CN117875506A - Method for predicting and processing aquaculture tail water based on LSTM neural network model - Google Patents

Method for predicting and processing aquaculture tail water based on LSTM neural network model Download PDF

Info

Publication number
CN117875506A
CN117875506A CN202410059054.9A CN202410059054A CN117875506A CN 117875506 A CN117875506 A CN 117875506A CN 202410059054 A CN202410059054 A CN 202410059054A CN 117875506 A CN117875506 A CN 117875506A
Authority
CN
China
Prior art keywords
data
sewage
water quality
water
tail water
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202410059054.9A
Other languages
Chinese (zh)
Inventor
江旭
乔椋
陈天明
宗朝鸿
征海钢
夏志斌
营中园
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
Yancheng Institute of Technology Technology Transfer Center Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Yancheng Institute of Technology, Yancheng Institute of Technology Technology Transfer Center Co Ltd filed Critical Yancheng Institute of Technology
Priority to CN202410059054.9A priority Critical patent/CN117875506A/en
Publication of CN117875506A publication Critical patent/CN117875506A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • 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
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0637—Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
    • G06Q10/06375—Prediction of business process outcome or impact based on a proposed change
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/02—Agriculture; Fishing; Forestry; Mining
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A20/00—Water conservation; Efficient water supply; Efficient water use
    • Y02A20/152—Water filtration

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Health & Medical Sciences (AREA)
  • Marketing (AREA)
  • General Health & Medical Sciences (AREA)
  • Development Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Educational Administration (AREA)
  • Biomedical Technology (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Biophysics (AREA)
  • Mathematical Physics (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Primary Health Care (AREA)
  • Computational Linguistics (AREA)
  • Computing Systems (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Game Theory and Decision Science (AREA)
  • Molecular Biology (AREA)
  • Mining & Mineral Resources (AREA)
  • Marine Sciences & Fisheries (AREA)
  • Animal Husbandry (AREA)
  • Agronomy & Crop Science (AREA)
  • Purification Treatments By Anaerobic Or Anaerobic And Aerobic Bacteria Or Animals (AREA)

Abstract

The invention provides a method and a system for predicting and processing aquaculture tail water based on an LSTM neural network model, wherein the method comprises the following steps: acquiring water quality monitoring data of the culture tail water, and preprocessing the water quality monitoring data; the water quality monitoring data comprises: ammonia nitrogen, pH, total nitrogen, total phosphorus, COD parameters; acquiring an LSTM neural network model; taking the pretreated water quality monitoring data as the input of the LSTM neural network model, and predicting the water quality parameter values of the culture tail water at a plurality of moments in the future; the method comprises the steps that the water quality parameter values of the culture tail water at a plurality of moments in the future are utilized, and the on-line regulation and control of the culture tail water equipment are carried out through a computer control module; after the online regulation and control are completed, real-time data of the culture tail water are collected through the collecting device to carry out feedback verification. According to the invention, parameters such as ammonia nitrogen, pH and the like in the device are predicted by utilizing the LSTM neural network prediction model, the water quality parameters are controlled, and the water quality is monitored in real time by the water quality acquisition device, so that the water quality of the effluent is improved.

Description

Method for predicting and processing aquaculture tail water based on LSTM neural network model
Technical Field
The invention relates to the technical field of aquaculture tail water treatment, in particular to an aquaculture tail water prediction treatment method and system based on an LSTM neural network model.
Background
In recent years, along with the rapid development of aquaculture industry in China, the aquaculture mode and technology are continuously improved, but the pollution problem caused by the aquaculture tail water is also increasingly serious, and the wastewater generated in the aquaculture process contains a large amount of pollutants such as organic matters, ammonia nitrogen and the like, so that the water environment is seriously influenced; finding an efficient and economical aquaculture tail water treatment technology has important significance in promoting sustainable development of aquaculture industry.
So far, the treatment process of the culture tail water in China comprises physical treatment, biochemical treatment and comprehensive treatment, while the common treatment process and the use mode of the related pond are that a sedimentation tank is used for the physical treatment, and suspended matters and larger solid particles are removed through gravity sedimentation; the aeration tank is used for biochemical treatment, and promotes the growth and metabolism of microorganisms through aeration, so that organic matters are degraded; the activated sludge tank is used for biochemical treatment, contains a large amount of active microorganisms, and can effectively decompose organic matters and remove nutrient substances such as nitrogen, phosphorus and the like; the air floatation tank is used for physical treatment, and suspended matters and grease are floated and separated from a water body by injecting tiny bubbles; the filter tank is used for physical treatment, and smaller suspended matters and impurities are removed by utilizing the filtering effect of the filter material; the photochemical pool is used for comprehensive treatment, and ultraviolet radiation or other photochemical processes are utilized for disinfection and sterilization to remove organic matter residues; meanwhile, different technologies have different characteristics, and the comprehensive treatment can effectively remove pollutants such as organic matters, nitrogen, phosphorus and the like, so that the treatment effect of the wastewater is improved; the physical treatment process is simple, the investment is low, and the method is suitable for primarily removing larger particles and suspended matters; the biochemical treatment process relies on degradation and treatment of organisms by microorganisms, so that the organisms in the wastewater can be effectively treated.
There are also disadvantages in these aquaculture tail water treatment processes. In terms of cost, large treatment facilities are usually required to be built for the treatment of the culture tail water, the investment and operation cost are high, and a large amount of energy, chemicals, equipment maintenance and other expenses are required for the traditional physical, chemical or biological treatment method. In technical terms, conventional tail water treatment techniques generally require complex process flows and expertise. Operators need to have special skills and training to ensure the effectiveness and safety of the tail water treatment process. In terms of treatment effect, the traditional tail water treatment method has a certain limitation on the removal effect of pollutants such as organic matters, nitrogen, phosphorus and the like. Especially for pollutants such as organic matters, trace heavy metals and the like which are difficult to degrade, the effect of the traditional method may be not ideal. In terms of environmental impact, by-products such as sludge, wastewater and the like generated in the traditional cultivation tail water treatment process, if the treatment is improper, secondary pollution can be caused to the environment. Furthermore, the chemicals and agents used in traditional treatment methods may create potential risks to the surrounding environment and the ecosystem; in terms of energy consumption and emission, the conventional tail water treatment process requires a large amount of energy supply, such as electricity and fuel, etc. The consumption of energy not only increases the economic burden, but also causes a certain carbon emission to the environment. These disadvantages are often affected by problems such as delays in detection and monitoring, blurring of process control, etc.
At present, with the development of artificial intelligence, a guarantee is provided in terms of solving the detection hysteresis by establishing an accurate model. And the establishment of an accurate model is to establish an accurate process model, and the change and response of process parameters are predicted by modeling and simulating the process. The control method based on the model can assist operators to make more accurate adjustment measures, and hysteresis is reduced. Proper algorithm is needed to ensure the accuracy of model establishment, and the treatment of the aquaculture tail water is a complex process, and involves time sequence relation and dynamic change among a plurality of variables. The RNN is a neural network model designed for processing time sequence data; RNNs can predict future outcomes by memorizing previous information. However, because factors such as parameter initialization, nonlinear activation function and the like of RNN in water quality prediction can generate gradient explosion, gradient index can be decreased, and long-term dependency of a prediction model is difficult to learn. The LSTM can effectively relieve the gradient vanishing problem by introducing a gating mechanism, so that the network can capture long-term dependency; LSTM controls the flow of information through a gating mechanism that can selectively memorize and forget information of an input sequence. This enables LSTM to better handle long-term dependencies, which is very effective for information that needs to be remembered for longer time intervals; LSTM can better process sequence data by adaptively learning features of different time scales in an input sequence.
LSTM is used in the treatment of the aquaculture tail water, because the LSTM has strong self-adaptive learning and nonlinear mapping capability, and can be widely applied to the fields of simulation prediction and real-time control of the aquaculture tail water treatment. Meanwhile, the cultivation tail water treatment equipment of the LSTM neural network can construct an efficient simulation model, and the relation between nonlinear characteristics of a treatment system and complex multivariable is solved. The equipment predicts and regulates the treatment of the culture tail water through the LSTM network so as to ensure that the quality of the effluent is stable and reaches the standard. By utilizing the self-adaptive learning capability and the nonlinear mapping capability of the LSTM network, the equipment can accurately predict and control key indexes in the treatment process, and the effect and the efficiency of the treatment of the aquaculture tail water are improved. Thereby realizing the optimization and optimization of the cultivation tail water treatment process. The method provides a reliable and efficient solution for the treatment of the aquaculture tail water, and promotes sustainable development and water environment protection.
Disclosure of Invention
The invention aims at providing a method for predicting and processing the aquaculture tail water based on an LSTM neural network model.
The invention provides a method for predicting and processing aquaculture tail water based on an LSTM neural network model, which comprises the following steps:
Acquiring water quality monitoring data of the culture tail water, and preprocessing the water quality monitoring data; the water quality monitoring data comprises: ammonia nitrogen, pH, total nitrogen, total phosphorus, COD parameters;
acquiring an LSTM neural network model;
taking the pretreated water quality monitoring data as the input of the LSTM neural network model, and predicting the water quality parameter values of the culture tail water at a plurality of moments in the future;
the method comprises the steps that the water quality parameter values of the culture tail water at a plurality of moments in the future are utilized, and the on-line regulation and control of the culture tail water equipment are carried out through a computer control module;
after the online regulation and control are completed, real-time data of the culture tail water are collected through the collecting device to carry out feedback verification.
Preferably, the preprocessing of the water quality monitoring data includes:
missing value processing:
acquiring the average value of sewage characteristic data at n times before the current time:
acquiring an average value of sewage characteristic data at the current moment in the previous m days:
filling the missing values as follows: f (f) k (t)＝αf 1 (t)+(1-α)f 2 (t)；
Outlier rejection:
the 3 sigma criterion is used to reject redundant and erroneous data, assuming x 1 ,x 2 ,...x n N samples of data, x is used as its average,the formula used to represent the deviation, standard deviation σ, is:
data normalization:
wherein f 1 (t) is the average value of the sewage characteristic data at n times before the current time, f 2 (t) is the average value of the sewage characteristic data at the current moment of the previous m days, f k (t) is the data at time t on the kth day, σ is the standard deviation, X * Representing the normalized data; alpha is used to represent the weighting factor, which has a value in the range of 0.5-1; v i (1. Ltoreq. I. Ltoreq. N) is used to represent the sample data x i Deviation; x is X max X is the maximum value of X in all data min X is the minimum value of X in all data n The new value obtained after normalization processing.
Preferably, the LSTM neural network model is implemented by using a forgetting gate, an input gate and an output gate structure:
the forgetting door can acquire the information h of the last moment t-1 Data x from the current time t The output value is between 0 and 1, the closer to 0 is to forget, the closer to 1 is to remember, and the following operation formula is:
f t ＝λ(ω f ·[h t-1 ,x t ]+b f )；
the input gate containing sigmoid function can update the information to be updated, C t Generating a new value by the tanh layer;
i t ＝λ(ω i ·[h t-1 ,x t ]+b i )；
the joint update is calculated as follows:
the cell state of the cell of the output gate determines the output result of the output gate, a part of the cell state is output through a sigmoid function, then the cell state is processed by a tanh function, and finally the output h at the current moment can be obtained t The method comprises the steps of carrying out a first treatment on the surface of the The output result of the output gate is determined by the cell state, a part of the cell state is output by a sigmoid function, the cell state is processed by a tanh function, and finally the current output h can be obtained t ；
O t ＝λ(ω o ·[h t-1 ,x t ]+b o )；
h t ＝O t *tanh(C t )；
In the formula, h t-1 For the output of the previous cell unit, x t For the input of the current cell unit, w and b are respectively a weight matrix and a bias vector in a forgetting gate, and lambda is an activation function sigmoid;generating a tan h layer; h is a t Is the output of the current moment.
Preferably, the cultivation tail water device comprises: a water collecting tank, an ABR tank, an aeration tank, a secondary sedimentation tank and an active carbon adsorption tank;
the water collecting tank, the ABR tank, the aeration tank, the secondary sedimentation tank and the active carbon adsorption tank are respectively provided with a water quality acquisition device, and the water quality acquisition devices are connected with the computer control module, wherein the water collecting tank can be used for collecting and storing the original sewage entering the sewage treatment system, balancing flow and water quality fluctuation and ensuring stable work of the subsequent treatment process;
the ABR pool is an anaerobic reactor and is mainly used for anaerobic degradation treatment, and the culture tail water stays in the ABR pool temporarily, so that organic substances are decomposed into relatively stable products such as methane, carbon dioxide and the like through the action of microorganisms;
An aeration tank is a key treatment unit in which metabolism of organisms is promoted by supplying oxygen to a water body, and in which bacteria and other microorganisms can degrade organic matters in sewage better by increasing concentration of dissolved oxygen;
the secondary sedimentation tank is used for sedimentation of suspended matters, suspended colloid and biological flocculate in the sewage, solid particles in the sewage are settled to the bottom to form sludge in the secondary sedimentation tank, and then clear water flows out from the upper part;
the activated carbon adsorption tank adopts activated carbon to adsorb and remove organic matters, dissolved matters and trace harmful substances in the sewage, and meanwhile, the activated carbon has strong adsorption capacity, so that refractory organic matters and peculiar smell substances in the sewage can be effectively removed.
Preferably, the water quality acquisition device is characterized by comprising various sensors for monitoring ammonia nitrogen, pH, total nitrogen, total phosphorus and COD parameters in water.
The embodiment of the invention provides a culture tail water prediction processing system based on an LSTM neural network model, which comprises the following steps:
the first acquisition module is used for acquiring water quality monitoring data of the culture tail water and preprocessing the water quality monitoring data; the water quality monitoring data comprises: ammonia nitrogen, pH, total nitrogen, total phosphorus, COD parameters;
The second acquisition module is used for acquiring the LSTM neural network model;
the prediction module is used for taking the pretreated water quality monitoring data as the input of the LSTM neural network model and predicting the water quality parameter values of the aquaculture tail water at a plurality of moments in the future;
the regulation and control module is used for carrying out online regulation and control on the cultivation tail water equipment by utilizing the water quality parameter values of the cultivation tail water at a plurality of moments in the future through the computer control module;
and the feedback verification module is used for collecting real-time data of the culture tail water through the collecting device to perform feedback verification after the online regulation and control are completed.
Preferably, the first obtaining module performs preprocessing on the water quality monitoring data, including:
missing value processing:
acquiring the average value of sewage characteristic data at n times before the current time:
acquiring an average value of sewage characteristic data at the current moment in the previous m days:
filling the missing values as follows: f (f) k (t)＝αf 1 (t)+(1-α)f 2 (t)；
Outlier rejection:
the 3 sigma criterion is used to reject redundant and erroneous data, assuming x 1 ,x 2 ,...x n N samples of data, x is used as its average,the formula used to represent the deviation, standard deviation σ, is:
data normalization:
wherein f 1 (t) is the average value of the sewage characteristic data at n times before the current time, f 2 (t) is the average value of the sewage characteristic data at the current moment of the previous m days, f k (t) is the data at time t on the kth day, σ is the standard deviation, X * Representing the normalized data; alpha is used to represent the weighting factor, which has a value in the range of 0.5-1; v i (1. Ltoreq. I. Ltoreq. N) is used to represent the sample data x i Deviation; x is X max X is the maximum value of X in all data min X is the minimum value of X in all data n The new value obtained after normalization processing.
Preferably, the LSTM neural network model is realized by forgetting gate, input gate and output gate structures, and the utilization is realized by:
the forgetting door can acquire the information h of the last moment t-1 Data x from the current time t The output value is between 0 and 1, the closer to 0 is to forget, the closer to 1 is to remember, and the following operation formula is:
f t ＝λ(ω f ·[h t-1 ,x t ]+b f )；
the input gate containing sigmoid function can update the information to be updated, C t Generating a new value by the tanh layer;
i t ＝λ(ω i ·[h t-1 ,x t ]+b i )；
the joint update is calculated as follows:
the unit state of the output gate cells determines the outputOutputting a part of the cell unit state through a sigmoid function, and then processing the cell unit state through a tanh function to finally obtain the output h at the current moment t The method comprises the steps of carrying out a first treatment on the surface of the The output result of the output gate is determined by the cell state, a part of the cell state is output by a sigmoid function, the cell state is processed by a tanh function, and finally the current output h can be obtained t ；
O t ＝λ(ω o ·[h t-1 ,x t ]+b o )；
h t ＝O t *tanh(C t )；
In the formula, h t-1 For the output of the previous cell unit, x t For the input of the current cell unit, w and b are respectively a weight matrix and a bias vector in a forgetting gate, and lambda is an activation function sigmoid;generating a tan h layer; h is a t Is the output of the current moment.
Preferably, the cultivation tail water device comprises: a water collecting tank, an ABR tank, an aeration tank, a secondary sedimentation tank and an active carbon adsorption tank;
the water collecting tank, the ABR tank, the aeration tank, the secondary sedimentation tank and the active carbon adsorption tank are respectively provided with a water quality acquisition device, and the water quality acquisition devices are connected with the computer control module, wherein the water collecting tank can be used for collecting and storing the original sewage entering the sewage treatment system, balancing flow and water quality fluctuation and ensuring stable work of the subsequent treatment process;
the ABR pool is an anaerobic reactor and is mainly used for anaerobic degradation treatment, and the culture tail water stays in the ABR pool temporarily, so that organic substances are decomposed into relatively stable products such as methane, carbon dioxide and the like through the action of microorganisms;
An aeration tank is a key treatment unit in which metabolism of organisms is promoted by supplying oxygen to a water body, and in which bacteria and other microorganisms can degrade organic matters in sewage better by increasing concentration of dissolved oxygen;
the secondary sedimentation tank is used for sedimentation of suspended matters, suspended colloid and biological flocculate in the sewage, solid particles in the sewage are settled to the bottom to form sludge in the secondary sedimentation tank, and then clear water flows out from the upper part;
the activated carbon adsorption tank adopts activated carbon to adsorb and remove organic matters, dissolved matters and trace harmful substances in the sewage, and meanwhile, the activated carbon has strong adsorption capacity, so that refractory organic matters and peculiar smell substances in the sewage can be effectively removed.
Preferably, the water quality acquisition device contains various sensors for monitoring ammonia nitrogen, pH, total nitrogen, total phosphorus and COD parameters in water.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.
The technical scheme of the invention is further described in detail through the drawings and the embodiments.
Drawings
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention. In the drawings:
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a block diagram of an LSTM neural network of the present invention;
FIG. 3 is a graph showing the COD prediction of the LSTM neural network according to the present invention;
FIG. 4 is a graph of the total phosphorus prediction of the LSTM neural network of the present invention;
FIG. 5 is a graph of ammonia nitrogen prediction for an LSTM neural network of the present invention;
FIG. 6 is a graph of the pH prediction of an LSTM neural network according to the present invention;
FIG. 7 is a graph of the total nitrogen prediction of the LSTM neural network of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described below with reference to the accompanying drawings, it being understood that the preferred embodiments described herein are for illustration and explanation of the present invention only, and are not intended to limit the present invention.
The embodiment of the invention provides a method for predicting and processing aquaculture tail water based on an LSTM neural network model, which comprises the following steps:
acquiring water quality monitoring data of the culture tail water, and preprocessing the water quality monitoring data; the water quality monitoring data comprises: ammonia nitrogen, pH, total nitrogen, total phosphorus, COD parameters;
Acquiring an LSTM neural network model; training the LSTM neural network model according to the marked multiple groups of water quality monitoring data and water quality evaluation results in a one-to-one correspondence manner, so that the LSTM neural network model can predict future water quality according to the water quality monitoring data;
taking the pretreated water quality monitoring data as the input of the LSTM neural network model, and predicting the water quality parameter values of the culture tail water at a plurality of moments in the future;
the method comprises the steps that the water quality parameter values of the culture tail water at a plurality of moments in the future are utilized, and the on-line regulation and control of the culture tail water equipment are carried out through a computer control module;
after the online regulation and control are completed, real-time data of the culture tail water are collected through the collecting device to carry out feedback verification.
Preprocessing the water quality monitoring data, including:
missing value processing:
acquiring the average value of sewage characteristic data at n times before the current time:
acquiring an average value of sewage characteristic data at the current moment in the previous m days:
filling the missing values as follows: f (f) k (t)＝αf 1 (t)+(1-α)f 2 (t)；
Outlier rejection:
the 3 sigma criterion is used to reject redundant and erroneous data, assuming x 1 ,x 2 ,...x n A total of n sample data are provided,is used as its average value, < >>The formula used to represent the deviation, standard deviation σ, is:
Data normalization:
wherein f 1 (t) is the average value of the sewage characteristic data at n times before the current time, f 2 (t) is the average value of the sewage characteristic data at the current moment of the previous m days, f k (t) is the data at time t on the kth day, σ is the standard deviation, X * Representing the normalized data; alpha is used to represent the weighting factor, which has a value in the range of 0.5-1; v i (1. Ltoreq. I. Ltoreq. N) is used to represent the sample data x i Deviation; x is X max X is the maximum value of X in all data min X is the minimum value of X in all data n The new value obtained after normalization processing.
The LSTM neural network model realizes the utilization by forgetting a gate, inputting a gate and outputting a gate structure:
the forgetting door can acquire the information h of the last moment t-1 Data x from the current time t The output value is between 0 and 1, the closer to 0 is to forget, the closer to 1 is to remember, and the following operation formula is:
f t ＝λ(ω f ·[h t-1 ,x t ]+b f )；
the input gate containing sigmoid function can update the information to be updated, C t Generating a new value by the tanh layer;
i t ＝λ(ω i ·[h t-1 ,x t ]+b i )；
the joint update is calculated as follows:
the cell state of the cell of the output gate determines the output result of the output gate, a part of the cell state is output through a sigmoid function, then the cell state is processed by a tanh function, and finally the output h at the current moment can be obtained t The method comprises the steps of carrying out a first treatment on the surface of the The output result of the output gate is determined by the cell state, a part of the cell state is output by a sigmoid function, the cell state is processed by a tanh function, and finally the current output h can be obtained t ；
O t ＝λ(ω o ·[h t-1 ,x t ]+b o )；
h t ＝O t *tanh(C t )；
In the formula, h t-1 For the output of the previous cell unit, x t For the input of the current cell unit, w and b are respectively a weight matrix and a bias vector in a forgetting gate, and lambda is an activation function sigmoid;generating a tan h layer; h is a t Is the output of the current moment.
The aquaculture tail water device comprises: a water collecting tank, an ABR tank, an aeration tank, a secondary sedimentation tank and an active carbon adsorption tank;
the water collecting tank, the ABR tank, the aeration tank, the secondary sedimentation tank and the active carbon adsorption tank are respectively provided with a water quality acquisition device, and the water quality acquisition devices are connected with the computer control module, wherein the water collecting tank can be used for collecting and storing the original sewage entering the sewage treatment system, balancing flow and water quality fluctuation and ensuring stable work of the subsequent treatment process;
the ABR pool is an anaerobic reactor and is mainly used for anaerobic degradation treatment, and the culture tail water stays in the ABR pool temporarily, so that organic substances are decomposed into relatively stable products such as methane, carbon dioxide and the like through the action of microorganisms;
An aeration tank is a key treatment unit in which metabolism of organisms is promoted by supplying oxygen to a water body, and in which bacteria and other microorganisms can degrade organic matters in sewage better by increasing concentration of dissolved oxygen;
the secondary sedimentation tank is used for sedimentation of suspended matters, suspended colloid and biological flocculate in the sewage, solid particles in the sewage are settled to the bottom to form sludge in the secondary sedimentation tank, and then clear water flows out from the upper part;
the activated carbon adsorption tank adopts activated carbon to adsorb and remove organic matters, dissolved matters and trace harmful substances in the sewage, and meanwhile, the activated carbon has strong adsorption capacity, so that refractory organic matters and peculiar smell substances in the sewage can be effectively removed.
The water quality acquisition device contains various sensors for monitoring ammonia nitrogen, pH, total nitrogen, total phosphorus and COD parameters in water.
Acquiring relevant water quality monitoring data of the culture tail water, and carrying out data preprocessing on the relevant water quality monitoring data, wherein the relevant water quality monitoring data comprises ammonia nitrogen, pH, total nitrogen, total phosphorus and COD parameters;
determining parameters in the neural network model, and determining a proper neural network structure;
taking the pretreated water quality parameters as the input of the neural network model, and obtaining the water quality parameter values of the aquaculture tail water equipment at all moments through the neural network model;
The obtained water quality parameter predicted value is utilized to carry out online regulation and control on the cultivation tail water equipment through a computer control module;
after the regulation and control are finished, the feedback verification can be carried out through the real-time data of the acquisition device.
The data preprocessing comprises missing value processing, outlier rejection and data normalization, and the related water quality monitoring data after preprocessing is preprocessed by the following formula:
the missing data may make the prediction model unstable, the prediction result is not reliable enough, the missing data is filled as necessary operation, the filling method is a weighted average method, and the average value of the sewage characteristic data at n times before the current time is obtained:
obtaining an average value of sewage characteristic data at the current moment of the previous m days:
filling the missing values as follows:
f k (t)＝αf 1 (t)+(1-α)f 2 (t) equation 3;
abnormal data can generate data noise, which affects the accuracy of the model. The 3 sigma criterion is used to reject redundant and erroneous data, assuming x 1 ,x 2 ,...x n A total of n sample data are provided,is used as an average value thereof,the formula for the standard deviation sigma is as follows:
after normalization, the distribution of the data is a distribution with a mean value of 0 and a variance of 1, and is not easily affected by outliers, and the influence of a large amount of data and a large measurement range on the model prediction result is greatly reduced.
Wherein f 1 (t) is the average value of the sewage characteristic data at n times before the current time, f 2 (t) is the average value of the sewage characteristic data at the current moment of the previous m days, f k (t) is the data at time t on the kth day, σ is the standard deviation, X * Representing the normalized data; alpha is used to represent the weighting factor, which has a value in the range of 0.5-1; v i (1. Ltoreq. I. Ltoreq. N) is used to represent the sample data x i Deviation; x is X max X is the maximum value of X in all data min X is the minimum value of X in all data n The new value obtained after normalization processing.
The neural network is an LSTM neural network, and the LSTM neural network realizes the utilization through a forgetting gate, an input gate and an output gate structure:
the forgetting door can acquire the information h of the last moment t-1 Data x from the current time t The output value is between 0 and 1, the closer to 0 is to forget, the closer to 1 is to remember, and the following operation formula is:
f t ＝λ(ω f ·[h t-1 ,x t ]+b f ) Equation 6;
the input gate containing sigmoid function can update the information to be updated, C t Generating a new value by the tanh layer;
i t ＝λ(ω i ·[h t-1 ,x t ]+b i ) Equation 8;
the joint update of equation 8 and equation 9 is calculated as follows:
the cell state of the cell of the output gate determines the output result of the output gate, a part of the cell state is output through a sigmoid function, then the cell state is processed by a tanh function, and finally the output h at the current moment can be obtained t The method comprises the steps of carrying out a first treatment on the surface of the The output result of the output gate is determined by the cell state, a part of the cell state is output by a sigmoid function, the cell state is processed by a tanh function, and finally the current output h can be obtained t ；
O t ＝λ(ω o ·[h t-1 ,x t ]+b o ) Equation 11;
h t ＝O t *tanh(C t ) Equation 12;
in the formula, h t-1 For the output of the previous cell unit, x t For the input of the current cell unit, w and b are respectively a weight matrix and a bias vector in a forgetting gate, and lambda is an activation function sigmoid;generating a tan h layer; h is a t Is the output of the current moment.
The cultivation tail water equipment structure comprises: the sewage treatment system comprises a water collecting tank, an ABR tank, an aeration tank, a secondary sedimentation tank, an activated carbon adsorption tank and the like, wherein water quality collecting devices are arranged in the water collecting tank, the ABR tank, the aeration tank, the secondary sedimentation tank and the activated carbon adsorption tank and are connected with a computer control module, the water collecting tank can be used for collecting and storing raw sewage entering a sewage treatment system, balancing flow and water quality fluctuation, and ensuring stable work of a subsequent treatment process; the ABR pool is an anaerobic reactor and is mainly used for anaerobic degradation treatment, and the culture tail water stays in the ABR pool temporarily, so that organic substances are decomposed into relatively stable products such as methane, carbon dioxide and the like through the action of microorganisms; an aeration tank is a key treatment unit in which metabolism of organisms is promoted by supplying oxygen to a water body, and in which bacteria and other microorganisms can degrade organic matters in sewage better by increasing concentration of dissolved oxygen; the secondary sedimentation tank is used for sedimentation of suspended matters, suspended colloid and biological flocculate in the sewage, solid particles in the sewage are settled to the bottom to form sludge in the secondary sedimentation tank, and then clear water flows out from the upper part; the activated carbon adsorption tank adopts activated carbon to adsorb and remove organic matters, dissolved matters and trace harmful substances in the sewage, and meanwhile, the activated carbon has strong adsorption capacity, so that refractory organic matters and peculiar smell substances in the sewage can be effectively removed.
The water quality acquisition device is internally provided with various sensors for monitoring dissolved oxygen, pH, total nitrogen, total phosphorus and COD parameters in water.
The computer control module includes:
the data processing is used for monitoring and collecting water quality data of the culture tail water in real time, obtaining relevant water quality monitoring data and carrying out data preprocessing on the relevant water quality monitoring data;
the neural network model prediction is used for taking the water quality parameter of the pretreatment of the culture tail water as the input of the LSTM neural network model, and predicting the water quality parameter value of the culture tail water treatment equipment at each moment through the LSTM neural network model;
and the intelligent regulation and control system is used for carrying out online regulation and control on the water quality parameter predicted value obtained by the LSTM neural network model through the regulation and control system.
An LSTM neural network model-based aquaculture tail water prediction processing method further comprises the following steps:
when the local prediction treatment of the breeding tail water is finished, acquiring a conference session;
determining a prediction treatment experience of the cultivation tail water based on the conference session;
based on the prediction treatment experience of the culture tail water, performing supplementary training on an LSTM neural network model;
based on the conference session, determining a prediction treatment experience for the aquaculture tail, comprising:
Sequencing the conference dialogs according to a dialog time sequence to obtain a dialog sequence;
traversing the conference dialogue from the middle point of the dialogue sequence to the two ends of the sequence in sequence, and acquiring a first keyword set of the traversed conference dialogue every time;
acquiring a plurality of first assessment keyword sets for the prediction treatment question assessment of the culture tail water;
matching the first keyword set with any first evaluation keyword set, and if the matching is met, acquiring a plurality of second evaluation keyword sets which correspond to the matched first evaluation keyword sets and are used for carrying out prediction processing on the aquaculture tail water and answer evaluation;
acquiring a second keyword set of the conference session in a first time preset after the traversed conference session in the session sequence;
matching the second keyword set with any second evaluation keyword set, if the matching is met, acquiring a plurality of preset third evaluation keyword sets which are corresponding to the matched second evaluation keyword sets and are used for carrying out prediction processing on the aquaculture tail water and are used for asking for a non-divergence evaluation on the answer, and acquiring a third keyword set of the conference session in a second time preset after the second keyword set matched in the session sequence corresponds to the conference session;
Matching the third keyword set with any one of the third evaluation keyword sets, and if the third keyword sets are not matched, acquiring a first dialogue semantic corresponding to the conference dialogue from the third keyword set and acquiring a traversed second dialogue semantic of the conference dialogue;
and generating a template based on preset prediction processing experience of the aquaculture tail water, and generating the prediction processing experience of the aquaculture tail water according to the first dialogue semantics and the second dialogue semantics.
The conference session may be voice information in a conference, etc. The conversation timing is the chronological order in which conversations were generated. Firstly, a prediction processor of the breeding tail water needs to be evaluated to ask questions. Generally, since the beginning and end of a conference require a start speech, an end speech, and the like, a question is often in the middle of the conference. The conference session is traversed from the midpoint of the session sequence to both ends and the first set of keywords is extracted. A first set of assessment keywords is introduced for a predictive treatment questioning assessment of the aquaculture tail water, for example comprising "how", "predictive" and "water quality" keywords. If the first keyword set is matched with the first evaluation keyword set, the condition that the culture tail water prediction processing personnel conduct questioning is indicated. And secondly, other breeding tail water prediction processing personnel need to be evaluated to answer questions of the breeding tail water prediction processing personnel. Typically, the answer is after it, and thus, a second set of keywords for the conference session is extracted for a preset first time (e.g., 10 minutes) after the traversed conference session. The preset second evaluation keyword sets corresponding to the matched first evaluation keyword sets are used for the prediction processing of the aquaculture tail water to question and answer evaluation, and the evaluation comprises the steps of collecting keywords firstly and analyzing keywords secondly. If the second keyword set matches with the second evaluation keyword set, other aquaculture tail water prediction processing personnel are indicated to give an answer. And thirdly, evaluating whether other breeding tail water prediction processing personnel have diverged or not according to answers given by other breeding tail water prediction processing personnel. Typically, a divergence follows, and therefore, a third set of keywords in the dialog sequence that match the meeting for the conference dialog within a second time (e.g., 5 minutes) preset after the conference dialog is extracted. And a preset plurality of third evaluation keyword sets corresponding to the matched second evaluation keyword sets are used for carrying out prediction processing on the aquaculture tail water, and the question answer is evaluated without divergence, and comprise keywords such as 'like' and 'no line'. If the third keywords are not matched with the third evaluation keyword set, the method indicates that no ambiguity is provided by the culture tail water prediction processing personnel. After disambiguation, extracting a first dialogue semantic corresponding to the conference dialogue and a second dialogue semantic corresponding to the traversed conference dialogue from a third keyword set, wherein the second dialogue semantic is used as a problem, and the first dialogue semantic is used as a problem solving method to generate a culture tail water prediction processing experience. The three steps are used for respectively determining questions and questions answers without divergence, which are given by the breeding tail water prediction processing personnel in the conference, as the breeding tail water prediction processing experience, so that the applicability of determining the breeding tail water prediction processing experience based on the conference dialogue is improved. And (3) performing supplementary training on the LSTM neural network model, and improving the working capacity of the LSTM neural network model.
The embodiment of the invention provides a culture tail water prediction processing system based on an LSTM neural network model, which is characterized by comprising the following steps:
the first acquisition module is used for acquiring water quality monitoring data of the culture tail water and preprocessing the water quality monitoring data; the water quality monitoring data comprises: ammonia nitrogen, pH, total nitrogen, total phosphorus, COD parameters;
the second acquisition module is used for acquiring the LSTM neural network model;
the prediction module is used for taking the pretreated water quality monitoring data as the input of the LSTM neural network model and predicting the water quality parameter values of the aquaculture tail water at a plurality of moments in the future;
the regulation and control module is used for carrying out online regulation and control on the cultivation tail water equipment by utilizing the water quality parameter values of the cultivation tail water at a plurality of moments in the future through the computer control module;
and the feedback verification module is used for collecting real-time data of the culture tail water through the collecting device to perform feedback verification after the online regulation and control are completed.
The first acquisition module preprocesses the water quality monitoring data, and the first acquisition module comprises:
missing value processing:
acquiring the average value of sewage characteristic data at n times before the current time:
acquiring an average value of sewage characteristic data at the current moment in the previous m days:
Filling the missing values as follows: f (f) k (t)＝αf 1 (t)+(1-α)f 2 (t)；
Outlier rejection:
the 3 sigma criterion is used to reject redundant and erroneous data, assuming x 1 ,x 2 ,...x n A total of n sample data are provided,is used as its average value, < >>The formula used to represent the deviation, standard deviation σ, is:
data normalization:
wherein f 1 (t) is the average value of the sewage characteristic data at n times before the current time, f 2 (t) is the average value of the sewage characteristic data at the current moment of the previous m days, f k (t) is the data at time t on the kth day, σ is the standard deviation, X * Representing the normalized data; alpha is used to represent the weighting factor, which has a value in the range of 0.5-1; v i (1. Ltoreq. I. Ltoreq. N) is used to represent the sample data x i Deviation; x is X max X is the maximum value of X in all data min X is the minimum value of X in all data n The new value obtained after normalization processing.
The LSTM neural network model realizes the utilization by forgetting a gate, inputting a gate and outputting a gate structure:
forgetful doorAcquiring information h of the last moment t-1 Data x from the current time t The output value is between 0 and 1, the closer to 0 is to forget, the closer to 1 is to remember, and the following operation formula is:
f t ＝λ(ω f ·[h t-1 ,x t ]+b f )；
/>
the input gate containing the sigmoid function can update the information that needs to be updated, Generating a new value by the tanh layer;
i t ＝λ(ω i ·[h t-1 ,x t ]+b i )；
the joint update is calculated as follows:
the cell state of the cell of the output gate determines the output result of the output gate, a part of the cell state is output through a sigmoid function, then the cell state is processed by a tanh function, and finally the output h at the current moment can be obtained t The method comprises the steps of carrying out a first treatment on the surface of the The output result of the output gate is determined by the cell state, a part of the cell state is output by a sigmoid function, the cell state is processed by a tanh function, and finally the current output h can be obtained t ；
O t ＝λ(ω o ·[h t-1 ,x t ]+b o )；
h t ＝O t *tanh(C t )；
In the formula, h t-1 For the previous cellOutput of unit, x t For the input of the current cell unit, w and b are respectively a weight matrix and a bias vector in a forgetting gate, and lambda is an activation function sigmoid;generating a tan h layer; h is a t Is the output of the current moment.
The aquaculture tail water device comprises: a water collecting tank, an ABR tank, an aeration tank, a secondary sedimentation tank and an active carbon adsorption tank;
the water collecting tank, the ABR tank, the aeration tank, the secondary sedimentation tank and the active carbon adsorption tank are respectively provided with a water quality acquisition device, and the water quality acquisition devices are connected with the computer control module, wherein the water collecting tank can be used for collecting and storing the original sewage entering the sewage treatment system, balancing flow and water quality fluctuation and ensuring stable work of the subsequent treatment process;
The ABR pool is an anaerobic reactor and is mainly used for anaerobic degradation treatment, and the culture tail water stays in the ABR pool temporarily, so that organic substances are decomposed into relatively stable products such as methane, carbon dioxide and the like through the action of microorganisms;
an aeration tank is a key treatment unit in which metabolism of organisms is promoted by supplying oxygen to a water body, and in which bacteria and other microorganisms can degrade organic matters in sewage better by increasing concentration of dissolved oxygen;
the secondary sedimentation tank is used for sedimentation of suspended matters, suspended colloid and biological flocculate in the sewage, solid particles in the sewage are settled to the bottom to form sludge in the secondary sedimentation tank, and then clear water flows out from the upper part;
the activated carbon adsorption tank adopts activated carbon to adsorb and remove organic matters, dissolved matters and trace harmful substances in the sewage, and meanwhile, the activated carbon has strong adsorption capacity, so that refractory organic matters and peculiar smell substances in the sewage can be effectively removed.
The water quality acquisition device contains various sensors for monitoring ammonia nitrogen, pH, total nitrogen, total phosphorus and COD parameters in water.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims (10)

1. A method for predicting and treating aquaculture tail water based on an LSTM neural network model is characterized by comprising the following steps:
acquiring water quality monitoring data of the culture tail water, and preprocessing the water quality monitoring data; the water quality monitoring data comprises: ammonia nitrogen, pH, total nitrogen, total phosphorus, COD parameters;
acquiring an LSTM neural network model;
taking the pretreated water quality monitoring data as the input of the LSTM neural network model, and predicting the water quality parameter values of the culture tail water at a plurality of moments in the future;
the method comprises the steps that the water quality parameter values of the culture tail water at a plurality of moments in the future are utilized, and the on-line regulation and control of the culture tail water equipment are carried out through a computer control module;
after the online regulation and control are completed, real-time data of the culture tail water are collected through the collecting device to carry out feedback verification.
2. The method for predicting and processing the aquaculture tail water based on the LSTM neural network model as set forth in claim 1, wherein the preprocessing of the water quality monitoring data comprises the steps of:
missing value processing:
acquiring an average value f of sewage characteristic data at n times before the current time t 1 (t)：f k (t-1) is the sewage characteristic data of the first time before the current time t on the kth day, f k (t-2)f k (t-1) is the sewage characteristic data at the second time before the current time t on the kth day, f k (t-n) is sewage characteristic data at an nth time before the current time t of the kth day;
last m days of acquisitionAverage value f of sewage characteristic data at current time t 2 (t)：
f k-m (t) is the sewage characteristic data of the current time t m days before the kth day, f k-(m-1) (t) is the sewage characteristic data of the current time t m-1 days before the kth day, f k-1 (t) is the sewage characteristic data of the current moment t of the k day;
for the missing value f k (t) filling: f (f) k (t)＝αf 1 (t)+(1-α)f 2 (t); alpha is a preset constant;
outlier rejection:
the 3 sigma criterion is used to reject redundant and erroneous data, assuming x 1 ,x 2 ,...x n N samples of data, x is used as its average,the formula used to represent the deviation, standard deviation σ, is:
data normalization:
wherein f 1 (t) is the average value of the sewage characteristic data at n times before the current time t, f 2 (t) is the average value of the sewage characteristic data at the current moment of the previous m days, f k (t) is the data at time t on the kth day, σ is the standard deviation, X * Representing the normalized data; alpha is used to represent the weighting factor, which has a value in the range of 0.5-1; v i (1. Ltoreq. I. Ltoreq. N) is used to represent the sample data x i Deviation; x is X max X is the maximum value of X in all data min X is the minimum value of X in all data n For normalizingAnd (5) obtaining new values after the processing.
3. The method for predicting and processing the aquaculture tail water based on the LSTM neural network model as set forth in claim 1, wherein the LSTM neural network model is realized by forgetting gate, input gate and output gate structures:
the forgetting door can acquire the information h of the last moment t-1 Data x from the current time t The output value is between 0 and 1, the closer to 0 is to forget, the closer to 1 is to remember, and the following operation formula is:
f t ＝λ(ω f ·[h t-1 ,x t ]+b f )；
the input gate containing the sigmoid function can update the information that needs to be updated,generating a new value by the tanh layer;
i t ＝λ(ω i ·[h t-1 ,x t ]+b i )；
the joint update is calculated as follows:
the cell state of the cell of the output gate determines the output result of the output gate, a part of the cell state is output through a sigmoid function, then the cell state is processed by a tanh function, and finally the output h at the current moment can be obtained t The method comprises the steps of carrying out a first treatment on the surface of the The output of the output gate is determined by the cell state, a portion of the cell stateThe cell state is processed by the tanh function, and the current output h can be obtained t ；
O t ＝λ(ω o ·[h t-1 ,x t ]+b o )；
h t ＝O t *tanh(C t )；
In the formula, h t-1 For the output of the previous cell unit, x t For the input of the current cell unit, w and b are respectively a weight matrix and a bias vector in a forgetting gate, and lambda is an activation function sigmoid; Generating a tan h layer; h is a t Is the output of the current moment.
4. The method for predicting and processing the aquaculture tail water based on the LSTM neural network model as set forth in claim 1, wherein the aquaculture tail water device comprises: a water collecting tank, an ABR tank, an aeration tank, a secondary sedimentation tank and an active carbon adsorption tank;
the water collecting tank, the ABR tank, the aeration tank, the secondary sedimentation tank and the active carbon adsorption tank are respectively provided with a water quality acquisition device, and the water quality acquisition devices are connected with the computer control module, wherein the water collecting tank can be used for collecting and storing the original sewage entering the sewage treatment system, balancing flow and water quality fluctuation and ensuring stable work of the subsequent treatment process;
the ABR pool is an anaerobic reactor and is mainly used for anaerobic degradation treatment, and the culture tail water stays in the ABR pool temporarily, so that organic substances are decomposed into relatively stable products such as methane, carbon dioxide and the like through the action of microorganisms;
an aeration tank is a key treatment unit in which metabolism of organisms is promoted by supplying oxygen to a water body, and in which bacteria and other microorganisms can degrade organic matters in sewage better by increasing concentration of dissolved oxygen;
the secondary sedimentation tank is used for sedimentation of suspended matters, suspended colloid and biological flocculate in the sewage, solid particles in the sewage are settled to the bottom to form sludge in the secondary sedimentation tank, and then clear water flows out from the upper part;
The activated carbon adsorption tank adopts activated carbon to adsorb and remove organic matters, dissolved matters and trace harmful substances in the sewage, and meanwhile, the activated carbon has strong adsorption capacity, so that refractory organic matters and peculiar smell substances in the sewage can be effectively removed.
5. The method for predicting and treating the aquaculture tail water based on the LSTM neural network model as set forth in claim 1, wherein the water quality acquisition device contains various sensors for monitoring parameters of ammonia nitrogen, pH, total nitrogen, total phosphorus and COD in the water.
6. An LSTM neural network model-based aquaculture tail water prediction processing system is characterized by comprising:
the first acquisition module is used for acquiring water quality monitoring data of the culture tail water and preprocessing the water quality monitoring data; the water quality monitoring data comprises: ammonia nitrogen, pH, total nitrogen, total phosphorus, COD parameters;
the second acquisition module is used for acquiring the LSTM neural network model;
the prediction module is used for taking the pretreated water quality monitoring data as the input of the LSTM neural network model and predicting the water quality parameter values of the aquaculture tail water at a plurality of moments in the future;
the regulation and control module is used for carrying out online regulation and control on the cultivation tail water equipment by utilizing the water quality parameter values of the cultivation tail water at a plurality of moments in the future through the computer control module;
And the feedback verification module is used for collecting real-time data of the culture tail water through the collecting device to perform feedback verification after the online regulation and control are completed.
7. The system of claim 6, wherein the first acquisition module pre-processes the water quality monitoring data, comprising:
missing value processing:
acquiring the current timeMean value f of sewage characteristic data at n times before t 1 (t)：f k (t-1) is the sewage characteristic data of the first time before the current time t on the kth day, f k (t-2)f k (t-1) is the sewage characteristic data at the second time before the current time t on the kth day, f k (t-n) is sewage characteristic data at an nth time before the current time t of the kth day;
acquiring an average value f of sewage characteristic data of the current moment t of the previous m days 2 (t)：
f k-m (t) is the sewage characteristic data of the current time t m days before the kth day, f k-(m-1) (t) is the sewage characteristic data of the current time t m-1 days before the kth day, f k-1 (t) is the sewage characteristic data of the current moment t of the k day;
for the missing value f k (t) filling: f (f) k (t)＝αf 1 (t)+(1-α)f 2 (t); alpha is a preset constant;
outlier rejection:
the 3 sigma criterion is used to reject redundant and erroneous data, assuming x 1 ,x 2 ,...x n N samples of data, x is used as its average, (i=1, 2,..n) is used to represent the deviation, the standard deviation σ of which is given by:
data normalization:
wherein f 1 (t) is the average value of the sewage characteristic data at n times before the current time t, f 2 (t) is the average value of the sewage characteristic data at the current moment of the previous m days, f k (t) is the data at time t on the kth day, σ is the standard deviation, X * Representing the normalized data; alpha is used to represent the weighting factor, which has a value in the range of 0.5-1; v i (1. Ltoreq. I. Ltoreq. N) is used to represent the sample data x i Deviation; x is X max X is the maximum value of X in all data min X is the minimum value of X in all data n The new value obtained after normalization processing.
8. The system for predicting and processing the aquaculture tail water based on the LSTM neural network model as set forth in claim 6, wherein the LSTM neural network model is realized by forgetting gate, input gate and output gate structures:
the forgetting door can acquire the information h of the last moment t-1 Data x from the current time t The output value is between 0 and 1, the closer to 0 is to forget, the closer to 1 is to remember, and the following operation formula is:
f t ＝λ(ω f ·[h t-1 ,x t ]+b f )；
the input gate containing sigmoid function can update the information to be updated, C t Generating a new value by the tanh layer;
i t ＝λ(ω i ·[h t-1 ,x t ]+b i )；
the joint update is calculated as follows:
the cell state of the cell of the output gate determines the output result of the output gate, a part of the cell state is output through a sigmoid function, then the cell state is processed by a tanh function, and finally the output h at the current moment can be obtained t The method comprises the steps of carrying out a first treatment on the surface of the The output result of the output gate is determined by the cell state, a part of the cell state is output by a sigmoid function, the cell state is processed by a tanh function, and finally the current output h can be obtained t ；
O t ＝λ(ω o ·[h t-1 ,x t ]+b o )；
h t ＝O t *tanh(C t )；
In the formula, h t-1 For the output of the previous cell unit, x t For the input of the current cell unit, w and b are respectively a weight matrix and a bias vector in a forgetting gate, and lambda is an activation function sigmoid;generating a tan h layer; h is a t Is the output of the current moment.
9. The system for predicting and processing aquaculture tail water based on LSTM neural network model as set forth in claim 6, wherein said aquaculture tail water device comprises: a water collecting tank, an ABR tank, an aeration tank, a secondary sedimentation tank and an active carbon adsorption tank;
the water collecting tank, the ABR tank, the aeration tank, the secondary sedimentation tank and the active carbon adsorption tank are respectively provided with a water quality acquisition device, and the water quality acquisition devices are connected with the computer control module, wherein the water collecting tank can be used for collecting and storing the original sewage entering the sewage treatment system, balancing flow and water quality fluctuation and ensuring stable work of the subsequent treatment process;
The ABR pool is an anaerobic reactor and is mainly used for anaerobic degradation treatment, and the culture tail water stays in the ABR pool temporarily, so that organic substances are decomposed into relatively stable products such as methane, carbon dioxide and the like through the action of microorganisms;
an aeration tank is a key treatment unit in which metabolism of organisms is promoted by supplying oxygen to a water body, and in which bacteria and other microorganisms can degrade organic matters in sewage better by increasing concentration of dissolved oxygen;
the secondary sedimentation tank is used for sedimentation of suspended matters, suspended colloid and biological flocculate in the sewage, solid particles in the sewage are settled to the bottom to form sludge in the secondary sedimentation tank, and then clear water flows out from the upper part;
the activated carbon adsorption tank adopts activated carbon to adsorb and remove organic matters, dissolved matters and trace harmful substances in the sewage, and meanwhile, the activated carbon has strong adsorption capacity, so that refractory organic matters and peculiar smell substances in the sewage can be effectively removed.
10. The system for predicting and treating the aquaculture tail water based on the LSTM neural network model as set forth in claim 6, wherein the water quality acquisition device comprises various sensors for monitoring parameters of ammonia nitrogen, pH, total nitrogen, total phosphorus and COD in the water.
CN202410059054.9A 2024-01-16 2024-01-16 Method for predicting and processing aquaculture tail water based on LSTM neural network model Pending CN117875506A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202410059054.9A CN117875506A (en) 2024-01-16 2024-01-16 Method for predicting and processing aquaculture tail water based on LSTM neural network model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202410059054.9A CN117875506A (en) 2024-01-16 2024-01-16 Method for predicting and processing aquaculture tail water based on LSTM neural network model

Publications (1)

Publication Number Publication Date
CN117875506A true CN117875506A (en) 2024-04-12

Family

ID=90591732

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202410059054.9A Pending CN117875506A (en) 2024-01-16 2024-01-16 Method for predicting and processing aquaculture tail water based on LSTM neural network model

Country Status (1)

Country Link
CN (1) CN117875506A (en)

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108717586A (en) * 2018-05-10 2018-10-30 江南大学 A kind of culture environment of aquatic products dissolved oxygen prediction method based on long memory network in short-term
CN109508811A (en) * 2018-09-30 2019-03-22 中冶华天工程技术有限公司 Parameter prediction method is discharged based on principal component analysis and the sewage treatment of shot and long term memory network
CN212174735U (en) * 2020-04-21 2020-12-18 郑州亿众环境科技有限公司 System for be used for handling rubber auxiliary agent CBS waste water
CN112132333A (en) * 2020-09-16 2020-12-25 安徽泽众安全科技有限公司 Short-term water quality and water quantity prediction method and system based on deep learning
CN115330040A (en) * 2022-08-05 2022-11-11 江苏润和软件股份有限公司 Deep learning-based comprehensive energy distributed wind power generation prediction method and system
CN115536206A (en) * 2021-06-30 2022-12-30 中国石油化工股份有限公司 Advanced treatment combined process for chemical degradation-resistant sewage
CN115838222A (en) * 2022-11-22 2023-03-24 盐城工学院 Intelligent test wastewater treatment integrated device and treatment method
CN116306803A (en) * 2023-01-09 2023-06-23 北京工业大学 Method for predicting BOD concentration of outlet water of ILSTM (biological information collection flow) neural network based on WSFA-AFE
CN116432832A (en) * 2023-03-21 2023-07-14 燕山大学 Water quality prediction method based on XGBoost-LSTM prediction model
CN117171624A (en) * 2023-09-28 2023-12-05 扬州大学 Waterfowl cultivation water quality prediction method

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108717586A (en) * 2018-05-10 2018-10-30 江南大学 A kind of culture environment of aquatic products dissolved oxygen prediction method based on long memory network in short-term
CN109508811A (en) * 2018-09-30 2019-03-22 中冶华天工程技术有限公司 Parameter prediction method is discharged based on principal component analysis and the sewage treatment of shot and long term memory network
CN212174735U (en) * 2020-04-21 2020-12-18 郑州亿众环境科技有限公司 System for be used for handling rubber auxiliary agent CBS waste water
CN112132333A (en) * 2020-09-16 2020-12-25 安徽泽众安全科技有限公司 Short-term water quality and water quantity prediction method and system based on deep learning
CN115536206A (en) * 2021-06-30 2022-12-30 中国石油化工股份有限公司 Advanced treatment combined process for chemical degradation-resistant sewage
CN115330040A (en) * 2022-08-05 2022-11-11 江苏润和软件股份有限公司 Deep learning-based comprehensive energy distributed wind power generation prediction method and system
CN115838222A (en) * 2022-11-22 2023-03-24 盐城工学院 Intelligent test wastewater treatment integrated device and treatment method
CN116306803A (en) * 2023-01-09 2023-06-23 北京工业大学 Method for predicting BOD concentration of outlet water of ILSTM (biological information collection flow) neural network based on WSFA-AFE
CN116432832A (en) * 2023-03-21 2023-07-14 燕山大学 Water quality prediction method based on XGBoost-LSTM prediction model
CN117171624A (en) * 2023-09-28 2023-12-05 扬州大学 Waterfowl cultivation water quality prediction method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
PING LIU ET AL.: "Analysis and Prediction of Water Quality Using LSTM Deep Neural Networks in IoT Environment", SUSTAINABILITY, no. 11, 7 April 2019 (2019-04-07), pages 1 - 14 *

Similar Documents

Publication Publication Date Title
Gontarski et al. Simulation of an industrial wastewater treatment plant using artificial neural networks
Han et al. Prediction of activated sludge bulking based on a self-organizing RBF neural network
CN103744293B (en) Wastewater treatment monitoring method and system based on fuzzy neural network
CN102854296B (en) Sewage-disposal soft measurement method on basis of integrated neural network
Nasr et al. Artificial intelligence for greywater treatment using electrocoagulation process
Pisa et al. LSTM-based wastewater treatment plants operation strategies for effluent quality improvement
CN107402586A (en) Dissolved Oxygen concentration Control method and system based on deep neural network
KR102311657B1 (en) Smart management system for wastewater treatment
CN114690700A (en) PLC-based intelligent sewage treatment decision optimization method and system
Qambar et al. Prediction of municipal wastewater biochemical oxygen demand using machine learning techniques: a sustainable approach
Türkmenler et al. Performance assessment of advanced biological wastewater treatment plants using artificial neural networks
CN113627506A (en) Intelligent detection method for total phosphorus in effluent based on information fusion-interval two-type fuzzy neural network
Rustum Modelling activated sludge wastewater treatment plants using artificial intelligence techniques (fuzzy logic and neural networks)
CN117228894A (en) Accurate aeration sewage treatment process method
CN117875506A (en) Method for predicting and processing aquaculture tail water based on LSTM neural network model
Wang et al. Artificial intelligence algorithm application in wastewater treatment plants: Case study for COD load prediction
Soehartanto et al. Dissolved oxygen control system in polishing unit using logic solver
Han et al. Efficient economic model predictive control of water treatment process with learning-based Koopman operator
Huang et al. Modeling and optimization of the activated sludge process
Vyas et al. Artificial neural network based model in effluent treatment process
CN114861543A (en) Data-driven intelligent evaluation method for biodegradability of petrochemical sewage
CN115353223A (en) Laboratory wastewater treatment integrated method and device based on ANFIS model
Han et al. Intelligent modeling approach to predict effluent quality of wastewater treatment process
Huang et al. Modeling of a paper-making wastewater treatment process using a fuzzy neural network
CN113065242A (en) KPLSR model-based soft measurement method for total nitrogen concentration of effluent from sewage treatment

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination