Storage medium, microwave flash pyrolysis process optimization method, device and equipment

Abstract

The invention discloses a storage medium, a microwave flash pyrolysis process optimization method, a microwave flash pyrolysis process optimization device and microwave flash pyrolysis process optimization equipment. The microwave flash pyrolysis process optimization method comprises the following steps: generating a bio-oil yield prediction model: acquiring historical data including a plurality of parameter items of biomass microwave flash pyrolysis equipment; the parameter items comprise raw material data type parameter items, process data type parameter items and bio-oil yield type parameter items; generating modeling data for machine learning according to the time correspondence of the historical data; the modeling data is divided into a training set and a test set according to a preset division proportion; based on a random forest regression model, generating a bio-oil yield prediction model for predicting a bio-oil yield prediction value according to the modeling data; according to the method, the bio-oil yield can be rapidly and accurately predicted according to the current working condition of the biomass microwave flash pyrolysis equipment, and then effective data support is provided for production work such as process optimization and equipment maintenance of the biomass microwave flash pyrolysis equipment.

Description

Technical Field

[0001] The invention relates to the field of microwave heating, and in particular to a storage medium, a microwave flash pyrolysis process optimization method, a device and equipment. Background Art

[0002] Rapid pyrolysis of biomass belongs to the field of thermochemistry. Usually, the process only requires a single reaction unit. Therefore, it has the advantages of simple process and low cost, and has become the main way to obtain bio-oil products.

[0003] The method of preparing bio-oil by pyrolysis of biomass includes: after a large number of preliminary experiments to explore the properties of raw materials and process conditions, the biomass is subjected to microwave flash pyrolysis in an oxygen-free or low-oxygen environment, and then the condensable components in the volatile gas generated by the pyrolysis are quickly cooled to obtain the bio-oil component.

[0004] There are many factors that affect the amount of bio-oil production, but they are complex. There is currently no clear conclusion or demonstration system, only a few mathematical models or experimental-based exploration of regularities.

[0005] The inventors have found through research that the production process optimization scheme for preparing bio-oil by microwave flash pyrolysis of biomass in the prior art has at least the following defects:

[0006] The production process that screens raw materials and process parameters through experience and a large number of test calibration methods is not only inefficient, but also has poor optimization effects, and has the defects of low versatility and unstable quality control.

[0007] The information disclosed in this background technology section is only intended to enhance the understanding of the overall background of the invention and should not be regarded as an acknowledgment or any form of suggestion that the information constitutes the prior art already known to a person skilled in the art. Summary of the invention

[0008] The purpose of the present invention is to improve the optimization efficiency and optimization effect of the production process for preparing bio-oil by microwave flash pyrolysis of biomass.

[0009] The present invention provides a microwave flash pyrolysis process optimization method, comprising the steps of generating a bio-oil yield prediction model:

[0010] S11, acquiring historical data including multiple parameter items of a biomass microwave flash pyrolysis device; the parameter items include: raw material data parameter items, process data parameter items, and bio-oil output parameter items;

[0011] S12. Generate modeling data for machine learning based on the time correspondence of the historical data; the modeling data is divided into a training set and a test set according to a preset division ratio;

[0012] S13, based on the random forest regression model, generating a bio-oil yield prediction model for predicting the bio-oil yield prediction value according to the modeling data, including:

[0013] Establishing a plurality of decision trees according to the training set and a preset training method to establish a random forest regression model;

[0014] The determination coefficient of the random forest regression model is calculated according to the test set, and the bio-oil yield prediction model is generated by adjusting the parameters of the random forest regression model according to a preset parameter adjustment method.

[0015] Preferably, the present invention further comprises:

[0016] S14, taking the raw material data parameter items and the process data parameter items as input, periodically generating a bio-oil yield prediction value according to the bio-oil yield prediction model;

[0017] S15. When the predicted value of the bio-oil production is less than a preset production threshold, the controllable parameter items in the raw material data type parameter items and the process data type parameter items are traversed to obtain the maximum value of the predicted value of the bio-oil production, and the adaptation value of each controllable parameter is generated through the bio-oil production prediction model; the adaptation value of the controllable parameter is used to optimize the production process of the biomass microwave flash pyrolysis equipment.

[0018] In another aspect of the present invention, a microwave flash pyrolysis process optimization device is provided, comprising a model generation unit for a bio-oil yield prediction model; the model generation unit comprises:

[0019] A historical data acquisition module is used to acquire historical data including multiple parameter items of the biomass microwave flash pyrolysis equipment; the parameter items include: raw material data parameter items, process data parameter items and bio-oil output parameter items;

[0020] A modeling data generation module, used to generate modeling data for machine learning according to the time correspondence between the historical data; the modeling data is divided into a training set and a test set according to a preset division ratio;

[0021] A prediction model generation module is used to generate a bio-oil yield prediction model for predicting a bio-oil yield prediction value based on the modeling data based on a random forest regression model, comprising:

[0022] Establishing a plurality of decision trees according to the training set and a preset training method to establish a random forest regression model;

[0023] The determination coefficient of the random forest regression model is calculated according to the test set, and the bio-oil yield prediction model is generated by adjusting the parameters of the random forest regression model according to a preset parameter adjustment method.

[0024] Preferably, the present invention further comprises:

[0025] A result prediction unit, used for periodically generating a bio-oil yield prediction value according to the bio-oil yield prediction model;

[0026] A process optimization unit is used to traverse the controllable parameter items in the raw material data type parameter items and the process data type parameter items when the bio-oil yield prediction value is less than a preset yield threshold, so as to obtain the maximum value of the bio-oil yield prediction value as the goal, and generate adaptation values ​​of each controllable parameter through the bio-oil yield prediction model; the adaptation values ​​of the controllable parameters are used to optimize the production process of the biomass microwave flash pyrolysis equipment.

[0027] On the other hand, a microwave flash pyrolysis process optimization device is provided, which includes a computer program stored on a medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer executes the methods described in the above aspects and achieves the same technical effects.

[0028] On the other hand, a storage medium is provided on which a computer program is stored. When the computer program is executed by a processor, each step of the microwave flash pyrolysis process optimization method as described in any one of the above items is implemented.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] The present invention is based on the specific application scenario of preparing bio-oil by microwave flash pyrolysis of biomass, and establishes a prediction algorithm based on a random forest regression model; since the random forest regression model is based on previous experimental data, it fully considers the coupling relationship of multiple influencing factors; therefore, the bio-oil yield prediction model generated by the present invention can quickly and accurately predict the bio-oil yield according to the current working conditions of the biomass microwave flash pyrolysis equipment, thereby providing effective data support for production work such as process optimization and equipment maintenance of the biomass microwave flash pyrolysis equipment.

[0031] Furthermore, in the present invention, the goal is to obtain the maximum value of the predicted value of bio-oil yield, and the adaptation value of each controllable parameter is generated; thereby, a control strategy for optimizing the production process of biomass microwave flash pyrolysis equipment can be obtained; it can be seen that compared with the laboratory data analysis or human experience judgment in the prior art, the present invention not only has strong versatility and robustness, but also can effectively avoid the investment of a large amount of repetitive labor, thereby improving the efficiency and effect of microwave heating microwave flash pyrolysis process optimization.

[0032] Furthermore, in the present invention, during the model training process, the weight factors of the parameters are also taken into account when performing pre-pruning of the binary tree partitioning. By first calculating the optimal splitting points for high-weight parameters, unnecessary calculations are reduced, thereby effectively optimizing the response time of the calculation process.

[0033] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and implement it according to the contents of the specification, and to make the above and other purposes, technical features and advantages of the present invention easier to understand, one or more preferred embodiments are listed below and described in detail with the accompanying drawings as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a step diagram of the microwave flash pyrolysis process optimization method described in the present invention;

[0035] Figure 2 It is a structural schematic diagram of the microwave flash pyrolysis process optimization device of the present invention;

[0036] Figure 3 It is a structural schematic diagram of the microwave flash pyrolysis process optimization control equipment described in the present invention. DETAILED DESCRIPTION

[0037] The specific implementation modes of the present invention are described in detail below in conjunction with the accompanying drawings, but it should be understood that the protection scope of the present invention is not limited by the specific implementation modes.

[0038] Unless explicitly stated otherwise, throughout the specification and claims, the term “comprise” or variations such as “include” or “comprising”, etc., will be understood to include the stated elements or components but not to exclude other elements or components.

[0039] In this document, for the convenience of description, spatial relative terms such as "below", "below", "down", "above", "above", "upper", etc. may be used to describe the relationship between one element or feature and another element or feature in the accompanying drawings. It should be understood that the spatial relative terms are intended to include different orientations of the object in use or operation in addition to the orientation depicted in the figure. For example, if the object in the figure is turned over, the element described as being "below" or "below" other elements or features will be oriented "above" the element or feature. Therefore, the exemplary term "below" can include both below and above. Objects may also have other orientations (rotated 90 degrees or other orientations) and the spatial relative terms used in this document should be interpreted accordingly.

[0040] In this document, the terms "first", "second", etc. are used to distinguish two different elements or parts, and are not used to limit a specific position or relative relationship. In other words, in some embodiments, the terms "first", "second", etc. can also be interchangeable.

[0041] Embodiment 1

[0042] In order to improve the optimization efficiency and optimization effect of the production process of biomass microwave flash pyrolysis to prepare bio-oil, Figure 1 As shown, in an embodiment of the present invention, a method for optimizing a microwave flash pyrolysis process is provided, including the steps of generating a bio-oil yield prediction model:

[0043] S11, acquiring historical data including multiple parameter items of the biomass microwave flash pyrolysis equipment; the parameter items include: raw material data parameter items, process data parameter items and bio-oil output parameter items;

[0044] In practical applications, the raw material data parameter items in the embodiment of the present invention may specifically include elemental analysis data of biomass raw materials, industrial analysis data of biomass raw materials, three-element composition data of biomass raw materials, dielectric constant data of biomass raw materials, pyrolysis temperature data of biomass raw materials and other related data;

[0045] The process data parameters may specifically include microwave frequency, microwave power, material bulk density, material residence time, material initial particle size, heating rate, initial feed temperature, target pyrolysis temperature, microwave absorbing agent information, inert gas flux, pressure, and reactor height-to-diameter ratio.

[0046] In the embodiment of the present invention, the bio-oil yield parameter items mainly include bio-oil yield data.

[0047] S12. Generate modeling data for machine learning based on the time correspondence of historical data; divide the modeling data into a training set and a test set according to a preset division ratio;

[0048] In the embodiment of the present invention, the ratio of the training set to the test set can be determined by those skilled in the art based on experience or a limited number of experiments; preferably, 80% of the training set and 20% of the test set.

[0049] It should be noted that in the embodiments of the present invention, historical data can be used to generate modeling data through appropriate data preprocessing methods according to the needs of modeling; the specific data preprocessing method can be selected by technical personnel in this field based on the characteristics of the data from the data preprocessing methods in the prior art.

[0050] In the embodiment of the present invention, data preprocessing includes:

[0051] The integrity of the data is checked and corrected to prevent model training errors caused by inconsistent data dimensions due to data defaults. Specifically, in an embodiment of the present invention, the isnull().sum() function in the Python library can be used to scan and obtain the integrity of the data. If the value corresponding to the extracted parameter is not 0, the default value NaN corresponding to the parameter is replaced by the average value of the parameter in the selected data to ensure the robustness of the model training. At the same time, due to the differences in the unit orders of magnitude between the parameters, in order to prevent the order of magnitude of individual parameter data from being too large to affect the model judgment, the data corresponding to all parameters obtained by machine learning can also be standardized before training the model.

[0052] Specifically, assuming that there are 10,000 sets of data in the machine learning database, each set of data contains 52 parameters including raw material data package, process data package and bio-oil production information, then the data dimension of the influencing factor X subset is D X = [10000, 51], the data dimension of the target value Y subset is D Y =[10000,1]. When the preset ratio is 20%, the data dimension of the training set X_train is D x_train =[8000,51], the data dimension of the test set X_test is D x_test = [2000,51], the data dimension of the training set Y_train is D y_train =[8000,1], the data dimension of the test set Y_test is D Y_test =[2000,1].

[0053] S13. Based on the random forest regression model, a bio-oil yield prediction model for predicting a bio-oil yield prediction value is generated according to the modeling data, including:

[0054] On the one hand, a plurality of decision trees are established according to a preset training method based on the training set to establish a random forest regression model; on the other hand, the determination coefficient of the random forest regression model is calculated according to the test set, and the random forest regression model is adjusted according to a preset parameter adjustment method to generate a bio-oil yield prediction model.

[0055] In practical applications, in order to ensure the versatility of the model, the number of data groups in the training set of the embodiment of the present invention is preferably higher than 200 groups. At the same time, in order to ensure the differences between decision trees, the number of data groups captured by each decision tree preferably accounts for 40% to 60% of the number of data groups in the training set.

[0056] In an embodiment of the present invention, a plurality of decision trees are established according to a preset training method based on a training set to establish a random forest regression model. The specific process may include:

[0057] S21. Set the number of decision trees N. Each decision tree randomly grabs data with replacement from the training sets X_train and Y_train according to a preset ratio as the training data set T of the decision tree. i ;

[0058] S22, performing pre-pruning of binary tree partitioning, including:

[0059] For the i-th decision tree, the training data set T i The parameters in are divided according to the preset weight critical values ​​and stored in subsets Param_i-1 and Param_i-2 respectively; wherein, subset Param_i-1 is used to store parameters whose weights are greater than the weight critical value; subset Param_i-2 is used to store parameters whose weights are less than the weight critical value; it should be noted that the weight critical value in the embodiment of the present invention can be determined by those skilled in the art based on experience or a limited number of experiments, and no specific limitation is made here.

[0060] In the embodiment of the present invention, the subset Param_i-1 generally stores the parameter items that have a significant impact on the biomass microwave flash pyrolysis bio-oil yield, such as the maximum pyrolysis temperature, microwave power, etc. The advantage of doing so is that the parameters that affect the bio-oil yield are binary divided in advance, which can reduce the complexity of the decision tree and the response time of the system calculation loss function, and improve the prediction accuracy. Therefore, the provisions of the weight critical value must meet the above requirements.

[0061] For the i-th decision tree, when performing binary tree partitioning on its j-th layer, if the subset Param_i-1 containing k parameters still has parameters that are not marked for use, then all values ​​of all parameters in Param_i-1 are traversed, and the mean square error (MSE) is used to calculate its loss function; if all parameters in the subset Param_i-1 are marked for use, then all values ​​of all parameters in Param_i-2 are traversed, and the mean square error (MSE) is used to calculate its loss function; in this embodiment of the present invention, the traversal is used to calculate the optimal splitting variables and the optimal splitting points of the j-th layer.

[0062] In the embodiment of the present invention, when traversing and calculating the optimal segmentation variable and the optimal segmentation point of the j-th layer, all values ​​of all parameters in Param_i-1 are traversed, and the loss function thereof is calculated using the mean square error, which may specifically include:

[0063] Select parameter x k and its corresponding value S, and the parameter x k The corresponding set is divided into two subsets R1 and R2, where:

[0064] R1={x k ≤S}

[0065] R2={x k >S}

[0066] The calculation formula of the loss function includes:

[0067]

[0068] Among them, x k Represents the set of parameters in the subset Param_i-1 that are not marked for use; x i Represents all values ​​of the kth parameter; Represents the output mean of all samples in subset R1; represents the output mean of all samples in subset R2; y i represents the true output value corresponding to all values ​​of the kth parameter; n represents the number of data in the kth parameter;

[0069] For the i-th decision tree, select the parameter x with the smallest loss function k and its value S as the optimal segmentation variable and optimal segmentation point of the jth layer, and the parameter x k Mark it.

[0070] Furthermore, in the embodiment of the present invention, after determining the optimal partitioning variable and the optimal splitting point of each layer, the output mean corresponding to the partitioned subsets can also be brought in and calculated. The specific calculation formula is:

[0071]

[0072] Among them, N m For R m The number of data of the characteristic parameters in .

[0073] The above binary tree partitioning and the process of determining the optimal segmentation variables and the optimal segmentation points are repeatedly executed until the stopping condition is met.

[0074] In an embodiment of the present invention, the stopping condition refers to when a preset division depth is reached, or when a preset number of leaf nodes is reached, or when the data cannot be divided any further, the division of the binary tree is stopped, and the three stopping conditions are stopped when they are reached first, and they do not have to be met at the same time. In an embodiment of the present invention, the preset division depth can be preferably 20 layers, and the preset number of leaf nodes can be preferably 10,000. In an embodiment of the present invention, data cannot be divided any further means that the data can no longer be meaningfully divided, for example, there are no feature parameters for continued division, or the number of samples is insufficient to support further division, then the division of the binary tree is stopped.

[0075] In the embodiment of the present invention, when the stopping condition is met, the establishment of the decision tree is completed.

[0076] For each decision tree, the output value of the leaf node corresponding to its end is the bio-oil production information corresponding to the decision tree.

[0077] When all decision trees are built, the training of the random forest regression model is completed.

[0078] In an embodiment of the present invention, predicting the test set is a key step in verifying the training effect of the random forest regression model. Since the random forest regression model contains N decision trees with large differences, for each set of data in the test set, each decision tree established through training can output a corresponding prediction result. The random forest regression model then takes the average value based on the output results of all decision trees, which is the final prediction result of the model.

[0079] In the embodiment of the present invention, the determination coefficient R of the random forest regression model is calculated based on the test set. 2 The specific process of generating a bio-oil yield prediction model by adjusting the parameters of the random forest regression model according to a preset parameter adjustment method may include:

[0080] First, the determination coefficient R of the random forest regression model is calculated based on the test set. 2 ; The formula used to generate the predicted value of bio-oil production includes:

[0081]

[0082] in, represents the predicted value of bio-oil yield, N represents the number of decision trees in the random forest regression model, represents the output value of the i-th decision tree for the prediction of bio-oil production;

[0083] In the embodiment of the present invention, the determination coefficient R 2 It is the key parameter to judge the effectiveness of training, and the determination coefficient R 2 The closer it is to 1, the better the prediction effect is, and the closer it is to 0, the worse the prediction effect is. 2 The calculation formula includes:

[0084]

[0085] Among them, y i represents the actual value of the bio-oil yield in the prediction set, and n represents the number of groups of data in the prediction set.

[0086] Generally speaking, the coefficient of determination R 2Should not be less than 0.8, then the prediction effect of the model is considered to be good. In the embodiment of the present invention, the determination coefficient R 2 can be set to 0.85.

[0087] When the coefficient of determination R 2 When the value is less than a preset value, the bio-oil yield prediction model is generated by adjusting the parameters of the random forest regression model according to a preset parameter adjustment method, specifically including:

[0088] S31, updating the calculation standard of the loss function used to select the optimal splitting variable and the optimal splitting point of the decision tree, so as to reselect the optimal splitting variable and the optimal splitting point of the decision tree; the calculation standard of the loss function includes: mean absolute error (MAE), mean absolute percentage error (MAPE), mean square logarithmic error (MSLE) and Poisson loss (Poisson);

[0089] In the embodiment of the present invention, when the mean absolute error (MAE) is used as the calculation standard of the loss function, the specific formula is as follows:

[0090]

[0091] In the embodiment of the present invention, when the mean absolute percentage error (MAPE) is used as the calculation standard of the loss function, the specific formula is as follows:

[0092]

[0093] In the embodiment of the present invention, when the mean square logarithmic error (MSLE) is used as the calculation standard of the loss function, the specific formula is as follows:

[0094]

[0095] In the embodiment of the present invention, when Poisson loss is used as the calculation standard of the loss function, the specific formula is as follows:

[0096]

[0097] S32, updating the number of decision trees N, including: taking the original number of decision trees N as the minimum number N min , with the preset ratio of the original number of decision trees N as the maximum value N max , with (N min ,N max ] is the selection range, which determines the number of updated decision trees.

[0098] In the embodiment of the present invention, the preset ratio of the number of decision trees ranges from 10% to 100%, and is generally preferably 50%.

[0099] In the embodiment of the present invention, the selection range of the number of decision trees means that after the selection range is set, all possibilities within the range need to be traversed. For example, when 100 decision trees are trained, the determination coefficient R of the test result is 2 If the preset value is not reached, 100 trees can be used as the minimum number of decision trees, and 50% can be used as the preset ratio, then 150 trees can be used as the maximum number of decision trees under the current step size; in this way, (100,150] is used as the selection range of the number of decision trees, that is, from 101, 102, 103, 104, ..., 150, the determination coefficient R is calculated in sequence. 2 , and select the coefficient of determination R 2 The maximum value. If the value is greater than the preset value, the number of decision trees corresponding to the value is the number of decision trees N at the current time step after parameter adjustment.

[0100] Repeat steps S31 and S32 until the determination coefficient R of the test result is found. 2 Until it is greater than the preset value.

[0101] It should be noted that, in the embodiment of the present invention, the order of step S31 and step S32 is not specifically limited, but depends on the actual process conditions and the actual computing power of the computer. For example, when the mean square error (MSE) is used as the calculation standard of the loss function and 100 decision trees are trained, the determination coefficient R in the range of 150, 200, 300, 500 decision trees can be directly traversed and calculated. 2 ; You can also first traverse and calculate the coefficient of determination R when using 100 decision trees and different calculation standards as the loss function 2 ; You can also traverse and calculate the coefficient of determination R when using different calculation standards as the loss function within the range of 150, 200, 300, 500 decision trees. 2 ; In other words, step S31 and step S32 can be calculated independently, in parallel, or coupled.

[0102] If the actual computing power of the computer is insufficient, the time required to generate the prediction results will not meet the requirements of the response time and the preset actual step size, or the response time is too long due to the high complexity of the data. At this time, it is possible to affect the subsequent algorithm's control of the bio-oil yield in the actual biomass microwave flash pyrolysis process. Therefore, further, the embodiment of the present invention also makes corresponding restrictions on the response time for generating the prediction results. Specifically, in the embodiment of the present invention, when the determination coefficient R 2 When it is greater than a preset value, the following steps may also be included:

[0103] S41, determine and calculate the determination coefficient R 2Whether the response time exceeds the preset actual step length, if so, proceed to S42;

[0104] S42, for the established i-th decision tree i, traverse all leaf nodes in the decision tree i to obtain its maximum depth Depth_i;

[0105] S43, back up the optimal segmentation variables and optimal segmentation points of the Depth_i-1 layer, and store them in the data set Backup;

[0106] S44, perform a pruning operation on the decision tree i. Specifically, discard the leaf nodes in the Depth_i layer, and at the same time use the parent nodes in the Depth_i-1 layer as the leaf nodes in the decision tree i, and use Depth_i-1 as the maximum depth of the decision tree i;

[0107] S44, repeating steps S42-S44 until the response time is less than the preset actual step length;

[0108] S45. If the response time of decision tree i is less than the preset actual step within several time steps, the optimal splitting variables and optimal splitting points of the most recent backup in Backup are imported and assigned to the leaf nodes of decision tree i, and its maximum depth is set to +1.

[0109] It should be noted that the above steps (steps S41 to S45) may also be simplified, specifically including:

[0110] S51, when the determination coefficient R 2 When it is greater than the preset value, the determination coefficient R is determined and calculated. 2 Whether the response time exceeds the preset actual step length, if yes, proceed to S52;

[0111] S52, for the i-th decision tree i that has been established, prune all the nodes extending downward in the subset Param_i-1 corresponding to the decision tree i, and store the optimal splitting variables and optimal splitting points corresponding to all pruned parent nodes containing child nodes into the data set Backup for backup;

[0112] If the response time of decision tree i is less than the preset actual step within several time steps, the optimal splitting variable and optimal splitting point of the parameter with the largest weight value in Backup are imported and assigned to the leaf node of decision tree i, and its maximum depth is set to +1.

[0113] Through the above steps S11 to S13, the embodiment of the present invention establishes a prediction algorithm based on the random forest regression model based on the specific application scenario of preparing bio-oil by microwave flash pyrolysis of biomass; since the random forest regression model is based on previous experimental data, it fully considers the coupling relationship of various influencing factors; therefore, the bio-oil yield prediction model generated by the present invention can quickly and accurately predict the bio-oil yield according to the current operating conditions of the biomass microwave flash pyrolysis equipment, thereby providing effective data support for production work such as process optimization and equipment maintenance of the biomass microwave flash pyrolysis equipment.

[0114] Furthermore, in the embodiment of the present invention, during the model training process, the parameter weight factors are also taken into account when performing pre-pruning of the binary tree partitioning. By first calculating the optimal splitting points for high-weight parameters, unnecessary calculations are reduced, thereby effectively optimizing the response time of the calculation process.

[0115] After the bio-oil yield prediction model is constructed, the steps of optimizing the production process of preparing bio-oil by microwave flash pyrolysis of biomass using the bio-oil yield prediction model may specifically include:

[0116] S14, taking the raw material data parameter items and the process data parameter items as input, periodically generating a bio-oil yield prediction value according to the bio-oil yield prediction model;

[0117] In an embodiment of the present invention, in order to ensure that the bio-oil production meets expectations, the current values ​​of the raw material data parameter items and the process data parameter items in the biomass microwave flash pyrolysis equipment can be used as input, and the bio-oil production prediction model can be used to periodically generate the next time step (or multiple time steps later) bio-oil production prediction value, thereby realizing real-time monitoring of the bio-oil production of the biomass microwave flash pyrolysis equipment.

[0118] S15. When the predicted value of the bio-oil production is less than a preset production threshold, the controllable parameter items in the raw material data type parameter items and the process data type parameter items are traversed to obtain the maximum value of the predicted value of the bio-oil production, and the adaptation value of each controllable parameter is generated through the bio-oil production prediction model; the adaptation value of the controllable parameter is used to optimize the production process of the biomass microwave flash pyrolysis equipment.

[0119] In an embodiment of the present invention, the parameter items in the raw material data type parameter items and the process data type parameter items are divided into two types, one is a controllable parameter and the other is a fixed parameter; wherein, the controllable parameter refers to a parameter item that can be adjusted and controlled after the process starts; the bio-oil yield can be affected by adjusting the numerical value of the controllable parameter.

[0120] In an embodiment of the present invention, for controllable parameters, a traversal range can be determined, and the range can be the rated range that the actual device can reach, or the adjustment range manually set by the operator based on past experience. For a given range, the step size can be manually set, and the ranges of several controllable parameters are converted into a data set in the form of step size division, and input into the bio-oil yield prediction model. Finally, through the adaptation of different parameters, the specific prediction of the bio-oil yield by the bio-oil yield prediction model is respectively made, and then the optimal adaptation value of each controllable parameter under the next time step (or after multiple time steps) is obtained (that is, under each controllable parameter combination, the bio-oil yield can be higher than the preset value and the maximum). In this way, according to the optimal adaptation value of each controllable parameter, the control instructions of the biomass microwave flash pyrolysis device can be generated to make the biomass microwave flash pyrolysis device work in the best state.

[0121] From the above, it can be seen that in the embodiments of the present invention, the adaptation value of each controllable parameter is generated with the goal of obtaining the maximum value of the predicted value of bio-oil yield; thereby, a control strategy for optimizing the production process of biomass microwave flash pyrolysis equipment can be obtained; it can be seen that compared with the laboratory data analysis or human experience judgment in the prior art, the present invention not only has strong versatility and robustness, but also can effectively avoid the investment of a large amount of repetitive labor, thereby improving the efficiency and effect of the optimization of microwave heating microwave flash pyrolysis process.

[0122] Embodiment 2

[0123] In another aspect of the embodiment of the present invention, a microwave flash pyrolysis process optimization device is also provided. Figure 2 The schematic diagram of the structure of the microwave flash pyrolysis process optimization device provided by the embodiment of the present invention is shown. Figure 1 The device corresponding to the microwave flash pyrolysis process optimization method described in the corresponding embodiment, that is, implemented by means of a virtual device Figure 1 The microwave flash pyrolysis process optimization method in the corresponding embodiment and each virtual module constituting the microwave flash pyrolysis process optimization device can be executed by an electronic device, such as a network device, a terminal device, or a server. Specifically, the microwave flash pyrolysis process optimization device in the embodiment of the present invention includes:

[0124] A model generation unit comprising a bio-oil yield prediction model; the model generation unit comprises:

[0125] A historical data acquisition module is used to acquire historical data including multiple parameter items of the biomass microwave flash pyrolysis equipment; the parameter items include: raw material data parameter items, process data parameter items and bio-oil output parameter items;

[0126] A modeling data generation module, used to generate modeling data for machine learning according to the time correspondence between the historical data; the modeling data is divided into a training set and a test set according to a preset division ratio;

[0127] A prediction model generation module is used to generate a bio-oil yield prediction model for predicting a bio-oil yield prediction value based on the modeling data based on a random forest regression model, comprising:

[0128] Establishing a plurality of decision trees according to the training set and a preset training method to establish a random forest regression model;

[0129] The determination coefficient of the random forest regression model is calculated according to the test set, and the bio-oil yield prediction model is generated by adjusting the parameters of the random forest regression model according to a preset parameter adjustment method.

[0130] Preferably, in the embodiment of the present invention, it may also include

[0131] A result prediction unit, used for periodically generating a bio-oil yield prediction value according to the bio-oil yield prediction model;

[0132] A process optimization unit is used to traverse the controllable parameter items in the raw material data type parameter items and the process data type parameter items when the bio-oil yield prediction value is less than a preset yield threshold, so as to obtain the maximum value of the bio-oil yield prediction value as the goal, and generate adaptation values ​​of each controllable parameter through the bio-oil yield prediction model; the adaptation values ​​of the controllable parameters are used to optimize the production process of the biomass microwave flash pyrolysis equipment.

[0133] Since the working principle and beneficial effects of the microwave flash pyrolysis process optimization device in the embodiment of the present invention have been Figure 1 The corresponding microwave flash pyrolysis process optimization method is also recorded and described, so they can be cross-referenced and will not be repeated here.

[0134] Embodiment 3

[0135] Corresponding to the method embodiment, in the embodiment of the present invention, there is also provided a microwave flash pyrolysis process optimization device, such as a terminal, a server, etc. Among them, the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc., but is not limited thereto.

[0136] An example diagram of a hardware structure block diagram of a microwave flash pyrolysis process optimization device provided in an embodiment of the present invention is shown in FIG. Figure 3 As shown, it may include:

[0137] Processor 1, communication interface 2, memory 3 and communication bus 4;

[0138] The processor 1, the communication interface 2, and the memory 3 communicate with each other via the communication bus 4;

[0139] Optionally, the communication interface 2 may be an interface of a communication module, such as an interface of a GSM module;

[0140] The processor 1 may be a central processing unit CPU, or an application-specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present application.

[0141] The memory 3 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0142] The processor 1 is specifically used to execute the computer program stored in the memory 3 to perform the following steps:

[0143] The steps for generating the bio-oil yield prediction model include:

[0144] S11, acquiring historical data including multiple parameter items of a biomass microwave flash pyrolysis device; the parameter items include: raw material data parameter items, process data parameter items, and bio-oil output parameter items;

[0145] S12. Generate modeling data for machine learning based on the time correspondence of the historical data; the modeling data is divided into a training set and a test set according to a preset division ratio;

[0146] S13, based on the random forest regression model, generating a bio-oil yield prediction model for predicting the bio-oil yield prediction value according to the modeling data, including:

[0147] Establishing a plurality of decision trees according to the training set and a preset training method to establish a random forest regression model;

[0148] The determination coefficient of the random forest regression model is calculated according to the test set, and the bio-oil yield prediction model is generated by adjusting the parameters of the random forest regression model according to a preset parameter adjustment method.

[0149] Preferably, the present invention further comprises:

[0150] S14, periodically generating a bio-oil production prediction value according to the bio-oil production prediction model;

[0151] S15. When the predicted value of the bio-oil production is less than a preset production threshold, the controllable parameter items in the raw material data type parameter items and the process data type parameter items are traversed to obtain the maximum value of the predicted value of the bio-oil production, and the adaptation value of each controllable parameter is generated through the bio-oil production prediction model; the adaptation value of the controllable parameter is used to optimize the production process of the biomass microwave flash pyrolysis equipment.

[0152] The above product can execute the method provided in the embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not described in detail in this embodiment, please refer to the microwave flash pyrolysis process optimization method provided in the embodiment of the present invention.

[0153] Embodiment 4

[0154] In an embodiment of the present invention, a storage medium is further provided, which may store a program suitable for execution by a processor, wherein the program is used to:

[0155] The steps for generating the bio-oil yield prediction model include:

[0156] S11, acquiring historical data including multiple parameter items of a biomass microwave flash pyrolysis device; the parameter items include: raw material data parameter items, process data parameter items, and bio-oil output parameter items;

[0157] S12. Generate modeling data for machine learning based on the time correspondence of the historical data; the modeling data is divided into a training set and a test set according to a preset division ratio;

[0158] S13, based on the random forest regression model, generating a bio-oil yield prediction model for predicting the bio-oil yield prediction value according to the modeling data, including:

[0159] Establishing a plurality of decision trees according to the training set and a preset training method to establish a random forest regression model;

[0160] The determination coefficient of the random forest regression model is calculated according to the test set, and the bio-oil yield prediction model is generated by adjusting the parameters of the random forest regression model according to a preset parameter adjustment method.

[0161] Preferably, the present invention further comprises:

[0162] S14, periodically generating a bio-oil production prediction value according to the bio-oil production prediction model;

[0163] S15. When the predicted value of the bio-oil production is less than a preset production threshold, the controllable parameter items in the raw material data type parameter items and the process data type parameter items are traversed to obtain the maximum value of the predicted value of the bio-oil production, and the adaptation value of each controllable parameter is generated through the bio-oil production prediction model; the adaptation value of the controllable parameter is used to optimize the production process of the biomass microwave flash pyrolysis equipment.

[0164] Optionally, the detailed functions and extended functions of the program may refer to the above description.

[0165] The above product can execute the method provided by the embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not described in detail in this embodiment, please refer to the methods provided by other embodiments of the present invention.

[0166] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0167] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0168] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0169] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0170] It should be understood that in the embodiments of the present application, the various embodiments and features can be combined with each other to solve the aforementioned technical problems.

[0171] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0172] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A microwave flash pyrolysis process optimization method, characterized in that: The steps for generating the bio-oil yield prediction model include: S11, acquiring historical data including multiple parameter items of a biomass microwave flash pyrolysis device; the parameter items include: raw material data parameter items, process data parameter items, and bio-oil output parameter items; S12. Generate modeling data for machine learning based on the time correspondence of the historical data; the modeling data is divided into a training set and a test set according to a preset division ratio; S13, based on the random forest regression model, generating a bio-oil yield prediction model for predicting the bio-oil yield prediction value according to the modeling data, including: Establishing a plurality of decision trees according to the training set and a preset training method to establish a random forest regression model; The determination coefficient of the random forest regression model is calculated according to the test set, and the bio-oil yield prediction model is generated by adjusting the parameters of the random forest regression model according to a preset parameter adjustment method.

2. The microwave flash pyrolysis process optimization method according to claim 1, characterized in that: Also includes: S14, taking the raw material data parameter items and the process data parameter items as input, periodically generating a bio-oil yield prediction value according to the bio-oil yield prediction model; S15. When the predicted value of the bio-oil production is less than a preset production threshold, the controllable parameter items in the raw material data type parameter items and the process data type parameter items are traversed to obtain the maximum value of the predicted value of the bio-oil production, and the adaptation value of each controllable parameter is generated through the bio-oil production prediction model; the adaptation value of the controllable parameter is used to optimize the production process of the biomass microwave flash pyrolysis equipment.

3. The microwave flash pyrolysis process optimization method according to claim 1 or 2, characterized in that: The step of establishing a plurality of decision trees according to the training set and a preset training method to establish a random forest regression model includes: S21. Set the number of decision trees N. Each decision tree randomly grabs data with replacement from the training sets X_train and Y_train according to a preset ratio as the training data set T of the decision tree. i ; S22, performing pre-pruning of binary tree partitioning, including: For the i-th decision tree, the training data set T i The parameters in are divided according to the preset weight critical value and stored in subsets Param_i-1 and Param_i-2 respectively; the subset Param_i-1 is used to store parameters with weights greater than the weight critical value; the subset Param_i-2 is used to store parameters with weights less than the weight critical value; For the i-th decision tree, when performing binary tree partitioning on its j-th layer, if the subset Param_i-1 containing k parameters still has parameters that are not marked for use, then all values ​​of all parameters in Param_i-1 are traversed, and the mean square error is used to calculate its loss function; if all parameters in the subset Param_i-1 are marked for use, then all values ​​of all parameters in Param_i-2 are traversed, and the mean square error is used to calculate its loss function; the traversal is used to calculate the optimal splitting variable and the optimal splitting point of the j-th layer.

4. The microwave flash pyrolysis process optimization method according to claim 1, characterized in that: The calculating the coefficient of determination of the random forest regression model according to the test set comprises: The formula used by the random forest regression model to generate the predicted value of bio-oil production includes: in, represents the predicted value of bio-oil yield, N represents the number of decision trees in the random forest regression model, represents the output value of the i-th decision tree for the prediction of bio-oil production; The coefficient of determination R 2 The calculation formula includes: Among them, y i represents the actual value of the bio-oil yield in the prediction set, and n represents the number of groups of data in the prediction set.

5. The microwave flash pyrolysis process optimization method according to claim 4, characterized in that: include: When the determination coefficient R 2 When the value is less than a preset value, the bio-oil yield prediction model is generated by adjusting the parameters of the random forest regression model according to a preset parameter adjustment method, including: S31, updating the calculation criteria of the loss function used to select the optimal splitting variable and the optimal splitting point of the decision tree; the calculation criteria of the loss function include: mean absolute error, mean absolute percentage error, mean square logarithmic error and Poisson loss; S32, updating the number of decision trees N, including: taking the original number of decision trees N as the minimum number N min , with the preset ratio of the original number of decision trees N as the maximum value N max , with (N min ,N max ] is the selection range and determines the number of updated decision trees.

6. The microwave flash pyrolysis process optimization method according to claim 4, characterized in that: Also includes: S41, when the determination coefficient R 2 When it is greater than the preset value, the determination coefficient R is determined and calculated. 2 Whether the response time exceeds the preset actual step length, if so, proceed to S42; S42, for the established i-th decision tree i, traverse all leaf nodes in the decision tree i to obtain its maximum depth Depth_i; S43, back up the optimal segmentation variables and optimal segmentation points of the Depth_i-1 layer, and store them in the data set Backup; S44, perform a pruning operation on the decision tree i. Specifically, discard the leaf nodes in the Depth_i layer, and at the same time use the parent nodes in the Depth_i-1 layer as the leaf nodes in the decision tree i, and use Depth_i-1 as the maximum depth of the decision tree i; S44, repeating steps S42-S44 until the response time is less than the preset actual step length; S45. If the response time of decision tree i is less than the preset actual step within several time steps, the optimal splitting variables and optimal splitting points of the most recent backup in Backup are imported and assigned to the leaf nodes of decision tree i, and its maximum depth is set to +1.

7. The microwave flash pyrolysis process optimization method according to claim 4, characterized in that: Also includes: S51, when the determination coefficient R 2 When it is greater than the preset value, the determination coefficient R is determined and calculated. 2 Whether the response time exceeds the preset actual step length, if yes, proceed to S52; S52, for the i-th decision tree i that has been established, prune all the nodes extending downward in the subset Param_i-1 corresponding to the decision tree i, and store the optimal splitting variables and optimal splitting points corresponding to all pruned parent nodes containing child nodes into the data set Backup for backup; If the response time of decision tree i is less than the preset actual step within several time steps, the optimal splitting variable and optimal splitting point of the parameter with the largest weight value in Backup are imported and assigned to the leaf node of decision tree i, and its maximum depth is set to +1.

8. The microwave flash pyrolysis process optimization method according to claim 3, characterized in that: The optimal splitting variable and optimal splitting point of the jth layer are calculated by traversing, including: traversing all values ​​of all parameters in Param_i-1, and using the mean square error to calculate its loss function, including: Select parameter x k and its corresponding value S, and the parameter x k The corresponding set is divided into two subsets R1 and R2, where: R1={x k ≤S} R2={x k >S} The calculation formula of the loss function includes: Among them, x k Represents the set of parameters in the subset Param_i-1 that are not marked for use; x i Represents all values ​​of the kth parameter; Represents the output mean of all samples in subset R1; represents the output mean of all samples in subset R2; y i represents the true output value corresponding to all values ​​of the kth parameter; n represents the number of data in the kth parameter; For the i-th decision tree, select the parameter x with the smallest loss function k and its value S as the optimal segmentation variable and optimal segmentation point of the jth layer, and the parameter x k Mark it.

9. The microwave flash pyrolysis process optimization method according to claim 8, characterized in that: After determining the optimal partitioning variables and optimal splitting points for each layer, it also includes: Bring in and calculate the output mean corresponding to the divided subsets. The calculation formula includes: Among them, N m For R m The number of data of the characteristic parameters in .

10. A microwave flash pyrolysis process optimization device, characterized in that: A model generation unit comprising a bio-oil yield prediction model; the model generation unit comprises: A historical data acquisition module is used to acquire historical data including multiple parameter items of the biomass microwave flash pyrolysis equipment; the parameter items include: raw material data parameter items, process data parameter items and bio-oil output parameter items; A modeling data generation module, used to generate modeling data for machine learning according to the time correspondence between the historical data; the modeling data is divided into a training set and a test set according to a preset division ratio; A prediction model generation module is used to generate a bio-oil yield prediction model for predicting a bio-oil yield prediction value based on the modeling data based on a random forest regression model, comprising: Establishing a plurality of decision trees according to the training set and a preset training method to establish a random forest regression model; The determination coefficient of the random forest regression model is calculated according to the test set, and the bio-oil yield prediction model is generated by adjusting the parameters of the random forest regression model according to a preset parameter adjustment method.

11. The microwave flash pyrolysis process optimization device according to claim 10, characterized in that: Also includes: A result prediction unit, used to take raw material data parameter items and process data parameter items as input, and periodically generate a bio-oil yield prediction value according to the bio-oil yield prediction model; A process optimization unit is used to traverse the controllable parameter items in the raw material data type parameter items and the process data type parameter items when the bio-oil yield prediction value is less than a preset yield threshold, so as to obtain the maximum value of the bio-oil yield prediction value as the goal, and generate adaptation values ​​of each controllable parameter through the bio-oil yield prediction model; the adaptation values ​​of the controllable parameters are used to optimize the production process of the biomass microwave flash pyrolysis equipment.

12. A microwave flash pyrolysis process optimization device, characterized in that: include: Memory for storing computer programs; A processor is used to call and execute the computer program to implement the steps of the microwave flash pyrolysis process optimization method as described in any one of claims 1 to 9.

13. A storage medium, characterized in that: The method comprises a software program, wherein the software program is suitable for executing the steps of the microwave flash pyrolysis process optimization method as claimed in any one of claims 1 to 9 by a processor.

Images