Multi-objective optimization method for rapeseed oil processing parameters and related device

Abstract

The invention discloses a multi-objective optimization method for rapeseed oil processing parameters and a related device, and the method comprises the steps: determining the processing parameters in a rapeseed oil processing process, taking the processing parameters as input variables, and taking oil yield, energy consumption and volatile compound content under degumming as output variables to construct an objective function; initializing a group of populations comprising a plurality of particles according to a particle swarm algorithm, and setting a fitness function to evaluate the advantages and disadvantages of each particle; calculating the fitness value of each particle in the population according to the rapid non-dominated sorting and the congestion degree result; and selecting a filial generation population according to the fitness value of each particle, and carrying out crossover operation and mutation operation on the filial generation population to generate new particles to obtain an optimal solution set containing a plurality of non-dominated solutions. Particle swarm multi-objective optimization is carried out on processing parameters based on the oil yield, the energy consumption and the volatile compound content under degumming, and it is ensured that the optimal expected performance and benefits are achieved in the actual processing process.

Description

Technical Field

[0001] The invention relates to the technical field of rapeseed oil processing and multi-objective optimization, and in particular to a multi-objective optimization method and device, and computing equipment for rapeseed oil processing parameters. Background Art

[0002] During the actual processing of rapeseed oil, slight changes in processing parameters may lead to significant fluctuations in oil yield, energy consumption and volatile compound content, thus affecting the overall quality of the product.

[0003] Traditional rapeseed oil processing methods mainly rely on empirical optimization of processing parameters and lack precise optimization of processing parameters. Not only is the efficiency low, but it is also difficult to achieve the best processing effect. Secondly, traditional processing methods, especially in the steaming and frying process, lack the setting optimization of processing parameters, resulting in low energy utilization and product quality, while also increasing production costs. In addition, the content of volatile compounds during the degumming process is one of the important indicators for evaluating the quality of rapeseed oil. However, traditional processing methods are difficult to effectively control the generation of volatile compounds, which affects the taste and flavor of rapeseed oil. In response to the above problems, existing optimization methods often only focus on the optimization of a single objective (such as oil yield or energy consumption), while ignoring the balance between multiple objectives. In addition, in the solution process, they often fall into the local optimal solution and cannot obtain the global optimal combination of processing parameters.

[0004] To solve the above problems, the present invention proposes a multi-objective optimization method for rapeseed oil processing parameters, and performs particle swarm multi-objective optimization of the processing parameters based on oil yield, energy consumption and volatile compound content under degumming to ensure that the best expected performance and benefits are achieved in the actual processing process. Summary of the invention

[0005] In view of the above problems, the present invention proposes a multi-objective optimization method and device, and computing equipment for rapeseed oil processing parameters.

[0006] According to one aspect of the present invention, a multi-objective optimization method for rapeseed oil processing parameters is provided, comprising: Determine the processing parameters in the rapeseed oil processing process, take the processing parameters as input variables, and take the oil yield, energy consumption and volatile compound content under degumming as output variables to construct an objective function; wherein the processing parameters include crushing degree, steaming temperature, steaming time and oil pressing pressure, and the steaming temperature includes low heating rate, medium heating rate and high heating rate; the volatile compounds include polyphenols, tocopherols, sterols, glucosinolate degradation products, Maillard reaction products and medium lipid oxidation products; A population of multiple particles is initialized according to the particle swarm algorithm, and a fitness function is set to evaluate the quality of each particle; wherein each particle represents a combination of a set of processing parameters and a corresponding objective function value; According to a fast non-dominated sorting algorithm, multiple particles in a population are sorted and the crowding degree of each particle is calculated to obtain fast non-dominated sorting and crowding results; according to the fast non-dominated sorting and crowding results, the fitness value of each particle in the population is calculated; according to the fitness value of each particle, a child population is selected, and a crossover operation and a mutation operation are performed on the child population to generate new particles; the new particles are merged with the particles of the current population to obtain a new population; the individuals in the new population are subjected to constraint feasibility judgment, and the individuals violating the constraints are excluded to obtain an optimal solution set containing multiple non-dominated solutions, and a combination of processing parameters for a processing process is selected from the optimal solution set.

[0007] In an optional manner, the particle includes an iterative velocity vector, which is used to represent the updated moving direction and velocity of the particle in the search space, wherein the calculation formula of the iterative velocity vector is: in, is the k+1th iteration value of the velocity vector of the dth dimension of the ith particle; is the inertia weight; , , They are particle learning factor, population learning factor and neighbor influence factor respectively; , , are random numbers in the interval [0,1] respectively; is the optimal position of the ith particle in the dth dimension at the kth iteration; The value of the position of the i-th particle in the d-th dimension at the k-th iteration; is the value of the global optimal position in the dth dimension; The value of the optimal position of the neighbor of the i-th particle in the d-th dimension; is the adaptive adjustment coefficient; is the velocity change of the ith particle in the dth dimension.

[0008] In an optional manner, the crossover operation randomly selects a crossover point and swaps the genes of two parent particles to generate a child particle; The mutation operation adjusts the gene of each particle with a preset probability, wherein the adjustment amplitude decreases as the number of iterations increases.

[0009] In an optional manner, the step of sorting the plurality of particles in the population according to the fast non-dominated sorting algorithm and calculating the congestion degree of each particle to obtain the fast non-dominated sorting and congestion degree results further comprises: Assign an initial rank to each particle, where all particles that are not dominated by other particles are assigned to the first frontier; For each particle in the population, find all the particles it dominates, starting from the first initial level, check the particles it dominates; assign the particles it dominates to the next initial level; In each initial level, particles are individually ranked according to the objective function value, and the overall ranking is determined based on the combination of the individual rankings; For the first and last particles of each initial level, set their crowding to infinity to ensure that they are not removed from the population; The congestion degree of other particles in each initial level is obtained by calculating the sum of the distances between the particle and its adjacent particles on each objective function.

[0010] In an optional manner, after each iteration of the crossover operation and mutation operation, particles whose fitness is greater than a preset fitness threshold are retained in the next generation population and no longer participate in the crossover and mutation operation.

[0011] In an optional manner, the calculation formula of the fitness threshold is: in, is the basic threshold; is the dynamic adjustment coefficient; is the average value of the fitness value of the population in the previous iteration; is the average fitness value of the current iteration population.

[0012] In an optional manner, the preset probability is an adaptive mutation probability to avoid premature convergence of particles. The calculation formula of the adaptive mutation probability is: in, is the mutation probability of the t+1th iteration; is the mutation probability of the tth iteration; To adjust the step size; is the average distance between all particles in the tth iteration; is the average distance between the ith particle and other particles in the tth iteration.

[0013] In an optional manner, the process of constructing the optimal solution set of the non-dominated solutions is: Compare the dominance relationship between different individuals. If one individual is not inferior to another individual in all objectives and is superior to another individual in at least one objective, the former individual is judged to dominate the latter individual. The individuals in the non-dominated state are grouped into a non-dominated solution set, the average distance between the solutions in the non-dominated solution set and the nearest solution on the true Pareto frontier is calculated, and the solution set with the smallest average distance is taken as the optimal solution set of the non-dominated solution.

[0014] According to another aspect of the present invention, a multi-objective optimization device for rapeseed oil processing parameters is provided, comprising: An objective function construction module is used to determine the processing parameters in the rapeseed oil processing process, and the processing parameters are used as input variables, and the oil yield, energy consumption and volatile compound content under degumming are used as output variables to construct the objective function; wherein the processing parameters include crushing degree, steaming and frying temperature, steaming and frying time and oil pressing pressure, and the steaming and frying temperature includes low heating rate, medium heating rate and high heating rate; the volatile compounds include polyphenols, tocopherols, sterols, glucosinolate degradation products, Maillard reaction products and medium lipid oxidation products; The particle swarm initialization module is used to initialize a population of multiple particles according to the particle swarm algorithm and set a fitness function to evaluate the quality of each particle; wherein each particle represents a combination of a set of processing parameters and a corresponding objective function value; The optimal solution set selection module is used to sort multiple particles in the population according to the fast non-dominated sorting algorithm and calculate the crowding degree of each particle to obtain fast non-dominated sorting and crowding results; calculate the fitness value of each particle in the population according to the fast non-dominated sorting and crowding results; select a child population according to the fitness value of each particle, perform crossover operations and mutation operations on the child population to generate new particles; merge the new particles with the current population particles to obtain a new population; perform constraint feasibility judgment on individuals in the new population, exclude individuals that violate constraints, obtain an optimal solution set containing multiple non-dominated solutions, and select a processing parameter combination as a processing process from the optimal solution set.

[0015] According to another aspect of the present invention, there is provided a computing device, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the multi-objective optimization method of rapeseed oil processing parameters.

[0016] According to the scheme provided by the present invention, the processing parameters in the rapeseed oil processing process are determined, and the processing parameters are used as input variables, and the oil yield, energy consumption and volatile compound content under degumming are used as output variables to construct an objective function; wherein the processing parameters include crushing degree, steaming temperature, steaming time and oil pressing pressure, and the steaming temperature includes low heating rate, medium heating rate and high heating rate; the volatile compounds include polyphenols, tocopherols, sterols, glucosinolate degradation products, Maillard reaction products and medium lipid oxidation products; a group of populations including multiple particles is initialized according to the particle swarm algorithm, and a fitness function is set to evaluate the quality of each particle; wherein each particle represents a group of processing parameters. combination and the corresponding objective function value; sorting multiple particles in the population according to a fast non-dominated sorting algorithm and calculating the crowding of each particle to obtain a fast non-dominated sorting and crowding result; calculating the fitness value of each particle in the population according to the fast non-dominated sorting and crowding result; selecting a child population according to the fitness value of each particle, performing a crossover operation and a mutation operation on the child population to generate new particles; merging the new particles with the particles of the current population to obtain a new population; performing constraint feasibility judgment on individuals in the new population, excluding individuals that violate the constraints, obtaining an optimal solution set containing multiple non-dominated solutions, and selecting a processing parameter combination as a processing process from the optimal solution set.

[0017] The present invention aims at the rapeseed oil processing process, comprehensively considers multiple performance indicators such as oil yield, energy consumption and volatile compound content under degumming, and performs multi-objective optimization, which can more comprehensively reflect the needs of the actual processing process. By introducing a particle swarm algorithm for global search, and combining a fast non-dominated sorting algorithm to sort the solution set and calculate the congestion, it can effectively avoid premature convergence while maintaining population diversity.

[0018] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings: Figure 1 A schematic flow chart of a multi-objective optimization method for rapeseed oil processing parameters according to an embodiment of the present invention is shown; Figure 2A schematic diagram of a rapeseed oil production process according to an embodiment of the present invention is shown; Figure 3 A schematic diagram of the heating of rapeseed at a heating rate according to an embodiment of the present invention is shown; Figure 4 A schematic diagram showing the substance content of rapeseed oil at a heating rate according to an embodiment of the present invention is shown; Figure 5 A schematic diagram showing the volatile compound content of rapeseed oil at a heating rate according to an embodiment of the present invention is shown; Figure 6 A schematic diagram showing the volatile compound content of rapeseed oil in a degumming unit according to an embodiment of the present invention is shown; Figure 7 A schematic diagram of a framework of a multi-objective optimization device for rapeseed oil processing parameters according to an embodiment of the present invention is shown; Figure 8 A schematic diagram of the structure of a computing device according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0020] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.

[0021] Figure 1 The flowchart of the multi-objective optimization method of rapeseed oil processing parameters according to the embodiment of the present invention is shown. Figure 1 As shown, the following steps are included: Step S101, determining the processing parameters in the rapeseed oil processing process, taking the processing parameters as input variables, and taking the oil yield, energy consumption and volatile compound content under degumming as output variables to construct an objective function.

[0022] In the present embodiment, the rapeseed oil processing process is as follows Figure 2 As shown, rapeseed is put into a drum seed frying machine for frying until the end temperature is reached to obtain fried rapeseed. Then, rapeseed cake and crude rapeseed oil are obtained by spiral pressing. After the oil temperature is cooled, boiling NaCl solution is added to the crude rapeseed oil for degumming. After natural sedimentation, the upper layer of oil is taken to obtain the degummed finished oil, i.e., rapeseed oil.

[0023] The processing parameters include crushing degree, steaming temperature, steaming time and oil pressing pressure. Figure 3 As shown, including low heating rate, medium heating rate and high heating rate; volatile compounds such as Figures 4 to 6As shown, these include polyphenols, tocopherols, sterols, glucosinolate degradation products, Maillard reaction products, and lipid oxidation products.

[0024] Step S102, a population including a plurality of particles is initialized according to the particle swarm algorithm, and a fitness function is set to evaluate the quality of each particle; wherein each particle represents a combination of a set of processing parameters and a corresponding objective function value.

[0025] In this embodiment, the particle swarm algorithm is used to search for the optimal solution in the entire solution space to avoid falling into the local optimum, thereby finding a better combination of processing parameters.

[0026] Specifically, the number of particles is set, where each particle represents a combination of processing parameters, including crushing degree, steaming temperature (low heating rate, medium heating rate, high heating rate), steaming time, and oil pressing pressure. The fitness function is used to evaluate the quality of each particle, that is, the objective function value corresponding to the processing parameter combination. The objective function represents the evaluation index of oil yield, energy consumption, and volatile compound content under degumming. The higher the oil yield, the lower the energy consumption, and the lower the volatile compound content, the higher the fitness value.

[0027] Step S103, sorting multiple particles in the population according to the fast non-dominated sorting algorithm and calculating the crowding of each particle to obtain fast non-dominated sorting and crowding results; calculating the fitness value of each particle in the population according to the fast non-dominated sorting and crowding results; selecting a child population according to the fitness value of each particle, performing crossover operations and mutation operations on the child population to generate new particles; merging the new particles with the current population particles to obtain a new population; performing constraint feasibility judgment on individuals in the new population, excluding individuals that violate the constraints, obtaining an optimal solution set containing multiple non-dominated solutions, and selecting a processing parameter combination as the processing process from the optimal solution set.

[0028] In this embodiment, all non-dominated particles in the population are first found and classified as the first layer (i.e., the optimal layer). Then, the particles in the first layer are ignored, and non-dominated particles are continuously found among the remaining particles and classified as the second layer. This process is repeated until all particles are assigned to the corresponding layers. The particles in the population are then layered according to their non-dominated relationships, i.e., particles in each layer are not dominated by particles in other layers.

[0029] For each particle, the difference in objective function values ​​between its adjacent particles at its level (i.e., crowding distance) is calculated. Particles with large crowding degrees indicate that the solution space around them is relatively rich. The relative merits of each particle at its level are evaluated based on the crowding degree to maintain the diversity of the population.

[0030] According to the fast non-dominated sorting and crowding results, a fitness value is assigned to each particle in the population. Particles with lower non-dominated levels (i.e. particles with smaller levels) have higher fitness values. In the same level, particles with higher crowding have higher fitness values.

[0031] Select particles according to the fitness value, and select a part of excellent particles from the current population as the offspring population. Select two parent particles and generate offspring particles by exchanging some gene fragments. Randomly adjust some genes of offspring particles to generate new solutions. Merge the particles generated by crossover and mutation operations with the current population particles. If the size of the new population exceeds the preset value, the truncation strategy or resampling strategy is used to reduce it, and the newly generated particles are merged with the current population particles to form a new population.

[0032] Constraint checks are performed on each particle in the new population to exclude particles that violate constraints, ensuring that the particles in the new population meet the constraints of the processing process. Based on the fast non-dominated sorting and crowding results, particles with lower non-dominated levels and higher crowding are selected as the optimal solution, and one or more processing parameter combinations are selected from the optimal solution set as the final optimization result.

[0033] Take a population of 10 particles as an example, each particle represents a combination of processing parameters (such as crushing degree, steaming temperature, steaming time, and oil pressing pressure). It is divided into 3 layers (i.e., 3 non-dominated levels) by fast non-dominated sorting. In the first layer, there are 2 particles (denoted as P1 and P2) that are non-dominated, that is, no other particles dominate P1 and P2. In the second layer, there are 4 particles (denoted as P3 to P6) that are dominated by particles in the first layer but not dominated by other particles. In the third layer, there are 4 particles (denoted as P7 to P10) that are dominated by particles in the first and second layers. Then, the crowding degree of each particle is calculated. In the first layer, the crowding degrees of P1 and P2 are 0.8 and 0.6 respectively, in the second layer, the crowding degrees of P3 to P6 are 0.5, 0.7, 0.4, and 0.6 respectively, and in the third layer, the crowding degrees of P7 to P10 are 0.3, 0.5, 0.2, and 0.4 respectively. According to the results of fast non-dominated sorting and crowding, the fitness value of each particle is calculated. Assuming that the particles in the first layer have higher fitness values, the particles with higher crowding in the same layer have higher fitness values. It can be seen that the fitness value of P1 is higher than that of P2, and the fitness value of P3 is higher than that of P4 and P6, but lower than that of P5. Finally, the offspring population is selected according to the fitness value, and the first four particles with the highest fitness values ​​are selected as the offspring population (i.e., P1, P2, P5, and P3). The four particles are crossover and mutation operations are performed to generate new particles, and the new particles are merged with the current population particles to form a new population. The particles in the new population are subjected to constraint feasibility judgment, and particles that violate the constraints are excluded. The optimal combination of processing parameters is selected from the new population after constraint feasibility judgment as the final optimization result.

[0034] In an optional manner, the particle includes an iterative velocity vector, which is used to represent the updated moving direction and velocity of the particle in the search space, wherein the calculation formula of the iterative velocity vector is: in, is the k+1th iteration value of the velocity vector of the dth dimension of the ith particle; is the inertia weight; , , They are particle learning factor, population learning factor and neighbor influence factor respectively; , , are random numbers in the interval [0,1] respectively; is the optimal position of the ith particle in the dth dimension at the kth iteration; The value of the position of the i-th particle in the d-th dimension at the k-th iteration; is the value of the global optimal position in the dth dimension; The value of the optimal position of the neighbor of the i-th particle in the d-th dimension; is the adaptive adjustment coefficient; is the velocity change of the ith particle in the dth dimension.

[0035] In this embodiment, by introducing adaptive adjustment coefficients and speed changes, the speed and direction of particles are dynamically adjusted according to the search status and progress of the particles, thereby improving the search efficiency. Among them, the particle learning factor reflects the ability of particles to learn from their own historical optimal positions, so that particles can maintain individual memory during the search process and avoid blind search. The population learning factor reflects the tendency of particles to learn from the global optimal position, so that particles can quickly converge to the global optimal area. The neighbor influence factor introduces the information of the neighbor's optimal position, enhances the information exchange between particles, and accelerates convergence while maintaining diversity.

[0036] In an optional manner, the crossover operation randomly selects a crossover point and swaps the genes of two parent particles to generate a child particle; The mutation operation adjusts the gene of each particle with a preset probability, wherein the adjustment amplitude decreases as the number of iterations increases.

[0037] In this embodiment, by randomly selecting a crossover point and swapping the genes of two parent particles, a daughter particle with new characteristics can be generated, thereby increasing the diversity of genes in the population. The crossover operation combines the advantages of the two parent particles to generate a daughter particle with a better position in the search space. The mutation operation adjusts the genes of the particles to prevent the algorithm from converging to the local optimal solution too early. As the number of iterations increases, the adjustment range gradually decreases, so that a wide search can be performed in the early stage of the iteration and a detailed adjustment can be performed in the later stage of the iteration, thereby improving the convergence speed of the algorithm.

[0038] In an optional manner, the step of sorting the plurality of particles in the population according to the fast non-dominated sorting algorithm and calculating the congestion degree of each particle to obtain the fast non-dominated sorting and congestion degree results further comprises: Assign an initial rank to each particle, where all particles that are not dominated by other particles are assigned to the first frontier; For each particle in the population, find all the particles it dominates, starting from the first initial level, check the particles it dominates; assign the particles it dominates to the next initial level; In each initial level, particles are individually ranked according to the objective function value, and the overall ranking is determined based on the combination of the individual rankings; For the first and last particles of each initial level, set their crowding to infinity to ensure that they are not removed from the population; The congestion degree of other particles in each initial level is obtained by calculating the sum of the distances between the particle and its adjacent particles on each objective function.

[0039] In this embodiment, the fast non-dominated sorting algorithm considers multiple objective functions simultaneously, can quickly find the Pareto optimal frontier, and further provide a series of trade-off solutions.

[0040] For example, the objective optimization functions are f1(x) and f2(x), and there are 5 particles in the population, and the objective function values ​​are: Particle 1: (f1=1, f2=2); Particle 2: (f1=1.5, f2=1.5); Particle 3: (f1=2, f2=1); Particle 4: (f1=2.5, f2=2.5); Particle 5: (f1=3, f2=3).

[0041] Perform a fast non-dominated sort. Particles 1 and 3 are located in different areas of the Pareto front, so they are both non-dominated and assigned to level 1. Particle 2 is dominated by particles 1 and 3 and is assigned to level 2. Particles 4 and 5 are both dominated by particles 1, 2, and 3 (or worse) and are assigned to level 3.

[0042] Crowding calculation: For particles 1 and 3 at level 1, set their crowding to infinity. For particle 2 at level 2, calculate the sum of its distances from particles 1 and 3 on f1 and f2 as the crowding. For particles 4 and 5 at level 3, calculate the crowding in the same way.

[0043] Selection operation, when selecting the next generation of particles, first select particles 1 and 3 of level 1. If more particles are needed, select particles in level 2 according to the crowding degree, and so on, until the required number of particles is reached.

[0044] Finally, the position and velocity of the particles are updated according to the update rule of PSO. Through the above process, the Pareto optimal frontier is gradually approached and a series of trade-off solutions are found.

[0045] In an optional manner, after each iteration of the crossover operation and mutation operation, particles whose fitness is greater than a preset fitness threshold are retained in the next generation population and no longer participate in the crossover and mutation operation.

[0046] In this embodiment, by retaining particles with higher fitness to the next generation, unnecessary crossover and mutation operations are reduced. Although particles with high fitness are directly retained, the remaining particles still generate new solutions through crossover and mutation operations, maintaining the diversity of the population, thereby exploring a wider solution space and avoiding premature convergence.

[0047] In an optional manner, the calculation formula of the fitness threshold is: in, is the basic threshold; is the dynamic adjustment coefficient; is the average value of the fitness value of the population in the previous iteration; is the average fitness value of the current iteration population.

[0048] In this embodiment, in each iteration, the threshold is dynamically adjusted according to the average fitness value of the current population, elite particles are screened, new candidate solutions are generated, and the population is updated. The positions of most particles in the population are concentrated at the global minimum point. By calculating the difference between the average values ​​of the population fitness values, unreasonable changes in the fitness threshold in extreme cases (such as a sudden increase or decrease in the population fitness value) are avoided. Therefore, when the population fitness changes significantly, the fitness threshold can respond quickly and adjust its value to adapt to the new population state.

[0049] In an optional manner, the preset probability is an adaptive mutation probability to avoid premature convergence of particles. The calculation formula of the adaptive mutation probability is: in, is the mutation probability of the t+1th iteration; is the mutation probability of the tth iteration; To adjust the step size; is the average distance between all particles in the tth iteration; is the average distance between the ith particle and other particles in the tth iteration.

[0050] In this embodiment, the calculation formula of the adaptive mutation probability dynamically adjusts the mutation probability according to the current state of the population (i.e., the average distance between particles). When the particles in the population are more concentrated, the mutation probability increases to promote the population to explore new solution space areas. On the contrary, when the particles are more dispersed, the mutation probability decreases to maintain the stability of the population. When the population has not fully explored the solution space, a higher mutation probability helps to discover new solutions. As the iteration proceeds, when the population gradually converges to a better area, the mutation probability gradually decreases to reduce unnecessary search overhead.

[0051] In an optional manner, the process of constructing the optimal solution set of the non-dominated solutions is: Compare the dominance relationship between different individuals. If one individual is not inferior to another individual in all objectives and is superior to another individual in at least one objective, the former individual is judged to dominate the latter individual. The individuals in the non-dominated state are grouped into a non-dominated solution set, the average distance between the solutions in the non-dominated solution set and the nearest solution on the true Pareto frontier is calculated, and the solution set with the smallest average distance is taken as the optimal solution set of the non-dominated solution.

[0052] In this embodiment, by comparing the dominance relations between different individuals, a non-dominated solution set can be quickly constructed, avoiding the tedious process of traversing all possible solution combinations in the traditional method. At the same time, the average distance between the solution in the non-dominated solution set and the nearest solution on the true Pareto front surface is calculated to further screen out the optimal solution set. Using the dominance relationship to construct a non-dominated solution set ensures that the selected solution performs well on multiple objectives and is better than other solutions on at least one objective.

[0053] To facilitate understanding of the technical solution provided by this embodiment, the following exemplary simulation experiment data is used for illustration, wherein the processing parameter settings are shown in Table 1, the objective function output variables are shown in Table 2, the particle swarm algorithm parameters are shown in Table 3, and the simulation experiment results are shown in Table 4.

[0054] Table 1 Table 2 Table 3 Table 4 In Table 4, solutions 1, 2, and 3 are all non-dominated solutions, that is, they achieve a good balance between the oil yield, energy consumption, and volatile compound content. According to actual needs, the most appropriate combination of processing parameters can be selected from the optimal solution set. For example, if the oil yield and volatile compound content are given priority, solution 2 can be selected; if energy consumption is reduced, solution 3 can be considered, but the oil yield and volatile compound content may be reduced.

[0055] According to the scheme provided by the present invention, the processing parameters in the rapeseed oil processing process are determined, and the processing parameters are used as input variables, and the oil yield, energy consumption and volatile compound content under degumming are used as output variables to construct an objective function; wherein the processing parameters include crushing degree, steaming temperature, steaming time and oil pressing pressure, and the steaming temperature includes low heating rate, medium heating rate and high heating rate; the volatile compounds include polyphenols, tocopherols, sterols, glucosinolate degradation products, Maillard reaction products and medium lipid oxidation products; a group of populations including multiple particles is initialized according to the particle swarm algorithm, and a fitness function is set to evaluate the quality of each particle; wherein each particle represents a group of processing parameters. combination and the corresponding objective function value; sort multiple particles in the population according to the fast non-dominated sorting algorithm and calculate the crowding of each particle to obtain the fast non-dominated sorting and crowding results; calculate the fitness value of each particle in the population according to the fast non-dominated sorting and crowding results; select the offspring population according to the fitness value of each particle, perform crossover operation and mutation operation on the offspring population to generate new particles; merge the new particles with the current population particles to obtain a new population; perform constraint feasibility judgment on the individuals in the new population, exclude the individuals that violate the constraints, obtain the optimal solution set containing multiple non-dominated solutions, and select the processing parameter combination as the processing process from the optimal solution set. The present invention is aimed at the rapeseed oil processing process, comprehensively considers multiple performance indicators such as oil yield, energy consumption and volatile compound content under degumming, and performs multi-objective optimization, which can more comprehensively reflect the needs of the actual processing process. By introducing the particle swarm algorithm for global search, and combining the fast non-dominated sorting algorithm to sort the solution set and calculate the crowding, it can effectively avoid premature convergence while maintaining population diversity.

[0056] Figure 7 The structure diagram of the multi-objective optimization device for rapeseed oil processing parameters according to the embodiment of the present invention is shown. The objective function construction module 710 is used to determine the processing parameters in the rapeseed oil processing process, and to construct the objective function with the processing parameters as input variables and the oil yield, energy consumption and volatile compound content under degumming as output variables; wherein the processing parameters include crushing degree, steaming temperature, steaming time and oil pressing pressure, and the steaming temperature includes low heating rate, medium heating rate and high heating rate; the volatile compounds include polyphenols, tocopherols, sterols, glucosinolate degradation products, Maillard reaction products and medium lipid oxidation products; The particle swarm initialization module 720 is used to initialize a population of multiple particles according to the particle swarm algorithm, and set a fitness function to evaluate the quality of each particle; wherein each particle represents a combination of a set of processing parameters and a corresponding objective function value; The optimal solution set selection module 730 is used to sort multiple particles in the population according to the fast non-dominated sorting algorithm and calculate the crowding of each particle to obtain fast non-dominated sorting and crowding results; calculate the fitness value of each particle in the population according to the fast non-dominated sorting and crowding results; select a child population according to the fitness value of each particle, perform crossover operations and mutation operations on the child population to generate new particles; merge the new particles with the current population particles to obtain a new population; perform constraint feasibility judgment on the individuals in the new population, exclude individuals that violate the constraints, obtain an optimal solution set containing multiple non-dominated solutions, and select a processing parameter combination as the processing process from the optimal solution set.

[0057] Figure 8 The schematic diagram of the structure of the computing device embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the computing device.

[0058] like Figure 8 As shown, the computing device may include: a processor (processor) 802 , a communication interface (Communications Interface) 804 , a memory (memory) 806 , and a communication bus 808 .

[0059] The processor 802, the communication interface 804, and the memory 806 communicate with each other via a communication bus 808. The communication interface 804 is used to communicate with other devices such as a client or other server network elements. The processor 802 is used to execute a program 810, which can specifically execute the relevant steps in the embodiment of the multi-objective optimization method for rapeseed oil processing parameters.

[0060] Specifically, the program 810 may include program codes, which include computer operation instructions.

[0061] The processor 802 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention. The one or more processors included in the computing device may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0062] The memory 806 is used to store the program 810. The memory 806 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0063] According to the scheme provided by the present invention, the processing parameters in the rapeseed oil processing process are determined, and the processing parameters are used as input variables, and the oil yield, energy consumption and volatile compound content under degumming are used as output variables to construct an objective function; wherein the processing parameters include crushing degree, steaming temperature, steaming time and oil pressing pressure, and the steaming temperature includes low heating rate, medium heating rate and high heating rate; the volatile compounds include polyphenols, tocopherols, sterols, glucosinolate degradation products, Maillard reaction products and medium lipid oxidation products; a group of populations including multiple particles is initialized according to the particle swarm algorithm, and a fitness function is set to evaluate the quality of each particle; wherein each particle represents a group of processing parameters. combination and the corresponding objective function value; sort multiple particles in the population according to the fast non-dominated sorting algorithm and calculate the crowding of each particle to obtain the fast non-dominated sorting and crowding results; calculate the fitness value of each particle in the population according to the fast non-dominated sorting and crowding results; select the offspring population according to the fitness value of each particle, perform crossover operation and mutation operation on the offspring population to generate new particles; merge the new particles with the current population particles to obtain a new population; perform constraint feasibility judgment on the individuals in the new population, exclude the individuals that violate the constraints, obtain the optimal solution set containing multiple non-dominated solutions, and select the processing parameter combination as the processing process from the optimal solution set. The present invention is aimed at the rapeseed oil processing process, comprehensively considers multiple performance indicators such as oil yield, energy consumption and volatile compound content under degumming, and performs multi-objective optimization, which can more comprehensively reflect the needs of the actual processing process. By introducing the particle swarm algorithm for global search, and combining the fast non-dominated sorting algorithm to sort the solution set and calculate the crowding, it can effectively avoid premature convergence while maintaining population diversity.

[0064] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and further may be divided into multiple submodules or subunits or subcomponents. All features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device disclosed in this manner may be combined in any combination, except that at least some of such features and / or processes or units are mutually exclusive. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature that provides the same, equivalent or similar purpose. In addition, those skilled in the art will appreciate that, although some embodiments herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the present invention and to form different embodiments. For example, in the following claims, any one of the claimed embodiments may be used in any combination. The present invention may be implemented by means of hardware including several different elements and by means of a suitably programmed computer. In a unit claim that lists several means, several of these means may be embodied by the same hardware item. The steps in the above embodiments should not be understood as limiting the execution order unless otherwise specified.

Claims

1. A multi-objective optimization method for rapeseed oil processing parameters, characterized in that: include: Determine the processing parameters in the rapeseed oil processing process, take the processing parameters as input variables, and take the oil yield, energy consumption and volatile compound content under degumming as output variables to construct an objective function; wherein the processing parameters include crushing degree, steaming temperature, steaming time and oil pressing pressure, and the steaming temperature includes low heating rate, medium heating rate and high heating rate; the volatile compounds include polyphenols, tocopherols, sterols, glucosinolate degradation products, Maillard reaction products and medium lipid oxidation products; A population of multiple particles is initialized according to the particle swarm algorithm, and a fitness function is set to evaluate the quality of each particle; wherein each particle represents a combination of a set of processing parameters and a corresponding objective function value; According to a fast non-dominated sorting algorithm, multiple particles in a population are sorted and the crowding degree of each particle is calculated to obtain fast non-dominated sorting and crowding results; according to the fast non-dominated sorting and crowding results, the fitness value of each particle in the population is calculated; according to the fitness value of each particle, a child population is selected, and a crossover operation and a mutation operation are performed on the child population to generate new particles; the new particles are merged with the particles of the current population to obtain a new population; the individuals in the new population are subjected to constraint feasibility judgment, and the individuals violating the constraints are excluded to obtain an optimal solution set containing multiple non-dominated solutions, and a combination of processing parameters for a processing process is selected from the optimal solution set.

2. The multi-objective optimization method for rapeseed oil processing parameters according to claim 1, characterized in that: The particle includes an iterative velocity vector, which is used to represent the updated moving direction and velocity of the particle in the search space, wherein the calculation formula of the iterative velocity vector is: in, is the k+1th iteration value of the velocity vector of the dth dimension of the ith particle; is the inertia weight; , , They are particle learning factor, population learning factor and neighbor influence factor respectively; , , are random numbers in the interval [0,1] respectively; is the optimal position of the ith particle in the dth dimension at the kth iteration; The value of the position of the i-th particle in the d-th dimension at the k-th iteration; is the value of the global optimal position in the dth dimension; The value of the optimal position of the neighbor of the i-th particle in the d-th dimension; is the adaptive adjustment coefficient; is the velocity change of the ith particle in the dth dimension.

3. The multi-objective optimization method for rapeseed oil processing parameters according to claim 1, characterized in that: The crossover operation randomly selects a crossover point and swaps the genes of two parent particles to generate a child particle; The mutation operation adjusts the gene of each particle with a preset probability, wherein the adjustment amplitude decreases as the number of iterations increases.

4. The multi-objective optimization method for rapeseed oil processing parameters according to claim 1, characterized in that: The step of sorting the plurality of particles in the population according to the fast non-dominated sorting algorithm and calculating the congestion degree of each particle to obtain the fast non-dominated sorting and congestion degree results further comprises: Assign an initial rank to each particle, where all particles that are not dominated by other particles are assigned to the first frontier; For each particle in the population, find all the particles it dominates, starting from the first initial level, check the particles it dominates; assign the particles it dominates to the next initial level; In each initial level, particles are individually ranked according to the objective function value, and the overall ranking is determined based on the combination of the individual rankings; For the first and last particles of each initial level, set their crowding to infinity to ensure that they are not removed from the population; The congestion degree of other particles in each initial level is obtained by calculating the sum of the distances between the particle and its adjacent particles on each objective function.

5. The multi-objective optimization method for rapeseed oil processing parameters according to claim 1, characterized in that: After each iteration of the crossover operation and mutation operation, particles whose fitness is greater than a preset fitness threshold are retained in the next generation population and no longer participate in the crossover and mutation operation.

6. The multi-objective optimization method for rapeseed oil processing parameters according to claim 5, characterized in that: The calculation formula of the fitness threshold is: in, is the basic threshold; is the dynamic adjustment coefficient; is the average value of the fitness value of the population in the previous iteration; is the average fitness value of the current iteration population.

7. The multi-objective optimization method for rapeseed oil processing parameters according to claim 3, characterized in that: The preset probability is the adaptive mutation probability to avoid premature convergence of particles. The calculation formula of the adaptive mutation probability is: in, is the mutation probability of the t+1th iteration; is the mutation probability of the tth iteration; To adjust the step size; is the average distance between all particles in the tth iteration; is the average distance between the ith particle and other particles in the tth iteration.

8. The multi-objective optimization method for rapeseed oil processing parameters according to claim 1, characterized in that: The construction process of the optimal solution set of the non-dominated solution is: Compare the dominance relationship between different individuals. If one individual is not inferior to another individual in all objectives and is superior to another individual in at least one objective, the former individual is judged to dominate the latter individual. The individuals in the non-dominated state are grouped into a non-dominated solution set, the average distance between the solutions in the non-dominated solution set and the nearest solution on the true Pareto frontier is calculated, and the solution set with the smallest average distance is taken as the optimal solution set of the non-dominated solution.

9. A multi-objective optimization device for rapeseed oil processing parameters, characterized in that: include: An objective function construction module is used to determine the processing parameters in the rapeseed oil processing process, and the processing parameters are used as input variables, and the oil yield, energy consumption and volatile compound content under degumming are used as output variables to construct the objective function; wherein the processing parameters include crushing degree, steaming and frying temperature, steaming and frying time and oil pressing pressure, and the steaming and frying temperature includes low heating rate, medium heating rate and high heating rate; the volatile compounds include polyphenols, tocopherols, sterols, glucosinolate degradation products, Maillard reaction products and medium lipid oxidation products; The particle swarm initialization module is used to initialize a population of multiple particles according to the particle swarm algorithm and set a fitness function to evaluate the quality of each particle; wherein each particle represents a combination of a set of processing parameters and a corresponding objective function value; The optimal solution set selection module is used to sort multiple particles in the population according to the fast non-dominated sorting algorithm and calculate the crowding degree of each particle to obtain fast non-dominated sorting and crowding results; calculate the fitness value of each particle in the population according to the fast non-dominated sorting and crowding results; select a child population according to the fitness value of each particle, perform crossover operations and mutation operations on the child population to generate new particles; merge the new particles with the current population particles to obtain a new population; perform constraint feasibility judgment on individuals in the new population, exclude individuals that violate constraints, obtain an optimal solution set containing multiple non-dominated solutions, and select a processing parameter combination as a processing process from the optimal solution set.

10. A computing device comprising: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the multi-objective optimization method of rapeseed oil processing parameters.

Images