drawings
FIG. 1 is a fuzzy width learning network architecture topology of the present invention;
FIG. 2 is a graph of the membrane fouling decision training effect of the present invention, wherein the red cross is the actual membrane fouling species and the blue circle is the output value of the knowledge-based fuzzy width learning network decision model;
fig. 3 is a diagram of the membrane fouling decision test effect of the present invention, wherein the red cross is the actual membrane fouling species, and the blue circle is the output value of the knowledge-based fuzzy width learning network decision model.
Detailed Description
Experimental data were from data of 2020 at a sewage treatment plant: taking actual detection data of water permeability, water permeability attenuation speed, water production flow, membrane scrubbing air flow, sludge concentration, transmembrane pressure difference, water production turbidity and water permeability recovery rate as process sample data, and removing 100 groups of usable data after abnormal experimental samples are removed, wherein 60 groups of usable data are used as training samples, and the rest 40 groups of usable data are used as test samples.
The invention adopts the following technical scheme and implementation steps:
1. a membrane pollution intelligent decision method based on knowledge fuzzy width learning is characterized in that a decision model based on a fuzzy width learning network is established to provide decision support for inhibiting membrane pollution, wherein a fuzzy neuron is adopted to replace a characteristic layer neuron of the width learning network, and parameters of the fuzzy width learning network decision model are set by using empirical knowledge, so that the problem of insufficient data in the actual membrane pollution decision process is solved, and accurate decision of membrane pollution is realized, and the method comprises the following steps:
(1) determining input and output variables of a membrane fouling decision model: one bottleneck problem of the membrane bioreactor sewage treatment process is the phenomenon of membrane pollution, which not only causes the reduction of effluent quality and the improvement of production energy consumption, but also can cause the collapse of the sewage treatment process, and seriously restricts the popularization and application of the membrane bioreactor. It is therefore necessary to provide decision support to inhibit membrane fouling. Taking a sewage treatment process of a membrane bioreactor as a research object, carrying out characteristic analysis on a sewage treatment process variable, and selecting a process variable related to membrane pollution as an input variable of an intelligent decision model: water permeability, water permeability attenuation speed, water production flow, membrane scrubbing air flow, sludge concentration, transmembrane pressure difference, water production turbidity and water permeability recovery rate; normalizing the acquired input variables to [0,1 ]; the operation suggestion is used as an output variable of the membrane pollution intelligent decision model; dividing all sample data into two groups, wherein one group comprises 60 training samples, and the other group comprises 40 test samples;
(2) establishing an intelligent decision model based on a fuzzy width learning network: establishing a membrane pollution decision model by using a fuzzy width learning network, wherein the fuzzy width learning network has four layers including an input layer, a fuzzy neural network layer, an enhancement layer and an output layer; the structure is a connection mode of 8-I-J-8, the number of neuron of model input layer for determining membrane pollution is determined to be 8, the number of sub-network of fuzzy neural network layer is determined to be I, and I is 1,10]Any positive integer in between; the number of the neuron groups in the enhancement layer is J, and the J is taken as [2,20]]Any positive integer in between; the number of neurons in an output layer is 8, 60 training samples are used for training a fuzzy width learning network, and the input quantity of a membrane pollution decision model is x (t) ═ x1(t),x2(t),…,x60(t)]，xn(t)＝[x1n(t),x2n(t),x3n(t),x4n(t),x5n(t),x6n(t),x7n(t),x8n(t)]For the nth sample in the t iteration after normalization, n is 1,2, …,60, x1n(t) is the water permeability, x, in the nth sample at the t iteration after normalization2n(t) is the water permeability attenuation speed x in the nth sample at the t iteration after normalization3n(t) is the water flow rate, x, in the nth sample at the t iteration after normalization4n(t) is the amount of membrane scrubbing gas in the nth sample at the tth iteration after normalization, x5n(t) is the sludge concentration in the nth sample at the t iteration after normalization, x6n(t) is the transmembrane pressure difference in the nth sample at the tth iteration after normalization, x7n(t) is the turbidity of water produced in the nth sample at the t iteration after normalization, x8n(t) is the water permeability recovery rate in the nth sample at the t iteration after normalization, and the output quantity of the membrane pollution decision model is the operation suggestion Yd(t)＝[Yd1(t),Yd2(t),…,Yd60(t)]，Ydn(t)＝[Y1dn(t),Y2dn(t),Y3dn(t),Y4dn(t),Y5dn(t),Y6dn(t),Y7dn(t),Y8dn(t)]For the operation suggestion of the nth sample at the t-th iteration, Y1dn(t) is an operation suggestion 1, namely that the operation is not suitable for more than 4 hours and the water yield is reduced by 5 to 10 percent, and Y is2dn(t) is an operation suggestion 2, namely, the transmembrane pressure difference is controlled to 10-40kPa, Y3dn(t) is an operation suggestion 3, namely the dissolved oxygen concentration is controlled to be 1.5-3.5mg/L and Y is controlled to increase the aeration quantity4dn(t) is an operation suggestion 4, namely, the water quality, the water production flow, the membrane scrubbing air quantity, whether the concentration of the sludge in the membrane tank is abnormal or not are checked, and Y is5dn(t) is an operation suggestion 5, namely the sludge discharge of the membrane pool is increased, the reflux ratio of the membrane pool is increased, and the sludge concentration of the membrane pool is controlled to be 8000-12000mg/L and Y6dn(t) is an operation suggestion 6, namely adjusting the operation parameters, reducing the water yield by 5-10 percent, or increasing the aeration rate, controlling the concentration of dissolved oxygen by 2-4mg/L, and Y7dn(t) is an operating recommendation 7, namely the recent development of an on-line physical cleaning, Y8dn(t) is the operating recommendation 8, i.e. on-line chemical cleaning is performed in 24 hours, based on blurThe layers in the decision model of the breadth learning network are represented as follows:
fuzzy width learning network input layer: this layer consists of 8 neurons, each outputting as follows:
us(t)＝xs(t),s＝1,2,...,8, (1)
wherein u iss(t) is the output value of the s-th neuron of the input layer of the fuzzy width learning network at the t-th iteration;
fuzzy neural network layer of fuzzy width learning network: the fuzzy neural network layer consists of I sub-networks, and each fuzzy neural network sub-network structure comprises four layers including an input layer, a hidden layer, a rule layer and an output layer; the structure is a connection mode of 8-L-L-V, the number of input layer neurons of each fuzzy neural network is determined to be 8, the number of hidden layer neurons is determined to be L, the number of regular layer neurons is determined to be L, and L is any positive integer between [2 and 20 ]; the number of neurons in the output layer is V, and the V is any positive integer between [2 and 10 ]; the individual layer outputs can be expressed as:
fuzzy neural network sub-network input layer: this layer consists of 8 neurons, each outputting as follows:
rs(t)＝us(t), (2)
wherein r iss(t) is the output value of the s-th neuron of the input layer of the fuzzy neural network sub-network at the t-th iteration;
fuzzy neural network sub-network hidden layer: the layer consists of L neurons, and the layer output can be expressed as:
wherein,
the output value of the ith neuron of the hidden layer of the fuzzy neural network sub-network at the t iteration, cls(t) is the s-th membership function of the l-th neuron of the hidden layer at the t-th iterationThe center of (a); sigmals(t) is the width of the s-th membership function of the l-th neuron of the hidden layer at the t-th iteration;
fuzzy neural network sub-network rule layer: the layer consists of L neurons, and the layer output can be expressed as:
wherein f isl(t) is the output value of the ith neuron of the fuzzy neural network sub-network rule layer at the tth iteration;
fuzzy neural network sub-network output layer: the layer is composed of V neurons, and the layer output can be expressed as:
wherein o isv(t) is the output value of the v-th neuron of the output layer of the fuzzy neural network sub-network at the t-th iteration, wvl(t) is the weight between the output layer vth neuron and the rule layer lth neuron in the tth iteration;
fuzzy width learning network enhancement layer: the layer consists of J groups of neurons, and the layer output can be expressed as:
H(t)＝[h1(t),h2(t),...,hJ(t)], (6)
hj(t)＝ζj(O(t)whj(t)+βhj(t)),j＝1,2,...,J, (7)
wherein, H (t) is the output value of the fuzzy width learning network enhancement layer at the t iteration, hj(t) is the output value of the j group of neurons in the enhancement layer at the t iteration, ζj() Activation function for the jth group of neurons in the enhancement layer, o (t) ═ o1(t),o2(t),…,oI(t)]Is the output value of the fuzzy neural network layer at the t iteration, oi(t)＝[o1(t),o2(t),…,oV(t)]For the fuzzy spirit at the t-th iterationOutput value via I-th sub-network of network layer, I-1, 2, …, I, whj(t) and betahj(t) weight and deviation values between the fuzzy neural network layer and the jth group of neurons in the enhancement layer during the tth iteration;
fuzzy width learning network output layer: the layer consists of 8 neurons, and the layer output can be expressed as:
wherein Y (t) is the output value of the fuzzy width learning network output layer at the t iteration, wi(t) is the weight between the output layer of the fuzzy width learning network and the ith sub-network of the fuzzy neural network layer at the t iteration, wj(t) is the weight between the output layer of the fuzzy width learning network and the jth group of neurons in the enhancement layer during the tth iteration;
(3) setting an initial structure and parameters of the fuzzy width learning network intelligent decision model by using empirical knowledge: converting empirical knowledge into the form of fuzzy rules:
wherein,
and
as the water permeability x in rule q1nThe lower and upper bound values of (a),
and
for a regular q middle membrane scrubbing gas volume x4nThe lower and upper bound values of (a),
and
for the water permeability recovery x in rule q8nThe lower and upper bound values of (a),
for the output value of the v-th neuron of the output layer of the fuzzy neural network sub-network in the rule Q, Q is 1, …, and Q is the number of rules. Setting the number L of neurons of a hidden layer and a regular layer of each fuzzy neural network layer sub-network to be Q;
setting initial parameters of the fuzzy width learning network intelligent decision model by using formula (9):
wherein, cqsn(t) is the center of the s-th membership function of the q-th neuron of the hidden layer in the nth sample at the t-th iteration,
and
the lower bound and the upper bound of the input layer s neuron corresponding to the q neuron of the hidden layer during the t iteration; sigmaqsn(t) is the width of the s-th membership function of the q-th neuron of the hidden layer in the nth sample at the tth iteration, rho is a normal number, and the value is randomly selected in an interval (0, 1);
(4) training a decision model based on a fuzzy width learning network, specifically:
setting the current iteration time t as 1, and setting the maximum iteration time K of a decision model based on a fuzzy width learning network as 30;
giving the number of initial fuzzy neural network layer sub-networks of the fuzzy width learning network as I, the number of enhancement layer neuron groups as J, the number of hidden layer and regular layer neurons of each fuzzy neural network layer sub-network as Q, and the number of output layer neurons as V;
thirdly, calculating the output Y (t) of the fuzzy width learning network according to the formulas (1) to (10), wherein the accuracy of the output of the fuzzy width learning network is represented as:
A(t)＝TF(t)/P, (12)
where a (t) is the output accuracy of the fuzzy width learning network at the t-th iteration, and tf (t) is the number of samples that the fuzzy width learning network outputs the same as the actual value at the t-th iteration. And (3) adjusting parameters of the two-type fuzzy width learning network by applying a pseudo-inverse algorithm:
W(t)＝D(t)+Yd(t), (13)
wherein W (t) [ < W >i(t)，Wj(t)]For the weight, W, between the output layer of the fuzzy width learning network and the fuzzy neural network layer and the enhancement layer at the t-th iterationi(t) is the weight between the fuzzy width learning network output layer and the fuzzy neural network layer at the t-th iteration, Wj(t) is the weight between the fuzzy width learning network output layer and the enhancement layer at the tth iteration; d (t) ([ O (t), H (t))]λ is a normal number in the interval (0,1)]Taking a value at random, and E is an identity matrix;
fourthly, if t is not more than K or A (t) is more than 0.05, returning to the third step; if t is more than K and A (t) is less than or equal to 0.05, stopping calculating and jumping out of the cycle to finish training;
(5) membrane fouling intelligent decision making
Learning a network membrane pollution decision model by using the trained fuzzy width to obtain the optimal network parameters and structure of the model, as shown in FIG. 1; the intelligent decision-making method for membrane pollution is shown in fig. 2, wherein the X axis: number of training samples, Y-axis: training output values of membrane pollution types, wherein the red cross is an actual membrane pollution type, and the blue circle is an output value of a knowledge fuzzy width-based learning network decision model;
by utilizing a trained membrane pollution decision-making model, the water permeability attenuation speed, the water production flow, the membrane scrubbing air flow, the sludge concentration, the transmembrane pressure difference, the water production turbidity and the water permeability recovery rate of 40 test samples are used as input variables of the model to obtain the membrane pollution class output value of the model, namely the operation suggestion corresponding to the class to which the sample to be tested belongs, and the operation suggestion comprises the following steps: operation advice 1: the operation is not suitable for more than 4 hours, the water yield is reduced by 5-10%, and the operation suggestion is 2: reducing the water yield of the membrane pool, controlling the transmembrane pressure difference to 10-40kPa, and recommending 3: increasing aeration quantity, controlling the dissolved oxygen concentration to be 1.5-3.5mg/L, and recommending 4: and (3) checking whether the water quality, the water production flow, the membrane scrubbing gas quantity and the membrane pool sludge concentration are abnormal or not, and recommending the operation 5: increasing the sludge discharge amount of the membrane pool, increasing the reflux ratio of the membrane pool, controlling the sludge concentration of the membrane pool to be 8000-one 12000mg/L, and recommending the operation to be 6: adjusting operation parameters, reducing the water yield by 5-10%, or increasing the aeration rate, controlling the dissolved oxygen concentration to be 2-4mg/L, and recommending the operation to be 7: recent development of online physical cleaning, operation recommendation 8: implementing on-line chemical cleaning within 24 hours; the test results of the membrane pollution intelligent decision method are shown in fig. 3, wherein the X axis: number of test samples, Y-axis: the membrane pollution type test output value, the red cross is the actual membrane pollution type, and the blue circle is the output value of the knowledge fuzzy width-based learning network decision model; the experimental result shows the effectiveness of the learning network decision method based on the knowledge fuzzy width.
Claims (1)
1. A membrane pollution intelligent decision method based on knowledge fuzzy width learning is characterized by comprising the following steps:
(1) determining input and output variables of a membrane fouling decision model: taking a sewage treatment process of a membrane bioreactor as a research object, carrying out characteristic analysis on a sewage treatment process variable, and selecting a process variable related to membrane pollution as an input variable of an intelligent decision model: water permeability, water permeability attenuation speed, water production flow, membrane scrubbing air flow, sludge concentration, transmembrane pressure difference, water production turbidity and water permeability recovery rate; normalizing the acquired input variables to [0,1 ]; the operation suggestion is used as an output variable of the membrane pollution intelligent decision model; dividing all sample data into two groups, wherein one group comprises P training samples, and the other group comprises M test samples, and P is required to be more than M;
(2) establishing an intelligent decision model based on a fuzzy width learning network: establishing a sludge bulking decision model by using a fuzzy width learning network, wherein the fuzzy width learning network has four layers including an input layer, a fuzzy neural network layer, an enhancement layer and an output layer; the structure is a connection mode of 8-I-J-8, the number of neuron of model input layer for determining membrane pollution is determined to be 8, the number of sub-network of fuzzy neural network layer is determined to be I, and I is 1,10]Any positive integer in between; the number of the neuron groups in the enhancement layer is J, and the J is taken as [2,20]]Any positive integer in between; the number of neurons in an output layer is 8, a fuzzy width learning network is trained by using P training samples, and the input quantity of a membrane pollution decision model is x (t) ═ x1(t),x2(t),…,xP(t)]，xn(t)＝[x1n(t),x2n(t),x3n(t),x4n(t),x5n(t),x6n(t),x7n(t),x8n(t)]For the nth sample in the t iteration after normalization, n is 1,2, …, P, x1n(t) is the water permeability, x, in the nth sample at the t iteration after normalization2n(t) is the water permeability attenuation speed x in the nth sample at the t iteration after normalization3n(t) is the water flow rate, x, in the nth sample at the t iteration after normalization4n(t) is the amount of membrane scrubbing gas in the nth sample at the tth iteration after normalization, x5n(t) is the sludge concentration in the nth sample at the t iteration after normalization, x6n(t) is the transmembrane pressure difference in the nth sample at the tth iteration after normalization, x7n(t) is the turbidity of water produced in the nth sample at the t iteration after normalization, x8n(t) is the water permeability recovery rate in the nth sample at the t iteration after normalization, and the output quantity of the membrane pollution decision model is the operation suggestion Yd(t)＝[Yd1(t),Yd2(t),…,YdP(t)]，Ydn(t)＝[Y1dn(t),Y2dn(t),Y3dn(t),Y4dn(t),Y5dn(t),Y6dn(t),Y7dn(t),Y8dn(t)]For the operation suggestion of the nth sample at the t-th iteration, Y1dn(t) is an operation suggestion 1, namely that the operation is not suitable for more than 4 hours and the water yield is reduced by 5 to 10 percent, and Y is2dn(t) is an operation suggestion 2, namely, the transmembrane pressure difference is controlled to 10-40kPa, Y3dn(t) is an operation suggestion 3, namely the dissolved oxygen concentration is controlled to be 1.5-3.5mg/L and Y is controlled to increase the aeration quantity4dn(t) is an operation suggestion 4, namely, the water quality, the water production flow, the membrane scrubbing air quantity, whether the concentration of the sludge in the membrane tank is abnormal or not are checked, and Y is5dn(t) is an operation suggestion 5, namely the sludge discharge of the membrane pool is increased, the reflux ratio of the membrane pool is increased, and the sludge concentration of the membrane pool is controlled to be 8000-12000mg/L and Y6dn(t) is an operation suggestion 6, namely adjusting the operation parameters, reducing the water yield by 5-10 percent, or increasing the aeration rate, controlling the concentration of dissolved oxygen by 2-4mg/L, and Y7dn(t) is an operating recommendation 7, i.e. to perform an on-line physical cleaning, Y8dn(t) is an operation suggestion 8, that is, for implementing online chemical cleaning within 24 hours, each layer in the decision model based on the fuzzy width learning network is represented as follows:
fuzzy width learning network input layer: this layer consists of 8 neurons, each outputting as follows:
us(t)＝xs(t),s＝1,2,...,8, (1)
wherein u iss(t) is the output value of the s-th neuron of the input layer of the fuzzy width learning network at the t-th iteration;
fuzzy neural network layer of fuzzy width learning network: the fuzzy neural network layer consists of I sub-networks, and each fuzzy neural network sub-network structure comprises four layers including an input layer, a hidden layer, a rule layer and an output layer; the structure is a connection mode of 8-L-L-V, the number of input layer neurons of each fuzzy neural network is determined to be 8, the number of hidden layer neurons is determined to be L, the number of regular layer neurons is determined to be L, and L is any positive integer between [2 and 20 ]; the number of neurons in the output layer is V, and the V is any positive integer between [2 and 10 ]; the individual layer outputs can be expressed as:
fuzzy neural network sub-network input layer: this layer consists of 8 neurons, each outputting as follows:
rs(t)＝us(t), (2)
wherein r iss(t) is the output value of the s-th neuron of the input layer of the fuzzy neural network sub-network at the t-th iteration;
fuzzy neural network sub-network hidden layer: the layer consists of L neurons, and the layer output can be expressed as:
wherein,
the output value of the ith neuron of the hidden layer of the fuzzy neural network sub-network at the t iteration, cls(t) is the center of the s-th membership function of the l-th neuron of the hidden layer at the t-th iteration; sigmals(t) is the width of the s-th membership function of the l-th neuron of the hidden layer at the t-th iteration;
fuzzy neural network sub-network rule layer: the layer consists of L neurons, and the layer output can be expressed as:
wherein f isl(t) is the output value of the ith neuron of the fuzzy neural network sub-network rule layer at the tth iteration;
fuzzy neural network sub-network output layer: the layer is composed of V neurons, and the layer output can be expressed as:
wherein o isv(t) fuzzy neural network at the t-th iterationOutput value, w, of the vth neuron of the subnetwork output layervl(t) is the weight between the output layer vth neuron and the rule layer lth neuron in the tth iteration;
fuzzy width learning network enhancement layer: the layer consists of J groups of neurons, and the layer output can be expressed as:
H(t)＝[h1(t),h2(t),...,hJ(t)], (6)
hj(t)＝ζj(O(t)whj(t)+βhj(t)),j＝1,2,...,J, (7)
wherein, H (t) is the output value of the fuzzy width learning network enhancement layer at the t iteration, hj(t) is the output value of the j group of neurons in the enhancement layer at the t iteration, ζj() Activation function for the jth group of neurons in the enhancement layer, o (t) ═ o1(t),o2(t),…,oI(t)]Is the output value of the fuzzy neural network layer at the t iteration, oi(t)＝[o1(t),o2(t),…,oV(t)]For the output value of ith sub-network of fuzzy neural network layer at the t iteration, I is 1,2, …, I, whj(t) and betahj(t) weight and deviation values between the fuzzy neural network layer and the jth group of neurons in the enhancement layer during the tth iteration;
fuzzy width learning network output layer: the layer consists of 8 neurons, and the layer output can be expressed as:
wherein Y (t) is the output value of the fuzzy width learning network output layer at the t iteration, wi(t) is the weight between the output layer of the fuzzy width learning network and the ith sub-network of the fuzzy neural network layer at the t iteration, wj(t) is the weight between the output layer of the fuzzy width learning network and the jth group of neurons in the enhancement layer during the tth iteration;
(3) setting an initial structure and parameters of the fuzzy width learning network intelligent decision model by using empirical knowledge: converting empirical knowledge into the form of fuzzy rules:
wherein,
and
as the water permeability x in rule q1nThe lower and upper bound values of (a),
and
for a regular q middle membrane scrubbing gas volume x4nThe lower and upper bound values of (a),
and
for the water permeability recovery x in rule q8nThe lower and upper bound values of (a),
the output value of the v-th neuron of the output layer of the fuzzy neural network sub-network in the rule Q is 1, …, and Q is the number of the rules; setting the number L of neurons of a hidden layer and a regular layer of each fuzzy neural network layer sub-network to be Q;
setting initial parameters of the fuzzy width learning network intelligent decision model by using formula (9):
wherein, cqsn(t) is the center of the s-th membership function of the q-th neuron of the hidden layer in the nth sample at the t-th iteration,
and
the lower bound and the upper bound of the input layer s neuron corresponding to the q neuron of the hidden layer during the t iteration; sigmaqsn(t) is the width of the s-th membership function of the q-th neuron of the hidden layer in the nth sample at the tth iteration, rho is a normal number, and the value is randomly selected in an interval (0, 1);
(4) training a decision model based on a fuzzy width learning network, specifically:
setting the current iteration time t as 1, and setting the maximum iteration time K of a decision model based on a fuzzy width learning network as 30;
giving the number of initial fuzzy neural network layer sub-networks of the fuzzy width learning network as I, the number of enhancement layer neuron groups as J, the number of hidden layer and regular layer neurons of each fuzzy neural network layer sub-network as Q, and the number of output layer neurons as V;
thirdly, calculating the output Y (t) of the fuzzy width learning network according to the formulas (1) to (10), wherein the accuracy of the output of the fuzzy width learning network is represented as:
A(t)＝TF(t)/P, (12)
wherein, a (t) is the output accuracy of the fuzzy width learning network at the t iteration, and tf (t) is the number of samples which are output by the fuzzy width learning network at the t iteration and are the same as the actual value; and (3) adjusting parameters of the two-type fuzzy width learning network by applying a pseudo-inverse algorithm:
W(t)＝D(t)+Yd(t), (13)
wherein W (t) [ < W >i(t)，Wj(t)]For the weight, W, between the output layer of the fuzzy width learning network and the fuzzy neural network layer and the enhancement layer at the t-th iterationi(t) is the weight between the fuzzy width learning network output layer and the fuzzy neural network layer at the t-th iteration, Wj(t) is the weight between the fuzzy width learning network output layer and the enhancement layer at the tth iteration; d (t) ([ O (t), H (t))]λ is a normal number in the interval (0,1)]Taking a value at random, and E is an identity matrix;
fourthly, if t is not more than K or A (t) is more than 0.05, returning to the third step; if t is more than K and A (t) is less than or equal to 0.05, stopping calculating and jumping out of the cycle to finish training;
(5) membrane fouling intelligent decision making
Learning a network membrane pollution decision model by using the trained fuzzy width, and obtaining a membrane pollution class output value of the model by using the water permeability, the water permeability attenuation speed, the water production flow, the membrane scrubbing air quantity, the sludge concentration, the transmembrane pressure difference, the water production turbidity and the water permeability recovery rate of M test samples as input variables of the model, namely an operation suggestion corresponding to the class to which the sample to be tested belongs, wherein the operation suggestion comprises the following steps: operation advice 1: the operation is not suitable for more than 4 hours, the water yield is reduced by 5-10%, and the operation suggestion is 2: reducing the water yield of the membrane pool, controlling the transmembrane pressure difference to 10-40kPa, and recommending 3: increasing aeration quantity, controlling the dissolved oxygen concentration to be 1.5-3.5mg/L, and recommending 4: and (3) checking whether the water quality, the water production flow, the membrane scrubbing gas quantity and the membrane pool sludge concentration are abnormal or not, and recommending the operation 5: increasing the sludge discharge amount of the membrane pool, increasing the reflux ratio of the membrane pool, controlling the sludge concentration of the membrane pool to be 8000-one 12000mg/L, and recommending the operation to be 6: adjusting operation parameters, reducing the water yield by 5-10%, or increasing the aeration rate, controlling the dissolved oxygen concentration to be 2-4mg/L, and recommending the operation to be 7: developing on-line physical cleaning, and operation suggestion 8: the on-line chemical cleaning was performed within 24 hours.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110421204.2A CN113313230B (en) | 2021-04-19 | 2021-04-19 | Film pollution intelligent decision-making method based on knowledge fuzzy width learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110421204.2A CN113313230B (en) | 2021-04-19 | 2021-04-19 | Film pollution intelligent decision-making method based on knowledge fuzzy width learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113313230A true CN113313230A (en) | 2021-08-27 |
| CN113313230B CN113313230B (en) | 2024-04-16 |
Family
ID=77372547
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110421204.2A Active CN113313230B (en) | 2021-04-19 | 2021-04-19 | Film pollution intelligent decision-making method based on knowledge fuzzy width learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113313230B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110071969A1 (en) * | 2008-05-23 | 2011-03-24 | Faiyaz Doctor | Neuro type-2 fuzzy based method for decision making |
| CN106096730A (en) * | 2016-06-09 | 2016-11-09 | 北京工业大学 | A kind of intelligent detecting method of MBR film permeability rate based on Recurrent RBF Neural Networks |
| CN108375534A (en) * | 2018-02-06 | 2018-08-07 | 北京工业大学 | MBR fouling membrane intelligent early-warning methods |
| CN109133351A (en) * | 2018-08-29 | 2019-01-04 | 北京工业大学 | Membrane bioreactor-MBR fouling membrane intelligent early-warning method |
| CN112101402A (en) * | 2020-07-22 | 2020-12-18 | 北京工业大学 | Membrane pollution early warning method based on knowledge fuzzy learning |
| CN112183719A (en) * | 2020-09-15 | 2021-01-05 | 北京工业大学 | Intelligent detection method for total nitrogen in effluent based on multi-objective optimization-fuzzy neural network |
-
2021
- 2021-04-19 CN CN202110421204.2A patent/CN113313230B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110071969A1 (en) * | 2008-05-23 | 2011-03-24 | Faiyaz Doctor | Neuro type-2 fuzzy based method for decision making |
| CN106096730A (en) * | 2016-06-09 | 2016-11-09 | 北京工业大学 | A kind of intelligent detecting method of MBR film permeability rate based on Recurrent RBF Neural Networks |
| CN108375534A (en) * | 2018-02-06 | 2018-08-07 | 北京工业大学 | MBR fouling membrane intelligent early-warning methods |
| CN109133351A (en) * | 2018-08-29 | 2019-01-04 | 北京工业大学 | Membrane bioreactor-MBR fouling membrane intelligent early-warning method |
| CN112101402A (en) * | 2020-07-22 | 2020-12-18 | 北京工业大学 | Membrane pollution early warning method based on knowledge fuzzy learning |
| CN112183719A (en) * | 2020-09-15 | 2021-01-05 | 北京工业大学 | Intelligent detection method for total nitrogen in effluent based on multi-objective optimization-fuzzy neural network |
Non-Patent Citations (3)
| Title |
|---|
| 赵晓霞;: "一种粗糙模糊神经网络识别器及其应用", 中北大学学报(自然科学版), no. 03, 15 June 2009 (2009-06-15) * |
| 闫宏英;李春青;: "MBR膜污染的智能模拟预测方法研究", 计算机测量与控制, no. 08, 25 August 2013 (2013-08-25) * |
| 马玉芩;姜涛;: "基于自组织模糊神经网络的污水处理出水水质预测建模", 信息技术与信息化, no. 04, 15 August 2013 (2013-08-15) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113313230B (en) | 2024-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105510546B (en) | A kind of biochemical oxygen demand (BOD) BOD intelligent detecting methods based on self-organizing Recurrent RBF Neural Networks | |
| CN102854296B (en) | Sewage-disposal soft measurement method on basis of integrated neural network | |
| CN103744293B (en) | Wastewater treatment monitoring method and system based on fuzzy neural network | |
| CN107358021B (en) | DO prediction model establishment method based on BP neural network optimization | |
| CN111354423B (en) | Method for predicting ammonia nitrogen concentration of effluent of self-organizing recursive fuzzy neural network based on multivariate time series analysis | |
| AU2021101438A4 (en) | Adaptive control method and system for aeration process | |
| CN101576734B (en) | Dissolved oxygen control method based on dynamic radial basis function neural network | |
| CN102662040B (en) | Ammonian online soft measuring method for dynamic modularized nerve network | |
| US20140052422A1 (en) | Method and system for wastewater treatment based on dissolved oxygen control by fuzzy neural network | |
| CN109473182B (en) | MBR (Membrane biological reactor) membrane water permeability intelligent detection method based on deep belief network | |
| CN109657790B (en) | PSO-based recursive RBF neural network effluent BOD prediction method | |
| CN107402586A (en) | Dissolved Oxygen concentration Control method and system based on deep neural network | |
| CN106096730A (en) | A kind of intelligent detecting method of MBR film permeability rate based on Recurrent RBF Neural Networks | |
| CN108536106B (en) | Aeration system dissolved oxygen online regulation and control method based on Kalman filtering-extreme learning machine | |
| CN114275912A (en) | Aeration system dissolved oxygen control method based on adaptive neural network model | |
| CN110928187A (en) | Sewage treatment process fault monitoring method based on fuzzy width self-adaptive learning model | |
| CN106706491B (en) | Intelligent detection method for membrane bioreactor-MBR water permeability | |
| CN109408896B (en) | Multi-element intelligent real-time monitoring method for anaerobic sewage treatment gas production | |
| CN109978024A (en) | A kind of water outlet BOD prediction technique based on interconnection module neural network | |
| CN110412878B (en) | PID-RBF steady state estimation and LSSVM soft measurement sewage BOD control method | |
| CN110991616B (en) | Method for predicting BOD of effluent based on pruning feedforward small-world neural network | |
| CN105372995A (en) | Measurement and control method for sewage disposal system | |
| Fu et al. | Dissolved oxygen control system based on the TS fuzzy neural network | |
| CN113313230B (en) | Film pollution intelligent decision-making method based on knowledge fuzzy width learning | |
| CN110837886A (en) | Effluent NH4-N soft measurement method based on ELM-SL0 neural network |
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 |