A Smart Control Method for Unconventional Feed Production Line

Abstract

This invention discloses an intelligent control method for an unconventional feed production line, relating to the field of intelligent control. It acquires user-defined weights for conventional nutrition, functional nutrition, and safety risks; uses raw material ratio vectors and production process parameter vectors as decision variables; takes maximizing the comprehensive performance score function of the feed product as the optimization objective; and uses the ranges of nutritional indicators, safety indicators, raw material ratios, and process parameters as constraints to construct a multi-objective optimization model. The multi-objective optimization model is solved using an optimization algorithm to obtain the optimal raw material ratio vector and the optimal production process parameter vector. Based on the optimal raw material ratio vector and the optimal production process parameter vector, the unconventional feed production line is controlled to produce feed. This method can adaptively adjust the ratios of various raw materials and key production process parameters to stably produce unconventional feed products that achieve comprehensive optimization in terms of conventional nutrition, functional nutrition, and safety.

Description

Technical Field

[0001] This invention relates to the field of intelligent control, and in particular to an intelligent control method for an unconventional feed production line. Background Technology

[0002] With the rapid development of the livestock industry, the shortage of feed resources has become increasingly prominent. Unconventional feeds (such as agricultural by-products, industrial by-products, and wild plant resources) have gradually become an important supplement to the feed industry due to their abundant resources and low cost. However, compared with traditional feeds, unconventional feeds have complex compositions, fluctuating nutritional values, and may contain harmful substances such as heavy metals, toxins, and anti-nutritional factors. These characteristics make the process control of unconventional feed production a crucial issue in feed science, necessitating the establishment of systematic and scientific evaluation methods to ensure the health and stable production performance of farmed animals.

[0003] However, the large-scale industrial application of unconventional feeds faces severe technical challenges. First, the composition of unconventional feed ingredients is complex and highly variable, with significant differences in nutritional components (such as protein, energy, and minerals) compared to conventional ingredients. Simply relying on linear programming for nutrient formulation makes precise matching difficult. Second, the value of unconventional feeds lies not only in their conventional nutrients but also in their functional nutrients, such as polysaccharides with immune-enhancing effects and phenolic substances with antioxidant properties. Existing feed formulation technologies primarily focus on meeting basic nutritional needs, rarely offering systematic quantitative modeling and synergistic optimization of these functional nutrients. More critically, unconventional feed ingredients commonly contain anti-nutritional factors (such as phytates and tannins), natural toxins (such as gossypol and sinigrin), and heavy metal contamination risks. These harmful substances interact complexly with nutritional components and process parameters. For example, certain processing techniques (such as heat treatment) may degrade toxins while simultaneously causing functional protein denaturation or vitamin loss. Existing feed production control methods typically treat formulation design and process control as relatively independent links, or only control a single risk factor in a simple way. There is a lack of a holistic solution that can coordinate raw material ratios and production processes to systematically control safety risks while maximizing nutritional functions.

[0004] Specifically, the existing technology has the following main drawbacks: 1) Disconnect between formulation and process: Traditional linear programming formulation software mainly optimizes the proportion of raw materials, but cannot dynamically correlate and optimize the matching production process parameters (such as particle size, conditioning temperature, expansion intensity, etc.), while process parameters have a decisive impact on nutrient retention, enhancement of functional active substances and deactivation of anti-nutritional factors.

[0005] 2) Insufficient multi-objective optimization capability: Existing methods struggle to balance the often conflicting objectives of conventional nutrition, functional nutrition, and safety indicators within a unified mathematical framework. This results in feed products that excel in one aspect but are not optimal in overall performance.

[0006] 3) Poor adaptability: Due to the large fluctuations in the composition of unconventional feed ingredients between batches, the production method based on fixed nutritional parameters and fixed process settings lacks the ability to adapt and adjust, and cannot dynamically feedback and optimize according to the actual quality of the raw materials, resulting in unstable final product quality.

[0007] Therefore, an intelligent control method for unconventional feed production lines is proposed. Summary of the Invention

[0008] The main objective of this invention is to provide an intelligent control method for unconventional feed production lines, which can connect the entire production chain from formulation to process. Through advanced modeling and optimization technology, it adaptively adjusts the proportions of various raw materials and key production process parameters, thereby stably producing unconventional feed products that achieve comprehensive optimality in terms of conventional nutrition, functional nutrition, and safety. This can effectively solve the problems in the background technology.

[0009] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for intelligent control of an unconventional feed production line includes the following steps: a) Obtain the target weight coefficients set by the user, the target weight coefficients including conventional nutrition weights, functional nutrition weights, and safety risk weights; b) Using the raw material ratio vector and the production process parameter vector as decision variables, the optimization objective is to maximize the comprehensive performance score function of the feed product, and the ranges of nutritional indicators, safety indicators, raw material ratios, and process parameters are used as constraints to construct a multi-objective optimization model; wherein, the comprehensive performance score function is a weighted sum of the conventional nutritional performance function, the functional nutritional performance function, and the safety risk function; c) Solve the multi-objective optimization model using an optimization algorithm to obtain the optimal raw material ratio vector and the optimal production process parameter vector; d) Control the unconventional feed production line to produce according to the optimal raw material ratio vector and the optimal production process parameter vector.

[0010] An intelligent control system for an unconventional feed production line includes: The modeling module is optimized and configured to construct the multi-objective optimization model. The intelligent solution module is configured to run the optimization algorithm and solve the multi-objective optimization model. The control execution module is configured to generate control commands and send them to the execution equipment on the production line based on the solved optimal raw material ratio vector and optimal production process parameter vector. The data acquisition and feedback module is configured to collect process parameter data during the production process and quality inspection data of the final product.

[0011] The system also includes a model management module configured for storing, training, and updating the indicator prediction model.

[0012] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements an intelligent control method for an unconventional feed production line.

[0013] Furthermore, the raw material proportioning vector is denoted as... Where n is the number of raw material types, For the first The percentage ratio of each raw material, and meets the following requirements. The production process parameter vector is denoted as... Where m is the number of process parameters, For the first The values ​​of a process parameter include at least one of the following: particle size, conditioning temperature, and degree of expansion.

[0014] Furthermore, the comprehensive performance scoring function The expression is: ;in, For standard nutritional weighting, This is a standard nutritional performance function; For functional nutrition weight, Functional nutritional performance function; For safety risk weighting, For the safety risk function, the weights satisfy... .

[0015] Furthermore, the conventional nutritional performance function The calculation is based on the degree of closeness between the predicted and target values ​​of crude protein, metabolizable energy, and mineral nutrient indicators in the feed. The expression is as follows: Let be the predicted value of the k-th nutritional indicator. Let k be the weight of the k-th nutritional indicator. The target value for the kth nutritional indicator; The functional nutritional performance function The calculation is based on the predicted values ​​of the activity or content of immune enhancers and antioxidant functional components in feed, expressed as follows: For the first Predicted values ​​of various functional components, For the first The weights of the functional components; The security risk function The formula is calculated based on the ratio of the predicted levels of heavy metals, toxins, and anti-nutritional factors in feed to their safety thresholds, and is expressed as follows: For the first Predicted values ​​for various risky substances, For the first The weight of each type of risky substance.

[0016] Furthermore, the predicted values ​​of the nutrition and safety indicators in the multi-objective optimization model are obtained by inputting the decision variables into a pre-trained indicator prediction model. The indicator prediction model includes at least one of the following: a conventional nutrition prediction sub-model for predicting conventional nutritional indicators, a functional nutrition prediction sub-model for predicting functional active ingredients, and a safety risk prediction sub-model for predicting hazardous substances. The indicator prediction model is a machine learning model trained using historical production data.

[0017] Furthermore, the method also includes a model self-learning step: After production is completed based on the optimal decision variables, the actual quality test data of the finished feed product is obtained; The optimal decision variables and the actual quality detection data are combined to form new sample data, which are then added to the training dataset. The indicator prediction model is periodically retrained using the updated training dataset to update the model parameters.

[0018] Furthermore, the expression for the multi-objective optimization model is: ; ; in, Let be the objective function. As constraints, These are the lower and upper limits of the k-th nutritional indicator, respectively. This represents the upper limit for the qth type of hazardous substance; The first The lower and upper limits of the percentage ratio of the raw materials. For the first The lower and upper limits of each process parameter; The optimization algorithm is an intelligent optimization algorithm, including genetic algorithm, particle swarm optimization algorithm or simulated annealing algorithm.

[0019] The present invention has the following beneficial effects: Compared to existing technologies, this solution establishes a comprehensive performance scoring function that unifies and quantifies three dimensions: conventional nutrition, functional nutrition, and safety risks. Optimization is then performed based on this function, fundamentally changing the traditional approach that focuses on only one or a few indicators. This method can automatically find the optimal balance between nutritional balance, functional enhancement, and controllable safety in a production process, thereby systematically improving the overall quality and value of the final product.

[0020] Compared with existing technologies, this solution integrates the optimization of raw material ratios and production process parameters as decision variables. This enables the system to proactively utilize the influence of process parameters on nutritional components, functional activities, and anti-nutritional factors (e.g., maximizing the retention of functional components while degrading toxins by precisely controlling the conditioning temperature), achieving deep synergy between "formulation and process" and generating a synergistic effect of "1+1>2".

[0021] Compared to existing technologies, this solution introduces a machine learning-based indicator prediction model and a closed-loop feedback and model self-learning mechanism. This allows the system to automatically adapt to significant fluctuations in the composition of unconventional raw materials from different batches and sources. Even if the initial quality of the raw materials is unstable, the system can ensure the continuous stability and reliability of the output product quality through real-time or near-real-time optimization adjustments, greatly reducing reliance on human experience.

[0022] Compared to existing technologies, this solution incorporates safety indicators such as heavy metals, toxins, and anti-nutritional factors as hard constraints or risk penalties into the optimization model, thus eliminating the possibility of unsafe formulations from the design stage. This allows for the full exploitation of the potential of unconventional feed resources while controlling their potential safety risks within an absolutely safe range, solving the core bottleneck problem that restricts their large-scale application.

[0023] Compared to existing technologies, this solution transforms the complex, multi-factor decision-making process into a computable mathematical model, which is then solved using an efficient intelligent optimization algorithm. This significantly shortens the R&D cycle for new formulas and processes, enabling rapid response to market changes and raw material supply conditions, and achieving a shift from experience-driven to data- and model-driven scientific decision-making. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating an intelligent control method for an unconventional feed production line according to the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. Example 1

[0026] See Figure 1 The flowchart shown below illustrates an intelligent control method for an unconventional feed production line according to the present invention, which includes the following steps: a) Obtain the target weight coefficient set by the user.

[0027] The target weighting coefficients include conventional nutrient weights. Functional Nutritional Weight and security risk weight And each weight satisfies .

[0028] b) Using the raw material ratio vector and the production process parameter vector as decision variables, the optimization objective is to maximize the comprehensive performance score function of the feed product, and the range of nutritional indicators, safety indicators, raw material ratio and process parameters are used as constraints to construct a multi-objective optimization model. Wherein, the raw material proportioning vector is denoted as Where n is the number of raw material types, For the first The percentage ratio of each raw material, and meets the following requirements. The production process parameter vector is denoted as... Where m is the number of process parameters, For the first The values ​​of a process parameter include at least one of the following: particle size, conditioning temperature, and degree of expansion.

[0029] The comprehensive performance scoring function is a weighted sum of the conventional nutritional performance function, the functional nutritional performance function, and the safety risk function. The expression is: ;in, For standard nutritional weighting, This is a standard nutritional performance function; For functional nutrition weight, Functional nutritional performance function; For safety risk weighting, This is a safety risk function.

[0030] Conventional nutritional performance functions The calculation is based on the degree of closeness between the predicted and target values ​​of crude protein, metabolizable energy, and mineral nutrient indicators in the feed. The expression is as follows: Let be the predicted value of the k-th nutritional indicator. Let k be the weight of the k-th nutritional indicator. The target value for the kth nutritional indicator; Functional nutritional performance function The calculation is based on the predicted values ​​of the activity or content of immune enhancers and antioxidant functional components in feed, expressed as follows: For the first Predicted values ​​of various functional components, For the first The weights of the functional components; Security risk function The formula is calculated based on the ratio of the predicted levels of heavy metals, toxins, and anti-nutritional factors in feed to their safety thresholds, and is expressed as follows: For the first Predicted values ​​for various risky substances, For the first The weight of each type of risky substance.

[0031] It should be noted that the predicted values ​​of nutritional indicators and safety indicators are obtained by inputting decision variables into a pre-trained indicator prediction model; the indicator prediction model includes at least one of the following: a conventional nutrition prediction sub-model for predicting conventional nutritional indicators, a functional nutrition prediction sub-model for predicting functional active ingredients, and a safety risk prediction sub-model for predicting hazardous substances; the indicator prediction model is a machine learning model trained using historical production data.

[0032] In one possible implementation, the following machine learning model is provided for routine nutritional indicators, functional active ingredients, and hazardous substances: ; in, These are, respectively, a conventional nutrition prediction sub-model for predicting conventional nutritional indicators, a functional nutrition prediction sub-model for predicting functional active ingredients, and a safety risk prediction sub-model for predicting hazardous substances. Using historical data on raw material ratios and process parameters as input, the system outputs conventional nutritional index data. Using historical data on raw material ratios and process parameters as input, the output is the content of functional active ingredients; Using historical data on raw material ratios and process parameters as input, the output is the content of hazardous substances.

[0033] The expression for the multi-objective optimization model is:

[0034] ; in, Let be the objective function. As constraints, These are the lower and upper limits of the k-th nutritional indicator, respectively. This represents the upper limit for the qth type of hazardous substance; The first The lower and upper limits of the percentage ratio of the raw materials. For the first The lower and upper limits of each process parameter.

[0035] c) Solve the multi-objective optimization model using optimization algorithms to obtain the optimal raw material ratio vector and the optimal production process parameter vector.

[0036] The optimization algorithm is an intelligent optimization algorithm, including genetic algorithm, particle swarm optimization algorithm or simulated annealing algorithm.

[0037] d) Control the unconventional feed production line to produce based on the optimal raw material ratio vector and the optimal production process parameter vector.

[0038] e) After production is completed based on the optimal decision variables, obtain the actual quality test data of the finished feed product; f) Combine the optimal decision variables with the actual quality inspection data to form new sample data and add it to the training dataset; g) Use the updated training dataset to periodically retrain the metric prediction model to update the model parameters.

[0039] Specifically, the control flow of the control method in this application is as follows: Initialization: Given an initial formula and process, set the target weights and all constraints.

[0040] Optimization solution: Over time t, the objective function is solved using an optimization engine (such as a genetic algorithm or a particle swarm optimization algorithm); Execution and Feedback: The obtained optimal solution is sent to the production line for execution.

[0041] After production is completed, obtain the actual quality indicators of the finished feed product.

[0042] Data storage: Add the new optimal decision variables and actual quality inspection data to the historical database.

[0043] Model update: Set a fixed training period and retrain at the set period to achieve continuous improvement of the system.

[0044] Example 2: This invention also provides an intelligent control system for an unconventional feed production line, comprising: Optimize the modeling module and configure it for building multi-objective optimization models; The intelligent solver module is configured to run optimization algorithms and solve multi-objective optimization models. The control execution module is configured to generate control commands and send them to the execution equipment on the production line based on the solved optimal raw material ratio vector and optimal production process parameter vector. The data acquisition and feedback module is configured to collect process parameter data during the production process and quality inspection data of the final product.

[0045] It also includes a model management module, which is configured for storing, training, and updating metric prediction models.

[0046] Example 3: The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can implement the steps of the intelligent control method for unconventional feed production lines in Embodiment 1 above.

[0047] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent control of an unconventional feed production line, characterized in that, Includes the following steps: a) Obtain the target weight coefficients set by the user, the target weight coefficients including conventional nutrition weights, functional nutrition weights, and safety risk weights; b) Using the raw material ratio vector and the production process parameter vector as decision variables, the optimization objective is to maximize the comprehensive performance score function of the feed product, and the ranges of nutritional indicators, safety indicators, raw material ratios, and process parameters are used as constraints to construct a multi-objective optimization model; wherein, the comprehensive performance score function is a weighted sum of the conventional nutritional performance function, the functional nutritional performance function, and the safety risk function; c) Solve the multi-objective optimization model using an optimization algorithm to obtain the optimal raw material ratio vector and the optimal production process parameter vector; d) Control the unconventional feed production line to produce according to the optimal raw material ratio vector and the optimal production process parameter vector; The comprehensive performance scoring function The expression is: ; in, For standard nutritional weighting, This is a standard nutritional performance function; For functional nutrition weight, Functional nutritional performance function; For safety risk weighting, For the safety risk function, the weights satisfy... .

2. The intelligent control method for an unconventional feed production line according to claim 1, characterized in that, The raw material proportioning vector is denoted as Where n is the number of raw material types, Let be the percentage of the i-th raw material, and satisfy . The production process parameter vector is denoted as... Where m is the number of process parameters, The value of the j-th process parameter is given, and the process parameter includes at least one of the following: particle size, conditioning temperature, and degree of expansion.

3. The intelligent control method for an unconventional feed production line according to claim 1, characterized in that, The conventional nutritional performance function The calculation is based on the degree of closeness between the predicted and target values ​​of crude protein, metabolizable energy, and mineral nutrient indicators in the feed. The expression is as follows: Let be the predicted value of the k-th nutritional indicator. Let k be the weight of the k-th nutritional indicator. The target value for the kth nutritional indicator; The functional nutritional performance function The calculation is based on the predicted values ​​of the activity or content of immune enhancers and antioxidant functional components in feed, expressed as follows: This is the predicted value for the l-th functional component. The weight of the l-th functional component; The security risk function The formula is calculated based on the ratio of the predicted levels of heavy metals, toxins, and anti-nutritional factors in feed to their safety thresholds, and is expressed as follows: Let be the predicted value for the q-th hazardous substance. Let q be the weight of the q-th type of risk substance.

4. The intelligent control method for an unconventional feed production line according to claim 1, characterized in that, The predicted values ​​of the nutrition and safety indicators in the multi-objective optimization model are obtained by inputting the decision variables into a pre-trained indicator prediction model. The indicator prediction model includes at least one of the following: a conventional nutrition prediction sub-model for predicting conventional nutritional indicators, a functional nutrition prediction sub-model for predicting functional active ingredients, and a safety risk prediction sub-model for predicting hazardous substances. The indicator prediction model is a machine learning model trained using historical production data.

5. The intelligent control method for an unconventional feed production line according to claim 4, characterized in that, The method also includes a model self-learning step: After production is completed based on the optimal decision variables, the actual quality test data of the finished feed product is obtained; The optimal decision variables and the actual quality detection data are combined to form new sample data, which are then added to the training dataset. The indicator prediction model is periodically retrained using the updated training dataset to update the model parameters.

6. The intelligent control method for an unconventional feed production line according to claim 1, characterized in that, The expression for the multi-objective optimization model is: ; ； in, Let be the objective function. As constraints, These are the lower and upper limits of the k-th nutritional indicator, respectively. This represents the upper limit for the qth type of hazardous substance; These are the lower and upper limits of the percentage ratio of the i-th raw material, respectively. Let be the lower and upper limits of the j-th process parameter; The optimization algorithm is an intelligent optimization algorithm, including genetic algorithm, particle swarm optimization algorithm or simulated annealing algorithm.

7. An intelligent control system for an unconventional feed production line, characterized in that, The system for implementing the method as described in any one of claims 1 to 6, the system comprising: The modeling module is optimized and configured to construct the multi-objective optimization model. The intelligent solution module is configured to run the optimization algorithm and solve the multi-objective optimization model. The control execution module is configured to generate control commands and send them to the execution equipment on the production line based on the solved optimal raw material ratio vector and optimal production process parameter vector. The data acquisition and feedback module is configured to collect process parameter data during the production process and quality inspection data of the final product.

8. The intelligent control system for an unconventional feed production line according to claim 7, characterized in that, The system also includes a model management module configured for storing, training, and updating the indicator prediction model.

9. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the intelligent control method for an unconventional feed production line as described in any one of claims 1 to 6.

Images