A multi-objective optimization control method for heat pump drying system

Abstract

The present invention relates to the field of optimization control, and in particular to a multi-objective optimization control method for a heat pump drying system. The specific contents include: first, real-time data parameters are collected during the heat pump drying process, and the collected data parameters are pre-processed, feature extracted, and dimension reduced to obtain key feature data; further, based on the key feature data, a hierarchical iterative dynamic multi-objective optimization algorithm is used to comprehensively optimize energy consumption and product quality to obtain a multi-objective optimization result; based on the multi-objective optimization result, the operating parameters of the heat pump drying process are adjusted to perform multi-objective optimization control. The method solves the technical problems of inaccurate data processing, inability to provide accurate and reliable data, and incomplete and poor adaptability when implementing optimization control during the heat pump drying process.

Description

Technical Field

[0001] The invention relates to the field of optimization control, and in particular to a multi-objective optimization control method for a heat pump drying system. Background Art

[0002] Heat pump drying systems are widely used in food, medicine, agriculture and other fields. By combining heat pump technology with drying process, energy efficiency and drying quality can be improved. However, traditional heat pump drying systems face a series of technical challenges in practical applications, such as how to reduce energy consumption while ensuring the quality of dried products, how to adapt to different drying needs for real-time adjustment, and how to achieve a balance between energy consumption and product quality in multi-objective optimization control. These technical difficulties have seriously affected the widespread application and promotion of heat pump drying systems.

[0003] In the heat pump drying process, real-time monitoring and optimization control of environmental parameters (such as temperature, humidity, etc.) and product quality parameters (such as color, hardness, etc.) are crucial. Most existing heat pump drying systems rely on experience adjustment and simple feedback control, and cannot make full use of real-time data for intelligent analysis and optimization control, resulting in high energy consumption, large fluctuations in product quality, and other problems. In addition, most existing optimization algorithms focus on the optimization of a single objective (such as energy consumption or product quality), lacking a multi-objective optimization method that comprehensively considers energy consumption and product quality, making it difficult to achieve ideal results in practical applications.

[0004] The existing heat pump drying system has the following technical problems: in the heat pump drying process, the data processing is not accurate enough, and accurate and reliable data cannot be provided; and the optimization control is not comprehensive and has poor adaptability. Summary of the invention

[0005] The present invention provides a multi-objective optimization control method for a heat pump drying system to solve the technical problems of inaccurate data processing, inability to provide accurate and reliable data, and incomplete and poor adaptability when implementing optimization control during the heat pump drying process.

[0006] A multi-objective optimization control method for a heat pump drying system of the present invention specifically includes the following technical solutions:

[0007] A multi-objective optimization control method for a heat pump drying system comprises the following steps:

[0008] S1. Real-time data parameter collection during heat pump drying process, and preprocessing, feature extraction and dimensionality reduction of the collected data parameters to obtain key feature data;

[0009] S2. Based on the key feature data, a hierarchical iterative dynamic multi-objective optimization algorithm is used to comprehensively optimize energy consumption and product quality to obtain multi-objective optimization results; based on the multi-objective optimization results, the operating parameters of the heat pump drying process are adjusted to perform multi-objective optimization control.

[0010] Preferably, the S1 specifically includes:

[0011] When performing dimensionality reduction processing on the extracted features, a principal component analysis algorithm with adaptive fidelity optimization is introduced.

[0012] Preferably, the S1 specifically includes:

[0013] In the process of implementing the principal component analysis algorithm with adaptive fidelity optimization, the extracted feature matrix is ​​subjected to information gain weighting processing to obtain a weighted feature matrix; the weighted feature matrix is ​​nonlinearly mapped using a nonlinear mapping function to obtain a weighted feature matrix after nonlinear mapping.

[0014] Preferably, the S1 specifically includes:

[0015] The covariance matrix of the weighted feature matrix after nonlinear mapping and the eigenvalues ​​and eigenvectors of the covariance matrix of the weighted feature matrix after nonlinear mapping are calculated.

[0016] Preferably, the S1 specifically includes:

[0017] Define the fidelity function, maximize the fidelity function, select the optimal feature vector for dimensionality reduction, and obtain key feature data.

[0018] Preferably, the S2 specifically includes:

[0019] The hierarchical iterative dynamic multi-objective optimization algorithm comprehensively optimizes energy consumption and product quality by introducing a hierarchical structure and iterative dynamic optimization steps.

[0020] Preferably, the S2 specifically includes:

[0021] In the process of implementing the hierarchical iterative dynamic multi-objective optimization algorithm, an initial population is randomly generated, and the population is composed of individuals, each of which is composed of key feature data; a hierarchical structure is introduced to divide the population into levels, and the relationship between levels is set through the mutual influence calculation formula between levels; the target values ​​of energy consumption and product quality for each individual are calculated, and the comprehensive optimization objective function is defined; further, selection operations within the level, crossover operations between levels, and mutation operations within the level are performed.

[0022] Preferably, the S2 specifically includes:

[0023] In the process of implementing the hierarchical iterative dynamic multi-objective optimization algorithm, the newly generated offspring individuals in the crossover operation are added to the population, and the hierarchical structure is updated to form a new generation of population, and the hierarchical adjustment is performed according to the calculation formula of the mutual influence between the levels.

[0024] Preferably, the S2 specifically includes:

[0025] In the process of implementing the hierarchical iterative dynamic multi-objective optimization algorithm, the optimization weights and inter-level parameters are dynamically adjusted, and the multi-objective optimization results are obtained through continuous iteration.

[0026] The beneficial effects of the technical solution of the present invention are:

[0027] 1. The principal component analysis algorithm with adaptive fidelity optimization is introduced to ensure that the feature data after dimensionality reduction retains sufficient information. The optimized feature extraction method can improve the efficiency and effect of data processing and provide a reliable data foundation for subsequent multi-objective optimization.

[0028] 2. Through the hierarchical iterative dynamic multi-objective optimization algorithm, optimization is performed at different levels, which improves the global search capability and the diversity of solutions. The hierarchical iterative dynamic multi-objective optimization algorithm achieves the comprehensive optimization of energy consumption and product quality by introducing a hierarchical structure and iterative dynamic optimization steps.

[0029] 3. The formula for calculating the mutual influence between levels is introduced. Through the information exchange and interaction in the hierarchical structure, the diversity and global search ability of the population are enhanced. The multi-level evaluation comprehensively considers the optimization effects of different levels, and improves the globality and stability of the optimization effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a flow chart of a multi-objective optimization control method for a heat pump drying system described in the present invention. DETAILED DESCRIPTION

[0031] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0033] The specific scheme of the multi-objective optimization control method for a heat pump drying system provided by the present invention is described in detail below with reference to the accompanying drawings.

[0034] See attached Figure 1 , which shows a flow chart of a multi-objective optimization control method for a heat pump drying system provided by an embodiment of the present invention, the method comprising the following steps:

[0035] S1. Collect data parameters in the heat pump drying process in real time, and perform preprocessing, feature extraction and dimensionality reduction on the collected data parameters to obtain key feature data.

[0036] Data parameters in the heat pump drying process are collected in real time through data collection devices such as temperature sensors, humidity sensors and quality detection sensors, such as environmental data parameters (such as temperature, humidity, etc.) and product quality related data parameters (such as color, hardness, etc.); the collected data parameters are preprocessed and feature extracted to obtain feature-extracted data; the specific implementation process is as follows:

[0037] First, the collected data parameters are preprocessed including data cleaning and data normalization to obtain preprocessed data. The preprocessing method is a technical means well known to those skilled in the art and will not be described in detail here;

[0038] The preprocessed data is subjected to feature extraction using existing feature engineering techniques to obtain a feature matrix, and the extracted feature matrix is ​​subjected to dimensionality reduction processing using an existing principal component analysis method to obtain feature data after dimensionality reduction processing.

[0039] When performing dimensionality reduction processing on the extracted feature matrix, in order to ensure that the feature data after dimensionality reduction can retain sufficient information, the traditional principal component analysis method is optimized and the principal component analysis algorithm with adaptive fidelity optimization is introduced. The specific implementation process is as follows:

[0040] In order to ensure the representativeness and effectiveness of the feature data after dimensionality reduction, the extracted feature matrix is ​​weighted by information gain to obtain the weighted feature matrix , and then calculate the weighted feature matrix The weighted feature matrix after nonlinear mapping introduces the nonlinear mapping function , and obtain the weighted feature matrix after nonlinear mapping; the formula is:

[0041] ,

[0042] in, is the weighted feature matrix after nonlinear mapping; , , They are the activation functions of the nonlinear mapping of the first layer, the second layer, and the third layer (such as ReLU, Sigmoid, Tanh, etc.), which can be set according to the specific implementation scenario and are not limited here; , , They are the weight matrices of the nonlinear mapping of the first, second, and third layers respectively; , , They are the bias vectors of the nonlinear mapping of the first, second and third layers respectively.

[0043] Further, the covariance matrix of the weighted feature matrix after nonlinear mapping is calculated , the formula is:

[0044] ,

[0045] in, For the A single eigenvector in the weighted feature matrix after nonlinear mapping; is the mean vector of the weighted feature matrix after nonlinear mapping; It is the total number of samples in the weighted feature matrix after nonlinear mapping, that is, the number of samples in the weighted feature matrix corresponding to the data parameters collected at different time points or under different conditions; Represents the transpose of a matrix.

[0046] Further, the covariance matrix of the weighted feature matrix after nonlinear mapping is calculated The eigenvalue of and the eigenvector , the formula is:

[0047] ,

[0048] in, For the eigenvalues, For the feature vectors.

[0049] In order to ensure that the feature data can retain enough information after dimension reduction, the fidelity function is defined , the formula is:

[0050] ,

[0051] in, It is the fidelity function, which indicates the degree of information retention of the feature data after dimensionality reduction; is a single eigenvector in the weighted feature matrix after nonlinear mapping; is the mutual information term, which represents the features after nonlinear mapping With features The mutual information between them is used to measure the correlation between two features; is the weight coefficient, which controls the mutual information term in the fidelity function The impact of is the regularization parameter that controls the impact of the correlation constraint between eigenvectors; is the absolute value of the dot product of the eigenvectors, indicating the correlation between different eigenvectors; For the feature vectors; is the number of features after dimensionality reduction.

[0052] The optimization goal is: ;

[0053] By maximizing the fidelity function , select the optimal feature vector for dimensionality reduction processing, and obtain the feature data after dimensionality reduction, that is, the key feature data, which can include temperature, humidity, energy consumption, product humidity, product color, product hardness, etc., as the basis for subsequent multi-objective optimization to ensure the comprehensive optimization of energy consumption and product quality of the heat pump drying system.

[0054] S2. Based on the key feature data, a hierarchical iterative dynamic multi-objective optimization algorithm is used to comprehensively optimize energy consumption and product quality to obtain multi-objective optimization results; based on the multi-objective optimization results, the operating parameters of the heat pump drying process are adjusted to perform multi-objective optimization control.

[0055] Based on key feature data, a hierarchical iterative dynamic multi-objective optimization algorithm is used to comprehensively optimize energy consumption and product quality, and a multi-objective optimization result is obtained; the hierarchical iterative dynamic multi-objective optimization algorithm achieves comprehensive optimization of energy consumption and product quality by introducing a hierarchical structure and iterative dynamic optimization steps. The specific implementation process is as follows:

[0056] The population and hierarchy are initialized to optimize at different levels to improve the global search capability and diversity of solutions.

[0057] An initial population is randomly generated, the initial population is composed of individuals, each individual is composed of key feature data, a hierarchical structure is introduced, the initial population is divided into different levels, each level is composed of individuals, and the relationship between the levels is set. The specific number of levels and the individuals contained in each level can be set according to the specific implementation scenario, and are not limited here. In order to set the relationship between the levels, a calculation formula for the mutual influence between the levels is introduced, and the diversity and global search ability of the population are enhanced through information exchange and interaction in the hierarchical structure;

[0058] No. Individual level By the Level and The influence of hierarchical individuals is as follows:

[0059] ,

[0060] in, It is Level individual; It is Level individual; It is Level individual; and It is the influencing factor between levels, controlling the interaction between different levels.

[0061] Furthermore, the fitness is evaluated and multi-level evaluation is introduced. For each individual, the target values ​​of energy consumption and product quality are calculated, and a multi-level evaluation is performed according to the hierarchical structure. The multi-level evaluation is to comprehensively consider the optimization effects at different levels and improve the globality and stability of the optimization effect.

[0062] Define the comprehensive optimization objective function for:

[0063] ,

[0064] in, represents the key feature vector; It is Key characteristic data, representing the key characteristic data of the heat pump drying system; are the coefficients and powers related to energy consumption, which respectively represent the weight and mode of influence of each characteristic data on energy consumption; It is Key feature data; are the coefficients and powers related to product quality, which respectively represent the weight and mode of influence of each characteristic data on product quality; It is the optimization weight, which indicates the balance between energy consumption and product quality, and is dynamically adjusted to meet different optimization requirements; is the global energy consumption objective function, is the number of key features related to energy consumption; is the global product quality objective function; is the number of key features related to product quality; is the current level; is the hierarchical balancing factor, used to adjust the optimization impact of different levels in the hierarchy; is the total number of levels; It is a combination of hierarchical objective functions, including local energy consumption, product quality and robustness optimization. By averaging the objective functions at different levels, it ensures that the requirements of each level can be balanced in the overall optimization process. It is The hierarchical objective function specifically refers to the sub-optimization objective at each level, and its physical meaning can cover local energy consumption optimization, local quality optimization, and performance optimization under other specific operating conditions.

[0065] At the same time, constraints are set according to expert experience.

[0066] Perform selection operations within the hierarchy; select excellent individuals in each hierarchy to enter the next generation population according to the fitness value. The selection operation can adopt existing roulette selection or tournament selection methods. The selection operation within the hierarchy is to eliminate the weak and retain individuals with higher fitness to the next generation population to improve the overall quality within the hierarchy.

[0067] After the selection operation, a crossover operation is performed between levels; a crossover operation is performed between levels to generate new offspring individuals. The crossover operation can adopt existing single-point crossover or multi-point crossover methods; the crossover operation between levels is to generate new solutions through gene recombination, thereby increasing the diversity and search capability of the population.

[0068] After the crossover operation, mutation operations are performed within the layer. Mutation operations are performed on the newly generated offspring individuals within each layer to increase the diversity of the population. The mutation operation can adopt existing random mutation or Gaussian mutation methods. The mutation operation within the layer is to introduce new genes to prevent the population from falling into the local optimal solution and improve the global search capability.

[0069] Furthermore, the hierarchical structure is updated; the newly generated offspring individuals are added to the population, and the hierarchical structure is updated to form a new generation of population, and the hierarchical adjustment is performed according to the calculation formula of the mutual influence between the hierarchical levels:

[0070] Furthermore, in order to adaptively adjust the optimization strategy during the optimization process and improve the flexibility and effect of the optimization process, the weights and inter-level parameters are dynamically adjusted; the optimization weights are dynamically adjusted according to the feedback information during the optimization process. and inter-level impact factors and , the formula is as follows:

[0071] ,

[0072] ,

[0073] ,

[0074] in, is the number of iterations, , , , It is an adjustment parameter used to control the change speed and center point of weights and influencing factors; , It is the adjustment factor of feedback information, which controls the sensitivity of feedback information to the adjustment of weights and influencing factors; , It is The real-time energy consumption feedback and real-time product quality feedback corresponding to the iteration are obtained based on the deviation between the real-time energy consumption and product quality and the target energy consumption and product quality. , , Replace optimization weights and inter-level impact factors and .

[0075] Finally, when the preset number of iterations is reached or the optimization goal is met, the optimization process is terminated; otherwise, the iteration continues.

[0076] Through the hierarchical iterative dynamic multi-objective optimization algorithm, it is possible to achieve comprehensive optimization of energy consumption and product quality in the heat pump drying system, and gradually obtain the optimal solution, i.e., the multi-objective optimization result. The multi-objective optimization result includes the values ​​of key characteristic data for achieving comprehensive optimization of energy consumption and product quality, i.e., the values ​​of key operating parameters in the heat pump drying system.

[0077] According to the key operating parameters in the heat pump drying system, the existing PID control technology is used to adjust the operating parameters of the heat pump drying process to achieve multi-objective optimization control.

[0078] In summary, a multi-objective optimization control method for a heat pump drying system was completed.

[0079] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0080] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0081] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A multi-objective optimization control method for a heat pump drying system, characterized in that: The following steps are involved: S1. Real-time data collection of heat pump drying process data parameters, and pre-processing, feature extraction and dimension reduction of the collected data parameters. When the extracted features are reduced in dimension, the principal component analysis algorithm with adaptive fidelity optimization is introduced. In the process of implementing the principal component analysis algorithm with adaptive fidelity optimization, the extracted feature matrix is ​​weighted by information gain to obtain a weighted feature matrix. ; Using nonlinear mapping function The weighted feature matrix is ​​nonlinearly mapped to obtain the weighted feature matrix after nonlinear mapping. The specific formula is: ， in, is the weighted feature matrix after nonlinear mapping; , , They are the activation functions of the nonlinear mapping of the first, second and third layers respectively; , , They are the weight matrices of the nonlinear mapping of the first, second, and third layers respectively; , , are the bias vectors of the nonlinear mapping of the first, second and third layers respectively; And calculate the covariance matrix of the weighted feature matrix after nonlinear mapping , the specific formula is: ， in, For the A single eigenvector in the weighted feature matrix after nonlinear mapping; is the mean vector of the weighted feature matrix after nonlinear mapping; is the total number of samples in the weighted feature matrix after nonlinear mapping; Represents the transpose of a matrix; Calculate the covariance matrix of the weighted feature matrix after nonlinear mapping The eigenvalue of and the eigenvector , the specific formula is: ， in, For the eigenvalues, For the feature vectors; Defining the fidelity function , the specific formula is: ， in, is the fidelity function; is the mutual information term, which represents the features after nonlinear mapping With features The mutual information between is the weight coefficient; is the regularization parameter; is the absolute value of the dot product of the eigenvectors; For the feature vectors; is the number of features after dimensionality reduction; By maximizing the fidelity function, the optimal feature vector is selected for dimensionality reduction to obtain key feature data; S2. Based on the key feature data, a hierarchical iterative dynamic multi-objective optimization algorithm is used to comprehensively optimize energy consumption and product quality. In the process of implementing the hierarchical iterative dynamic multi-objective optimization algorithm, an initial population is randomly generated. The population consists of individuals, and each individual consists of key feature data. A hierarchical structure is introduced to divide the population into levels, and the relationship between the levels is set through the calculation formula for the mutual influence between the levels. Individual level By the Level and The influence of hierarchical individuals is as follows: ， in, It is Level individual; It is Level individual; It is Level individual; and is the influencing factor between levels; And calculate the target values ​​of each individual energy consumption and product quality, and define the comprehensive optimization objective function; further, perform selection operations within the level, crossover operations between levels, and variation operations within the level to obtain multi-objective optimization results; based on the multi-objective optimization results, adjust the operating parameters of the heat pump drying process and perform multi-objective optimization control.

2. The multi-objective optimization control method for the heat pump drying system according to claim 1, characterized in that: The S2 specifically includes: The hierarchical iterative dynamic multi-objective optimization algorithm comprehensively optimizes energy consumption and product quality by introducing a hierarchical structure and iterative dynamic optimization steps.

3. The multi-objective optimization control method of the heat pump drying system according to claim 2, characterized in that: The S2 specifically includes: In the process of implementing the hierarchical iterative dynamic multi-objective optimization algorithm, the newly generated offspring individuals in the crossover operation are added to the population, and the hierarchical structure is updated to form a new generation of population, and the hierarchical adjustment is performed according to the calculation formula of the mutual influence between the levels.

4. The multi-objective optimization control method for a heat pump drying system according to claim 3, characterized in that: The S2 specifically includes: In the process of implementing the hierarchical iterative dynamic multi-objective optimization algorithm, the optimization weights and inter-level parameters are dynamically adjusted, and the multi-objective optimization results are obtained through continuous iteration.

Images