Biomass gasification product prediction method based on mechanism and data fusion driving

Abstract

The invention discloses a biomass gasification product prediction method based on mechanism and data fusion driving, and the method comprises the steps: constructing a biomass gasification product prediction mechanism model, and inputting biomass gasification operation parameters into the model to obtain a prediction result; calculating a residual vector between a prediction result and actual experimental data; establishing a support vector regression model between the biomass gasification operation parameters and the residual vectors; and taking the output of the support vector regression model as an error compensation item of a biomass gasification product prediction mechanism model, constructing a biomass gasification product prediction model based on mechanism and data fusion driving, and predicting a biomass gasification product by using the model. Through fusion of a mechanism model and data driving, the problem of prediction deviation caused by model simplification hypothesis in a traditional method is solved, high-precision prediction of biomass gasification products is achieved, a theoretical basis is provided for real-time optimization of process parameters, and the method is suitable for industrial control of biomass gasification systems such as fluidized beds and fixed beds.

Description

Technical Field

[0001] The present invention relates to biomass gasification prediction technology, and specifically relates to a method for predicting biomass gasification products driven by the fusion of mechanism and data. Background Art

[0002] Biomass gasification is a core technology for converting renewable resources such as agricultural and forestry waste into clean syngas (H2, CO, etc.), which is of great significance for reducing dependence on fossil energy and lowering carbon emissions. Accurately predicting gasification products is the key to optimizing gasification processes and improving energy conversion efficiency, directly affecting the design of gasifiers, the regulation of operating parameters, and the operating stability of downstream energy systems. The biomass gasification process involves multi-phase flow, multi-stage complex reactions such as pyrolysis-gasification-tar cracking, and its kinetic characteristics are affected by raw material components (moisture content, ash content), operating conditions (temperature, oxygen-fuel ratio), and equipment structure. However, in the biomass gasification process, traditional mechanism models have large prediction errors due to simplified assumptions, and the optimization of operating parameters depends on static rules and cannot adapt to dynamic operating conditions.

[0003] The current biomass gasification modeling technology mainly has the following bottlenecks: Mechanism models based on platforms such as Aspen Plus need to make simplified assumptions for complex gasification reactions (such as tar formation, ash melting), resulting in significant prediction errors for the volume fractions of key gas components such as CO, H2, CH4, N2, CO2, etc., especially when raw material components fluctuate or equipment parameters change, the errors are amplified; Pure data-driven methods (such as neural networks) rely on a large amount of experimental data and are difficult to capture the inherent physical laws of gasification reactions. When the operating conditions exceed the range of the training set, the prediction reliability drops sharply.

[0004] The problems existing in the prior art mainly focus on the following two aspects: First, traditional mechanism models simplify the complex gasification reaction process and are not sufficient to comprehensively capture key physical and chemical behaviors, easily resulting in systematic prediction errors under dynamic operating conditions and raw material fluctuations; Second, pure data-driven models rely heavily on a large amount of high-quality experimental data and are prone to the problem of insufficient generalization ability when the operating conditions exceed the training range. These problems have restricted the application effect of gasification systems in real-time control and process optimization. Summary of the Invention

[0005] Object of the Invention: The object of the present invention is to provide a method for predicting biomass gasification products driven by the fusion of mechanism and data. By the fusion of the mechanism model and data driving, the problem of prediction deviation caused by the simplified assumptions of traditional methods is solved, and high-precision prediction of biomass gasification products is achieved.

[0006] Technical Solution: A method for predicting biomass gasification products driven by the fusion of mechanism and data according to the present invention includes:

[0007] Build a prediction mechanism model for biomass gasification products through the Aspen Plus platform; input the biomass gasification operation parameters into the prediction mechanism model for biomass gasification products to obtain the prediction results output by the prediction mechanism model for biomass gasification products; wherein, the biomass gasification operation parameters at least include gasification temperature, feed rate, and air equivalence ratio;

[0008] Calculate the deviation between the prediction results output by the prediction mechanism model for biomass gasification products and the actual experimental data to obtain the residual vector predicted by the prediction mechanism model for biomass gasification products;

[0009] Establish a support vector regression model using the support vector regression algorithm to represent the relationship between the biomass gasification operation parameters and the residual vector; input the biomass gasification operation parameters into the support vector regression model, and the support vector regression model outputs the prediction residuals of biomass gasification products;

[0010] Use the prediction residuals of biomass gasification products output by the support vector regression model as the error compensation term of the prediction mechanism model for biomass gasification products, and construct a prediction model for biomass gasification products driven by the fusion of mechanism and data. The prediction model for biomass gasification products driven by the fusion of mechanism and data is used to dynamically correct the prediction deviation of the prediction mechanism model for biomass gasification products;

[0011] Use the prediction model for biomass gasification products driven by the fusion of mechanism and data to predict biomass gasification products.

[0012] Furthermore, the prediction mechanism model for biomass gasification products includes a stoichiometric reactor DRY, a first separator SEP-1, a yield reactor PYRO, and a second separator SEP-2 connected in sequence along the biomass transmission direction; wherein, the output end of the second separator SEP-2 is divided into two paths, one path is connected to an input end of a first mixing unit MIX-1 through a Gibbs reactor PYRO-2, and the other path is connected to an input end of a third mixing unit MIX-3 through a third separator SEP-3;

[0013] The output end of the first mixing unit MIX-1 is connected to the other input end of the third mixing unit MIX-3 through a second mixing unit MIX-2;

[0014] The output end of the third mixing unit MIX-3 is sequentially connected with a first plug flow reactor GASIFLY1, a second plug flow reactor GASIFLY2, and a fourth separator SEP-4.

[0015] Furthermore, the residual vector includes the prediction deviation values of the volume fractions of gas components such as CO, H2, CH4, N2, and CO2.

[0016] Further, a support vector regression model for representing the relationship between biomass gasification operation parameters and the residual vector is established using the support vector regression algorithm; the biomass gasification operation parameters are input into the support vector regression model, and the support vector regression model outputs the prediction residual of the biomass gasification product, including:

[0017] Model initialization: Select the radial basis function RBF as the kernel function of the support vector regression model, and the biomass gasification operation parameters are mapped to the high-dimensional feature space φ(x) through the radial basis function RBF; Set Hyperparameters of the support vector regression model; Global optimization is performed through a grid search strategy combined with a cross-validation method, Determine The optimal hyperparameters of the support vector regression model;

[0018] Model training and validation: Train the support vector regression model. In the training stage, solve the support vector weights based on the Lagrangian dual problem Screen the support vectors, achieve residual prediction and output, and generate residuals through the following formula:

[0019]

[0020] Obtain the non-linear mapping relationship from biomass gasification operation parameters to the residual vector by training the support vector regression model;

[0021] Residual estimate calculation: Input the biomass gasification operation parameters into the prediction function of the support vector regression model, and the support vector regression model outputs the prediction residual of the biomass gasification product.

[0022] Further, the expression of the radial basis function RBF is as follows:

[0023]

[0024] where x i -x j is the input feature vector.

[0025] Further, the objective function expression of the support vector regression model is as follows:

[0026]

[0027] The constraint conditions are as follows:

[0028]

[0029] Further, the hyperparameters of the support vector regression model include the regularization coefficient C, the kernel bandwidth γ, and the insensitive loss threshold ε.

[0030] Further, global optimization is performed through a grid search strategy combined with a cross-validation method,Determine The optimal hyperparameters of the support vector regression model include:

[0031] According to the distribution characteristics of the biomass gasification data, reasonable ranges of candidate values are set for each hyperparameter. Set C ∈ [0.1, 100], γ ∈ [10 -3 , 10], ε ∈ [0.01, 0.5];

[0032] The training dataset is randomly divided into 5 non-overlapping subsets. Using the 5-fold cross-validation method, in each fold, a part of the data is used as the validation set in turn, and the remaining data is used as the training set. The model is trained for each combination of hyperparameters, and the mean squared error MSE is used as the evaluation index to calculate the average MSE of each fold and the stability of the prediction residuals, ensuring the generalization of the support vector regression model to the dynamic conditions of the gasifier;

[0033] Traverse the parameter combinations, and select the combination of hyperparameters with the smallest MSE of the validation set and stable distribution of prediction residuals as the optimal hyperparameters of the support vector regression model.

[0034] Furthermore, take the residual estimation value output by the support vector regression model as the error compensation term of the biomass gasification product prediction mechanism model, and construct a biomass gasification product prediction model driven by the fusion of mechanism and data, including:

[0035] Algebraically superimpose the prediction result of the biomass gasification product prediction mechanism model and the residual estimation value output by the support vector regression model to establish a biomass gasification product prediction model driven by the fusion of mechanism and data with error compensation function. The expression is as follows:

[0036] y_hybrid = y_mechanism + SVR(x)

[0037] Among them, x is the input vector of gasification operation parameters; through the above equation, the systematic deviation of the biomass gasification product prediction mechanism model is dynamically corrected.

[0038] Furthermore, during the process of constructing the biomass gasification product prediction mechanism model, set the kinetic parameters and thermodynamic equilibrium conditions of the biomass gasification product prediction mechanism model.

[0039] Beneficial effects: Compared with the prior art, the significant technical effects of the present invention are as follows:

[0040] The present invention adopts a fusion method of mechanism model and data-driven model. By constructing a mechanism model for predicting biomass gasification products and using the support vector regression algorithm to correct the prediction error in real time, it effectively solves the systematic prediction deviation problem caused by simplified assumptions in traditional mechanism models. At the same time, online updating of the residual regression parameters enables the model to learn and compensate for the unmodeled dynamic characteristics in the gasification process in real time, thus significantly improving the prediction accuracy and operating condition adaptability. It can be applied to the dynamic optimization control of gasifiers, providing a feasible technical route for the efficient conversion of biomass energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a schematic flow chart of the present invention;

[0042] Figure 2 is a schematic diagram of the mechanism model for predicting biomass gasification products in the present invention;

[0043] Figure 3 is a flow chart for constructing the support vector regression model in the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0044] The technical solutions of the present invention will be introduced in detail below in conjunction with the specific embodiments and the drawings in the specification.

[0045] As Figure 1 shown, a method for predicting biomass gasification products based on the fusion drive of mechanism and data of the present invention includes the following steps:

[0046] S1. Construct a mechanism model for predicting biomass gasification products through the Aspen Plus platform; input the biomass gasification operation parameters into the mechanism model for predicting biomass gasification products to obtain the prediction results output by the mechanism model for predicting biomass gasification products. Among them, the biomass gasification operation parameters at least include gasification temperature, feed rate, and air equivalence ratio.

[0047] Establish a mechanism model for predicting biomass gasification products through the Aspen Plus platform, including biomass drying, biomass pyrolysis, and coke oxidation and reduction.

[0048] As Figure 2As shown, the prediction mechanism model of biomass gasification products includes a stoichiometric reactor DRY, a first separator SEP-1, a yield reactor PYRO, and a second separator SEP-2, which are sequentially connected along the biomass transmission direction; wherein the output end of the second separator SEP-2 is divided into two paths, one of which is connected to an input end of the first mixing unit MIX-1 through the Gibbs reactor PYRO-2, and the other is connected to an input end of the third mixing unit MIX-3 through the third separator SEP-3. The output end of the first mixing unit MIX-1 is connected to the other input end of the third mixing unit MIX-3 through the second mixing unit MIX-2. The output end of the third mixing unit MIX-3 is sequentially connected to the first plug flow reactor GASIFIG-1, the second plug flow reactor GASIFIG-2, and the fourth separator SEP-4.

[0049] In the stoichiometric reactor RStoic (DRY), the biomass stream first undergoes a drying treatment to remove the moisture contained in the raw material. Subsequently, the first separator SEP-1 is used to effectively separate the water vapor and the dehydrated biomass. The dried DRY-BIO material flow enters the yield reactor RYield (PYRO) for pyrolysis conversion. During this process, the raw material is cracked into elemental substances such as C, S, O2, H2, N2 and ash in proportion to the mass. After being treated in the second separator SEP-2, the pyrolysis product is divided into a TAR gaseous flow and a CHAR solid flow. The subsequent CHAR material is further purified of ash through the third separator SEP-3 to obtain a pyrolysis solid product (ELEM-SLD).

[0050] The TAR gaseous stream is introduced into the Gibbs reactor RGIBBS (PYRO-2), and the chemical equilibrium and phase equilibrium of volatile substances are achieved through the principle of minimization of Gibbs free energy. Three mixing units, MIX-1, MIX-2 and MIX-3, are set up in the process system. After the air (AIR), the steam (STEAM) generated by the drying section and the pyrolysis solid product (ELEM-SLD) are mixed in proportion, they are jointly input into the plug flow reactor RPlug (GASIFLY1, GASIFLY2) for gasification reaction. The redox process eventually generates primary crude synthesis gas, and after the residual water vapor and other components are removed by the fourth separator SEP-4, a dry and clean combustible gas product that meets the requirements is finally obtained.

[0051] In the process of constructing the prediction mechanism model of biomass gasification products, it is necessary to set the kinetic parameters and thermodynamic equilibrium conditions of the prediction mechanism model of biomass gasification products, and input the biomass gasification operation parameters for simulation calculation.

[0052] Biomass includes agricultural and forestry waste, energy crops or organic solid waste. In this embodiment, agricultural and forestry waste corn stalks are specifically used as an example.

[0053] S2. Calculate the deviation between the predicted result output by the prediction mechanism model of biomass gasification products and the actual experimental data, and obtain the residual vector predicted by the prediction mechanism model of biomass gasification products.

[0054] In this embodiment, the residual vector includes at least the predicted deviation values of the volume fractions of gas components such as CO, H2, CH4, N2, and CO2.

[0055] S3. Establish a support vector regression model for representing the relationship between biomass gasification operation parameters and the residual vector by using the support vector regression algorithm; input the biomass gasification operation parameters into the support vector regression model, and the support vector regression model outputs the prediction residual of biomass gasification products.

[0056] As Figure 3 shown, the specific implementation process of step S3 is as follows:

[0057] S3.1. Model initialization: Select the radial basis function RBF as the kernel function of the support vector regression model, and map the biomass gasification operation parameters (including gasification temperature, feed rate, and air equivalence ratio) to the high-dimensional feature space φ(x) through the radial basis function RBF; set the hyperparameters of the support vector regression model; perform global optimization through the grid search strategy combined with the cross-validation method to determine the optimal hyperparameters of the support vector regression model.

[0058] In this embodiment, the expression of the radial basis function RBF is as follows:

[0059] k(x i ,x j ) = exp(-γ||x i -x j || 2 )

[0060] where x i -x j is the input feature vector (including gasification temperature, feed rate, and air equivalence ratio).

[0061] Hyperparameter optimization strategy: The hyperparameters of the support vector regression model include the regularization coefficient C, the kernel bandwidth γ, and the insensitive loss threshold ε.

[0062] Perform global optimization through the grid search strategy combined with the cross-validation method to determine the optimal hyperparameters of the support vector regression model. The specific process is as follows:

[0063] According to the distribution characteristics of biomass gasification data, set reasonable value ranges for each hyperparameter. Set C ∈ [0.1, 100] (logarithmic uniform sampling), γ ∈ [10 -3 , 10], and ε ∈ [0.01, 0.5];

[0064] Randomly divide the training dataset into 5 non - overlapping subsets. Adopt the 5 - fold cross - validation method. In each fold, successively use a part of the data as the validation set and the remaining data as the training set. Train the model for each set of hyperparameter combinations, and use the mean squared error (MSE) as the evaluation index to calculate the average MSE and the stability of the prediction residuals for each fold, ensuring the generalization of the support vector regression model to the dynamic conditions of the gasifier.

[0065] Traverse the parameter combinations, and select the parameter combination with the minimum MSE of the validation set and a stable distribution of prediction residuals as the optimal hyperparameters of the support vector regression model.

[0066] S3.2. Model training and validation: Train the support vector regression model. In the training stage, solve for the support vector weights based on the Lagrangian dual problem. Screen the support vectors to achieve residual prediction and output.

[0067] The biomass gasification operation parameters (including gasification temperature, feed rate, and air equivalence ratio) are mapped to a high - dimensional feature space through the RBF kernel function to solve the problem of inseparable data caused by multi - variable coupling and non - linear dynamic characteristics during the biomass gasification process. The objective function expression of the support vector regression (SVR) model is as follows:

[0068]

[0069] The constraint conditions are as follows:

[0070]

[0071] After converting to the Lagrangian dual problem, solve to obtain the Lagrange multipliers α i ,

[0072] When the data is non - linearly separable, map the input x to the high - dimensional feature space φ(x), and construct the prediction function (or decision function) of the support vector regression model:

[0073] f(x) = w T φ(x)+b

[0074] Through Lagrangian dual transformation, w in the original problem is implicitly expressed as a linear combination of support vectors:

[0075]

[0076] After substituting w into the prediction function of the support vector regression model:

[0077]

[0078] Generate residuals through the above formula.

[0079] The non - linear mapping relationship from biomass gasification operation parameters to the residual vector is obtained by training a support vector regression model.

[0080] S3.3. Calculation of the residual estimated value: Input the biomass gasification operation parameters into the prediction function of the support vector regression model, and the support vector regression model outputs the prediction residual of the biomass gasification products.

[0081] In step S3, the dual transformation solution of the objective function of the support vector regression model is carried out to obtain the prediction function parameters of the support vector regression model, then the prediction function of the support vector regression model is obtained, and finally the prediction function of the support vector regression model obtained by using the solved parameters is used to calculate the prediction residual of the biomass gasification products.

[0082] S4. Use the prediction residual of the biomass gasification products output by the support vector regression model as the error compensation term of the prediction mechanism model of the biomass gasification products, and construct a prediction model of the biomass gasification products driven by the fusion of mechanism and data; the prediction model of the biomass gasification products driven by the fusion of mechanism and data is used to dynamically correct the prediction deviation of the prediction mechanism model of the biomass gasification products.

[0083] Algebraically superimpose the prediction result of the prediction mechanism model of the biomass gasification products and the residual estimated value output by the support vector regression model to establish a prediction model of the biomass gasification products driven by the fusion of mechanism and data with an error compensation function. The expression is as follows:

[0084] y_hybrid = y_mechanism + SVR(x)

[0085] Among them, x is the input gasification operation parameter vector (temperature, feed rate, air equivalence ratio); through the above equation, the systematic deviation of the prediction mechanism model of the biomass gasification products is dynamically corrected.

[0086] S5. Use the prediction model of the biomass gasification products driven by the fusion of mechanism and data to predict the biomass gasification products.

[0087] The present invention constructs a prediction model of the biomass gasification products driven by the fusion of mechanism and data through the prediction mechanism model of the biomass gasification products and real - time observation data, effectively making up for the limitations of a single modeling method.

[0088] The present invention outputs fitting data through a prediction mechanism model of biomass gasification products, dynamically corrects the residuals of the prediction mechanism model of biomass gasification products by combining support vector regression (SVR), and constructs a prediction model of biomass gasification products driven by the fusion of mechanism and data. This method can significantly improve the prediction accuracy by enabling the data-driven module to learn the unmodeled dynamic characteristics in real time, providing reliable technical support for the efficient conversion of biomass energy.

[0089] Through the fusion of the mechanism model and data driving, the present invention solves the problem of prediction deviation caused by model simplification assumptions in traditional methods, realizes high-precision prediction of biomass gasification products, and is applicable to industrial control of biomass gasification systems such as fluidized beds and fixed beds.

Claims

1. A prediction method for biomass gasification products driven by the fusion of mechanism and data, characterized in that, Including: Construct a prediction mechanism model of biomass gasification products through the Aspen Plus platform; input the biomass gasification operation parameters into the prediction mechanism model of biomass gasification products to obtain the prediction results output by the prediction mechanism model of biomass gasification products; wherein, the biomass gasification operation parameters at least include gasification temperature, feed rate, and air equivalence ratio; Calculate the deviation between the prediction results output by the prediction mechanism model of biomass gasification products and the actual experimental data to obtain the residual vector predicted by the prediction mechanism model of biomass gasification products; Establish a support vector regression model using the support vector regression algorithm to represent the relationship between the biomass gasification operation parameters and the residual vector; input the biomass gasification operation parameters into the support vector regression model, and the support vector regression model outputs the prediction residuals of biomass gasification products; Take the prediction residuals of biomass gasification products output by the support vector regression model as the error compensation term of the prediction mechanism model of biomass gasification products, and construct a prediction model of biomass gasification products driven by the fusion of mechanism and data; the prediction model of biomass gasification products driven by the fusion of mechanism and data is used to dynamically correct the prediction deviation of the prediction mechanism model of biomass gasification products; Use the prediction model of biomass gasification products driven by the fusion of mechanism and data to predict biomass gasification products.

2. The biomass gasification product prediction method based on mechanism and data fusion drive according to claim 1, wherein: The prediction mechanism model of biomass gasification products includes a stoichiometric reactor DRY, a first separator SEP-1, a yield reactor PYRO, and a second separator SEP-2 connected in sequence along the biomass transmission direction; wherein, the output end of the second separator SEP-2 is divided into two paths, one path is connected to an input end of the first mixing unit MIX-1 through the Gibbs reactor PYRO-2, and the other path is connected to an input end of the third mixing unit MIX-3 through the third separator SEP-3; The output end of the first mixing unit MIX-1 is connected to the other input end of the third mixing unit MIX-3 through the second mixing unit MIX-2; The output end of the third mixing unit MIX-3 is sequentially connected with a first plug flow reactor GASIFLY1, a second plug flow reactor GASIFLY2, and a fourth separator SEP-4.

3. The biomass gasification product prediction method driven by the fusion of mechanism and data according to claim 1, wherein: The residual vector includes the prediction deviation values of the volume fractions of gas components such as CO, H2, CH4, N2, and CO2.

4. The biomass gasification product prediction method based on mechanism and data fusion drive according to claim 1, wherein Establish a support vector regression model using the support vector regression algorithm to represent the relationship between the biomass gasification operation parameters and the residual vector; Input the biomass gasification operation parameters into the support vector regression model, and the support vector regression model outputs the prediction residuals of biomass gasification products, including: Model initialization: Select the radial basis function RBF as the kernel function of the support vector regression model, and map the biomass gasification operation parameters to the high-dimensional feature space φ(x) through the radial basis function RBF; set the hyperparameters of the support vector regression model; perform global optimization through the grid search strategy combined with the cross-validation method to determine the optimal hyperparameters of the support vector regression model; Model training and validation: Train a support vector regression model. During the training phase, solve for the support vector weights based on the Lagrangian dual problem Screen the support vectors, achieve residual prediction and output, and generate the residuals through the following formula: Obtain the non-linear mapping relationship from the biomass gasification operation parameters to the residual vector by training the support vector regression model; Residual Estimation Value Calculation: Input the biomass gasification operation parameters into the prediction function of the support vector regression model, and the support vector regression model outputs the prediction residual of the biomass gasification products.

5. The biomass gasification product prediction method based on mechanism and data fusion drive according to claim 4, wherein The expression of the radial basis function RBF is as follows: where x i -x j is the input feature vector.

6. The biomass gasification product prediction method based on mechanism and data fusion drive according to claim 4, characterized in that, The objective function expression of the support vector regression model is as follows: The constraint conditions are as follows:

7. The biomass gasification product prediction method based on mechanism and data fusion drive according to claim 4, wherein: The hyperparameters of the support vector regression model include the regularization coefficient C, the kernel bandwidth γ, and the insensitive loss threshold ε.

8. The method for predicting biomass gasification products driven by mechanism and data fusion according to claim 7, characterized in that, Global optimization is carried out through a grid search strategy combined with a cross-validation method, Determine for the optimal hyperparameters of the support vector regression model, including: According to the distribution characteristics of biomass gasification data, reasonable ranges of candidate values are set for each hyperparameter, where \(C\in[0.1,100]\), \(\gamma\in[10 -3 ,10]\), and \(\varepsilon\in[0.01,0.5]\); Randomly divide the training data set into 5 non-overlapping subsets. Using the 5-fold cross-validation method, in each fold, sequentially take a part of the data as the validation set and the remaining data as the training set to train the model for each combination of hyperparameters. Using the mean square error MSE as the evaluation index, calculate the average MSE of each fold and the stability of the prediction residuals to ensure the generalization of the support vector regression model to the dynamic conditions of the gasifier. Traverse the parameter combinations and select the combination of hyperparameters with the smallest MSE of the validation set and a stable distribution of prediction residuals as the optimal hyperparameters of the support vector regression model.

9. The biomass gasification product prediction method based on mechanism and data fusion drive according to claim 1, wherein Take the residual estimation value output by the support vector regression model as the error compensation term of the biomass gasification product prediction mechanism model, and construct a biomass gasification product prediction model driven by the fusion of mechanism and data, including: Algebraically superimpose the prediction result of the biomass gasification product prediction mechanism model and the residual estimation value output by the support vector regression model to establish a biomass gasification product prediction model driven by the fusion of mechanism and data with an error compensation function. The expression is as follows: y_hybrid = y_mechanism + SVR(x) where x is the input vector of gasification operation parameters; through the above equation, the systematic deviation of the biomass gasification product prediction mechanism model is dynamically corrected.

10. The biomass gasification product prediction method based on mechanism and data fusion drive according to claim 1, wherein, During the construction of the biomass gasification product prediction mechanism model, set the kinetic parameters and thermodynamic equilibrium conditions of the biomass gasification product prediction mechanism model.

Images