Optimization method for multiple process parameters of grain dryer

Through gray correlation analysis and KNN regression model chain, the process parameters of the grain dryer are optimized, the problems of low mechanized drying efficiency and energy waste are solved, and the efficient and green grain drying process is achieved, which reduces grain losses and production costs.

CN118963118APending Publication Date: 2024-11-15WUHAN POLYTECHNIC UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411006498.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-25
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The efficiency of mechanized drying grains is caused by improper equipment parameters and mismatch of physical characteristics, which reduces drying efficiency and increases energy consumption and food loss.

Method used

Using multi-process parameter optimization method, through gray correlation analysis and KNN regression model chain, the relationship between the physical characteristics of grain, environmental parameters and drying processing process parameters is established, the optimal processing parameter group is determined, and independent control and parameter group selection is realized.

Benefits of technology

It improves the efficiency of grain drying, reduces grain losses, reduces production costs, and promotes the green and intelligent development of grain drying technology.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118963118A_ABST
    Figure CN118963118A_ABST
Patent Text Reader

Abstract

The invention relates to an optimization method for multi-process parameters of a grain dryer, which can quickly and effectively extract a significant relationship between grain drying processing process parameters and processing benefits, is suitable for multivariable and multi-objective structure optimization design, can effectively reduce the selection quantity of parameter groups for grain drying processing, and improves the processing efficiency. The method has a good guiding effect on grain drying processing. According to the method, the function relationship among the grain physical characteristic parameters, the environment parameters and the grain drying processing technological parameters is established, and autonomous control over the grain drying processing parameters is met. According to the method, the optimal processing parameter group is selected from the multiple groups of qualified grain drying processing technological parameters, so that the grain drying processing efficiency is the highest, the loss caused by the fact that grains are not dried in time and the loss caused during drying processing are reduced, the production cost is reduced, and the green and intelligent development of the grain drying technology is promoted.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of grain drying technology, and in particular to a method for optimizing multiple process parameters of a grain dryer. Background Art

[0002] Grain drying technology is an important link to ensure a good harvest of grain, and it is of great significance to ensure national food security. Data show that due to untimely processing or improper storage and processing, my country loses about 70 billion jin of grain every year in the storage, transportation and processing links. Grain drying technology is the main technology for controlling grain moisture. It is a technology that reduces the moisture content of harvested wet grains and controls it within a safe range by heating them through mechanical equipment. Compared with traditional drying methods, mechanized drying improves the drying efficiency of grains, but the efficiency of mechanized drying of grains will be unbalanced due to improper equipment parameters and mismatch with the current physical properties of grains, resulting in reduced drying efficiency. This operation method that cannot be adjusted in time not only greatly reduces the utilization rate of the dryer, increases energy consumption, and may even increase grain losses. For this reason, there is an urgent need for a grain drying method that improves grain drying efficiency and reduces grain losses.

[0003] Improve the drying capacity of grain production areas and follow the trend of green and intelligent development. The intelligence level of drying facilities and equipment should be improved, energy-saving and efficient green technologies and equipment should be developed and used, and existing grain dryers should be upgraded and renovated to be environmentally friendly and energy-saving to make them more in line with the development requirements of green agriculture. Summary of the invention

[0004] In view of the above problems, the present invention provides a method for optimizing multiple process parameters of a grain dryer.

[0005] The present invention adopts the following technical solutions:

[0006] A method for optimizing multiple process parameters of a grain dryer comprises the following steps:

[0007] Step 1, collecting multiple grain dryer processing parameters corresponding to different processing benefits, where there are n processing benefits in total, and each processing benefit corresponds to m processing parameters, and performing average processing on the collected data to obtain averaged data;

[0008] Step 2: Perform grey correlation analysis on the averaged processing parameters and processing benefits, take each processing parameter as an influencing factor, and obtain the correlation coefficient matrix between each processing parameter and processing benefit. According to the obtained correlation coefficient matrix, calculate the correlation degree r of each influencing factor. i ;

[0009] Step 3: Sort the m processing parameters according to their correlation from large to small to obtain the main sequence [R1, R2, …, …, R m ];

[0010] Step 4: For the pth processing parameter, set the processing parameter R p , put it in the first position, and the rest of the order is the same as the main sequence, and the sorting sequence Q corresponding to the pth processing parameter is obtained p =[R p ,…,R p-1 ,R p+1 ,…,R m ], and further obtain the sorting sequence corresponding to m processing parameters;

[0011] Step 5: Each parameter of each processing forms a set of samples, namely {z k =(x k ,y k ),k=1,2,…,n}; using the sample {(x k ,y k ),k=1,2,…,n} construct about the sequence Q p The m-level KNN regression model chain is obtained according to the m sequences. In each level of each model chain, x k is the input data, y k is the output data. For the mth level of the pth model chain, in each model, x k is the input data, y k is the output data. For the p-th regression model chain, in the m-th level model, x k Contains basic decision variables x0 and additional decision variables x rk ; is the initial setting value, d is the number of basic decision variables; is the actual value of each processing parameter in the kth sample, x rk The order of the processing parameters in is the sorting sequence Q corresponding to the pth processing parameter p ; Where m is the number of drying process parameters;

[0012] Step 6: Determine the optimal K value of each level of the m KNN regression model chains through cross-validation method, and then obtain the m most suitable KNN regression model chains;

[0013] Step 7: The basic decision variable x0 during the current processing ′ As the initial data input, the first level of the first KNN regression model chain in the model chain obtained in step 6, the model outputs the predicted value Will As an additional variable and the basic decision variable x0 ′ Together as the input of the second level of the first KNN regression model chain, get the corresponding output, traverse each level of the first model chain in turn, the input of the mth level of the first model chain is x0 ′ and After traversing the m-level models of the first KNN regression model chain, the output vector of the corresponding model chain is

[0014] Traverse m model chains in sequence to obtain m output vectors to form a vector group

[0015] Step 8: According to Pareto dominance, perform non-dominated hierarchical sorting on each row of the output vector group in the previous step. r , and the vector Pareto improvement margin corresponding to each layer of vector groups increases from top to bottom;

[0016] Step 9: Take the output vector set of the first layer, calculate the processing benefit for each row of vectors, and take the row with the largest processing benefit as the output vector, which is the optimal processing parameter vector F(p).

[0017] Furthermore, the step 2 specifically includes the following steps:

[0018] Step 2.1: Perform grey correlation analysis on each drying process parameter and processing benefit to obtain the correlation coefficient matrix. The elements in the matrix are ξ i (k),ξ i The calculation formula of (k) is:

[0019]

[0020] x0(k) is the processing benefit, k = 1, 2, ..., n, n is the number of samples; x i (k) is the processing parameter, i = 1, 2, ..., m, m is the number of processing parameters; ρ is the preset resolution coefficient;

[0021] Step 2.2: According to the obtained correlation coefficient matrix, calculate the correlation degree r of each influencing factor i , that is, the correlation degree between each processing parameter and processing benefit, the formula is:

[0022]

[0023] Furthermore, in step 5, the processing parameters in each sample are standardized, and the formula is:

[0024]

[0025] Among them, x new is the data after standardized processing, mean(x) is the mean value, and std(x) is the standard deviation.

[0026] After adopting the above technical solution, the present invention has the following advantages compared with the prior art:

[0027] The present invention can quickly and effectively extract the significant relationship between the grain drying process parameters and the processing benefits, is suitable for multi-variable and multi-objective structural optimization design, can effectively reduce the selection amount of parameter groups for grain drying processing, and has a good guiding role in grain drying processing.

[0028] The present invention establishes a functional relationship between grain physical property parameters, environmental parameters and grain drying process parameters to meet the autonomous control of grain drying process parameters.

[0029] The present invention selects the optimal processing parameter group from multiple groups of qualified grain drying process parameters, so that the efficiency of grain drying processing is the highest, the loss of grain not dried in time and the loss during drying processing are reduced, the production cost is reduced, and the green and intelligent development of grain drying technology is promoted.

[0030] The present invention is described in detail below with reference to the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is a flow chart of the method of the present invention;

[0032] Figure 2 is the construction flow chart of sequence k;

[0033] Figure 3 Flowchart for building the KNN regression model chain. DETAILED DESCRIPTION

[0034] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0035] like Figure 1 As shown, the method for optimizing multiple process parameters of a grain dryer of the present invention comprises the following steps:

[0036] 1. Data Preprocessing

[0037] Collect the drying process parameters and processing efficiency data in the grain drying process, and standardize the data. The formula is:

[0038]

[0039] x i′ (k) is the averaged data, x i (k) is the initial sample data of the i-th category, The mean of the i-th type of variable. n is the number of samples, and m is the number of influencing factors.

[0040] 2. Correlation Analysis

[0041] The grey correlation analysis of each drying process parameter and processing benefit is carried out to obtain the correlation coefficient matrix, the formula is:

[0042]

[0043] x0(k) is the processing benefit, k = 1, 2, ..., n, n is the number of samples; x i (k) is the influencing factor such as the inlet air temperature, i = 1, 2, ..., m, m is the number of influencing factors; ρ is the resolution coefficient, usually taken as 0.5.

[0044] According to the obtained correlation coefficient matrix, the correlation degree of each influencing factor (processing parameters) and the processing benefit r are obtained. i , the formula is:

[0045]

[0046] Sort the m processing parameters according to their correlation from large to small and get the main sequence [R1, R2, …, …, R m ], for the pth processing parameter, the processing parameter R p , put it in the first position, and the rest of the order is the same as the main sequence, and the sorting sequence Q corresponding to the pth processing parameter is obtained p =[R p ,…,R p-1 ,R p+1 ,…,R m ], and further obtain the sorting sequence corresponding to m processing parameters;

[0047] 3. Establishing KNN regression model chain

[0048] like Figure 3 As shown, the physical property parameters of grains during the drying process of the grain dryer are collected, including grain type, grain temperature, grain moisture, porosity of the grain pile, and impurity rate; environmental parameters, including external temperature and external humidity; process parameters, including inlet air volume, inlet air temperature, and grain pile feeding speed, etc., are collected to construct the sample space.

[0049] The collected grain physical property parameters, environmental parameters, and process parameters are standardized, and the formula is:

[0050]

[0051] Among them, x new is the data after standardized processing, mean(x) is the mean value, and std(x) is the standard deviation.

[0052] Each parameter of each processing constitutes a set of samples, namely {z k =(x k ,y k ),k=1,2,…,n};

[0053] Using the sample {(x k ,y k ),k=1,2,…,n} construct about the sequence Q p The m-level KNN regression model chain is obtained according to the m sequences. In each level of each model chain, x k is the input data, y k is the output data. For the mth level of the pth model chain, in each model, x k is the input data, y k is the output data. For the p-th regression model chain, in the m-th level model, x k Contains basic decision variables x0 and additional decision variables x rk ; is the initial setting value, d is the number of basic decision variables; is the actual value of each processing parameter in the kth sample, x rk The order of the processing parameters in is the sorting sequence Q corresponding to the pth processing parameter p ; Where m is the number of drying process parameters;

[0054] The optimal K value of each level of the m KNN regression model chains is determined by cross-validation method, and then the m most suitable KNN regression model chains are obtained;

[0055] That is, calculate the distance D(z a ,z b ),in n is the number of variables;

[0056] Calculate the test sample z k The average of the regression values ​​of the samples with the smallest distance D The sample z k of With the average By comparison, when the error is minimum, K is the optimal value.

[0057] 4. Output vector group

[0058] The basic decision variable x0 during the current processing ′ The first level of the first KNN regression model chain in the model chain obtained as the initial data input, the model outputs the predicted value Will As an additional variable and the basic decision variable x0 ′ Together as the input of the second level of the first KNN regression model chain, get the corresponding output, traverse each level of the first model chain in turn, the input of the mth level of the first model chain is x0 ′ and After traversing the m-level models of the first KNN regression model chain, the output vector of the corresponding model chain is Traverse m model chains in sequence to obtain m output vectors to form a vector group

[0059] 5. Choose the optimal processing parameters

[0060] According to Pareto dominance, the above output vector groups are sorted in a non-dominated hierarchical manner. r , which makes the space for Pareto improvement of the vectors in the first layer smaller, and the space for Pareto improvement of the vectors in the last layer larger.

[0061] Take the output vector set of the first layer, calculate the processing benefit for each row of vectors, and take the row with the largest processing benefit as the output vector, which is the optimal processing parameter vector F(p).

[0062] The above is an example of the best implementation of the present invention, and the parts not described in detail are common knowledge of ordinary technicians in the field. The protection scope of the present invention shall be based on the content of the claims, and any equivalent transformation based on the technical enlightenment of the present invention is also within the protection scope of the present invention.

Claims

1. A method for optimizing multiple process parameters of a grain dryer, characterized in that: The following steps are involved: Step 1, collecting multiple grain dryer processing parameters corresponding to different processing benefits, wherein there are n processing benefits in total, and each processing benefit corresponds to m processing parameters, and performing average processing on the collected data to obtain averaged data; Step 2: Perform grey correlation analysis on the averaged processing parameters and processing benefits, take each processing parameter as an influencing factor, obtain the correlation coefficient matrix between each processing parameter and processing benefit, and calculate the correlation degree r of each influencing factor based on the obtained correlation coefficient matrix. i ; Step 3: Sort the m processing parameters according to their correlation from large to small to obtain the main sequence [R1, R2, …, …, R m ]; Step 4: For the pth processing parameter, set the processing parameter R p , put it in the first position, and the rest of the order is the same as the main sequence, and the sorting sequence Q corresponding to the pth processing parameter is obtained p =[R p ,…,R p-1 ,R p+1 ,…,R m ], and further obtain the sorting sequence corresponding to m processing parameters; Step 5: Each parameter of each processing forms a set of samples, namely {z k =(x k ,y k ),k=1,2,…,n}; using the sample {(x k ,y k ),k=1,2,…,n} construct about the sequence Q p The m-level KNN regression model chain is obtained according to the m sequences. In each level of each model chain, x k is the input data, y k is the output data. For the mth level of the pth model chain, in each model, x k is the input data, y k is the output data. For the p-th regression model chain, in the m-th level model, x k Contains basic decision variables x0 and additional decision variables x rk ; is the initial setting value, d is the number of basic decision variables; is the actual value of each processing parameter in the kth sample, x rk The order of the processing parameters in is the sorting sequence Q corresponding to the pth processing parameter p ; Where m is the number of drying process parameters; Step 6: Determine the optimal K value of each level of the m KNN regression model chains through cross-validation method, and then obtain the m most suitable KNN regression model chains; Step 7: The basic decision variable x0 during the current processing ′ As the initial data input, the first level of the first KNN regression model chain in the model chain obtained in step 6, the model outputs the predicted value Will As an additional variable and the basic decision variable x0 ′ Together as the input of the second level of the first KNN regression model chain, get the corresponding output, traverse each level of the first model chain in turn, the input of the mth level of the first model chain is x0 ′ and After traversing the m-level models of the first KNN regression model chain, the output vector of the corresponding model chain is Traverse m model chains in sequence to obtain m output vectors to form a vector group Step 8: According to Pareto dominance, perform non-dominated hierarchical sorting on each row of the output vector group in the previous step. r , and the vector Pareto improvement margin corresponding to each layer of vector groups increases from top to bottom; Step 9: Take the output vector set of the first layer, calculate the processing benefit for each row of vectors, and take the row with the largest processing benefit as the output vector, which is the optimal processing parameter vector F(p).

2. The method for optimizing multiple process parameters of a grain dryer according to claim 1, characterized in that: The step 2 specifically includes the following steps: Step 2.1: Perform grey correlation analysis on each drying process parameter and processing benefit to obtain the correlation coefficient matrix. The elements in the matrix are ξ i (k),ξ i The calculation formula of (k) is: x0(k) is the processing benefit, k = 1, 2, ..., n, n is the number of samples; x i (k) is the processing parameter, i = 1, 2, ..., m, m is the number of processing parameters; ρ is the preset resolution coefficient; Step 2.2: According to the obtained correlation coefficient matrix, calculate the correlation degree r of each influencing factor i , that is, the correlation degree between each processing parameter and processing benefit, the formula is:

3. The method for optimizing multiple process parameters of a grain dryer according to claim 1, characterized in that: In step 5, the processing parameters in each sample are standardized, and the formula is: Among them, x new is the data after standardized processing, mean(x) is the mean value, and std(x) is the standard deviation.