A method for optimizing raw material conditioning process

Abstract

The present invention discloses a method for optimizing a raw material tempering process, and relates to the technical field of data model optimization. The method first generates a first parameter set based on different working conditions of variable parameters, and calculates a function value based on a tempering process model and actual values. The first parameter set is screened by function value, non-dominated sorting, and crowding distance to generate a parameter queue. The parameter queue is crossover and mutated by crossover rate and mutation rate to generate a second parameter set. The first parameter set is updated in combination with the parameter queue and the second parameter set to participate in the next iterative calculation, and the target parameter set is output at the end of the iteration. The interval of the variable parameters is determined according to the mean and standard deviation of the target parameter set, and the target parameter combination is determined based on the weight coefficient. Furthermore, the crossover rate and mutation rate are dynamically adjusted according to different iteration stages to avoid premature convergence and loss of the target parameter combination.

Description

Technical Field

[0001] The present invention relates to the technical field of data model optimization, and in particular to a method for optimizing a raw material tempering process. Background Art

[0002] During the conditioning process of feed production, the structure of the raw materials can be changed by changing the moisture content or temperature of the feed, making it easier for animals to digest and absorb, thereby improving the nutritional value of the feed. However, in the actual production process, on the same production line, with the same raw material formula, different process parameters will affect the quality of the product. The Chinese patent application with publication number CN118348865A discloses an automated control system and method for livestock feed processing. The method uses deep learning technology to monitor and analyze parameters such as material temperature, humidity and flow rate in the pellet mill in real time, analyzes the feed status and equipment operating status characteristics based on these parameters, and adaptively optimizes the feed production process. However, this method requires a large amount of historical data to train the deep learning model, and frequent real-time adjustments are required to find the target process parameters. Therefore, it is necessary to further improve the existing technology. Summary of the Invention

[0003] To address these issues, the present invention discloses a method for optimizing a raw material conditioning process. This method pregenerates parameter combinations for different operating conditions, then iterates the variable parameters based on the existing parameter combinations to generate new parameter combinations, thereby increasing the diversity of parameter combinations. Ultimately, the optimal variable parameters under current production conditions are quickly identified, effectively optimizing the parameter combinations and improving the quality of the conditioned material.

[0004] The invention objectives of this application can be achieved through the following technical solutions:

[0005] A method for optimizing a raw material conditioning process comprises the following steps:

[0006] Step 1: Extract the fixed parameters of the tempering production line, the first variable parameter of the raw material, and the second variable parameter of the inlet steam, generate a tempering process model, generate a parameter combination including the first variable parameter and the second variable parameter, create a first parameter set consisting of multiple parameter combinations, and initialize the number of iterations q = 0;

[0007] Step 2: feeding raw materials and inlet steam into the tempering production line, the tempering production line produces tempered material, controlling the tempering production line based on different parameter combinations in the first parameter set, and collecting the moisture content and temperature of the tempered material;

[0008] Step 3: Obtain the target moisture content and target temperature based on the quenching and tempering process model, and establish the objective functions of the quenching and tempering material moisture content and moisture content deviation, as well as the quenching and tempering material temperature and temperature deviation;

[0009] Step 4: Divide the first parameter set into multiple non-dominated sets according to the function value of the objective function and generate non-dominated queues. Calculate the crowding distance of the parameter combination in each non-dominated set and obtain the mean crowding distance of the non-dominated sets.

[0010] Step 5: Extract at least one parameter combination based on the mean of the crowding distances of the non-dominated sets in the non-dominated queue to generate a parameter queue;

[0011] Step 6: Based on the current number of iterations q and the maximum number of iterations q max The crossover rate and mutation rate are dynamically adjusted based on the crossover rate and mutation rate, and the parameter queue is processed based on the crossover rate and mutation rate to generate a second parameter set. If q max , go to step 7, otherwise go to step 8;

[0012] Step 7: Update the first parameter set according to the parameter queue and the second parameter set, q=q+1, and return to step 2;

[0013] Step 8: Take the first parameter set as the target parameter set, generate intervals of the first variable parameter and the second variable parameter according to the target parameter set, and select a target parameter combination from the target parameter set based on the weight coefficients of the moisture content and temperature of the tempering material.

[0014] In the present invention, in step 1, the fixed parameters include the effective length of the tempering chamber, the effective diameter of the tempering chamber, the blade installation angle, and the rated speed of the blade; the first variable parameters include the raw material feed rate, raw material humidity, and raw material temperature; and the second variable parameters include the inlet steam flow rate, inlet steam pressure, and inlet steam temperature.

[0015] In the present invention, in step 1, the output parameters of the tempering production line are collected, the steam latent heat is generated according to the second variable parameter and the output parameter, the heat energy transfer efficiency is generated according to the fixed parameter, and the tempering process model is generated according to the energy change equation and the mass change equation.

[0016] In the present invention, in step 2, the raw materials are one or more of corn flour, wheat bran powder, soybean meal or meat and bone meal, and the conditioning material is starch gelatinized product of corn flour, wheat bran powder, soybean meal or meat and bone meal.

[0017] ​In the present invention, in step 3, the first variable parameter, the second variable parameter, the heat energy transfer efficiency, the steam latent heat and the output parameters are substituted into the tempering process model to obtain the target moisture content and the target temperature. The output parameters include the outlet steam flow rate M3, the outlet steam pressure P2, the tempering material temperature T2, the outlet steam temperature T4, the steam latent heat H2=R / [ln(P2 / P1)(1 / T4-1 / T3)], and the energy change equation: M2c2(T2-T1)=η1[( M1-M3)c1(100-T2)+M1H2], mass change equation: (W1-W2)M2=η2(M1-M3), where R is the ideal gas constant, P1 is the inlet steam pressure, T3 is the inlet steam temperature, M1 is the inlet steam flow rate, M2 is the raw material feed amount, c1 is the water specific heat capacity, c2 is the raw material specific heat capacity, T1 is the raw material temperature, η1 is the heat energy transfer efficiency, η2 is the steam dryness, W1 is the moisture content of the tempering material, and W2 is the raw material humidity.

[0018] In the present invention, in step 3, the objective function of the moisture content of the tempering material and the moisture content deviation is ΔA1=|A1-A1'|, and the objective function of the tempering material temperature and the temperature deviation is ΔA2=|A2-A2'|, wherein A1 is the moisture content of the tempering material, A2 is the tempering material temperature, A1' is the target moisture content, A2' is the target temperature, the function value ΔA1 is the moisture content deviation, and the function value ΔA2 is the temperature deviation.

[0019] In the present invention, in step 4, non-dominated parameter combinations are extracted from the first parameter set according to the function values, different non-dominated sets are created according to the extraction order of the non-dominated parameter combinations, and the non-dominated queue is filled according to the creation order of the non-dominated sets.

[0020] In the present invention, in step 4, in the same non-dominated set, a moisture sub-queue and a temperature sub-queue are generated respectively according to the function value, the crowding distance of the parameter combination is calculated according to the normalized difference between adjacent parameter combinations in the moisture sub-queue and the temperature sub-queue, and then the mean crowding distance of the non-dominated set is calculated.

[0021] In the present invention, in step 6, if q max / 3, then C=C+ΔC and O=O+ΔO, if q max / 3≤q≤2q max / 3, then C=C-ΔC and O=O-ΔO, if 2q max / 3 <q<q max , then O=O-ΔO, where C is the crossover rate, O is the mutation rate, ΔC is the crossover rate adjustment, ΔO is the mutation rate adjustment, and C, ΔC, O, and ΔO are all greater than 0 and less than 1.

[0022] ​In the present invention, in step 7, the first parameter set is initialized, at least part of the parameter combinations in the parameter queue are added to the first parameter set, at least part of the parameter combinations in the parameter queue and the second parameter set are merged into a temporary set, and then the parameter queue of the temporary set is generated, and the parameter queue of the temporary set is added to the first parameter set.

[0023] The beneficial effect of implementing the optimization method of the raw material tempering process of the present invention is that the present invention generates a first parameter set based on the variable range of the variable parameters, and controls the tempering production line according to the parameter settings of the first parameter set. The actual moisture content and temperature of the tempering material are collected, and the target moisture content and target temperature are calculated based on the tempering process model. The deviation between the moisture content and the target temperature is used as the objective function, and the variable parameters are continuously iterated and optimized. New parameter combinations are generated based on the existing parameter combinations to improve the diversity of parameter combinations. The interval of the variable parameters is determined by the first parameter set completed by the final iteration, and the target parameter combination is determined according to the weight coefficient. The preferred variable parameters under the current production conditions are quickly found, thereby improving the quality of the finished product. Furthermore, the crossover rate and mutation rate are dynamically adjusted according to different iteration stages to avoid premature convergence and loss of the target parameter combination. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 Flowchart of the optimization method of the raw material conditioning process of the present invention;

[0025] Figure 2 is a schematic diagram of a quenching and tempering production line;

[0026] Figure 3 A process diagram for generating a non-dominated queue of the present invention;

[0027] Figure 4 A schematic diagram of updating a first parameter set according to the present invention;

[0028] Figure 5 This is a structural diagram of the tempering chamber of the tempering production line of the present invention;

[0029] Figure 6 Schematic diagram of the tempering chamber of the tempering production line of the present invention.

[0030] Explanation of some reference numerals: tempering equipment 11 , raw material warehouse 12 , steam warehouse 13 , pressure pump 14 , waste gas pool 15 , storage warehouse 16 . DETAILED DESCRIPTION

[0031] The technical solutions in the embodiments of the present invention will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present invention.

[0032] Feed conditioning involves physically modifying the properties of feed ingredients, particularly by changing their moisture content and temperature, to facilitate subsequent processing and improve feed quality, nutritional value, and digestion and absorption efficiency. However, controlling the moisture content of the conditioning material also affects its temperature. Therefore, to optimize both the moisture content and temperature, this paper combines the fast non-dominated sorting and crowding distance methods of a multi-objective genetic algorithm to iteratively optimize variable parameters based on a relatively small amount of data, identifying a target parameter combination that simultaneously satisfies both optimization objectives. Example 1

[0033] Reference Figure 1 The method for optimizing the raw material conditioning process of the present invention described in detail in this embodiment comprises the following steps:

[0034] Step 1: Extract the fixed parameters of the tempering production line, the first variable parameter of the raw material, and the second variable parameter of the inlet steam, generate a tempering process model, generate a parameter combination including the first variable parameter and the second variable parameter, create a first parameter set consisting of multiple parameter combinations, and initialize the number of iterations q = 0. The fixed parameters include the effective length and diameter of the tempering chamber, the blade mounting angle, and the rated blade speed. The first variable parameter includes the raw material feed rate, raw material humidity, and raw material temperature. The second variable parameter includes the inlet steam flow rate, inlet steam pressure, and inlet steam temperature. The parameter combination refers to the combination of specific data of the raw material feed rate, raw material humidity, raw material temperature, inlet steam flow rate, inlet steam pressure, and inlet steam temperature. Collect the output parameters of the tempering production line, generate steam latent heat based on the second variable parameter and the output parameters, generate heat energy transfer efficiency based on the fixed parameters, and generate the tempering process model based on the energy change equation and the mass change equation.

[0035] In this embodiment, starting from the minimum values of each of the first and second variable parameters, specific values of each of the first and second variable parameters are selected in equal steps to generate parameter combinations. These multiple parameter combinations constitute a first parameter set. In a further embodiment, the data of the first or second variable parameters are divided into multiple operating ranges based on the rated operating conditions of the quenching and tempering production line. Specific values are then selected from these operating ranges, ensuring that at least one specific value is selected from each operating range, to generate parameter combinations. The number of parameter combinations in the first parameter set ranges, for example, from 50 to 100, needs to be considered, taking into account the processing conditions of the quenching and tempering production line and the sample requirements of the iterative algorithm.

[0036] Step 2: Add raw materials and inlet steam to the conditioning production line, the conditioning production line produces conditioning material, the conditioning production line is controlled based on different parameter combinations in the first parameter set, and the moisture content and temperature of the conditioning material are collected. The raw materials are one or more of corn flour, wheat bran powder, soybean meal or meat and bone meal, and the conditioning material is a starch gelatinized product of corn flour, wheat bran powder, soybean meal or meat and bone meal. The conditioning production line is as follows: Figure 2 As shown, inlet steam enters the tempering equipment 11 from the steam bin 13 through the pressure pump 14 , raw materials enter the tempering equipment 11 from the raw material bin 12 , outlet steam enters the exhaust gas pool 15 , and the tempered material enters the storage bin 16 .

[0037] Step 3: According to the tempering process model, the target moisture content and target temperature are obtained, and the target function of the moisture content of the tempering material and the moisture content deviation, as well as the target function of the temperature of the tempering material and the temperature deviation, are established. The first variable parameter, the second variable parameter, the heat energy transfer efficiency, the latent heat of steam and the output parameter are substituted into the tempering process model to obtain the corresponding target moisture content and target temperature. For the specific implementation method, refer to Example 2. The target function of the moisture content of the tempering material and the moisture content deviation is ΔA1=|A1-A1'|, and the target function of the temperature of the tempering material and the temperature deviation is ΔA2=|A2-A2'|, wherein A1 is the moisture content of the tempering material, A2 is the temperature of the tempering material, A1' is the target moisture content, A2' is the target temperature, the function value ΔA1 is the moisture content deviation, and the function value ΔA2 is the temperature deviation.

[0038] Step 4: Based on the function values of the objective function, the first parameter set is divided into multiple non-dominated sets and a non-dominated queue is generated. The crowding distance of each parameter combination in the non-dominated set is calculated, and the mean crowding distance of the non-dominated sets is obtained. For parameter combinations B1 and B2, if the function values of all objective functions of B1 are no greater than the corresponding function values of B2, and at least one function value of B1 is less than the corresponding function value of B2, then B1 is considered to dominate B2. If no other parameter combination can dominate a parameter combination, then the parameter combination is considered non-dominated.

[0039] First create an empty non-dominated queue. Figure 3 , then traverse the first parameter set, extract non-dominated parameter combinations from the first parameter set, create a non-dominated set consisting of these non-dominated parameter combinations, delete the non-dominated parameter combinations in the first parameter set, and insert the non-dominated set into the non-dominated queue until the first parameter set is completely traversed. Repeat the above steps, retraverse the remaining parameter combinations in the first parameter set, extract non-dominated parameter combinations and generate a non-dominated set, and then insert the non-dominated set into the non-dominated queue. This continues until the first parameter set is empty, and the non-dominated queue is finally filled.

[0040] In the same non-dominated set, the parameter combinations of the non-dominated set are arranged from small to large according to the function value ΔA1 of the moisture content deviation to generate a moisture content sub-queue. Calculate any parameter combination B in the moisture content sub-queue j Combined with adjacent parameters B j+1 and B j-1 The normalized difference d 1j =[ΔA 1(j+1) –ΔA 1(j-1) ] / (ΔA1 max -ΔA1 min ). Where, ΔA 1(j+1) is the function value of the moisture content deviation of the j+1th parameter combination, ΔA 1(j-1) is the function value of the moisture content deviation of the j-1th parameter combination, ΔA1 max is the maximum function value of the moisture content deviation in the non-dominated queue, ΔA1 min is the function value with the minimum moisture content deviation in the non-dominated queue.

[0041] Then, according to the function value of temperature deviation ΔA2, the parameter combinations of the non-dominated set are arranged from small to large to generate a temperature sub-queue. Calculate any parameter combination B in the temperature sub-queue j Combined with adjacent parameters B j+1 and B j-1 The normalized difference d 2j =[ΔA 2(j+1) –ΔA 2(j-1) ] / (ΔA2 max -ΔA2 min ). Where, ΔA 2(j+1) is the function value of the temperature deviation of the j+1th parameter combination, ΔA 2(j-1) is the function value of the temperature deviation of the j-1th parameter combination, ΔA2 max is the maximum function value of temperature deviation in the non-dominated queue, ΔA2 min is the function value that minimizes the temperature deviation in the non-dominated queue.

[0042] According to the normalized difference d 1j and d 2j Calculate the parameter combination B j crowding distance Then the mean crowding distance of the non-dominated set is calculated by the crowding distance of multiple parameter combinations , J is the number of parameter combinations in the non-dominated set. The crowding distances of the first and last parameter combinations in the moisture content and temperature sub-queues are usually preset to a constant greater than 1, and this constant is greater than all crowding distances.

[0043] Step 5: Extract at least one parameter combination based on the average crowding distance of the non-dominated sets in the non-dominated queue to generate a parameter queue. In this implementation, real number encoding is used to represent parameter combinations. Searching starts from the first non-dominated set according to the order of the non-dominated sets in the non-dominated queue, and the parameter combinations in the non-dominated set with a crowding distance greater than the corresponding average crowding distance are added to the parameter queue until all non-dominated sets are searched or the capacity limit of the parameter queue is reached, and finally a parameter queue in the form of a stack structure is obtained.

[0044] Step 6: Dynamically adjust the crossover rate and mutation rate based on the ratio of the current iteration number q and the maximum iteration number q max , and process the parameter queue based on the crossover rate and mutation rate to generate a second parameter set. If q < q max , go to Step 7, otherwise go to Step 8. If q < q max / 3, at this time it is the initial stage of iteration, and higher crossover rate and mutation rate are required to quickly generate diverse parameter combinations, so C = C + ΔC, and O = O + ΔO. If q max / 3 ≤ q ≤ 2q max / 3, at this time it is the middle stage of iteration, gradually reducing the probabilities of crossover and mutation can promote convergence, so C = C - ΔC and O = O - ΔO. If 2q max / 3 < q < q max , at this time it is the late stage of iteration, which is in the stage of concentrating on optimizing the discovered parameter combinations, and the crossover rate needs to be maintained while further appropriately reducing the mutation rate, so O = O - ΔO, where C is the crossover rate, O is the mutation rate, ΔC is the crossover rate adjustment amount, ΔO is the mutation rate adjustment amount, and C, ΔC, O, and ΔO are all greater than 0 and less than 1.

[0045] In this embodiment, 0.5 < C < 0.9, ΔC = 0.005, 0.01 < O < 0.1, ΔO = 0.001. In the initial stage of iteration, when the crossover rate reaches 0.9, it will not increase anymore, and when the mutation rate reaches 0.1, it will not increase anymore; in the middle stage of iteration, when the crossover rate reaches 0.5, it will not decrease anymore, and when the mutation rate reaches 0.01, it will not decrease anymore; in the late stage of iteration, the crossover rate remains unchanged. If O > 0.01, the mutation rate continues to decrease until the iteration ends or the mutation rate reaches 0.01. Crossover operation: Randomly select two parameter combinations from the parameter queue, and decide whether to perform the crossover operation according to the crossover rate to generate the first offspring parameter combination. Mutation operation: Decide whether the first offspring parameter combination mutates according to the mutation rate to generate the second offspring parameter combination, and generate the second parameter set according to the second offspring parameter combination, specifically referring to that described in Embodiment 3.

[0046] Step 7: Update the first parameter set according to the parameter queue and the second parameter set, q = q + 1, and return to Step 2. Refer to Figure 4, initialize the first parameter set. The parameter combinations at the top of the parameter queue are of higher importance and are necessary to participate in the next genetic iteration. In this embodiment, the parameter combinations at the top 20% to 30% of the parameter queue are directly retained and added to the first parameter set. Then, the remaining parameter combinations in the parameter queue and all parameter combinations in the second parameter set are merged into a temporary set. If there are duplicate parameter combinations, one of the parameter combinations is retained. Referring to steps 4 and 5, a parameter queue of the temporary set is generated using a method based on non-dominated sorting and crowding distance, and the parameter queue of the temporary set is added to the first parameter set until the first parameter set reaches the capacity limit or the parameter queue of the temporary set is empty.

[0047] Step 8: Take the first parameter set as the target parameter set, generate the intervals of the first variable parameter and the second variable parameter based on the target parameter set, and select the target parameter combination from the target parameter set based on the weight coefficients of the moisture content of the tempering material and the tempering material temperature. Use a method based on mean and standard deviation to generate the intervals of the first variable parameter and the second variable parameter. Calculate the mean and standard deviation of each first variable parameter and the mean and standard deviation of each second variable parameter in the target parameter set. The upper limit of the interval of the first variable parameter and the second variable parameter is the corresponding mean plus twice the standard deviation, and the lower limit of the interval of the first variable parameter and the second variable parameter is the corresponding mean minus twice the standard deviation. In actual production, the first variable parameter and the second variable parameter can be fine-tuned based on the interval, which helps to quickly respond to small deviations in the production process and ensure that the production line always operates in the best state.

[0048] The weight coefficients of the moisture content of the tempering material and the temperature of the tempering material can be determined according to the emphasis of the actual production demand, and the target parameter combination is selected from the target parameter set based on the weight coefficient. Since the temperature of the tempering material can be adjusted again in the subsequent granulation process, but the water content directly affects the product quality, in a simpler embodiment, the weight coefficients of the preset moisture content of the tempering material and the temperature of the tempering material are 0.6 and 0.4 respectively. The present invention can also use a multi-objective comprehensive model to evaluate the parameter combination in the target parameter set, that is, using, for example, the CRITIC algorithm to quantify the conflict of two objective functions, determine the weight coefficient of each objective function, each parameter combination corresponds to a group of objective function function values, and further based on the TOPSIS algorithm and the weight coefficient, the function value is comprehensively scored. The greater the comprehensive score, the smaller the comprehensive error of the corresponding parameter combination. Select the parameter combination with the largest score as the target parameter combination. Example 2

[0049] This embodiment further discloses a method for obtaining a target moisture content and a target temperature based on a tempering process model.

[0050] Flow sensors, pressure sensors, temperature sensors and other devices are installed on the quenching and tempering production line to collect the output parameters of the quenching and tempering production line, including outlet steam flow M3, outlet steam pressure P2, quenching and tempering material temperature T2, and outlet steam temperature T4.

[0051] The latent heat of steam is generated based on the second variable parameter and the output parameter. According to the Clausius-Clapeyron equation, the latent heat of steam H2 = R / [ln(P2 / P1)(1 / T4-1 / T3)], where P1 is the inlet steam pressure, T3 is the inlet steam temperature, and R is the ideal gas constant, which is 8.314 J / (mol·K). Temperature is expressed in Kelvin.

[0052] Generate heat transfer efficiency based on fixed parameters. Heat conversion efficiency η1=1-exp(-t / t0), effective tempering time t=L1 / (2NtanαL2f), such as Figure 5 Where L1 is the effective length of the tempering chamber, L2 is the effective diameter of the tempering chamber, α is the blade installation angle, N is the rated blade speed, and f is the material damping coefficient, typically ranging from 0.2 to 0.4. t0 is the baseline tempering time, which can range from 0.6 to 1 minute depending on the blade structure and material type of the tempering production line. In another embodiment, the fixed parameters include heat transfer efficiency and humidity absorption rate. Heat transfer efficiency is calibrated through testing before the tempering production line leaves the factory, with a typical heat conversion efficiency of 60%.

[0053] The tempering process model is generated based on the energy change equation and mass change equation. Energy change equation: M2c2(T2-T1)=η1[(M1-M3)c1(100-T2)+M1H2]. During the tempering process, the raw materials and inlet steam enter the tempering chamber, and the outlet steam and tempering materials are discharged from the tempering chamber, such as Figure 6 The raw materials and inlet steam introduce moisture, while the outlet steam and quenching and tempering materials remove moisture. The quenching and tempering materials absorb condensed water. The change in moisture mass is absorbed by the quenching and tempering chamber, as shown by the mass change equation: (W1-W2)M2=η2(M1-M3), where M1 is the inlet steam flow rate, M2 is the raw material feed rate, c1 is the specific heat capacity of water, c2 is the specific heat capacity of the raw material, T1 is the raw material temperature, W1 is the moisture content of the quenching and tempering material, W2 is the raw material humidity, and η2 is the steam dryness, typically between 70% and 90%. Temperatures in the quenching and tempering process model are expressed in degrees Celsius, and all other units are in the International System of Units.

[0054] The first variable parameter, the second variable parameter, the heat energy transfer efficiency, the humidity absorption rate, the steam latent heat, and the output parameters of the current parameter combination are substituted into the tempering process model to obtain the target moisture content and target temperature under the current parameter combination. Example 3

[0055] This embodiment further discloses a method for processing the parameter queue using a crossover rate and a mutation rate and generating a second parameter set in step 6.

[0056] Parameter initialization: Preset a crossover distribution index λ1, and the range of λ1 is [1, 20]. A smaller value of λ1 will cause the offspring parameter combinations to be more closely distributed around the parameter combination, reducing the diversity of the parameter combinations. A larger value of λ1 will make the offspring parameter combinations more widely distributed, increasing the diversity of the parameter combinations but reducing the convergence speed. At the same time, preset a mutation distribution index λ2 to control the magnitude of the mutation, and the range of λ2 is [1, 20]. In this embodiment, λ1 is set to 10 and λ2 is set to 20.

[0057] Crossover parameter preset: Select a random parameter.

[0058] Parameter crossover operation: Randomly select two parameter combinations B1 and B2 from the parameter queue. If u1 < C, use the crossover algorithm to generate the first offspring parameter combinations B1' and B2'. Otherwise, copy the parameter combinations B1 and B2 as the first offspring parameter combinations, that is, B1' = B1, B2' = B2.

[0059] Mutation parameter preset: Select a random parameter.

[0060] Parameter mutation operation: If u2 < O, use the mutation algorithm to generate new second offspring parameter combinations B1'' and B2'' for the first offspring parameter combinations B1' and B2'. Otherwise, copy the first offspring parameter combinations B1' and B2' as the second offspring parameter combinations, that is, B1'' = B1', B2'' = B2'.

[0061] If the remaining parameter combinations in the parameter queue are not empty, return to the crossover parameter preset step. If the remaining parameter combinations in the parameter queue are empty, merge all the second offspring parameter combinations into the second parameter set.

[0062] Among them, the preferred crossover algorithm in this embodiment is as follows.

[0063] Calculate the crossover parameter . As described in Embodiment 1, the parameter combination B1 is a set composed of a first variable parameter and a second variable parameter. Any element b in B1 1n The corresponding element in the first offspring parameter combination B1' is b 1n ', . Any element b in B2 2n The corresponding element in the first offspring parameter combination B2' is b 2n ', .

[0064] The preferred mutation algorithm in this embodiment is as follows.

[0065] Calculate the variation parameter Any element b in B1' 1n 'The corresponding element in the second generation parameter combination B1'' is b 1n '', , any element b in B2' 2n 'The corresponding element in the second generation parameter combination B2'' is b 2n '', , where b 1n and b 2n are the nth elements in B1 and B2 respectively, b n max is the maximum value of the nth element in the parameter queue, b n min The minimum value of the nth element in the parameter queue.

[0066] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for optimizing a raw material conditioning process, characterized in that: The following steps are involved: Step 1: Extract the fixed parameters of the tempering production line, the first variable parameter of the raw material, and the second variable parameter of the inlet steam, generate a tempering process model, generate a parameter combination including the first variable parameter and the second variable parameter, create a first parameter set consisting of multiple parameter combinations, and initialize the number of iterations q = 0; Step 2: feeding raw materials and inlet steam into the tempering production line, the tempering production line produces tempered material, controlling the tempering production line based on different parameter combinations in the first parameter set, and collecting the moisture content and temperature of the tempered material; Step 3: Obtain the target moisture content and target temperature based on the quenching and tempering process model, and establish the objective functions of the quenching and tempering material moisture content and moisture content deviation, as well as the quenching and tempering material temperature and temperature deviation; Step 4: Divide the first parameter set into multiple non-dominated sets according to the function value of the objective function and generate non-dominated queues. Calculate the crowding distance of the parameter combination in each non-dominated set and obtain the mean crowding distance of the non-dominated sets. Step 5: Extract at least one parameter combination based on the mean crowding distance of the non-dominated sets in the non-dominated queue to generate a parameter queue. The search begins with the first non-dominated set according to the order of the non-dominated sets in the non-dominated queue, and adds parameter combinations whose crowding distances in the non-dominated sets are greater than the mean crowding distances to the parameter queue until all non-dominated sets are searched or the capacity limit of the parameter queue is reached, thereby obtaining a stack-structured parameter queue. Step 6: Based on the current number of iterations q and the maximum number of iterations q max The crossover rate and mutation rate are dynamically adjusted based on the crossover rate and mutation rate, and the parameter queue is processed based on the crossover rate and mutation rate to generate a second parameter set. If q max , go to step 7, otherwise go to step 8;​ Step 7: Update the first parameter set according to the parameter queue and the second parameter set, q=q+1, and return to step 2; Step 8: Take the first parameter set as the target parameter set, generate the intervals of the first variable parameter and the second variable parameter according to the target parameter set, and select the target parameter combination from the target parameter set based on the weight coefficients of the moisture content of the quenched and tempered material and the quenched and tempered material temperature, wherein, In step 1, the output parameters of the tempering production line are collected, the steam latent heat is generated according to the second variable parameter and the output parameter, the heat energy transfer efficiency is generated according to the fixed parameter, and the tempering process model is generated according to the energy change equation and the mass change equation. In step 3, the first variable parameter, the second variable parameter, the heat energy transfer efficiency, the steam latent heat and the output parameters are substituted into the tempering process model to obtain the target moisture content and the target temperature. The output parameters include the outlet steam flow rate M3, the outlet steam pressure P2, the tempering material temperature T2 and the outlet steam temperature T4. In step 3, the objective function of the moisture content of the tempering material and the moisture content deviation is ΔA1=|A1-A1'|, and the objective function of the tempering material temperature and the temperature deviation is ΔA2=|A2-A2'|, wherein A1 is the moisture content of the tempering material, A2 is the tempering material temperature, A1' is the target moisture content, A2' is the target temperature, the function value ΔA1 is the moisture content deviation, and the function value ΔA2 is the temperature deviation.

2. The method for optimizing the raw material tempering process according to claim 1, characterized in that: In step 1, the fixed parameters include the effective length of the tempering chamber, the effective diameter of the tempering chamber, the blade installation angle, and the rated speed of the blade; the first variable parameters include the raw material feed rate, raw material humidity, and raw material temperature; and the second variable parameters include the inlet steam flow rate, inlet steam pressure, and inlet steam temperature.

3. The method for optimizing the raw material tempering process according to claim 1, characterized in that: In step 2, the raw materials are one or more of corn flour, wheat bran powder, soybean meal or meat and bone meal, and the conditioning material is starch gelatinized product of corn flour, wheat bran powder, soybean meal or meat and bone meal.

4. The method for optimizing the raw material tempering process according to claim 1, characterized in that: In step 3, the energy change equation is: M2c2(T2-T1)=η1[(M1-M3)c1(100-T2)+M1H2], and the mass change equation is: (W1-W2)M2=η2(M1-M3), where R is the ideal gas constant, P1 is the inlet steam pressure, T3 is the inlet steam temperature, M1 is the inlet steam flow rate, M2 is the raw material feed rate, c1 is the water specific heat capacity, c2 is the raw material specific heat capacity, T1 is the raw material temperature, η1 is the heat energy transfer efficiency, η2 is the steam dryness, W1 is the moisture content of the tempering material, and W2 is the raw material humidity.

5. The method for optimizing the raw material tempering process according to claim 1, characterized in that: Non-dominated parameter combinations are sequentially extracted from the first parameter set according to the function values, different non-dominated sets are created according to the extraction order of the non-dominated parameter combinations, and the non-dominated queue is filled according to the creation order of the non-dominated sets.

6. The method for optimizing the raw material tempering process according to claim 1, characterized in that: In step 4, in the same non-dominated set, a moisture content sub-queue and a temperature sub-queue are generated according to the function value. The crowding distance of the parameter combination is calculated according to the normalized difference between adjacent parameter combinations in the moisture content sub-queue and the temperature sub-queue, and then the mean crowding distance of the non-dominated set is calculated.

7. The method for optimizing the raw material tempering process according to claim 1, characterized in that: In step 6, if q max / 3, then C=C+ΔC and O=O+ΔO, if q max / 3≤q≤2q max / 3, then C=C-ΔC and O=O-ΔO, if 2q max / 3 <q<q max , then O=O-ΔO, where C is the crossover rate, O is the mutation rate, ΔC is the crossover rate adjustment, ΔO is the mutation rate adjustment, and C, ΔC, O, and ΔO are all greater than 0 and less than 1.​ 8. The method for optimizing the raw material tempering process according to claim 1, characterized in that: In step 7, the first parameter set is initialized, at least part of the parameter combinations in the parameter queue are added to the first parameter set, at least part of the parameter combinations in the parameter queue and the second parameter set are merged into a temporary set, a parameter queue of the temporary set is generated, and the parameter queue of the temporary set is added to the first parameter set.

Images