A large model and cuckoo algorithm-based grinding system control method

Abstract

The application relates to the technical field of industrial automation, and discloses a grinding system control method based on a large model and a cuckoo algorithm. The method comprises the following steps: setting parameters of the cuckoo algorithm, and generating candidate solutions based on grinding system control parameters; scoring the candidate solutions by the large model, and recording the fitness of all candidate solutions; adding global disturbance to the candidate solutions in combination with a Levy flight mechanism, recalculating the fitness, and performing an elimination mechanism based on the fitness score until a preset number of times is reached, recording the candidate solution corresponding to the optimal fitness score in the two fitness score processes; applying the obtained candidate solution to the grinding system, collecting actual feedback, calculating a deviation value based on the large model, updating the candidate solution based on the deviation value, returning to step 2 to score the fitness of all candidate solutions again until the grinding task is completed. The method solves the problem that the algorithm model used by the existing grinding system control method cannot meet the industrial control requirements.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of industrial automation, and in particular to a control method for a grinding system using a large model and a cuckoo algorithm. BACKGROUND

[0002] In actual production, the operation of the grinding system involves multiple key process parameters such as ore feed amount, ball addition amount, water feed amount, and ore discharge concentration. These parameters have significant nonlinear coupling relationships, and are subject to various uncertain factors such as ore properties, equipment status, and environmental disturbances. The system exhibits strong dynamics and complexity. For a long time, traditional grinding control systems have mainly relied on manual experience or rule-based control methods, which have limitations in responding to complex conditions. Even though some ore processing companies have introduced automatic monitoring and process control systems to achieve real-time data collection and preliminary adjustment, the existing systems still lack intelligence and refinement in management due to insufficient core modeling and optimization capabilities. Specifically, the current grinding system optimization and control mainly have the following problems:

[0003] Traditional mechanism models cannot fully describe the multivariable, strong coupling, and nonlinear dynamic characteristics of the grinding process, and have high parameter uncertainty, which often leads to model mismatch. The quality of historical data and the representativeness of samples are highly dependent, and the generalization ability for extreme or new conditions is limited, resulting in limited modeling capability.

[0004] Existing optimization methods are mostly single-objective or simple multi-objective optimization. Genetic algorithms and particle swarm optimization commonly used in high-dimensional constraint spaces are prone to local optimization, and the optimization ability and convergence speed are limited, making it difficult to meet the global optimization needs of complex industrial scenarios, resulting in limitations of optimization algorithms.

[0005] The optimization and control process is mostly open-loop or semi-closed-loop, and cannot fully utilize actual operation feedback for parameter adaptive adjustment. The system is difficult to achieve long-term efficient and dynamic stable operation, resulting in a lack of feedback and adaptive mechanisms.

[0006] Existing systems lack effective adaptive compensation and fault-tolerant mechanisms for sensor failures, model distortions, and extreme disturbances, which can cause system efficiency to drop sharply or even become unstable, affecting production continuity and process safety, and have weak robustness for abnormal and extreme conditions.

[0007] Existing methods mostly separate parameter optimization and process control, and lack closed-loop coordination between model evaluation, parameter optimization, and actual execution, making it difficult to effectively implement optimization results, restricting the intelligent upgrade of production processes, and lacking integration of optimization and control.

[0008] Therefore, there is an urgent need for a grinding system control method to meet the above industrial control requirements. SUMMARY

[0009] The application provides a large model and cuckoo algorithm-based grinding system control method to solve the problem that the algorithm model used in the existing grinding system control method cannot meet the industrial control requirements.

[0010] To achieve the above-mentioned purpose, the application is implemented through the following technical solutions:

[0011] In the first aspect, the application provides a large model and cuckoo algorithm-based grinding system control method, including the following steps:

[0012] Step 1: Set the parameters of the cuckoo algorithm, and generate candidate solutions based on grinding system control parameters;

[0013] The generated candidate solutions are a set of grinding system control parameter combinations;

[0014] Step 2: Score all candidate solutions through a large model, and record the fitness of all candidate solutions;

[0015] In step 2, the large model used is an industrial-level large model, which comprehensively models the multi-variable, nonlinear, and strongly coupled process of the grinding system, realizes accurate evaluation of parameter combinations under different working conditions, and provides a scientific and reliable objective function basis for optimization.

[0016] Step 3: Add global disturbance to all candidate solutions based on the Levy flight mechanism, re-execute step 2, and perform the elimination mechanism based on the fitness score, repeat step 2 until the preset number of times is reached, and record the optimal fitness score corresponding to the candidate solution in the two fitness score processes;

[0017] Step 4: Apply the candidate solution obtained in step 3 to the grinding system, collect actual feedback, calculate the deviation value based on the large model, and update the candidate solution based on the deviation value, return to step 2 to re-score all candidate solutions until the grinding task is completed;

[0018] The collection of actual feedback, the calculation of deviation value based on the large model, and the update of the candidate solution based on the deviation value include: collecting the grinding fineness and grinding efficiency under the actual working condition of the grinding system, scoring based on the grinding fineness and grinding efficiency combined with the predetermined rules, calculating the deviation value based on the actual score and the fitness score corresponding to the candidate solution, and updating the candidate solution according to the deviation value combined with the preset feedback calibration weight and uniform disturbance term;

[0019] The actual score based on the combination of the grinding fineness and the grinding efficiency according to the predetermined rule includes: if the collected grinding fineness reaches the predetermined fineness index and the grinding efficiency is improved, the actual score output is +100 points; if the collected grinding fineness reaches the predetermined fineness index and the grinding efficiency is unchanged, the actual score output is +50 points; if the collected grinding fineness reaches the predetermined fineness index and the grinding efficiency is reduced, the actual score output is 0 point; if the collected grinding fineness does not reach the predetermined fineness index and the grinding efficiency is improved, the actual score output is -50 points; and if the collected grinding fineness does not reach the predetermined fineness index and the grinding efficiency is reduced, the actual score output is -100 points.

[0020] The updating of the candidate solution according to the deviation value, the preset feedback calibration weight and the uniform disturbance term is represented by the following formula:

[0021] ;

[0022] Wherein, represents the updated candidate solution; represents the current optimal candidate solution; is the feedback calibration weight, and is a preset value, specifically 0.2; is the difference between the feedback and the model score, is a Gaussian or uniform disturbance term.

[0023] Through the above operation, a closed-loop optimization mechanism combining virtual model evaluation and actual process feedback is established, and through dynamic calibration of actual production feedback, continuous self-learning and self-adaptive adjustment of the parameter optimization scheme are realized, and the adaptability of the system to real disturbances such as working condition changes and equipment aging is significantly enhanced.

[0024] Further, the parameters of the cuckoo algorithm include: population size, maximum iteration number, basic elimination probability, capacity of historical memory pool, local perturbation coefficient, step initial value, step adaptive factor, elimination probability adjustment coefficient and Levy distribution parameter.

[0025] Through the above operation, combined with the global search ability, adaptive step, dynamic elimination and local search mechanism of the cuckoo optimization algorithm, the local optimum can be efficiently jumped out in a high-dimensional and multi-constrained parameter space, and the optimal parameter combination can be quickly obtained, thereby improving the optimization speed and result stability.

[0026] Further, the grinding system control parameters include the ore supply amount, the ball addition amount, the water supply amount and the ore discharge concentration.

[0027] The candidate solution is generated based on the grinding system control parameters, which includes: randomly generating a predetermined number of candidate solutions within the physical boundary of the grinding system control parameters combined with the historical memory pool.

[0028] Further, the combination of Levy flight mechanism adds global disturbance to all candidate solutions, which is expressed by the following formula:

[0029] ;

[0030] wherein, is a new solution of the i-th candidate solution in the j-th generation, is the i-th candidate solution in the j-th generation; is the j-th generation step size, is a Levy distribution random variable, satisfying . Further, the large model adopts an adaptive compensation adjustment mechanism to adjust the step size and improve the convergence efficiency. The adaptive compensation adjustment mechanism is expressed by the following formula:

[0031] ;

[0032] wherein,

[0033] ;

[0034] wherein, is a step size initial value, is a difference between optimal scores of adjacent two generations, is a historical maximum score, is a step size adaptive factor.

[0035] Further, the elimination mechanism based on the fitness score includes: comparing the fitness score of the candidate solution added with global disturbance and the fitness score of the corresponding candidate solution, if the fitness score of the candidate solution added with global disturbance is higher than the fitness score of the corresponding candidate solution, then the corresponding candidate solution is eliminated according to a predetermined elimination probability, otherwise the candidate solution added with global disturbance is removed.

[0036] Further, in the process of the elimination mechanism, the predetermined elimination probability is adjusted based on the fitness fluctuation, which is expressed by the following formula:

[0037] ;

[0038] wherein, is a predetermined elimination probability, is an elimination probability adjustment coefficient, is a fitness score standard deviation of the j-th generation, is an optimal score of the j-th generation.

[0039] ​​​​​Further, in step 3, the perturbation processing is further included to generate a local solution by perturbing the candidate solution corresponding to the optimal fitness score, and if the fitness score of the local solution is higher than the fitness score of the candidate solution, the local solution is used to replace the candidate solution.

[0040] The perturbation processing is represented by the following formula:

[0041] ;

[0042] wherein, represents the local solution, represents the candidate solution corresponding to the optimal fitness score of the nth generation, is a normal distribution perturbation with a mean of 0 and a variance of .

[0043] Further, in the process of executing steps 3 and 4, if the control parameters of the grinding system corresponding to any candidate solution exceed the physical boundary, the projection method is used to pull back to the physical boundary.

[0044] The projection method is represented by the following formula:

[0045] ;

[0046] wherein, is the value of the mth parameter of the nth solution in the nth generation, and are the upper and lower physical boundaries of the parameter.

[0047] Through the above operation, the robustness of the grinding system is improved by out-of-bound projection, which can effectively deal with model errors, sensor failures, extreme working conditions and other abnormal situations, and ensure the reliability and safety of the grinding system operation.

[0048] Beneficial effects:

[0049] The grinding system control method based on the large model and the cuckoo algorithm provided by the application realizes the whole-process integrated closed-loop management of large model evaluation, intelligent optimization and process parameter execution, and the optimization result can directly guide the actual production operation, thereby improving the intelligentization, automation and green level of the grinding system.

[0050] Without relying on specific equipment or hardware environment, the parameter setting is flexible, suitable for different types and different scales of grinding systems, and has good engineering implementability and industry promotion prospect. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 ​​​​A flowchart of a large model and cuckoo algorithm-based grinding system control method is provided in an embodiment of the present application. DETAILED DESCRIPTION

[0052] The technical solutions of the present application will be described below in a clear and complete manner. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0053] Unless otherwise defined, the technical terms or scientific terms used in the present application shall be understood as the usual meanings understood by those skilled in the art to which the present application belongs. The terms "first", "second", and similar terms used in the present application do not represent any order, quantity, or importance, but are only used to distinguish different components. Similarly, "one" or "a" and similar terms do not represent a quantity limitation, but represent the existence of at least one. The terms "connected" or "connected" and similar terms are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up", "down", "left", "right", and the like are only used to represent relative positional relationships, which change accordingly when the absolute position of the described object changes.

[0054] Please refer to Figure 1 The present application provides a large model and cuckoo algorithm-based grinding system control method, which comprises the following steps:

[0055] Step 1: Set the parameters of the cuckoo algorithm, and generate candidate solutions based on the grinding system control parameters;

[0056] In a high-dimensional, multi-constrained, and dynamically changing grinding system, existing optimization algorithms are prone to local optimization, have low optimization efficiency, and are difficult to quickly obtain a global optimal parameter combination. The present application uses the cuckoo optimization algorithm, combined with adaptive step size, dynamic elimination, local search, and other mechanisms, to significantly improve the global ability and convergence speed of multi-parameter joint optimization.

[0057] The parameters of the cuckoo algorithm include: population size, maximum iteration number, basic elimination probability, capacity of the history memory pool, local perturbation coefficient, step size initial value, step size adaptive factor, elimination probability adjustment coefficient, and Levy distribution parameter.

[0058] In the present embodiment, the population size is set to 20, the maximum iteration number is set to 50, the basic elimination probability is set to 0.2, the capacity of the history memory pool is set to 5, the local perturbation coefficient is set to 0.01, the step size initial value is set to 0.5, the step size adaptive factor is set to 2, the elimination probability adjustment coefficient is set to 0.5, and the Levy distribution parameter is set to 1.5.

[0059] The grinding system control parameters include the feed quantity, the ball loading quantity, the water feeding quantity, and the discharge concentration;

[0060] The generating the candidate solutions based on the grinding system control parameters includes: randomly generating a predetermined number of candidate solutions within the preset physical boundary of the grinding system control parameters in combination with a historical memory pool constructed by historical grinding system control parameters;

[0061] The generated candidate solutions are a set of grinding system control parameter combinations, which can be defined as:

[0062] ;

[0063] Wherein, represents one of the generated candidate solutions; represents the feed quantity in the candidate solution; represents the ball loading quantity in the candidate solution; represents the water feeding quantity in the candidate solution; represents the discharge concentration in the candidate solution.

[0064] In the embodiment, the physical boundary set for each control parameter is as follows:

[0065] The feed quantity is greater than 80 tons / hour and less than 12080 tons / hour; the ball loading quantity is greater than 2 tons / hour and less than 5 tons / hour; the water feeding quantity is greater than 30 cubic meters / hour and less than 50 cubic meters / hour; and the discharge concentration is greater than 65% and less than 75%.

[0066] Step 2: Scoring the fitness of all candidate solutions by the large model, and recording the fitness corresponding to all candidate solutions;

[0067] The existing grinding system running process is affected by multiple process parameters, and there is a complex nonlinear coupling relationship between variables. Traditional mechanism models and shallow data-driven models are difficult to accurately represent the dynamic behavior of the system, resulting in limited parameter optimization and adjustment effect. The present application introduces an industrial-level large model to realize high-precision evaluation and modeling of the efficiency of the grinding system under multiple parameters and complex conditions, providing a scientific and reliable objective function basis for intelligent optimization;

[0068] Wherein, the large model adopts an adaptive compensation adjustment mechanism to adjust the step length and improve the convergence efficiency;

[0069] The adaptive compensation adjustment mechanism is represented by the following formula:

[0070] ;

[0071] Wherein, is the initial value of the step length, The difference between the optimal fitness scores of two adjacent generations. This is the highest rating in history. This is the step size adaptive factor.

[0072] Step 3: Combine the Levy flight mechanism to add global perturbation to all candidate solutions, re-execute step 2, and perform the elimination mechanism based on fitness score. Repeat step 2 until the preset number of times is reached, and record the candidate solution corresponding to the best fitness score in the two fitness scoring processes.

[0073] Incorporating the Levy flight mechanism, a global perturbation is added to all candidate solutions, expressed by the following formula:

[0074] ;

[0075] in, For the first The generation A new solution for each candidate solution. For the first The generation There are 10 candidate solutions; For the first Motorcycle driver, Let be a Levy-distributed random variable, satisfying , for Distribution parameters.

[0076] Specifically, the fitness scores of the candidate solutions with global perturbation are compared with those of the corresponding candidate solutions. If the fitness score of the candidate solution with global perturbation is higher than that of the corresponding candidate solution, the corresponding candidate solution is eliminated according to a predetermined elimination probability; otherwise, the candidate solution with global perturbation is removed.

[0077] During the elimination mechanism, the predetermined elimination probability is adjusted based on fitness fluctuations, as expressed by the following formula:

[0078] ;

[0079] in, To determine the probability of elimination, This is the elimination probability adjustment coefficient. For the first Standard deviation of generation fitness score For the first The optimal score is determined.

[0080] After selecting the candidate solution corresponding to the optimal fitness score, the candidate solution corresponding to the optimal fitness score is further perturbed to generate a local solution. If the fitness score of the local solution is higher than that of the candidate solution, the local solution is directly used to replace the candidate solution.

[0081] The perturbation processing is represented by the following formula:

[0082]

[0083] wherein, represents a local solution, represents the candidate solution corresponding to the optimal fitness of the th generation, is a normal distribution perturbation with a mean of 0 and a variance of .

[0084] Step 4: Apply the candidate solution obtained in step 3 to the grinding system, collect the actual feedback, calculate the deviation value based on the large model, and update the candidate solution based on the deviation value, return to step 2 to re-score the fitness of all candidate solutions until the grinding task is completed.

[0085] During the execution of steps 3 and 4, if the control parameters of any candidate solution of the grinding system exceed the physical boundary, the projection method is used to pull back to the physical boundary;

[0086] The projection method is represented by the following formula:

[0087]

[0088] wherein, is the value of the th parameter of the th solution in the th generation, and are the upper and lower physical boundaries of the parameter.

[0089] Collecting actual feedback and calculating deviation value based on the large model to update the candidate solution includes: collecting the grinding fineness and grinding efficiency under the actual working condition of the grinding system, scoring the actual score based on the grinding fineness and grinding efficiency combined with the predetermined rule, calculating the deviation value based on the actual score and the fitness score corresponding to the candidate solution, and updating the candidate solution according to the deviation value combined with the preset feedback calibration weight and uniform perturbation term;

[0090] Scoring the actual score based on the grinding fineness and grinding efficiency combined with the predetermined rule includes: if the collected grinding fineness reaches the predetermined fineness index and the grinding efficiency improves, the actual score output is +100 points; if the collected grinding fineness reaches the predetermined fineness index and the grinding efficiency remains unchanged, the actual score output is +50 points; if the collected grinding fineness reaches the predetermined fineness index and the grinding efficiency decreases, the actual score output is 0 points; if the collected grinding fineness does not reach the predetermined fineness index and the grinding efficiency improves, the actual score output is -50 points; if the collected grinding fineness does not reach the predetermined fineness index and the grinding efficiency decreases, the actual score output is -100 points.​​

[0091] The candidate solution is updated according to the deviation value, a preset feedback calibration weight and a uniform disturbance term, and is represented by the following formula:

[0092]

[0093] wherein, represents the updated candidate solution; represents the current optimal candidate solution; is a feedback calibration weight, is a feedback and model score difference, is a Gaussian or uniform disturbance term, in the embodiment, is set to 0.2, obeys a normal distribution with a mean of 0 and a standard deviation of 0.01, that is .

[0094] At present, the optimization and control process of the grinding system is mostly open loop or semi-closed loop, and lacks dynamic self-adaptive ability to feedback of actual operation, and lacks effective compensation and fault-tolerant mechanism for abnormal scores and extreme working conditions. The present application establishes a closed-loop self-learning optimization framework combining virtual and real, uses actual production feedback to continuously calibrate and optimize parameters, realizes robust regulation of abnormality and disturbance, and guarantees long-term stable and efficient operation of the system.

[0095] Finally, in the embodiment, the above parameter settings are adopted, and comparative experiments are carried out in combination with two typical optimization methods, i.e., a large model and a particle swarm optimization algorithm (PSO) and a large model and a cuckoo search algorithm (CSA). The final fitness score, actual grinding efficiency improvement rate and fitness fluctuation amplitude are selected as evaluation indexes, and the specific results are shown in Table 1.

[0096] Table 1: Comparative experiment result summary

[0097]

[0098] ​From the comparative experimental results in Table 1, it can be seen that the method of the present application is superior to the large model combined with the particle swarm algorithm and the large model combined with the traditional cuckoo algorithm in the three key indicators of the final fitness score, the actual grinding efficiency improvement rate and the fitness fluctuation amplitude. Specifically, by introducing the adaptive step adjustment, the dynamic elimination mechanism and the closed-loop calibration based on actual feedback, the method of the present application effectively improves the precision and convergence speed of parameter optimization, and the final fitness score reaches 95 points, which is significantly higher than 83 points and 91 points of the comparative methods. At the same time, in terms of the actual grinding efficiency improvement, the method achieves an improvement of 7%, which is significantly higher than the traditional method, verifying the practical application value of the optimization result. In addition, the fitness fluctuation amplitude of the method of the present application is the lowest, only 5%, which shows strong robustness and tolerance to disturbance, and helps the grinding system to maintain stable operation under complex and variable working conditions. In summary, the experimental results fully prove the significant advantages of the method of the present application in improving the optimization effect and system stability.

[0099] The foregoing describes in detail the preferred embodiments of the present application. It should be understood that those skilled in the art can make many modifications and changes without creative labor based on the concept of the present application. Therefore, any technical solutions obtained by logical analysis, reasoning or limited experiments based on the prior art according to the concept of the present application shall be within the protection scope determined by the claims.

Claims

1. A large model and cuckoo algorithm-based grinding system control method, characterized in that, The method comprises the following steps: Step 1: setting parameters of the cuckoo algorithm, generating candidate solutions based on grinding system control parameters; Step 2: scoring all candidate solutions by a large model, and recording the fitness of all candidate solutions; Step 3: adding global disturbance to all candidate solutions by combining the Levy flight mechanism, re-executing step 2, and eliminating the candidate solutions based on the fitness score, repeating step 2 until a preset number of times is reached, and recording the candidate solution corresponding to the optimal fitness score in the fitness score process; Step 4: applying the candidate solution obtained in step 3 to the grinding system, collecting actual feedback, calculating a deviation value based on the large model, and updating the candidate solution based on the deviation value, returning to step 2 to score all candidate solutions until the grinding task is completed; The collecting actual feedback, calculating a deviation value based on the large model, and updating the candidate solution based on the deviation value comprises: collecting the grinding fineness and grinding efficiency under the actual working condition of the grinding system, scoring the actual score based on the grinding fineness and grinding efficiency combined with the predetermined rule, calculating the deviation value based on the actual score and the fitness score corresponding to the candidate solution, and updating the candidate solution according to the deviation value combined with the preset feedback calibration weight and uniform disturbance term; The actual score based on the grinding fineness and grinding efficiency combined with the predetermined rule comprises: if the collected grinding fineness reaches the predetermined fineness index and the grinding efficiency is improved, the actual score output is +100 points; if the collected grinding fineness reaches the predetermined fineness index and the grinding efficiency remains unchanged, the actual score output is +50 points; if the collected grinding fineness reaches the predetermined fineness index and the grinding efficiency decreases, the actual score output is 0 point; if the collected grinding fineness does not reach the predetermined fineness index and the grinding efficiency improves, the actual score output is -50 points; and if the collected grinding fineness does not reach the predetermined fineness index and the grinding efficiency decreases, the actual score output is -100 points; The updating of the candidate solution according to the deviation value combined with the preset feedback calibration weight and the uniform disturbance term is represented by the following formula: ； wherein, denotes the updated candidate solution; denotes the current best candidate solution; is the feedback calibration weight, is the feedback and model score difference, is a Gaussian or uniform perturbation term.

2. The large model and cuckoo algorithm-based grinding system control method according to claim 1, characterized in that, The parameters of the cuckoo algorithm comprise: population size, maximum iteration number, basic elimination probability, capacity of historical memory pool, local perturbation coefficient, step initial value, step adaptive factor, elimination probability adjustment coefficient, and Levy distribution parameter.

3. The large model and cuckoo algorithm-based grinding system control method according to claim 1, characterized in that, The grinding system control parameters comprise feed quantity, ball loading quantity, water quantity, and discharge concentration; The generating of the candidate solution based on the grinding system control parameters comprises: randomly generating a predetermined number of candidate solutions within the physical boundary of the grinding system control parameters preset by the historical grinding system control parameters combined with the historical memory pool constructed by the historical grinding system control parameters.

4. The large model and cuckoo algorithm-based grinding system control method according to claim 1, characterized in that, The adding of global disturbance to all candidate solutions by combining the Levy flight mechanism is represented by the following formula: ； wherein, is the new solution of the th iteration of the th candidate solution, is the new solution of the th iteration of the th candidate solution; is the new solution of the th iteration of the is a Levy distributed random variable satisfying .

5. The large model and cuckoo algorithm based grinding system control method according to claim 4, characterized in that, The large model adopts an adaptive compensation adjustment mechanism to adjust the step length and improve the convergence efficiency; The adaptive compensation adjustment mechanism is represented by the following formula: ； wherein, is a step initial value, is a difference between the optimal scores of two adjacent generations, is a historical maximum score, is a step adaptation factor.

6. The large model and cuckoo algorithm-based grinding system control method according to claim 1, characterized in that, The elimination mechanism based on the fitness score comprises: comparing the fitness score of the candidate solution added with the global disturbance with the fitness score of the corresponding candidate solution, if the fitness score of the candidate solution added with the global disturbance is higher than the fitness score of the corresponding candidate solution, eliminating the corresponding candidate solution according to a predetermined elimination probability, and otherwise, removing the candidate solution added with the global disturbance.

7. The large model and cuckoo algorithm based grinding system control method according to claim 6, characterized in that, In the process of the elimination mechanism, the predetermined elimination probability is adjusted based on the fitness fluctuation, and is represented by the following formula: ； wherein, is a predetermined elimination probability, is an elimination probability adjustment coefficient, is the fitness score of the i-th individual in the j-th generation, is the standard deviation of the fitness scores of the individuals in the j-th generation, is the optimal fitness score of the j-th generation, and is the optimal fitness score of the j-th generation.

8. The large model and cuckoo algorithm based grinding system control method according to claim 1, wherein, In step 3, the candidate solution corresponding to the optimal fitness score is also subjected to a perturbation process to generate a local solution, and if the fitness score of the local solution is higher than the fitness score of the candidate solution, the candidate solution is replaced by the local solution; The perturbation process is represented by the following formula: ； wherein, denotes a local solution, denotes the th best candidate solution, is a normal distribution perturbation with mean 0 and variance .

9. The large model and cuckoo algorithm based grinding system control method according to claim 1, wherein, In the process of executing steps 3 and 4, if the control parameter of the grinding system corresponding to any candidate solution exceeds the physical boundary, a projection method is used to pull back to the physical boundary; The projection method is represented by the following formula: ； wherein is the th dimensional parameter at the th iteration, is the physical lower bound for this parameter.

Images