CN115759437A - BP neural network sewage index prediction method based on HGS - Google Patents
BP neural network sewage index prediction method based on HGS Download PDFInfo
- Publication number
- CN115759437A CN115759437A CN202211482168.1A CN202211482168A CN115759437A CN 115759437 A CN115759437 A CN 115759437A CN 202211482168 A CN202211482168 A CN 202211482168A CN 115759437 A CN115759437 A CN 115759437A
- Authority
- CN
- China
- Prior art keywords
- neural network
- individual
- fitness
- population
- optimal
- 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
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 67
- 239000010865 sewage Substances 0.000 title claims abstract description 39
- 238000000034 method Methods 0.000 title claims abstract description 36
- 239000013598 vector Substances 0.000 claims abstract description 55
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 15
- 238000003062 neural network model Methods 0.000 claims abstract description 14
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 claims abstract description 10
- OAICVXFJPJFONN-UHFFFAOYSA-N Phosphorus Chemical compound [P] OAICVXFJPJFONN-UHFFFAOYSA-N 0.000 claims abstract description 5
- XKMRRTOUMJRJIA-UHFFFAOYSA-N ammonia nh3 Chemical compound N.N XKMRRTOUMJRJIA-UHFFFAOYSA-N 0.000 claims abstract description 5
- 229910052757 nitrogen Inorganic materials 0.000 claims abstract description 5
- 229910052698 phosphorus Inorganic materials 0.000 claims abstract description 5
- 239000011574 phosphorus Substances 0.000 claims abstract description 5
- 235000003642 hunger Nutrition 0.000 claims description 18
- 210000002569 neuron Anatomy 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 description 16
- 238000005457 optimization Methods 0.000 description 10
- 238000012549 training Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 210000004205 output neuron Anatomy 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 239000007789 gas Substances 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000001301 oxygen Substances 0.000 description 1
- 229910052760 oxygen Inorganic materials 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 239000002351 wastewater Substances 0.000 description 1
Images
Classifications
-
- 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
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a BP neural network sewage index prediction method based on HGS, which comprises the following steps: step S1: constructing an initial BP neural network; step S2: optimizing the network structure and network parameters of the initial BP neural network by using an HGS algorithm to obtain a population individual vector with optimal fitness; and step S3: and optimizing the BP neural network structure and parameters by using the obtained individual vector with the optimal fitness to obtain an optimal BP neural network model, predicting the sewage data to be detected by using the optimal BP neural network model and obtaining a prediction result. The BP neural network based on the HGS is characterized by simple structure and stable performance, and compared with the traditional BP neural network, the BP neural network based on the HGS has lower requirement on data volume, and can realize high-precision prediction of four water quality indexes of COD, ammonia nitrogen, total phosphorus and total nitrogen in water under the condition of low data volume.
Description
Technical Field
The invention belongs to the field of sewage index big data prediction and analysis, and particularly relates to a BP neural network sewage index prediction method based on HGS.
Background
China strictly requires water quality indexes of inlet and outlet water of sewage, so that the real-time monitoring of key water quality indexes in the sewage treatment process is very important, and the current sewage treatment station in China mainly uses traditional water quality detection means such as manual detection equipment for partial water quality indexes, and the like, and is lack of good real-time performance. In addition, the traditional method does not have the long-term prediction capability on the water quality index in a future period of time.
The sewage treatment is a very complex nonlinear system, has the characteristics of large hysteresis, strong coupling and the like, is difficult to establish a reliable and effective prediction model through process mechanism analysis, and intelligent algorithms such as neural networks, machine learning and the like do not depend on the mechanism model, can actively learn through the existing data, have strong nonlinear approximation capability, and can be applied to the modeling prediction research of the sewage treatment system.
Back-propagation neural networks (BPNN), one of the most commonly used machine learning algorithms at present, has been widely used in a variety of fields. In the field of sewage index prediction, a BP neural network establishes a mapping relation between input and output by utilizing a large amount of data in a training mode, and multi-index real-time prediction can be effectively realized. However, the traditional BP neural network is prone to get into a problem of a local optimal solution in a training process, so that the trained network is not ideal in performance, and especially, when the training data amount is insufficient and the index prediction is too much, the error of the BP neural network is large.
The Hunger Game Search (HGS) is a novel intelligent optimization algorithm, which is designed according to animal Hunger driving activities and behaviors, has the characteristics of simple structure, strong optimizing capability, high convergence speed and the like, has special stability characteristics, has very competitive performance, and can more effectively solve the problems of constraint and non-constraint.
Disclosure of Invention
In view of the above, the invention aims to provide a BP neural network sewage index prediction method based on an HGS (hybrid gas sensor), which optimizes a BP neural network structure and parameters through a hunger game search algorithm (HGS), establishes an HGS-BPNN algorithm, and finds a better global optimal solution to obtain a BP neural network model with better prediction capability by virtue of the characteristics of easy convergence, high convergence precision and strong capability of escaping from a local optimal solution of the HGS, thereby realizing successful prediction of water quality indexes under the condition of low data volume.
In order to achieve the purpose, the technical scheme of the invention is realized as follows: a BP neural network sewage index prediction method based on HGS comprises the following steps:
step S1: constructing an initial BP neural network;
step S2: optimizing the network structure and network parameters of the initial BP neural network by utilizing an HGS algorithm to obtain a population individual vector with optimal fitness;
and step S3: and optimizing the BP neural network structure and parameters by using the obtained individual vector with the optimal fitness to obtain an optimal BP neural network model, predicting the sewage data to be detected by using the optimal BP neural network model and obtaining a prediction result.
Further, in order to meet the requirement of prediction of main indexes of sewage of a sewage treatment station, in the step S1, the initial neural network is set to be a 3-layer structure, wherein the number m of neurons in an input layer, the number n of neurons in an output layer, and the number S of neurons in a hidden layer are determined by a formula (1);
where the function floor () represents a floor.
Further, the step S2 specifically includes:
s2.1, initializing population individuals and initializing parameters; initializing population individuals comprises initializing population individual number N and population individual vector
The initialization parameters include: number of data samples K and maximum number of iterations Max iter Total hunger degree SHungry and initial hunger degree hungry;
s2.2, calculating the fitness of each population individual by using a fitness function; the fitness function RMSD is the deviation between the output of the neural network optimized by utilizing the corresponding population individual vector and actual data:
wherein K =1,2, \8230;, K, Y Pred (k) Is the output value of the neural network, Y Ture (k) Is the actual data value;
s2.3, sequencing the fitness of the population individuals obtained by calculation from small to large, wherein the individual fitness with the optimal fitness is bF, and the worst individual fitness is wF; if the bF is better than the global optimal fitness value BF, updating the BF to be the bF, and storing the individual vector as an individual optimal vector; if the wF is worse than the global worst fitness value WF, updating the WF to the wF; then, calculating the hunger degree hungry (i) of the population individuals according to formulas (3) to (5);
wherein i =1,2, \8230;, N, allFitness (i) represents the fitness value of each individual, and r 6 Is [0,1 ]]UB and LB represent the upper and lower limits of the search space, respectively; LH is the lower bound of H;
step S2.4, calculating the hunger degree weight of the individual population according to the formula (6) and the formula (7)
And
where SHungry represents the sum of hunger levels of all individuals, r 3 、r 4 、r 5 Is [0,1 ]]A random number in between; l is a set constant;
s2.5, calculating a new position of each individual according to formulas (8) to (10), and updating an individual vector;
E＝sech(|AllFitness(i)-BF|) (9)
wherein, the first and the second end of the pipe are connected with each other,
representing the current individual vector, randn representing the normal score satisfying the criteriaThe random number of the cloth, t is the current iteration number,
a globally optimal individual vector is represented and,
is between [ -a, a [ -a]Sech is a hyperbolic function,
a＝2*(1-t/Max iter )，r 1 、r 2 and rand are each [0,1 ]]A random number in between;
s2.6, judging whether the iteration frequency reaches the maximum iteration frequency, if so, ending the iteration, and calculating the fitness of the updated individual vector to obtain an individual vector with the optimal fitness; otherwise, returning to the step S2.2 to repeat iterative calculation.
Furthermore, all information to be optimized must be contained in the population individual vector, and data needing to be optimized in the invention is mainly divided into two types, one type is a network structure; the other is the optimal initial weight under the specified network structure, therefore, initializing the population individual vector in step S2.1 includes: defining population individual vectors
Randomly generating N population individual vectors; defining a population individual vector as a p-dimensional vector, wherein the first dimension represents the number of hidden layers of a neural network; the second dimension to the fourth dimension represent the quantity s E of each hidden layer neuron of the neural network [1,13]](ii) a The fourth dimension and beyond represents weights and offsets between layers.
Further, in step S2.2, the fitness function is a multi-input single-output function, and the function value is the degree of adaptability of each individual to the environment. The input of the fitness function of the HGS algorithm is the basic parameters of individual vector and BP neural network training, such as learning rate and maximum iteration times; the main body of the fitness function is a BP neural network defined by a network structure and initial parameters analyzed by individual vectors and basic parameters of the BP neural network; the output is the deviation RMSD of the output of the BP neural network optimized by the individual vectors from the actual data.
Further, in order to realize real-time processing of data, the invention also uses a timer, and the timer can scan the number of files in a data source folder (a folder for storing the data collected by the lower computer) at intervals. If the number of the files is increased, the lower computer sensor acquires new data, and then the newly acquired data is input into the trained neural network model for prediction. And finally, storing the predicted result in a local appointed folder according to a naming rule.
Compared with the prior art, the BP neural network sewage index prediction method based on the HGS has the following advantages:
(1) The sewage index prediction method disclosed by the invention is added with the advanced optimization algorithm HGS on the basis of the BP neural network, so that the self-adaptive optimization setting of a large number of parameters such as a hidden layer in the BP neural network algorithm is realized, the algorithm operation efficiency is improved, the global optimal solution can be found, a BP neural network model with better prediction capability is obtained, and the prediction precision is improved.
(2) Compared with the traditional BPNN method, the HGS-BPNN sewage index prediction method provided by the invention needs less training data, realizes sewage index prediction under the condition of low data volume, and realizes output of a plurality of sewage prediction indexes under the condition of ensuring precision, thereby being beneficial to reducing manual intervention in the sewage treatment process, reducing the operation cost of a sewage treatment station and improving the intelligence degree of the sewage treatment station.
(3) The prediction method can realize automatic scanning and judgment of newly acquired data by setting the timer, and input the newly acquired data into the neural network model for prediction and storage, thereby realizing real-time data processing and automatic sewage quality index prediction.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention, illustrate embodiments of the invention and together with the description serve to explain the invention and do not constitute a limitation of the invention.
FIG. 1 is a technical route diagram of a BP neural network sewage index prediction method based on HGS according to an embodiment of the present invention;
FIG. 2 is a flowchart of a BP neural network sewage index prediction method based on HGS according to an embodiment of the present invention;
FIG. 3 is a graph showing the convergence curves of two optimization algorithms, HGS-BPNN and PSO-BPNN;
FIG. 4 is a diagram showing the prediction results of two optimization algorithms, HGS-BPNN and PSO-BPNN; FIG. 4 (a) -FIG. 4 (d) HGS-BPNN prediction results, FIG. 4 (e) -FIG. 4 (h) PSO-BPNN prediction results.
Detailed Description
It should be noted that the embodiments and features of the embodiments of the present invention may be combined with each other without conflict. The invention will be described in detail below with reference to the drawings and embodiments.
The data used by the method is derived from actually measured data of a sewage treatment station, and four water quality indexes of COD (chemical oxygen demand), ammonia nitrogen, total phosphorus and total nitrogen are predicted through six water quality indexes of PH value, turbidity, conductivity, URP (Universal Serial bus), UV254 and UV280, so that the multi-target prediction of 6 input and 4 output is realized. As shown in fig. 1-2, the BP neural network sewage index prediction method based on the HGS, which is established by the present invention, is mainly composed of an HGS algorithm and a BPNN algorithm, wherein the HGS algorithm is used for searching for an optimal network structure and network parameters of the BP neural network, and the BPNN algorithm uses the network structure and parameters obtained by the HGS algorithm to perform model training, thereby obtaining an optimized neural network model, and finally uses the trained model to predict new data and store the prediction result. The method specifically comprises the following steps:
step S1: constructing an initial BP neural network;
in order to meet the requirement of main sewage index prediction of a sewage treatment station, the embodiment of the invention sets an initial BP neural network to be a 3-layer structure, the number m of neurons in an input layer of the neural network is 6, the number n of output neurons is 4, and the initial BP neural network and the output neurons do not participate in HGS optimization; the number of hidden layers (to-be-optimized amount) of the neural network is 1, and the number s of neurons of the hidden layers is determined by a formula (1);
where the function floor () represents a floor. Here, a =5, and the number of hidden layer neurons is 8.
Step S2: optimizing the network structure and network parameters of the initial BP neural network by using an HGS algorithm to obtain a population individual vector with optimal fitness;
and step S3: and optimizing the BP neural network structure and parameters by using the obtained individual vector with the optimal fitness to obtain an optimal BP neural network model, predicting the sewage data to be detected by using the optimal BP neural network model and obtaining a prediction result.
Further, the step S2 specifically includes:
s2.1, initializing population individuals and initializing parameters; initializing population individuals comprises initializing population individual quantity N and population individual vector
The initialization parameters include: number of data samples K and maximum number of iterations Max iter Total hunger degree SHungry and initial hunger degree hungry;
all information to be optimized must be contained in the population individual vector, the data to be optimized is mainly divided into two types, one type is a network structure; the other is the optimal initial weight under a given network structure, thus defining population individual vectors
A p-dimensional vector is adopted, and the first dimension represents the number of hidden layers of the neural network; the second dimension to the fourth dimension represent the quantity s E of each hidden layer neuron of the neural network [1,13]](ii) a Representing the weight and bias among all layers after the fourth dimension, and randomly generating N population individual vectors;
s2.2, calculating the fitness of each population individual by using a fitness function; the fitness function of the HGS algorithm is a multi-input single-output function, and the input is an individual vector and basic parameters of BP neural network training, such as a learning rate and the maximum iteration times; the function body is a BP neural network defined by a network structure and initial parameters analyzed by individual vectors and basic parameters of the BP neural network; the output is the deviation RMSD of the output of the BP neural network optimized by the individual vector from the actual data:
wherein K =1,2, \8230;, K, Y Pred (k) Is the output value of the neural network, Y Ture (k) Is the actual data value;
s2.3, sequencing the fitness of the population individuals obtained by calculation from small to large, wherein the individual fitness with the optimal fitness is bF, and the worst individual fitness is wF; if the bF is better than the global optimal fitness value BF, updating the BF to be the bF, and storing the individual vector as an individual optimal vector; if the wF is worse than the global worst fitness value WF, updating the WF to the wF; then, calculating the hunger degree hungry (i) of the population individuals according to formulas (3) to (5);
wherein i =1,2, \8230;, N, allFitness (i) represents the fitness value of each individual, and r 6 Is [0,1 ]]UB and LB represent the upper and lower limits of the search space, respectively; LH is the lower bound of H;
step S2.4 calculating population individuals according to formula (6) and formula (7)Hunger level weight
And
where SHungry represents the sum of hunger levels of all individuals, r 3 、r 4 、r 5 Is [0,1 ]]A random number in between; l is a set constant;
step S2.5, updating the individual vector according to formulas (8) to (10);
E＝sech(|AllFitness(i)-BF|) (9)
wherein the content of the first and second substances,
representing the current individual vector, randn representing a random number satisfying a standard normal distribution, t being the current number of iterations,
a globally optimal individual vector is represented and,
is between [ -a, a [ -a]Sech is a hyperbolic function,
a＝2*(1-t/Max iter )，r 1 、r 2 and rand are each [0,1 ]]A random number in between;
step S2.6, judging whether the iteration times reach the maximum iteration times, if so, ending the iteration, and calculating the fitness of the updated individual vector to obtain an individual vector with the optimal fitness; otherwise, returning to the step S2.2 to repeat iterative calculation.
In order to realize real-time processing of data, the invention also uses a timer which scans the number of files in a data source folder (a folder for storing data collected by the lower computer) at intervals. If the number of the files is increased, the lower computer sensor acquires new data, and then the newly acquired data is input into the trained neural network model for prediction. And finally, storing the predicted result in a local appointed folder according to a naming rule.
Fig. 3 shows performance analysis of the HGS optimization algorithm, and by comparing the convergence curves of the PSO optimization algorithm and the HGS algorithm, it can be seen that the HGS optimization speed is fast, and the output result of the fitness function obtained by the optimization is better than the result obtained by the PSO algorithm.
The predicted results are analyzed, and FIGS. 4 (a) - (d) are HGS-BP neural network predicted results; FIGS. 4 (e) - (h) are the predicted results of the PSO-BP neural network. The comparison of fig. 4 (a) and 4 (e) shows that both can predict the trend of the COD content in the sewage well, but the prediction result of the HGS-BPNN algorithm is closer to the actually measured COD content in the sewage; comparing fig. 4 (b) with fig. 4 (f), it can be seen that the ammonia nitrogen prediction result is closest to the real result at each sample point; FIG. 4 (c) is compared with FIG. 4 (g) to show that although the prediction accuracy of the total phosphorus content of the HGS-BPNN method at sample point 6 is lower than that of the PSO-BPNN method, the overall accuracy is better than that of the PSO-BPNN method; comparing fig. 4 (d) with fig. 4 (h), it is known that HGS-BPNN can better predict the trend of change of the total nitrogen content in wastewater, and the prediction accuracy is higher than that of the PSO-BPNN method.
According to the embodiment of the invention, the BP neural network is optimized by utilizing the HGS algorithm, compared with the traditional BP neural network, the BP neural network sewage quality index prediction method based on the HGS has lower demand on data volume, and can realize high-precision prediction of four water quality indexes of COD, ammonia nitrogen, total phosphorus and total nitrogen in water under the condition of low data volume.
It will be understood by those skilled in the art that all or part of the steps in the method for implementing the above embodiments may be implemented by relevant hardware instructed by a program, and the program may be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc.
While the invention has been described in connection with what is presently considered to be the most practical and preferred embodiment, it is to be understood that the invention is not to be limited to the disclosed embodiment, but on the contrary, is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.
Claims (8)
1. The BP neural network sewage index prediction method based on HGS is characterized in that: the method comprises the following steps:
step S1: constructing an initial BP neural network;
step S2: optimizing the network structure and network parameters of the initial BP neural network by using an HGS algorithm to obtain a population individual vector with optimal fitness;
and step S3: and optimizing the BP neural network structure and the network parameters by using the obtained population individual vector with the optimal fitness to obtain an optimal BP neural network model, predicting the sewage data to be detected by using the optimal BP neural network model and obtaining a prediction result.
2. The method according to claim 1, wherein the initial BP neural network is set to a 3-layer structure in step S1, wherein the input layer neuron number m, the output layer neuron number n, and the hidden layer neuron number S are determined by formula (1);
where the function floor () represents a floor.
3. The method of claim 2, wherein the number of input layer neurons m in the initial BP neural network is 6, the number of output layer neurons n is 4, q is 5, and the number of hidden layer neurons s is 8.
4. The method according to claim 1, wherein the step S2 specifically comprises:
s2.1, initializing population individuals and initializing parameters; initializing population individuals comprises initializing population individual quantity N and population individual vector
The initialization parameters include: number of data samples K, maximum number of iterations Max iter Total hunger degree SHungry and initial hunger degree hungry;
s2.2, calculating the fitness of each population individual by using a fitness function RMSD;
s2.3, sequencing the fitness of the population individuals obtained by calculation from small to large, wherein the individual fitness with the optimal fitness is bF, and the worst individual fitness is wF; if the bF is better than the global optimal fitness value BF, updating the BF to be bF, and storing the individual vector as an individual optimal vector; if the wF is worse than the global worst fitness value WF, updating the WF to the wF; then, calculating the hunger degree hungry (i) of the individual population according to a formula;
wherein i =1,2, \8230;, N, allFitness (i) represents the fitness value of each individual, and r 6 Is [0,1 ]]UB and LB represent the upper and lower limits of the search space, respectively; LH is the lower bound of H;
step S2.4 of calculating the hunger degree weight of individual population
And
where SHungry represents the sum of hunger of all individuals, r 3 、r 4 、r 5 Is [0,1 ]]A random number in between; l is a set constant;
step S2.5, updating individual vectors according to the hunger degree weight of the population individuals;
E＝sech(|AllFitness(i)-BF|) (9)
wherein, the first and the second end of the pipe are connected with each other,
representing the current individual vector, randn represents a random number satisfying a standard normal distribution, t is the current iteration number,
a globally optimal individual vector is represented and,
is between [ -a, a [ ]]Sech is a hyperbolic function,
a＝2*(1-t/Max iter )，r 1 、r 2 and rand are each [0,1 ]]A random number in between;
s2.6, judging whether the iteration frequency reaches the maximum iteration frequency, if so, ending the iteration, and calculating the fitness of the updated individual vector to obtain a population individual vector with the optimal fitness; otherwise, returning to the step S2.2 to repeat iterative calculation.
5. The method of claim 4, wherein initializing a population individual vector in step S2.1 comprises: defining population individual vectors
Randomly generating N population individual vectors;
defining a population individual vector as a p-dimensional vector, wherein the first dimension represents the quantity of hidden layers of a neural network; the second dimension to the fourth dimension represent the number s of the neurons of each hidden layer of the neural network, which belongs to [1,13]; the fourth dimension and beyond represents weights and offsets between layers.
6. The method according to claim 4, characterized in that in step S2.2, the fitness function RMSD is the deviation of the output of the neural network optimized with the corresponding population individual vector from the actual data:
wherein K =1,2, \8230;, K, Y Pred (k) Is the output value of the neural network, Y Ture (k) Is the actual data value.
7. The method of claim 1, further comprising a timer for scanning and obtaining the number of files in the data source folder at intervals; if the number of the files is increased, new data are collected, and the newly collected data are input into the optimized BP neural network for prediction.
8. The method according to claim 1, wherein the step S3 of predicting the sewage data to be measured comprises: four water quality indexes of COD, ammonia nitrogen, total phosphorus and total nitrogen are predicted according to six water quality indexes of PH value, turbidity, conductivity, URP, UV254 and UV280, and multi-target prediction of 6 input and 4 output is achieved.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211482168.1A CN115759437B (en) | 2022-11-24 | 2022-11-24 | BP neural network sewage index prediction method based on HGS |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211482168.1A CN115759437B (en) | 2022-11-24 | 2022-11-24 | BP neural network sewage index prediction method based on HGS |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115759437A true CN115759437A (en) | 2023-03-07 |
| CN115759437B CN115759437B (en) | 2024-03-01 |
Family
ID=85337621
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211482168.1A Active CN115759437B (en) | 2022-11-24 | 2022-11-24 | BP neural network sewage index prediction method based on HGS |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115759437B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113743011A (en) * | 2021-08-31 | 2021-12-03 | 华中科技大学 | Device life prediction method and system based on PSO-BP neural network |
| CN114037163A (en) * | 2021-11-10 | 2022-02-11 | 南京工业大学 | Sewage treatment effluent quality early warning method based on dynamic weight PSO (particle swarm optimization) optimization BP (Back propagation) neural network |
| US20220082545A1 (en) * | 2020-09-15 | 2022-03-17 | Beijing University Of Technology | Total Nitrogen Intelligent Detection Method Based on Multi-objective Optimized Fuzzy Neural Network |
| CN115271167A (en) * | 2022-06-23 | 2022-11-01 | 合肥工业大学 | BP neural network-based tire vulcanization quality RFV index prediction method and system |
-
2022
- 2022-11-24 CN CN202211482168.1A patent/CN115759437B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20220082545A1 (en) * | 2020-09-15 | 2022-03-17 | Beijing University Of Technology | Total Nitrogen Intelligent Detection Method Based on Multi-objective Optimized Fuzzy Neural Network |
| CN113743011A (en) * | 2021-08-31 | 2021-12-03 | 华中科技大学 | Device life prediction method and system based on PSO-BP neural network |
| CN114037163A (en) * | 2021-11-10 | 2022-02-11 | 南京工业大学 | Sewage treatment effluent quality early warning method based on dynamic weight PSO (particle swarm optimization) optimization BP (Back propagation) neural network |
| CN115271167A (en) * | 2022-06-23 | 2022-11-01 | 合肥工业大学 | BP neural network-based tire vulcanization quality RFV index prediction method and system |
Non-Patent Citations (2)
| Title |
|---|
| YANG YUTAO等: "Hunger games search: Visions, conception, implementation, deep analysis, perspectives, and towards performance shifts", EXPERT SYSTEMS WITH APPLICATIONS, vol. 177 * |
| 江敏;钱磊;苏学满;: "基于PSO优化的灰色神经网络预测算法的研究", 宁夏师范学院学报, no. 03, pages 54 - 60 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115759437B (en) | 2024-03-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Inferring gene regulatory networks using the improved Markov blanket discovery algorithm | |
| CN110110862A (en) | A kind of hyperparameter optimization method based on adaptability model | |
| CN110032755B (en) | Multi-objective optimization method for urban sewage treatment process under multiple working conditions | |
| CN116542382A (en) | Sewage treatment dissolved oxygen concentration prediction method based on mixed optimization algorithm | |
| CN113611356B (en) | Drug relocation prediction method based on self-supervision graph representation learning | |
| CN112131402A (en) | PPI knowledge graph representation learning method based on protein family clustering | |
| CN111914094A (en) | Knowledge graph representation learning method based on ternary interaction | |
| CN115329096A (en) | Interactive knowledge tracking method based on graph neural network | |
| CN112765902A (en) | RBF neural network soft measurement modeling method based on TentFWA-GD and application thereof | |
| CN115391385A (en) | Database query optimization method based on ant colony genetic dynamic fusion algorithm | |
| CN111832817A (en) | Small world echo state network time sequence prediction method based on MCP penalty function | |
| CN102521654A (en) | Supercritical water oxidation reaction kinetic model parameter estimation method employing RNA (Ribonucleic Acid) genetic algorithm | |
| CN101893852A (en) | Multi-target modeling method for complex industrial process | |
| CN115759437B (en) | BP neural network sewage index prediction method based on HGS | |
| CN111797363A (en) | Self-learning trajectory tracking method of sewage treatment system based on data | |
| CN110705704A (en) | Neural network self-organizing genetic evolution algorithm based on correlation analysis | |
| CN116662893A (en) | Water quality prediction method for optimizing SVM (support vector machine) based on improved goblet sea squirt algorithm | |
| CN111310974A (en) | Short-term water demand prediction method based on GA-ELM | |
| CN113704570B (en) | Large-scale complex network community detection method based on self-supervision learning type evolution | |
| Li et al. | Integrating expert knowledge for Bayesian network structure learning based on intuitionistic fuzzy set and Genetic Algorithm | |
| Pachuau et al. | Segmentation of composite signal into harmonic Fourier expansion using genetic algorithm | |
| CN114357869A (en) | Multi-objective optimization agent model design method and system based on data relation learning and prediction | |
| Furze et al. | Mathematical methods to quantify and characterise the primary elements of trophic systems | |
| CN116844645B (en) | Gene regulation network inference method based on multi-view layered hypergraph | |
| CN114841472B (en) | GWO optimization Elman power load prediction method based on DNA hairpin variation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: No.29 Zhujiang Road, Hexi District, Tianjin Applicant after: Tianjin Water Conservancy Engineering Group Co.,Ltd. Address before: No.29 Zhujiang Road, Hexi District, Tianjin Applicant before: TIANJIN MUNICIPAL WATER LIMITED CONSERVANCY |
|
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |