CN116822753A - Electric leaching repair sediment prediction optimization method and system based on neural network - Google Patents

Abstract

The invention discloses a neural network-based prediction and optimization method and system for electric leaching sediment repair, which belongs to the technical field of lead and cadmium contaminated sediment repair. The method includes the following steps: conduct electric leaching repair experiments on lead-cadmium contaminated sediments under various influencing factors to obtain the actual lead removal rate and actual cadmium removal rate under different influencing factors; construct a BP neural network model to use multiple Different influencing factors are used as input parameters of the BP neural network model, and the actual lead removal rate and actual cadmium removal rate are used as the output parameters of the BP neural network model to train the BP neural network model; the trained BP neural network model is used to be repaired The lead removal rate and cadmium removal rate of contaminated sediment were predicted. Through the present invention, the prediction results of electric leaching repair sediment under various influencing factors can be obtained.

Description

Neural network-based prediction and optimization method and system for electric leaching sediment remediation

Technical field

The present invention relates to the technical field of lead and cadmium contaminated sediment remediation technology, and more specifically to a neural network-based prediction and optimization method and system for electrolytic leaching remediation of sediment remediation.

Background technique

Electric remediation technology is often used in the remediation work of low-permeability clay and silt soil. By applying a DC electric field on both sides of the heavy metal-contaminated soil, the heavy metal ions in the soil are caused to migrate through electromigration, electroosmotic flow or electrophoresis under the action of the electric field. Migrate to two levels to achieve the purpose of removing heavy metals from the soil.

Chemical leaching is one of the common methods to remediate heavy metals in sediments or soil. This technology can quickly remove heavy metals from sediments or soil and complete the remediation work in a short time.

However, in the process of combined electric leaching and remediation of heavy metals (lead and chromium) in the sediment, various factors such as the type of agent, combination ratio, combination concentration, and remediation time all have an impact on the final heavy metal removal rate.

Therefore, how to provide a method to predict the removal rate of heavy metals under a variety of different influencing factors is an urgent problem that those skilled in the art need to solve.

Contents of the invention

In view of this, the present invention provides a method and system for predicting and optimizing electric leaching repair sediments based on neural networks, which are used to solve at least some of the technical problems existing in the background technology.

In order to achieve the above objects, the present invention adopts the following technical solutions:

The present invention first discloses a neural network-based electric leaching repair sediment prediction and optimization method. The method includes the following steps:

Sample data acquisition:

Conduct electric leaching repair experiments on lead-cadmium contaminated sediment under various influencing factors, and obtain the actual lead removal rate and actual cadmium removal rate of lead-cadmium contaminated sediment under different influencing factors;

Build and train a BP neural network:

Construct a BP neural network model, use the various influencing factors as input parameters of the BP neural network model, use the obtained actual lead removal rate and actual cadmium removal rate as the output parameters of the BP neural network model, and train the BP neural network model;

Heavy metal removal rate prediction:

The trained BP neural network model was used to predict the lead removal rate and cadmium removal rate of the contaminated sediment to be remediated.

Further, the above method also includes using a genetic algorithm to solve the optimal value of the lead and chromium removal rate, and obtain the corresponding influencing factors under the highest lead removal rate and/or cadmium removal rate.

Further, in the sample data acquisition step, the lead-cadmium contaminated sediment is obtained by the following method:

The lead-cadmium contaminated river sediment was collected on site, and the collected river sediment was air-dried, ground and screened in sequence, and then water was added to prepare lead-cadmium contaminated sediment.

Further, in the sample data acquisition step, the various influencing factors include the components of the elution agent, the concentrations of different components of the elution agent, the proportions of different components of the elution agent, and the elution repair time.

Further, the ingredients of the elution agent include a mixture of tetrasodium glutamic acid diacetate and sodium chloride, a mixture of citric acid and sodium chloride, and a mixture of tetrasodium glutamic acid diacetate, citric acid and chlorine. A mixture of sodium chloride.

Further, in the step of constructing and training a BP neural network, the BP neural network model includes: an input layer, an intermediate hidden layer and an output layer connected in sequence, where the number of nodes in the input layer is four, and the intermediate hidden layer The number of nodes in the containing layer is ten, and the number of nodes in the output layer is two; the activation function of the intermediate hidden layer includes a relu function; and the activation function of the output layer includes a sigmoid function.

Further, in the step of constructing and training a BP neural network, training the BP neural network model specifically includes the following steps:

Normalize the input parameters of the BP neural network model;

Use the normalized input parameters as the data set of the BP neural network model;

The ten-fold cross-validation method was used to divide the data set into ten parts, and nine parts were used as training data and one part as test data in turn for training and verification.

Further, the input parameters of the BP neural network model are normalized, which specifically includes the following steps:

Calculate the average value of each input parameter;

Calculate the standard deviation corresponding to the input parameter based on the average value of the input parameter;

The normalized processing results of the input parameters are obtained based on the average value and standard deviation of the input parameters.

Further, the normalized processing result of the input parameters is obtained based on the average value and standard deviation of the input parameters, which specifically includes the following formula:

.

In the formula, Indicates the normalized processing result of the i-th influence value in the n-th influencing factor;/> Represents the i-th influence value among the n-th influencing factor;/> Represents the average value of the i-th influence value,/> Represents the standard deviation of the i-th influence value.

The invention also discloses a neural network-based electric leaching and repair sediment prediction and optimization system, which includes a computer system. When the computer system is running, it can realize any neural network-based electric leaching and repair sediment prediction and optimization of the present invention. method.

It can be seen from the above technical solutions that compared with the existing technology, the present invention provides a neural network-based electric leaching repair sediment prediction and optimization method and system, which has the following beneficial effects:

The present invention uses BP neural network to predict the lead and chromium removal rate of contaminated sediment, taking into account various different influencing factors of electric leaching, thereby improving the accuracy of predicting the removal rate of lead and cadmium by electric leaching in the actual application process.

The present invention uses a genetic algorithm to solve the optimal values of factors affecting the lead and chromium removal rate, which can effectively improve the stability and accuracy of the prediction of the removal rate of lead and cadmium in electric elution during actual application.

Description of the drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are only These are embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on the provided drawings without exerting creative efforts.

Figure 1 is a flow chart of the optimization method of the present invention based on the neural network electric-leaching joint remediation of lead and cadmium contaminated sediment.

Figure 2 is a schematic diagram of the prediction model of the electrodynamic-leaching joint remediation of lead and cadmium contaminated sediment based on neural network according to the present invention.

Figure 3 is a flow chart of the neural network model training of the present invention.

Figure 4 is a prediction diagram of the removal rate of heavy metal lead (Pb) by the BP network after training according to the embodiment of the present invention.

Figure 5 is a prediction diagram of the removal rate of heavy metal cadmium (Cd) by the BP network after training according to the embodiment of the present invention.

Figure 6 is a diagram illustrating optimization results of heavy metal lead (Pb) removal rate prediction results using a genetic algorithm according to an embodiment of the present invention.

Figure 7 is a diagram illustrating optimization results of heavy metal cadmium (Cd) removal rate prediction results using a genetic algorithm according to an embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

Embodiment 1

An optimization method for electrodynamic-leaching joint remediation of lead and cadmium contaminated sediment based on neural network is provided, as shown in Figure 1, which includes the following steps:

S0: Collect lead-cadmium contaminated river sediment on the spot, air-dry, grind and sieve, then add water to prepare lead-cadmium contaminated sediment;

S1: Conduct electric leaching repair experiments on the obtained lead-cadmium contaminated sediment under various influencing factors, and obtain the actual lead removal rate and actual cadmium removal rate of the lead-cadmium contaminated sediment under different influencing factors.

In this example, the components of the leaching agent, the concentration of different components of the leaching agent, the proportions of different components of the leaching agent, and the leaching repair time are selected as influencing factors to conduct an electrodynamic-leaching multi-factor experiment on lead and cadmium contaminated sediment. , obtain the actual lead and cadmium removal rate after the repair is completed, and accumulate multiple sets of sample data;

S2: Construct a BP neural network model, and determine the structural parameters of the BP neural network, based on a variety of different influencing factors, including the components of the eluent agent, the concentrations of different components of the eluent agent, the proportions of different components of the eluent agent, and the elution agent. The cleaning and repair time is used as the input parameter of the BP neural network, and the actual lead removal rate and the actual cadmium removal rate are used as the output parameters.

In this embodiment, the constructed BP neural network model is shown in Figure 2, including an input layer, an intermediate hidden layer, and an output layer connected in sequence. The number of nodes in the input layer is four, and the nodes in the intermediate hidden layer are four. The number is ten, and the number of nodes in the output layer is two.

This step also includes using multiple sets of sample data to train the BP neural network model. First, the input parameters of the BP neural network are normalized and pre-processed, and then the normalized input parameters are used as the BP neural network model. data set.

The data set is divided into ten parts using the ten-fold cross-validation method, and nine parts are used as training data and one part as test data in turn for training and verification. Specifically, the BP neural network is trained with the data of the training group, and the BP neural network is trained with the data of the verification group. The data is used to test the trained BP neural network model to verify the accuracy of the neural network model and adjust the BP neural network network. Specifically, the network initial parameters are used to calculate the network prediction output, and the mean square error is obtained by comparing it with the actual output. Then the network parameters are adjusted backwards through the gradient descent criterion.

In this embodiment, the normalization process of the input parameters adopts the zero mean variance normalization method. Specifically: in this embodiment, the input parameter data sample is a sample matrix composed of different specific influence values of multiple different influencing factors, where A set of sample matrices is shown below,

Among them/> Represents the i-th influence value in the n-th influencing factor.

First calculate the average of the different influence values of each factor in multiple sets of sample data. The specific calculation formula is as follows: , where m represents m groups of sample data, k=1,2,3…m. Then calculate the standard deviation of the influence value data of each factor. The calculation formula is as follows:/> .

Finally, zero-mean normalization is performed on each influence value data of each factor. The specific formula is as follows: , in the formula,/> Indicates the normalized processing result of the i-th influence value in the n-th influencing factor;/> Represents the i-th influence value among the n-th influencing factor;/> Represents the average value of the i-th influence value,/> Represents the standard deviation of the i-th influence value.

S3: Use the trained BP neural network model to predict the removal efficiency of heavy metal lead and heavy metal cadmium in the contaminated sediment to be remediated.

As a preferred embodiment, the above method also includes using a genetic algorithm to solve the optimal value of the trained BP neural network simulation model to obtain the highest heavy metal (lead/chromium) removal efficiency corresponding to the electric-leaching combined process. experimental conditions.

The method steps and principles of the present invention will be further elaborated below in conjunction with specific data.

The lead (Pb) and cadmium (Cd) contaminated sediment obtained in step S0 is pretreated and used as special soil for remediation experiments.

In order to ensure the lead and chromium content in the contaminated sediment during the experimental verification process, the lead (Pb) and cadmium (Cd) content in the soil to be repaired for the experiment can be measured after sampling and digestion. If the lead and cadmium content does not meet the set threshold, the Lead and chromium content can be artificially added to the soil.

Then sample data is obtained. In this step, the combination method, concentration and proportion of different elution combination agents can be controlled to prepare 300ml of eluent and 500g of special soil to be repaired for the experiment to prepare lead and cadmium contaminated sediment with a moisture content of 60%. Mix and set aside for more than 24 hours before use. Prepare another 1000ml of the same electrolyte as the eluent and install it in the electrolyte bottles on both sides of the electrode chamber.

Place 480g of fully mixed sediment in the sediment chamber. EKG electrodes are installed on both sides of the electrode chamber and connected to the power supply through wires. The yin and yang two-stage peristaltic pump realizes real-time circulation of electrolyte.

When running for 24h, 48h, and 72h respectively, the sediment chamber was divided into five areas A1-A5 according to the position from the anode electrode chamber, and sampling was carried out in each zone. After digestion, the lead (Pb) and cadmium (Cd) in the repaired sediment were measured respectively. ) content, average the heavy metal content in the five zones, and define this value as the content of lead (Pb) and cadmium (Cd) in the sediment after restoration.

The actual values of lead and cadmium removal rates were obtained through the contents of lead (Pb) and cadmium (Cd) in the remediation sediment and the contents of lead (Pb) and cadmium (Cd) in the experimental contaminated soil.

In the embodiment of the present invention, the ingredients of the elution agent are any one or a mixture of two of tetrasodium glutamic acid diacetate GLDA and citric acid CA. Citric acid CA is a small molecule organic acid that can reduce soil environmental pollution. The pH value enables heavy metals to be desorbed from the surface of particles in the soil in a short time. Due to its advantages of low price and environmental friendliness; tetrasodium glutamate diacetate (GLDA) is a biodegradable agent with better biodegradability. Chelating agents can form complexes with heavy metal ions to desorb heavy metals from the surface of sediment particles, thereby enhancing the mobility of metals in electric repairs; in addition, in the embodiment of the present invention, sodium chloride (NaCl) is added to HIA to increase the conductivity of the elution agent. Rate. The specific components of the elution agent, the concentration of different components of the elution agent, the proportions of different components of the elution agent, and the elution repair time are shown in Table 1.

Table 1 Combination drug experimental design

Based on a large amount of accumulated experimental data, a BP network model was established to refine the input and output of the prediction method for the electrodynamic-leaching combined remediation of lead and cadmium contaminated sediments, and introduce the combination method, concentration, ratio, and repair of the leaching combination agent. The four parameters of time are used as the system input, and the predicted values of the two factors of lead (Pb) removal rate and cadmium (Cd) removal rate are introduced as the system output.

BP neural network is an artificial neural network based on the BP algorithm, which uses the BP algorithm to adjust weights and thresholds. The BP neural network consists of an input layer, an intermediate hidden layer and an output layer. The present invention contains four input signals, so the number of nodes in the input layer is four; the hidden layer is provided with ten nodes, that is, the intermediate hidden layer Contains ten neurons; contains two output signals, so the number of nodes in the output layer is set to two.

The training process of BP neural network is as follows. First, the sample data of all input parameters are normalized. The purpose of normalization is to eliminate dimensions and avoid some unimportant numerical problems. It can speed up convergence, avoid neuron saturation, and ensure that small values in the data are not swallowed up.

After normalization, the multiple sets of data were divided into a training group and a validation group using 10-fold cross-validation. The ratio between the two was 9:1, and the average was repeated 10 times. Among them, as shown in Figure 2, the BP neural network model includes a 3-layer structure:

Input layer, hidden layer and output layer

The number of input layer nodes is 4, the number of hidden layer nodes is 10, the number of input layer nodes is 2, the linear relu function is the activation function of the intermediate hidden layer neurons, and the S-type sigmoid function is the activation function of the output layer neurons. function.

As shown in Figure 3, use the data in the training group to train the BP neural network model, and adjust the network parameters of the BP neural network model including:

Use the neural network toolbox in MATLAB to train the network:

Set the model to a neural network with a 3-layer structure;

Set the number of input layer nodes to 4, the number of intermediate hidden layer nodes to 10, and the number of output layer nodes to 2;

Set the activation functions of the middle hidden layer and output layer of the network to relu and sigmoid functions respectively;

Set the network training function to trainbfg and the network performance function to mean square error MSE;

Set network parameters, network iteration number epochs, expected error goal, and learning rate lr;

After setting the parameters, bring in the training set data to start training the neural network.

As shown in Figure 3, after the network model training is completed, the verification group data is brought in to verify the accuracy of the neural network model. If the expected error of electrodynamic-leaching removal rate prediction model verification is within 5%, then the neural network model can closely approximate the real model and achieve the prediction of output heavy metal removal rate.

As shown in Figure 4, the fitting error of the successfully trained BP neural network for the removal rate of heavy metal lead (Pb) is 0.00096641, which is less than 0.05. The network model can predict the heavy metal lead (Pb).

As shown in Figure 5, the fitting error of the successfully trained BP neural network for the heavy metal lead (Pb) removal rate is 0.00023, which is less than 0.05. The network model can predict the heavy metal cadmium (Cd).

The above-mentioned embodiment of the present invention also includes solving the optimal value of the lead and chromium removal rate through a genetic algorithm to obtain the experimental conditions corresponding to the electrodynamic-leaching combined process under the highest heavy metal (lead/chromium) removal efficiency.

In this embodiment, the process of solving the optimal value through genetic algorithm is as follows:

Step S001: Initialize various parameters of the genetic algorithm, including evolutionary algebra, population size, crossover probability and mutation probability;

Step S002: Use the values of various influencing factors that affect lead and chromium removal efficiency (mainly including the components of the leaching agent, the concentrations of different components of the leaching agent, the proportions of different components of the leaching agent, and the leaching repair time) as genetic The chromosome encoding string individuals of the algorithm randomly generate a certain number of individuals within the value range of each influencing factor to form the first generation population;

Step S003: Calculate the fitness value of each individual in the first generation population according to the fitness function;

Step S004: Based on the roulette selection algorithm, select according to the fitness value of each individual. The smaller the fitness value, the closer the self-calculated results of each different influencing factors are to the standard calculation results, and the greater the probability of the corresponding individual being selected. ;

Step S005: Genetically evolve the selected outstanding individuals according to the set crossover probability and mutation probability to generate the next generation population;

Step S006: Repeat steps S003 to S005 until the set evolutionary generations are reached. The individual with the smallest fitness value in the last generation population is the optimal solution for the values of each influencing factor that affects lead and chromium removal efficiency;

Step S007: According to the obtained optimal solution of the values of each influencing factor, obtain the optimal parameters affecting the lead and chromium removal efficiency.

This invention establishes an optimization method for electrodynamic-leaching joint repair of lead-cadmium contaminated sediment based on neural networks. In the early stage, sample data are obtained by controlling different experimental variables to conduct large-scale experiments, and in the later stage, the accumulated large amounts of experimental data are used to train the neural network. Predict the removal efficiency of heavy metal lead and cadmium corresponding to other different experimental conditions, use genetic algorithm to solve the optimal value, and find the experimental conditions with the highest removal rate under this experimental system.

In specific applications, MATLAB's genetic algorithm toolbox can also be used to solve the optimal value of lead and chromium removal efficiency.

As shown in Figure 6, the genetic algorithm toolbox of MATLAB was used to optimize the heavy metal lead (Pb) removal rate prediction model, and the conditions corresponding to the optimal value of 80.098% were obtained as [7, 0.13402, 72], that is, the elution agent When the combination is CA+GLDA+NaCl, the concentration is 0.13402 mol/L, and the repair time is 72 h, the removal rate of heavy metal lead (Pb) under this repair system is optimal - 80.098%.

As shown in Figure 7, the genetic algorithm toolbox of MATLAB was used to optimize the heavy metal cadmium (Cd) removal rate prediction model, and the conditions corresponding to the optimal value of 87.098% were obtained as [8,0.1707,72], that is, the elution agent When the combination is 2CA+GLDA+NaCl, the concentration is 0.13402 mol/L, and the repair time is 72 h, the removal rate of heavy metal cadmium (Cd) under this repair system is optimal - 87.082%.

Embodiment 2

The present invention also discloses a computer system, which may include a computer program for implementing any neural network-based prediction and optimization method of electric leaching repair sediment sludge of the present invention.

Each embodiment in this specification is described in a progressive manner. Each embodiment focuses on its differences from other embodiments. The same and similar parts between the various embodiments can be referred to each other. As for the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple. For relevant details, please refer to the description in the method section.

The above description of the disclosed embodiments enables those skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be practiced in other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (10)

1. The electric leaching repair sediment prediction optimization method based on the neural network is characterized by comprising the following steps of:

sample data acquisition:

carrying out electric leaching repair experiments on the lead-cadmium polluted bottom mud under various different influence factors to obtain the actual lead removal rate and the actual cadmium removal rate of the lead-cadmium polluted bottom mud under the different influence factors;

building and training a BP neural network:

constructing a BP neural network model, taking the various different influencing factors as input parameters of the BP neural network model, taking the obtained actual lead removal rate and the obtained actual cadmium removal rate as output parameters of the BP neural network model, and training the BP neural network model;

and (3) predicting the heavy metal removal rate:

and predicting the lead removal rate and the cadmium removal rate of the polluted bottom mud to be repaired by using the trained BP neural network model.

2. The neural network-based electric leaching repair sediment prediction optimization method is characterized by further comprising the step of solving an optimal value of lead-chromium removal rate by utilizing a genetic algorithm to obtain a corresponding influence factor under the highest lead removal rate and/or cadmium removal rate.

3. The neural network-based electric leaching repair sediment prediction optimization method according to claim 1, wherein in the sample data acquisition step, the lead-cadmium polluted sediment is acquired by the following method:

and (3) collecting the lead-cadmium polluted river sediment in the field, and sequentially carrying out air drying, grinding and sieving treatment on the collected river sediment, and adding water to prepare the lead-cadmium polluted river sediment.

4. The neural network-based electric leaching repair sediment prediction optimization method according to claim 1, wherein in the sample data acquisition step, the plurality of different influencing factors include components of leaching agent, different component concentrations of leaching agent, different component proportions of leaching agent, and leaching repair time.

5. The neural network-based electric leaching repair sediment prediction optimization method according to claim 4, wherein the leaching agent comprises any one or two of tetra sodium glutamate diacetate and citric acid, a mixed component of tetra sodium glutamate diacetate and sodium chloride, a mixed component of citric acid and sodium chloride, and a mixed component of tetra sodium glutamate diacetate, citric acid and sodium chloride.

6. The neural network-based electric leaching repair sediment prediction optimization method according to claim 1, wherein the BP neural network model comprises: the device comprises an input layer, a middle hidden layer and an output layer which are sequentially connected, wherein the number of nodes of the input layer is four, the number of nodes of the middle hidden layer is ten, and the number of nodes of the output layer is two; the activation function of the intermediate hidden layer comprises a relu function; the activation function of the output layer comprises a sigmoid function.

7. The neural network-based electric leaching repair sediment prediction optimization method according to claim 1, wherein in the step of constructing and training a BP neural network, the BP neural network model is trained, and specifically comprises the following steps:

normalizing the input parameters of the BP neural network model;

taking the input parameters after normalization processing as a data set of the BP neural network model;

and dividing the data set into ten parts by adopting a ten-fold cross validation method, taking nine parts of the data set as training data and one part of the data set as test data in turn, and performing training validation.

8. The neural network-based electric leaching repair sediment prediction optimization method is characterized by carrying out normalization processing on input parameters of a BP neural network model, and specifically comprises the following steps:

calculating an average value of each input parameter;

calculating a standard deviation corresponding to the input parameter according to the average value of the input parameter;

and obtaining a normalization processing result of the input parameters according to the average value and the standard deviation of the input parameters.

9. The neural network-based electric leaching repair sediment prediction optimization method is characterized by obtaining a normalization processing result of input parameters according to the average value and standard deviation of the input parameters, and specifically comprises the following formulas:

；

in the method, in the process of the invention,representing the normalization processing result of the ith influence value in the nth influence factors; />Representing an ith influence value in the nth influence factors; />Mean value representing the ith influence value, +.>The standard deviation of the ith effect value is indicated.

10. The electric leaching repair sediment prediction optimization system based on the neural network is characterized by comprising a computer system, wherein the computer system can realize the electric leaching repair sediment prediction optimization method based on the neural network according to any one of claims 1-9 when in operation.