Anaerobic fermentation process soft measurement modeling method based on EWOA-LSSVM

Abstract

The invention discloses an anaerobic fermentation soft measurement modeling method based on EWOA-LSSVM. The method is used for accurately predicting the methane concentration which is difficult to directly measure in the anaerobic fermentation process. Comprising the following steps: 1) performing normalization processing after data acquisition, and eliminating the influence of units and dimensions of different variables; 2) initializing the population number, the inertia weight and the number of iterations of the whale optimization algorithm (EWOA); 3) establishing an EWOA-LSSVM soft measurement model for an anaerobic fermentation process, and optimizing parameters of the LSSVM model by using a whale optimization algorithm; and 4) substituting the obtained parameter optimal value into the LSSVM, extracting the characteristics of the anaerobic fermentation process data by using the LSSVM, and carrying out modeling. The modeling method provided by the invention has the advantages of good estimation performance, high accuracy and strong generalization ability, and is also suitable for soft measurement modeling in other complex chemical reaction processes.

Description

Technical Field

[0001] The present invention relates to the field of soft sensing in the anaerobic fermentation process, and more specifically, to a soft sensing modeling method for the anaerobic fermentation process based on WOA-LSSVM. Background Art

[0002] Energy shortage and environmental pollution caused by the combustion of fossil fuels have become global problems, and the development and utilization of renewable clean energy are imminent. As a large agricultural country, China is rich in straw resources, and the potential of anaerobic digestion (AD) of straw lignocellulosic biomass into bio-methane is huge. This is an environmentally friendly alternative energy with high calorific value and combustion efficiency. This process is crucial for reducing straw burning, reducing carbon emissions, and solving the energy shortage problem.

[0003] The AD process includes four interrelated stages: hydrolysis, acidification, acetogenesis, and methanogenesis, and each stage requires the complex cooperation of various microorganisms. The biological process involves a series of reactions. Therefore, the performance of AD is affected by operating conditions, microbial types, and quantities. These factors are extremely complex and non-linear. Therefore, it is difficult to monitor the AD process.

[0004] With the development of computational algorithms and the accessibility of computing power, machine learning (ML) has become a new data mining technology and modeling tool and has been applied to the prediction of the AD process. It can achieve output prediction based on the potential interaction between input and output variables. Predicting AD performance through ML does not require knowledge of the process mechanism. Therefore, it is an effective method for predicting biogas production. Currently, various ML algorithms, including artificial neural networks (ANN), adaptive neuro-fuzzy inference systems (ANFIS), and random forests (RF), have been used to simulate the complex non-linear relationships of AD progress.

[0005] Previous studies have demonstrated the positive role of machine learning (ML) in predicting and monitoring the anaerobic fermentation (AD) process, but there are still deficiencies. First, existing research mainly focuses on the performance evaluation of the ML model itself or the performance comparison between different models, while less attention is paid to how to further improve the prediction ability of the ML model through optimization algorithms. This results in limited optimization of the model in practical applications and difficulty in fully exploiting its prediction potential. Second, most current studies only validate the performance of the ML model based on a single or specific dataset, and this approach may limit the generalization ability of the model under different scenarios and conditions. Finally, most studies only focus on the prediction of biogas production in the AD process, while ignoring the importance of building a comprehensive early warning and monitoring system for the operation of the AD system, which to a certain extent limits the comprehensive application potential of ML technology in AD process management.

[0006] To address these deficiencies, the present invention proposes an innovative solution, namely, an LSSVM model combined with an improved enhanced whale optimization algorithm (EWOA). As an efficient optimization algorithm, EWOA can significantly improve the parameter optimization effect of the ML model, thereby enhancing the prediction accuracy and robustness of the model. The present invention not only focuses on the prediction of biogas production but also endeavors to construct a comprehensive early warning and monitoring system for the operation of the AD system. This system can real-time monitor the key parameters during the AD process, promptly warn of abnormal states, and comprehensively evaluate the performance of the system, providing more comprehensive and accurate technical support for the efficient, safe, and intelligent management of the AD process. Summary of the Invention

[0007] The objective of the present invention is to overcome the deficiencies of the prior art and provide a soft-sensing modeling method for the anaerobic fermentation process based on EWOA-LSSVM.

[0008] A soft-sensing modeling method for the anaerobic fermentation process based on EWOA-LSSVM according to the present invention comprises the following steps:

[0009] 1) Obtain sampling data of the anaerobic fermentation process with a time-tagged sequence through on-site operation or experiments;

[0010] 2) Divide the sample data of the extracted basic feature variables into a training set and a test set, and perform normalization processing to eliminate the influence of the dimensions of different feature variables;

[0011] 3) Set the initial parameters of the improved whale optimization algorithm (EWOA), namely, the number of iterations and the population size, and use the pooling mechanism to increase the diversity of the population; use the migration search strategy and the preferred search strategy to enhance the local and global search capabilities;

[0012] 4) Construct a least squares support vector machine (LSSVM) soft-sensing model for the anaerobic fermentation process, and use the objective function that minimizes an expression containing two parts, namely, the training error and the regularization loss, as the fitness function; the data loss part is used to evaluate the prediction error of the model on the training data through the root mean square error (RMSE); use EWOA to search for the optimal parameters of the LSSVM model, and these parameters include the kernel function parameter σ and the regularization parameter C;

[0013] 5) Substitute the obtained optimal parameter set into the LSSVM to form a high-precision soft-sensing model EWOA-LSSVM for the anaerobic fermentation process, and predict the methane concentration and biogas production.

[0014] The LSSVM in the soft-sensing modeling method for the anaerobic fermentation process based on EWOA-LSSVM changes the inequality constraint condition in the traditional SVM into an equality constraint condition.

[0015] Assume that for l sample data (x i , y i ), the corresponding category y i ∈ (-1, 1). A regression function is established through the following formula:

[0016] y = ω·φ(x) + b (1)

[0017] where ω is the normal vector of the hyperplane; b is the bias; x i is the input; y i is the output; φ(x) is the mapping function.

[0018] The optimization mathematical model of LSSVM is as follows:

[0019]

[0020] where e i is the error; γ represents the regularization parameter.

[0021] Introduce the Lagrange multiplier α i to Equation (2) to establish the Lagrange equation and transform the optimization problem into a minimum problem. The formula is as follows:

[0022]

[0023] At this time, the regression expression of LSSVN is as follows:

[0024]

[0025] where K(x i , x j ) is the radial basis kernel function; σ is the parameter of the radial basis kernel function.

[0026] In the soft sensor modeling method for anaerobic fermentation process based on EWOA-LSSVM, the pooling mechanism crosses and saves the worst solution and the promising solution of each iteration to the matrix Pool through the following formula to increase the diversity of the population:

[0027]

[0028] where is the i-th solution of the iteration in the matrix Pool; is the binary random vector; is 's inverse vector; is the random position near the best humpback whale; is the worst solution obtained after Δ iterations of the current iteration.

[0029] In the described soft sensor modeling method for anaerobic fermentation process based on EWOA-LSSVM, the enhanced whale optimization algorithm (EWOA) uses a migration search strategy to randomly separate a part of the humpback whales through the following formula to cover the unvisited areas and increase the diversity of the population, reducing the local optimal solution:

[0030]

[0031] Where is the position of the i-th humpback whale at t + 1; is a random position within the search space range; is a random position near the best humpback whale; rand is a uniformly distributed random number between 0 and 1; δ max and δ min are the upper and lower bounds of the problem; δ best_max and δ best_min are the upper and lower bounds of the best humpback whale.

[0032] In the described soft sensor modeling method for anaerobic fermentation process based on EWOA-LSSVM, the enhanced whale optimization algorithm (EWOA) uses a preferential selection strategy to disperse the whales in different regions of the search space through the following formula to discover diverse solutions and improve the search ability of EWOA in the prey search method:

[0033]

[0034] Where is the position of the i-th humpback whale at the, moment; is the Cauchy distribution sampling; and are randomly selected from the matrix Pool after iterating, times; is the vector coefficient of the i-th humpback whale at the, moment.

[0035] The parameter settings of EWOA in the described soft sensor modeling method for anaerobic fermentation process based on EWOA-LSSVM include the number of iterations, the convergence factor, and the inertia weight. The parameter settings of LSSVM include the kernel function parameter σ and the regularization parameter C. The steps for EWOA algorithm to optimize LSSVM parameters are as follows:

[0036] 1) Initialize the population of the enhanced whale optimization algorithm (EWOA), set the population size and the maximum number of iterations; each individual represents a combination of LSSVM parameters, that is, the kernel function parameter σ and the regularization parameter C;

[0037] 2) Use the selected variables as input and output to train LSSVM, and use the objective function that minimizes the expression containing two items of training error and regularization loss as the fitness function;

[0038] 3) Evaluate whale individuals according to fitness to find the current optimal parameter combination;

[0039] 4) If the current iteration number t is less than the maximum iteration number (MaxIt), go to step 5) to continue optimization; otherwise, go to step 7);

[0040] 5) Adopt a pooling mechanism to cross the worst solutions and promising solutions of each iteration and save them to the matrix Pool. Randomly select a part of the individuals in Pool for migration search. When the individual fitness probability satisfies p′ i ≥ 0.5, go to step 6); otherwise, go to step 7);

[0041] 6) Use the preferential selection strategy to disperse the whales in different regions of the search space to discover diverse solutions and update the position information;

[0042] 7) If the maximum iteration number is reached, assign the found σ and C of the optimal LSSVM to the LSSVM; otherwise, repeat step 3). Description of the Drawings

[0043] Figure 1 It is a flowchart of the EWOA-LSSVM soft sensor model;

[0044] Figure 2 It is a schematic diagram of the overall process of the soft sensor system;

[0045] Figure 3 It is the methane concentration prediction result of the EWOA-LSSVM soft sensor model. Detailed Embodiments

[0046] Combined with the embodiments, the technical solutions of this modeling are clearly and completely described. Use the described soft sensor modeling method for anaerobic fermentation process based on BO-CNN-LSTM to perform soft sensor modeling for the actual same or similar processes, including the following steps:

[0047] 1) Collect sampling data of the anaerobic fermentation reaction process of the food waste treatment plant through on-site operations, including the percentage of solid content (TS), pH value, percentage of volatile suspended solid content (VS), chemical oxygen demand (COD), average flow rate, alkalinity (Alk), percentage of CO2 content, daily gas production, percentage of CH4 content, and VFA laboratory test values, a total of 475 groups;

[0048] 2) Take out 380 groups and 95 groups of data from the selected input variable sample data as the training set and the test set respectively, and then perform normalization to eliminate the influence of the units and dimensions of different variables. The mapping space is selected as (0, 1), and use as the normalization criterion, where X0 is the historical data, Xi is the minimum value in the historical data, X a is the maximum value, and X is the data sample after normalization; the data sample is divided into a training set and a test set;

[0049] 3) Set the parameters of EWOA and LSSVM: the number of iterations of EWOA is 20, the number of populations is 5, and the kernel function of LSSVM is the radial basis kernel function;

[0050] 4) Use EWOA to optimize the parameters of LSSVM and assign the optimal parameter solution to LSSVM;

[0051] 5) Run the proposed EWOA-LSSVM model to extract features from the sample and predict the methane concentration accordingly.

[0052] The soft-sensing model established by this method has good prediction accuracy for the biogas concentration in the embodiment, and the prediction results are as attached Figure 3 . The results show that the established soft-sensing model of methane concentration can accurately estimate the methane concentration in practical applications and has broad application prospects in the field of monitoring and controlling the anaerobic digestion process of food waste treatment.

Claims

1. A soft sensor modeling method for anaerobic fermentation process based on EWOA-LSSVM, characterized by including The following steps: 1) Obtain sampling data of the anaerobic fermentation process with a time-tagged sequence through on-site operation or experiment; 2) Divide the sample data of the extracted basic characteristic variables into a training set and a test set, and perform normalization processing to eliminate the influence of the dimensions of different characteristic variables; 3) Set the initial parameters of the enhanced whale optimization algorithm (EWOA), namely the number of iterations and the population size, and use the pooling mechanism to increase the diversity of the population; use the migration search strategy and the preferred search strategy to enhance the local and global search capabilities; 4) Construct a least squares support vector machine (LSSVM) soft-sensing model for the anaerobic fermentation process, and use the objective function that minimizes the expression containing two items, namely the training error and the regularization loss, as the fitness function; the data loss part is evaluated by the root mean square error (RMSE) to measure the prediction error of the model on the training data; use EWOA to search for the optimal parameters of the LSSVM model, and these parameters include the kernel function parameter σ and the regularization parameter C; 5) Substitute the obtained optimal parameter set into LSSVM to form a high-precision soft-sensing model EWOA-LSSVM for the anaerobic fermentation process, and predict the methane concentration and biogas production.

2. The soft sensor modeling method for anaerobic fermentation process based on EWOA-LSSVM according to claim 1, characterized in that The described pooling mechanism increases the diversity of the population by crossing and saving the worst solution and the promising solution of each iteration into the matrix Pool through the following formula: Among them is the i-th solution at the t-th iteration in the matrix Pool; is a binary random vector; is the inverse vector of; is a random position near the best humpback whale; is the worst solution obtained at the current t-th iteration.

3. A soft sensor modeling method for anaerobic fermentation process based on EWOA-LSSVM according to claim 1, characterized in that The described migration search strategy randomly separates a part of the humpback whales through the following formula to cover the unvisited areas to increase the diversity of the population and reduce the local optimal solution: Among them is the position of the i-th humpback whale at time t+1; is a random position in the search space range; rand is a uniformly distributed random number between 0 and 1; δ max and δ min are the upper and lower bounds of the problem; δ best_max and δ best_min are the upper and lower bounds of the best humpback whale.

4. A soft sensor modeling method for anaerobic fermentation process based on EWOA-LSSVM according to claim 1, characterized in that The described priority selection strategy disperses the whales in different regions of the search space through the following formula to discover diverse solutions and improve the search ability of EWOA in the prey search method: where is the position of the i-th humpback whale at time is the Cauchy distribution sampling; and are randomly selected from the matrix Pool after the k-th iteration; is the vector coefficient of the i-th humpback whale at time t.

5. A soft sensor modeling method for anaerobic fermentation process based on EWOA-LSSVM according to claim 1, characterized in that The described enhanced whale optimization algorithm is used to optimize the LSSVM parameters, and the specific steps are as follows: 1) Initialize the population of the enhanced whale optimization algorithm (EWOA), set the population size and the maximum number of iterations; each individual represents a combination of LSSVM parameters, namely the kernel function parameter σ and the regularization parameter C; 2) Use the selected variables as the input and output to train LSSVM, and use the objective function that minimizes the expression containing two items, namely the training error and the regularization loss, as the fitness function; 3) Evaluate the whale individuals according to the fitness, and find the current optimal parameter combination; 4) If the current iteration number t is less than the maximum iteration number (MaxIt), then go to step 5) to continue the optimization, otherwise go to step 8); 5) The pooling mechanism is adopted to cross the worst solution and the promising solution in each iteration and save them to the matrix Pool. When the individual fitness probability satisfies , go to step 6); otherwise, go to step 7). 6) Use the priority selection strategy to disperse the whales in different regions of the search space to discover diverse solutions, and update the position information of the humpback whales; 7) Use the migration search strategy to randomly separate a part of the humpback whales to cover the unvisited areas to increase the diversity of the population, and update the position information of the humpback whales; 8) If the maximum iteration number is reached, assign the obtained optimal σ and C of LSSVM to LSSVM, otherwise repeat step 3).

Images