Sugarcane squeezing process optimization method and device based on interval multi-objective optimization

Abstract

The invention discloses a sugarcane squeezing process optimization method based on interval multi-objective optimization, and belongs to the technical field of sugarcane squeezing process optimization, and the method comprises the steps: obtaining production data of a sugarcane squeezing process, carrying out the preprocessing of the production data, and screening out variables which are highly correlated with the power consumption and the extraction rate; taking the midpoint and the deviation of the power consumption and the extraction rate as a multi-objective optimization problem associated with a variable to represent a mean value and an uncertainty change interval of two squeezing indexes of the power consumption and the extraction rate; solving the multi-objective optimization problem by using an interval multi-objective optimization algorithm to obtain a Pareto solution set; and performing multi-attribute multi-target decision by using an approximate ideal solution sorting method based on entropy weight to obtain an optimal optimization scheme. According to the method, the extraction rate is increased, the energy consumption is reduced, meanwhile, the two indexes of the squeezing extraction rate and the production energy consumption are insensitive to the change of uncertain variables, and the problem that the index fluctuation range is large due to uncertain factors in the sugarcane squeezing process is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of sugarcane pressing process optimization, and more specifically, to a sugarcane pressing process optimization method and device based on interval multi-objective optimization. Background Art

[0002] Sugarcane squeezing is a link in sugar production. Squeezing extraction rate and production energy consumption are two important indicators of this section. Whether they meet the standards will affect the smooth operation and economic benefits of the entire sugar production. The Chinese patent with publication number CN113743652A and announcement date 2021.12.03 discloses a sugarcane squeezing process prediction method based on deep feature recognition, including: 1. Collecting several groups of production data; 2. Eliminating abnormal data and normalizing the production data collected in step one to obtain normalized data; 3. Performing multi-level screening on the normalized data obtained in step two to obtain feature vectors with high correlation and low redundancy with energy consumption and extraction rate; 4. Using a hybrid chicken swarm algorithm to search for the effects of different feature combinations and model parameters on a single data-driven model from the feature vector candidate set screened in step three, obtain the parameter variables, energy consumption and extraction rate under the optimal performance of a single model; 5. Establishing the first layer of deterministic prediction output; 6. Establishing a multi-model combination model to achieve deterministic and probabilistic prediction of extraction rate and energy consumption. However, it failed to solve the problem of large fluctuation range of indicators caused by uncertain factors in the sugarcane crushing process. Summary of the invention

[0003] The purpose of the present invention is to provide a sugarcane pressing process optimization method based on interval multi-objective optimization, which ensures that the two indicators of pressing extraction rate and production energy consumption are insensitive to changes in uncertain variables while maintaining the extraction rate and reducing energy consumption, thereby solving the problem of large fluctuation range of indicators caused by uncertain factors in the sugarcane pressing process.

[0004] In order to achieve these objectives of the present invention, the present invention provides a sugarcane squeezing process optimization method based on interval multi-objective optimization, comprising:

[0005] Step 1, obtaining production data of the sugarcane squeezing process; the production data includes the squeezing machine speed, the two double roller speeds, the primary and secondary belt speeds, the infiltration water flow, the mixed juice flow, the bagasse moisture, the fiber content and the sucrose content; the production data is obtained from several groups of historical production data and corresponding detection data of the monitoring system;

[0006] Step 2: pre-process the production data in step 1, and use the average impact value variable screening method to screen out variables with high correlation with power consumption and extraction rate; this pre-processing prevents the occurrence of large prediction errors due to large order of magnitude differences between production data;

[0007] Step 3: The midpoint (mean) and deviation (radius) of power consumption and extraction rate are used as a multi-objective optimization problem to represent the mean and uncertainty variation range of the squeezing index;

[0008] Step 4: Use interval multi-objective optimization algorithm to solve the multi-objective optimization problem and obtain the Pareto optimal solution set and Pareto frontier;

[0009] Step 5: Use TOPSIS-EW to make multi-attribute and multi-objective decisions to obtain the best optimization solution. TOPSIS stands for the approximate ideal solution ranking method, EW stands for the entropy method, also known as the entropy weight method, and TOPSIS-EW stands for the approximate ideal solution ranking method based on the entropy method (i.e., a comprehensive evaluation method combining the two).

[0010] Preferably, in the step 2, the preprocessing is normalization processing: the production data are normalized to the range of [0,1] according to the following formula;

[0011] y=(xx min ) / x max -x min ;

[0012] Among them, y is normalized data, x represents production data, and x min The minimum value of all production data, x max It is the maximum value among all production data.

[0013] Preferably, the method for screening the variables using the average impact value in step 2 is: the relative contribution rate ε is calculated according to the calculation formula of the relative contribution rate i , select the variables with high relative contribution rate as the variables with high correlation:

[0014]

[0015] Among them, MIV i is the MIV value (mean impact value) of the ith variable, and n is the number of variables.

[0016] Preferably, the optimization framework of the present invention is as follows: the midpoint (mean) and deviation (radius) of power consumption and extraction rate are taken as a multi-objective optimization problem, combined with MOPSO (multi-objective particle swarm optimization algorithm) and nonlinear least squares method (NLSM), and a two-stage optimization method is used to obtain the upper and lower limits of the objective function interval, the midpoint and radius of the interval, the Pareto solution set and the Pareto frontier; the entropy-based approach to ideal solution ranking method TOPSIS-EW is used for multi-objective decision-making to obtain the optimal solution for sugarcane squeezing. The present invention reduces energy consumption while ensuring the extraction rate, and ensures that the two indicators of extraction rate and energy consumption are insensitive to changes in uncertain variables, thereby solving the problem of large fluctuation range of the above indicators caused by uncertain factors in the sugarcane squeezing process.

[0017] Preferably, in step 3 of the present invention, the following function formula is used to express the optimization interval mean and deviation problem:

[0018]

[0019] Among them, U is the set of uncertain factors, X is the set of variables; f L (X) and f R (X) are the lower and upper bounds of the interval of the objective function; min f(X,U) and max f(X,U) are the upper and lower bounds of the objective function when X is fixed and U varies within the interval; mean f M (X,U) represents the mean and also the expected optimal solution, width f W (X,U) represents the deviation and also represents the degree of deviation (or degree of uncertainty) of the optimal solution.

[0020] Preferably, the interval multi-objective optimization algorithm in step 4 of the present invention is a two-stage optimization method combining a multi-objective particle swarm optimization algorithm with a nonlinear least squares method. An optimization system based on the interval multi-objective optimization algorithm is established, and uncertain factors are embedded in the optimization system. N uncertain factors establish N working conditions, and the optimization algorithm process is as follows:

[0021] (1) In SI (the first stage), an optimization algorithm in the optimization space is selected and an initial solution is generated, and then the solution is passed to S-II (the second stage);

[0022] (2) When the optimization variable takes a fixed value, the uncertainty factor is expressed as an interval number, and the upper and lower limits of the uncertain objective function (mean and deviation) are calculated using the NLSM (nonlinear least squares method) to perform S-II optimization;

[0023] (3) If converged, the uncertainty interval of the uncertain objective function is passed to SI optimization to find the optimization variable; if converged, stop the iteration and obtain the Pareto solution set; otherwise, continue to iterate until it converges.

[0024] Preferably, in step 5, the obtained Pareto solution set is subjected to multi-attribute multi-objective decision-making using the approximate ideal solution ranking method based on entropy weight. If there are Q evaluation objects and m evaluation indicators in the Pareto solution set, X is a matrix composed of the evaluation indicator values ​​of each evaluation object, and x ij Represents the jth indicator value of the i-th evaluation object. The specific steps include:

[0025] (1) Data normalization

[0026] In order to avoid the influence of the dimensions of each indicator, the indicator value should be normalized to obtain the standard value y ij ;

[0027]

[0028] (2) Determination of index entropy weight

[0029] According to the standard value y ij The information entropy H(y i ), information entropy is defined by the following formula:

[0030]

[0031] In the formula, 0ln0 is always equal to 0;

[0032] According to the information entropy H(y i ) Calculate the output entropy E j , the formula is as follows:

[0033] E j =H(y i ) / lnQ

[0034] According to the output entropy E j Calculate the difference G of the index j , the formula is as follows:

[0035] G j =1-E j (1≤j≤m)

[0036] According to the difference G j Calculate the entropy weight a j , the formula is as follows:

[0037]

[0038] Where, j = 1, 2, ..., m;

[0039] (3) Construct the weighted normalization matrix V

[0040]

[0041] (4) Determine the negative ideal solution V of the matrix V - and the ideal solution V +

[0042]

[0043] In the formula, J1 is a benefit-based indicator, and J2 is a cost-based indicator;

[0044] (5) Calculate distance

[0045] The distance between each evaluation object and the negative ideal solution The distance between each evaluation object and the ideal solution They are as follows:

[0046]

[0047] Where, i = 1, 2, ..., Q;

[0048] (6) Calculate the relative closeness C i

[0049]

[0050] Where, i = 1, 2, ..., Q;

[0051] Select C in the Pareto frontier of the Pareto solution set i The point with the highest value is taken as the optimal solution.

[0052] The present invention provides a device including a processor and a memory, wherein the memory stores executable code, and when the processor executes the executable code, any one of the methods described above is implemented.

[0053] The present invention has at least the following beneficial effects:

[0054] 1. The method of the present invention obtains data from the monitoring system of the pressing process, normalizes the data, and then screens the variables of the data to obtain variables with a high correlation with energy consumption and extraction rate; in the presence of uncertain parameters, the minimum power consumption and the maximum extraction rate are taken as the goals, and constraints are imposed to obtain the corresponding optimization plan.

[0055] 2. The present invention takes the midpoint (mean) and deviation (radius) of power consumption and extraction rate as a multi-objective optimization problem to represent the mean and uncertainty variation range of the pressing index; combines MOPSO (multi-objective particle swarm optimization algorithm) and nonlinear least squares method (NLSM), and uses a two-stage optimization method to obtain the Pareto optimal solution set; finally, TOPSIS-EW is used for multi-attribute decision-making to obtain the best optimization solution.

[0056] 3. The present invention ensures the extraction rate while reducing energy consumption, and ensures that the two indicators are insensitive to changes in uncertain variables, thereby solving the problem of large fluctuation range of indicators caused by uncertain factors in the sugarcane squeezing process.

[0057] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 is a schematic diagram of steps of a sugarcane squeezing process optimization method based on interval multi-objective optimization according to the present invention;

[0059] Figure 2 is a schematic diagram of an interval multi-objective optimization algorithm according to the present invention;

[0060] Figure 3 is a flow chart of an interval multi-objective optimization algorithm according to the present invention;

[0061] Figure 4 is the Pareto frontier based on TOPSIS-EW according to the present invention;

[0062] Figure 5 is the relative rate of change of energy consumption and extraction rate according to the present invention. DETAILED DESCRIPTION

[0063] The present invention is further described in detail below with reference to examples so that those skilled in the art can implement the invention with reference to the description.

[0064] It should be understood that the terms such as “having”, “including” and “comprising” used herein do not exclude the existence or addition of one or more other elements or combinations thereof.

[0065] like Figure 1 As shown, a sugarcane squeezing process optimization method based on interval multi-objective optimization includes:

[0066] Step 1, obtaining production data of the sugarcane squeezing process; preferably, the production data includes the squeezing machine speed, the two double roller speeds, the primary and secondary belt speeds, the infiltration water flow rate, the mixed juice flow rate, the bagasse moisture, the fiber content and the sucrose content; the production data is obtained from several groups of historical production data and corresponding detection data of the monitoring system. Different from the existing patent CN113743652A which uses the field detection data of 3 sugarcane cutter current loads, the primary belt current load and speed, the secondary belt current load and speed, 6 squeezing machine current loads and speeds, the actual value and target value of the material level on the conveyor belt, the infiltration water flow rate and the ratio of infiltration water to sugarcane, the present invention further considers the parameter variables such as the mixed juice flow rate, bagasse moisture, fiber content and sucrose content, which is closer to the actual situation.

[0067] Step 2: preprocess the production data in step 1, and filter out variables with high correlation with power consumption and extraction rate; the preferred preprocessing is normalization, which can prevent large prediction errors caused by large differences in the order of magnitude between production data; the variable is represented by X, and U represents a set of uncertain factors, and both X and U are vectors. For example, when bagasse moisture is taken as an uncertain factor and X takes a fixed value, the objective function is an interval value, and the mean and deviation of the interval can be calculated.

[0068] Step three, such as Figure 1 As shown, the midpoint (mean) and deviation (radius) of power consumption and extraction rate are taken as a multi-objective optimization problem to represent the mean and uncertainty variation range of the pressing index.

[0069] Step 4: Figures 1 to 3 As shown, the interval multi-objective optimization algorithm is used to optimize the Pareto optimal solution set (corresponding to the Pareto solution set in the figure) and its Pareto frontier.

[0070] Step 5: Use TOPSIS-EW to make multi-attribute and multi-objective decisions to obtain the best optimization solution. TOPSIS stands for the approximate ideal solution ranking method, EW stands for the entropy method, also known as the entropy weight method, and TOPSIS-EW stands for the approximate ideal solution ranking method based on the entropy method (i.e., a comprehensive evaluation method combining the two).

[0071] This implementation method takes the midpoint (mean) and deviation (radius) of power consumption and extraction rate as a multi-objective optimization problem, combines MOPSO (multi-objective particle swarm optimization algorithm) and nonlinear least squares method (NLSM), and uses a two-stage optimization method to obtain the upper and lower limits of the objective function interval, the midpoint and radius of the interval, the Pareto solution set, and the Pareto frontier; uses the entropy-based approximate ideal solution ranking method TOPSIS-EW to make multi-objective decisions and obtain the optimal solution for sugarcane crushing. By optimizing the sugarcane crushing process through this method, it is possible to ensure that the two crushing indicators of extraction rate and energy consumption reduction are insensitive to changes in uncertain variables while ensuring the extraction rate and reducing energy consumption, thereby solving the problem of large fluctuation range of indicators caused by uncertain factors in the sugarcane crushing process.

[0072] Furthermore, in the step 2, the preprocessing is normalization processing: the production data are normalized to the range of [0,1] according to the following formula;

[0073] y=(xx min ) / x max -x min ;

[0074] Among them, y is normalized data, x represents production data, and x min The minimum value of all production data, x max The above normalization process can prevent the occurrence of large prediction errors due to large differences in the order of magnitude between production data.

[0075] Further, in another embodiment, the average impact value variable screening method in step 2 is: the relative contribution rate ε is calculated according to the calculation formula of the relative contribution rate. i , select the variables with high relative contribution rate as the variables with high correlation:

[0076]

[0077] Among them, MIV i is the MIV value (mean impact value) of the ith variable, and n is the number of variables.

[0078] Furthermore, in the multi-objective optimization problem of step three of the present invention, the following function formula is specifically used to represent the optimization interval mean and deviation problem:

[0079]

[0080] Among them, U is the set of uncertain factors, X is the set of variables; f L (X) and f R(X) are the lower and upper bounds of the interval of the objective function; min f(X,U) and max f(X,U) are the upper and lower limits of the objective when X is determined and U varies within the interval; f M (X,U) represents the mean and also the expected optimal solution, f W (X,U) represents the deviation and also represents the degree of deviation from the optimal solution.

[0081] Further, such as Figure 2 and Figure 3 As shown, the optimization process of the multi-objective optimization algorithm in step 4 of the present invention is specifically as follows: the uncertainty factors are embedded into the optimization system using the interval analysis method, and N uncertain factors establish N working conditions. Taking one uncertain factor as an example, the U space contains the set of uncertain factors, and the X space contains the set of changes. The optimization process is as follows:

[0082] (1) In SI (the first stage), an optimization algorithm in the optimization space is selected to generate an initial solution X, and then the solution is passed to S-II (the second stage);

[0083] (2) When the optimization variable X takes a fixed value, the uncertainty factor U is expressed as an interval number, and the upper and lower limits of the uncertain objective function (mean and deviation) are calculated using the NLSM (nonlinear least squares method) to perform S-II optimization;

[0084] (3) If converged, the uncertainty interval of the uncertain objective function is passed to SI optimization to find the optimization variable; if converged, stop the iteration and obtain the Pareto solution set; otherwise, continue to iterate until it converges.

[0085] Furthermore, in step 5 of the present invention, the entropy-based approach to ideal solution sorting method is used to perform multi-objective decision-making on the Pareto solution set. An implementation process is as follows: if there are Q evaluation objects and m evaluation indicators in the Pareto solution set, X is a matrix composed of the evaluation indicator values ​​of each evaluation object, x ij Represents the jth indicator value of the i-th evaluation object. The specific steps include:

[0086] (1) Data normalization

[0087] In order to avoid the influence of the dimensions of each indicator, the indicator value should be normalized to obtain the standard value y ij ;

[0088]

[0089] (2) Determination of index entropy weight

[0090] According to the standard value y ij The information entropy H(y i ), information entropy is defined by the following formula:

[0091]

[0092] In the formula, 0ln0 is always equal to 0;

[0093] According to the information entropy H(y i ) Calculate the output entropy E j , the formula is as follows:

[0094] E j =H(y i ) / lnQ

[0095] According to the output entropy E j Calculate the difference G of the index j , the formula is as follows:

[0096] G j =1-E j (1≤j≤m)

[0097] According to the difference G j Calculate the entropy weight a j , the formula is as follows:

[0098]

[0099] Where, j = 1, 2, ..., m;

[0100] (3) Construct the weighted normalization matrix V

[0101]

[0102] (4) Determine the negative ideal solution V of the matrix V - and the ideal solution V +

[0103]

[0104] In the formula, J1 is a benefit-based indicator, and J2 is a cost-based indicator;

[0105] (5) Calculate distance

[0106] The distance between each evaluation object and the negative ideal solution The distance between each evaluation object and the ideal solution They are as follows:

[0107]

[0108] Where, i = 1, 2, ..., Q;

[0109] (6) Calculate the relative closeness C i

[0110]

[0111] Where, i = 1, 2, ..., Q;

[0112] Select C in the Pareto frontier i The point with the highest value is taken as the final optimal solution.

[0113] U represents the uncertain factors, where U1, U2, U3 and U4 represent the infiltration water flow, mixed juice flow, bagasse moisture and fiber content respectively; U1U3 means that these two uncertain factors are considered at the same time, and the rest are similar; S represents the objective function value obtained by optimization. Figure 4 The Pareto frontier of the optimization objective function when three uncertain factors (U1U2U3 and U1U3U4) exist is given. The optimal point is obtained by TOPSIS-EW decision, and its C i The value (closeness coefficient) reaches 0.5530, so it can be used as the optimal target solution. The closeness coefficients of points A, B, and C are lower than 0.5530. Figure 5 In (1), min is the lower limit of power consumption, mean is the average power consumption, max is the upper limit of power consumption, and S represents the objective function value obtained by optimization (6890kW h -1 ), it can be seen that whether it is a single factor or multiple uncertain factors, the power consumption range is better than the determined optimal value of 6890kW h -1 As the uncertainty factors increase, the power consumption increases, and the coupling effect of the uncertain parameters increases, but the power consumption value is always less than the determined optimization value of 6890kW h -1 , and in the range of 6300-6600kW h -1 The fluctuation range is small, which solves the problem of large fluctuation range of production energy consumption indicators caused by uncertain factors. Uncertainty optimization not only ensures the stability of the target, but also reduces energy consumption accordingly. Figure 5 In (2), min is the lower limit of the extraction rate, mean is the average value of the extraction rate, max is the upper limit of the extraction rate, and S represents the objective function value (97.15%) obtained by the optimization. It can be seen that no matter it is a single factor U1, U2, U3, or multiple uncertain factors U1U2, U1U3, U2U3, etc., the extraction rate is above 96%, which is slightly smaller than the determined optimization value (97.15%), but the extraction rate interval is between 96% and 96.9%, and the fluctuation range is very small, which also solves the problem of large fluctuation range of the extraction rate index caused by uncertain factors. With the increase of uncertain factors, the extraction rate decreases, the coupling effect of uncertain parameters increases, and the extraction rate values ​​obtained by interval optimization decrease, but they are still above 96%, which can meet the constraint requirements.

[0114] A device of the present invention includes a processor and a memory, wherein the memory stores executable code, and when the processor executes the executable code, the sugarcane pressing process optimization method based on interval multi-objective optimization is implemented.

[0115] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and the embodiments. They can be applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily realized.

Claims

1. A sugarcane squeezing process optimization method based on interval multi-objective optimization, characterized in that: include: Step 1: Obtain production data of the sugarcane pressing process; Step 2: pre-process the production data, and use the average impact value variable screening method to screen out variables that are highly correlated with power consumption and extraction rate; Step 3: The midpoint and deviation of power consumption and extraction rate are used as a multi-objective optimization problem associated with variables to represent the mean and uncertainty variation range of the two squeezing indicators of power consumption and extraction rate; Step 4: Use interval multi-objective optimization algorithm to solve the multi-objective optimization problem and obtain the Pareto optimal solution set and Pareto frontier; Step 5: Use the approximate ideal solution ranking method based on entropy weight to make multi-attribute and multi-objective decision-making to obtain the best optimization solution.

2. The sugarcane squeezing process optimization method based on interval multi-objective optimization according to claim 1, characterized in that: The production data include the press speed, the two twin roll speeds, the primary and secondary belt speeds, the percolation water flow, the mixed juice flow, the bagasse moisture, the fiber fraction and the sucrose fraction.

3. The sugarcane squeezing process optimization method based on interval multi-objective optimization according to claim 1, characterized in that: In the step 2, the preprocessing is performed according to the normalization formula y=(xx min ) / x max -x min Normalize all production data to the range of [0,1] to achieve normalization preprocessing of production data; where y is the normalized data, x is the production data, and x min Minimum value of production data, x min It is the maximum value of production data.

4. The sugarcane squeezing process optimization method based on interval multi-objective optimization according to claim 3, characterized in that: The method for screening the variables using the average impact value in step 2 is: the relative contribution rate ε is calculated according to the calculation formula of the relative contribution rate i , select variables with high relative contribution rates as highly correlated variables: Among them, MIV i is the MIV value of the ith variable, and n is the number of variables.

5. The sugarcane squeezing process optimization method based on interval multi-objective optimization according to claim 1, characterized in that: In the multi-objective optimization problem of step 3, the following function formula is used to express the optimization interval mean and deviation problem: Among them, U is the set of uncertain factors, X is the set of variables; f L (X) and f R (X) are the lower and upper bounds of the interval of the objective function; min f(X,U) and max f(X,U) are the upper and lower bounds of the objective function when X is fixed and U varies within the interval; f M (X,U) represents the mean and also the expected optimal solution, f W (X,U) represents the deviation and also represents the degree of deviation from the optimal solution.

6. The sugarcane squeezing process optimization method based on interval multi-objective optimization according to claim 5, characterized in that: The interval multi-objective optimization algorithm in step four is a two-stage optimization method that combines the multi-objective particle swarm optimization algorithm with the nonlinear least squares method.

7. The sugarcane squeezing process optimization method based on interval multi-objective optimization according to claim 6, characterized in that: An optimization system based on interval multi-objective optimization algorithm is established. N operating conditions are established corresponding to N uncertain factors. The interval analysis method is used to embed the uncertain factors into the optimization system, and the following optimization steps are performed: (1) In the first stage, an optimization algorithm in the optimization space is selected and an initial solution is generated, which is then passed to the second stage; (2) When the optimization variable takes a fixed value, the uncertainty factor is expressed as an interval number, and the upper and lower limits of the uncertain objective function are calculated using the nonlinear least squares method to perform the second stage optimization; (3) If converged, the uncertainty interval of the uncertain objective function is passed to the first stage optimization to find the optimization variable; If it converges, stop iterating and get the Pareto solution set; otherwise, continue iterating until it converges.

8. The sugarcane squeezing process optimization method based on interval multi-objective optimization according to claim 7, characterized in that: The point with the highest degree of closeness to the Pareto frontier in the Pareto solution set is selected as the optimal solution.

9. A device comprising a processor and a memory, characterized in that: The memory stores executable codes, and when the processor executes the executable codes, the method according to any one of claims 1 to 8 is implemented.

Images