A kind of portunid multi-step binding timing control method, system and computer readable medium

Abstract

The application discloses a kind of portunid multi-step bundling timing control method, system and computer readable medium, wherein the method comprises: S1, obtains the influence parameter of single bundling cycle time and bundling success rate;S2, respectively establishes the second-order response surface regression model of single bundling cycle time and bundling success rate;S3, with the bundling success rate predicted by the second-order response surface regression model of S2 is constructed constraint, with the predicted single bundling cycle time as optimization goal, according to the coding of S1, with the combination of all influence parameters obtained by S1 as particle, construct particle swarm optimization model and execute global optimization;S4, according to the optimal coding variable combination obtained by S3 global optimization, it is converted into actual control parameter.The application realizes the minimization of single bundling cycle time under the condition that the bundling success rate is not lower than the set threshold, and considers the optimization accuracy and test cost.

Description

Technical Field

[0001] This invention relates to the field of crab packaging, and more particularly to a method, system, and computer-readable medium for controlling the timing of multi-step binding of swimming crabs. Background Technology

[0002] Currently, the binding of swimming crabs is still mainly done manually, which generally suffers from prominent problems such as high labor intensity, low work efficiency, and poor binding consistency. Especially during the peak fishing season, there is a severe labor shortage, and manual binding has become a key bottleneck restricting the industrialization and distribution of swimming crabs. To replace manual labor, existing equipment has achieved basic mechanized actions such as automatic conveying and cutting of rubber tubes, transfer of grippers, and expansion binding, and uses programmable logic controllers to detect and compensate for the position status of each actuator. However, the control strategy of existing equipment is still limited to judging the position status of each actuator discretely and supplementing operation, without starting from the complete process flow from binding and feeding to unloading, and without systematic multi-step timing arrangement, and lacks state transition logic and timing connection design based on sensors and timers between actions.

[0003] Furthermore, the key motion and timing parameters in existing equipment are all fixed preset values ​​embedded in the control program. The setting of these parameters relies entirely on the operator's experience and trial and error, which has the following shortcomings:

[0004] (1) There is an inherent contradiction between binding efficiency and binding stability. Although simply increasing the movement speed of each actuator can shorten the cycle time of a single operation, excessive speed will reduce equipment reliability and binding success rate, and it is difficult to balance the two.

[0005] (2) There are complex interactive coupling effects between multiple motion parameters and time parameters, and it is difficult for operators to model and optimize the quantitative relationship in this high-dimensional multi-parameter space based on experience.

[0006] (3) When the batch of rubber hoses is replaced or the working conditions such as ambient temperature and humidity change, the fixed preset parameters lack the ability to adjust quickly and cannot achieve rapid response and dynamic matching of parameters. Summary of the Invention

[0007] Purpose of the invention: To address the shortcomings of existing control methods for binding equipment of swimming crabs, which only perform discrete positioning judgment and supplementary rotation of each actuator, lack complete multi-step timing arrangement, and use fixed preset values ​​for key process parameters, which cannot achieve quantitative modeling and global synergistic optimization of multi-parameter interaction effects, this invention provides a multi-step binding timing control method, system, and computer-readable medium for swimming crabs. It minimizes the single binding cycle time while ensuring that the binding success rate is not lower than a set threshold, thus balancing optimization accuracy and experimental cost.

[0008] Technical Solution: This invention provides a multi-step binding timing control method for swimming crabs, applied to a binding platform consisting of a rubber tube conveying mechanism, a cutting mechanism, a clamping mechanism, a support claw mechanism, and a material ejection mechanism arranged sequentially along a straight line in multiple steps, including:

[0009] S1. Obtain the parameters affecting the single binding cycle time and binding success rate, including the cutting parameters of the cutting mechanism, the lifting parameters of the clamping mechanism, the connection sequence of the lifting of the clamping mechanism and the translation of the support claw mechanism, and the expansion parameters of the support claw mechanism, and encode them according to the corresponding value range as low, medium and high levels;

[0010] S2. Based on the coding of S1, a prototype test was carried out using a four-factor central composite design method, thereby establishing second-order response surface regression models for single-bundle cycle time and bundle success rate.

[0011] S3. The binding success rate predicted by the second-order response surface regression model of S2 is used to construct constraints. The predicted single binding cycle time is used as the optimization objective. Based on the encoding of S1, the combination of all the influence parameters obtained by S1 is used as particles to construct a particle swarm optimization model and perform global optimization.

[0012] S4. Based on the global optimization in S3, obtain the optimal combination of encoded variables and convert it into actual control parameters.

[0013] Specifically, in S1, the value range of each influencing parameter is determined based on the rated parameters of the drive unit and the safety boundary obtained from the pre-test. The corresponding low, medium and high level values ​​are obtained from the boundary values ​​and center values ​​of the corresponding value range, and then encoded accordingly.

[0014] More specifically, in S2, a prototype test is conducted using a four-factor central composite design method based on the coding of S1, comprising three parts:

[0015] (1) 2 4 The full factorial design includes 16 points, each influencing parameter being located at the boundary value of its corresponding range.

[0016] (2) Each influencing parameter is located at a coding level of ±α, where α is within the coding value range, and the other influencing parameters are kept at a coding level of 0, for a total of 8 points;

[0017] (3) Repeat the experiment several times with all influencing parameters at the coding level of 0;

[0018] This results in a corresponding number of prototype tests.

[0019] Specifically, in S2, for the second-order response surface regression models for the single-bundle cycle time and the success rate of the bundle, least squares regression is used and insignificant terms are removed, thereby obtaining the final second-order response surface regression models for the single-bundle cycle time and the success rate of the bundle.

[0020] Specifically, in S3, the binding success rate predicted by the second-order response surface regression model in S2 is used to construct constraints, and the predicted single binding cycle time is used as the optimization objective, as follows:

[0021] ;

[0022] ;

[0023] Where X=[x1, x 2, x 3, x4],x1, x 2, x 3, x4 represents the encoded values ​​of the influence parameters of the single-bundle cycle time and binding success rate obtained in S1; T(X) and S(X) represent the single-bundle cycle time and binding success rate predicted by the second-order response surface regression model of the single-bundle cycle time and binding success rate established in S2, respectively; S min This represents the lowest possible success rate for bundling.

[0024] Specifically, in S3, the influencing parameters obtained in S1 are used as variables to construct a particle swarm optimization model and perform global optimization, as follows:

[0025] Each influencing parameter is randomly initialized within its encoding range, thereby obtaining several particles;

[0026] Set the optimal velocity range and maximum number of iterations for the particles;

[0027] The variation range of each influencing parameter is randomly initialized within the set optimization speed range of the particle, and global optimization is performed by using nonlinear adaptive decreasing inertial weights and time-varying asynchronously adjusted learning factors.

[0028] More specifically, when comparing two particles,

[0029] If both satisfy the constraints, the one with the smaller optimization objective is preferred;

[0030] If one is feasible while the other is not, the feasible particle is preferred.

[0031] If neither is feasible, the one with the lower degree of constraint violation is preferred; wherein the degree of constraint violation is obtained by the difference between the success rate and the lowest binding success rate.

[0032] Specifically, it also includes a verification step, in which the actual control parameters obtained in S4 are written into the control unit of the bundled platform, and several rounds of independent verification tests are carried out, with the test conditions being consistent with those of the prototype test.

[0033] The present invention also provides a multi-step binding timing control system for swimming crabs based on the aforementioned multi-step binding timing control method, comprising:

[0034] Bundled platform;

[0035] The parameter encoding module is used to obtain the parameters affecting the single binding cycle time and binding success rate, including the cutting parameters of the cutting mechanism, the lifting parameters of the clamping mechanism, the connection sequence of the lifting of the clamping mechanism and the translation of the support claw mechanism, and the expansion parameters of the support claw mechanism, and encodes them according to the corresponding value range as low, medium and high levels;

[0036] The prototype testing module is used to conduct prototype testing using a four-factor central composite design method based on the aforementioned coding, thereby establishing second-order response surface regression models for single-bundle cycle time and bundle success rate.

[0037] The optimization model building module is used to construct constraints based on the binding success rate predicted by the second-order response surface regression model, with the predicted single binding cycle time as the optimization objective. Combined with the encoding of the parameter encoding module, the combination of all influencing parameters is used as particles to construct a particle swarm optimization model and perform global optimization.

[0038] The result acquisition module is used to obtain the optimal combination of encoded variables based on global optimization, convert it into actual control parameters, and input them into the control unit of the bundled platform.

[0039] The present invention also provides a computer-readable medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the aforementioned multi-step binding timing control method for swimming crabs.

[0040] Beneficial effects: This invention uses key parameters of specific work steps as optimization variables to establish a data-driven optimization framework based on response surface regression model and adaptive constrained particle swarm optimization algorithm. Under the condition that the binding success rate is not lower than the set threshold, it minimizes the single binding cycle time, taking into account both optimization accuracy and experimental cost. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in this invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely embodiments of this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a flowchart of the multi-step binding timing control method for swimming crabs according to the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions and advantages of the present invention clearer, the present application will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0044] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of the present invention should have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0045] The multi-step binding timing control method for swimming crabs of the present invention is applied to a binding platform consisting of a rubber tube conveying mechanism, a cutting mechanism, a clamping mechanism, a support claw mechanism, and a material unloading mechanism arranged sequentially in a straight line. The specific process is as follows: Figure 1 As shown, it includes:

[0046] S1. Obtain the parameters affecting the single binding cycle time and binding success rate, including the cutting parameters of the cutting mechanism, the lifting parameters of the clamping mechanism, the connection sequence of the lifting of the clamping mechanism and the translation of the support claw mechanism, and the expansion parameters of the support claw mechanism, and encode them according to the corresponding value range as low, medium and high levels;

[0047] S2. Based on the coding of S1, a prototype test was carried out using a four-factor central composite design method, thereby establishing second-order response surface regression models for single-bundle cycle time and bundle success rate.

[0048] S3. The binding success rate predicted by the second-order response surface regression model of S2 is used to construct constraints. The predicted single binding cycle time is used as the optimization objective. Based on the encoding of S1, the combination of all the influence parameters obtained by S1 is used as particles to construct a particle swarm optimization model and perform global optimization.

[0049] S4. Based on the global optimization in S3, obtain the optimal combination of encoded variables and convert it into actual control parameters.

[0050] In this invention, the rubber tube conveying mechanism uses a drive unit such as a stepper motor driven by a reducer to drive an active synchronous pulley; the straight section of the synchronous belt clamps the rubber tube and pushes it forward, and the conveying amount is precisely determined by the number of pulses issued by the controller to control the rotation of the stepper motor, so as to provide a fixed length of rubber tube segment in preparation for cutting a fixed width rubber ring.

[0051] The cutting mechanism includes an inner support rod and a blade. The inner support rod is driven by an electromagnet. When energized, it extends horizontally into the inner cavity of the conveyed rubber tube, supporting the tube wall from the inside to prevent collapse and deformation during cutting. The blade is vertically mounted on the slider, allowing it to slide vertically. The drive unit, such as a stepper motor, is connected to the blade via a crank, which drives the blade to move downward vertically. With the cooperation of the inner support rod, the blade cuts the rubber tube, producing a single rubber ring with a width of 15-17 mm.

[0052] The clamping mechanism is located outside the cutting mechanism and consists of an electromagnet-driven split jaw and a drive unit such as a stepper motor-driven rotary lifting arm. The stepper motor-driven rotary lifting arm moves the split jaw into position. Before cutting, the jaw closes and clamps the outer wall of the rubber tube to cooperate with the inner support rod of the cutting mechanism to form double fixation inside and outside. After the cutting mechanism finishes cutting, the inner support rod retracts, and the jaw holds the rubber ring in the state of clamping. The drive unit such as a stepper motor drives the rotary lifting arm to rotate upward by 90° to achieve lifting. At the same time, the support claw mechanism starts to translate. When the rotary lifting arm completes the lifting, it moves to the position directly below the jaw. At this time, the jaw is released, and the rubber ring falls and is sleeved on the four expansion parts of the support claw mechanism.

[0053] The claw support mechanism includes a rotatable turntable with four vertically extending members evenly distributed circumferentially on it. The turntable has arc-shaped grooves corresponding to the extending members, which can be slidably installed within these grooves. When the drive unit, such as a servo motor, rotates, it causes the extending members to slide along the arc-shaped grooves, converting the rotation of the turntable into synchronous radial expansion motion of the extending members. This expands the rubber ring, and once fully expanded, the servo motor enters a torque locking mode, at which point the crab's claw can be inserted.

[0054] The unloading mechanism includes a lever mounted on the expansion member that is driven to rotate by a drive unit such as a stepper motor. The lever swings from the outside to the inside of the expansion member under the drive of the stepper motor, thereby pulling away the rubber ring sleeved on the outside of the expansion member. The rubber ring contracts and binds to the crab claw under the action of elastic restoring force.

[0055] In this invention, the bundling platform and its related structures can be based on the patent application CN121044104A, which will not be described in detail here.

[0056] In this invention, the multi-step process of the aforementioned bundling platform includes:

[0057] Conveying step: The drive unit of the rubber tube conveying mechanism, such as a stepper motor, precisely advances the rubber tube according to the set number of pulses, providing a quantitative amount of raw material for the cutting step;

[0058] Fixing step: The inner support rod of the cutting mechanism extends into the inner cavity of the rubber tube to support the tube wall from the inside. The gripper of the clamping mechanism clamps the outer wall of the rubber tube from the outside to cooperate with the inner support rod of the cutting mechanism to form double fixation inside and outside, preventing the tube wall from collapsing during cutting and ensuring a neat cross-section.

[0059] Cutting step: The blade of the cutting mechanism moves down along the guide rail by the drive unit such as a stepper motor to drive the slider, thereby cutting the rubber tube and cutting out a rubber ring of a fixed width;

[0060] Transfer preparation step: The push rod is retracted to avoid obstructing the lifting and transfer of the clamping mechanism. At this time, the jaws hold the rubber ring.

[0061] Transfer step: The clamping mechanism lifts and transfers. In this step, the clamping mechanism rotates and lifts the jaws by 90°, thereby driving the rubber ring to rotate and lift by 90°. At the same time, the support jaw mechanism starts to translate, and moves to the position directly below the jaws as the rotating lifting arm completes the lifting.

[0062] Rubber ring release step: The gripper releases the rubber ring, causing the rubber ring to fall into and be fitted onto the expansion part of the support claw mechanism;

[0063] Rubber ring opening step: The drive unit of the support claw mechanism, such as a servo motor, drives the turntable to rotate, causing the expansion component to expand radially in sync, thereby opening the rubber ring and locking the torque after it is opened to the position;

[0064] The process of detecting the placement of swimming crabs involves using a grating sensor mounted on the claw mechanism. The placement of a swimming crab is detected when its claws block the grating sensor.

[0065] Unloading and binding step: The drive unit of the unloading mechanism, such as a stepper motor, drives the lever to rotate, thereby pulling the rubber ring off the expansion piece, binding it to the crab claw, and then resetting it.

[0066] System reset step: All mechanisms return to their initial positions and wait for the next cycle.

[0067] In this invention, the arrival time of each step can be determined by the input signals of the detection components installed at the corresponding positions of the above-mentioned binding platform. For example, the start button on the operation panel, the lifting limit switch of the gripper mechanism, the translation limit switch of the support claw mechanism, the expansion limit switch of the expansion component, the grating sensor for detecting the insertion of the swimming crab, and the reset point limit switches of each mechanism.

[0068] In S1, in the aforementioned steps, the cutting parameter of the cutting mechanism is the rotational speed A when the driving unit of the cutting mechanism drives the blade to cut; the lifting parameter of the clamping mechanism is the rotational speed B when the driving unit of the clamping mechanism drives the rotating lifting arm to rotate upward; the connection sequence between the lifting of the clamping mechanism and the translation of the claw mechanism is the synchronization time difference C between the driving unit of the clamping mechanism driving the rotating lifting arm to rotate upward and the translation of the claw mechanism; and the expansion parameter of the claw mechanism is the rotational speed D when the driving unit of the claw mechanism drives the expansion component to expand. These four parameters not only independently affect the single binding cycle time T and the binding success rate S, but also have significant interaction effects with each other, and must be optimized collaboratively rather than tuned independently.

[0069] In this invention, in S1, the value range of each influencing parameter is determined based on the rated parameters of the drive unit and the safety boundary obtained from the pre-test. The boundary value and center value of the corresponding value range are used to obtain the corresponding low, medium and high level values, which are then encoded as -1, 0 and +1, respectively.

[0070] Specifically, the encoded value x and the actual value X real The conversion relationship is as follows:

[0071] X real =X0+xΔX;

[0072] Where X0 is the median level value of each influencing parameter, and ΔX is the step size of the level change of each influencing parameter.

[0073] In one embodiment of the present invention, the median value of each influencing parameter can be taken as the middle value between the low level and the high level, and the step size of the level change of each influencing parameter can be taken as half of the difference between the high level and the low level.

[0074] For example, the cutting parameters of the cutting mechanism are the rotational speed A of the cutting mechanism's drive unit driving the blade to cut, which can be set to [30, 120], in rpm. The lower limit ensures sufficient cutting torque to reliably cut the rubber tube, while the upper limit is limited by the dynamic balance of the crank-slider mechanism and the blade's lifespan. Therefore, the horizontal value is 75 rpm, and the horizontal change step is 45 rpm. The lifting parameters of the clamping mechanism are the rotational speed B of the clamping mechanism's drive unit driving the rotary lifting arm to rotate upwards, which can be set to [30, 100], in rpm. The lower limit ensures overcoming the weight of the rubber ring and the gripper, while the upper limit is limited by the torque-frequency characteristic curve of the stepper motor. Therefore, the horizontal value is 65 rpm, and the horizontal change step is 35 rpm. The connection sequence between the lifting of the clamping mechanism and the translation of the support claw mechanism is the synchronous time difference C between the upward rotation of the rotary lifting arm driven by the clamping mechanism's drive unit and the translation of the support claw mechanism, which can be set to [0, 120], in rpm. [0.4], in seconds, where the lower limit corresponds to the complete synchronization of the lifting of the clamping mechanism and the translation of the support claw mechanism, and the upper limit is limited by the physical avoidance conditions of the mechanism's motion envelope surface. In this case, the horizontal value is 0.2s and the horizontal change step is 0.2s. The expansion parameter of the support claw mechanism is the value range of the rotational speed D when the drive unit of the support claw mechanism drives the expansion component to expand. It can be set to [100, 800], in rpm. The lower limit ensures that the time taken to expand does not exceed the tolerance limit, and the upper limit is limited by the highest rotational speed of the drive unit and the material damage threshold of the rubber ring under high-speed tension. In this case, the horizontal value is 450 rpm and the horizontal change step is 350 rpm.

[0075] In this invention, in S2, a prototype test is conducted using a four-factor central composite design method based on the coding of S1, comprising three parts:

[0076] (1) 2 4 The full factorial design includes 16 corner points (i.e., the corner points corresponding to the boundary values ​​of each influencing parameter within its corresponding value range, i.e., the coding level ±1).

[0077] (2) Eight axis points, that is, each influencing parameter is located at the coding level of ±α, α is within the coding value range, and the other influencing parameters are kept at the coding level of 0; in this embodiment, α can be 1;

[0078] (3) All influencing parameters are at the coding level of 0 and the experiment is repeated several times, such as 6 times, to estimate the pure experimental error;

[0079] This resulted in 30 sets of prototype tests.

[0080] In one embodiment of the present invention, during the aforementioned test, the ambient temperature was maintained at 25±2℃, and the rubber tubing was from the same batch. Each group underwent 100 consecutive binding cycles. The time T for each binding cycle and the binding success rate S were recorded, where the binding success rate S=N. s / 100×100%, N s This indicates the number of times the binding was successful. The criterion for this is that the rubber ring falls into the target position of the crab claw and does not fall off for 30 seconds.

[0081] In this invention, the second-order response surface regression model for the single-bundle cycle time and binding success rate established in S2 is as follows:

[0082] ;

[0083] Where Y is the response value, which can be used to predict the single binding cycle time and binding success rate; using the aforementioned encoded influence parameters as dependent variables, then xᵢ, x j Let i and j be two independent variables that generate an interaction. Each variable is one of the aforementioned encoded influence parameters, and they are different. Specifically, i and j can both take the values ​​1, 2, 3, and 4, and x1 = A, x2 = B, x3 = C, and x4 = D; β0 and β2 are also variables that interact. i β ii β ij These are the coefficients of the corresponding constant term, linear term, quadratic term, and interaction term, respectively; ε is the random error.

[0084] Substituting the aforementioned encoded influence parameters, the second-order response surface regression model can be expressed as:

[0085] ;

[0086] Where β1, β2, β3, and β4 are coefficients of the linear term, β 11 β 22 β 33 and β 44 It is the coefficient of the quadratic term, β 12 β 13 β 14 β 23 β 24 and β 34 It is the coefficient of the interaction term.

[0087] In one embodiment of the present invention, the established second-order response surface regression model of single-bundle cycle time and binding success rate can be subjected to least squares regression and insignificant terms can be removed to obtain the final second-order response surface regression model of single-bundle cycle time and binding success rate, as follows:

[0088] Using the minimum sum of squared residuals as the criterion, a design matrix H is constructed from the aforementioned prototype test results. This matrix can then be obtained using the formula β=(H).T H) -1 H T Y is calculated to obtain the coefficients of the aforementioned 1 constant term, 4 linear terms, 4 quadratic terms, and 6 interaction terms. Here, Y is a column vector composed of the measured data obtained from the aforementioned experiment. The measured data are the single binding cycle time or binding success rate measured in each experiment. Thus, the final second-order response surface regression models of the single binding cycle time and binding success rate can be obtained respectively.

[0089] Design a table with matrix H consisting of 30 rows and 15 columns. Each row represents one experiment, and each column represents one constant term, four linear terms, four quadratic terms, and six interaction terms in the second-order response surface regression model.

[0090] The response vector Y, consisting of 30 rows and 1 column, is constructed from the measured single-bundle cycle time or binding success rate of each test.

[0091] Substituting the design matrix H and the response vector Y into the formula β=(H T H) -1 H T Y can be used to calculate 15 regression coefficients, which can then be used to obtain the final second-order response surface regression model of the single binding cycle time and binding success rate.

[0092] In this invention, a second-order response surface regression model containing all terms (one constant term, four linear terms, four quadratic terms, and six interaction terms) is fitted to determine the single-bundle cycle time and the success rate of the bundle. A t-test is performed on each term to obtain the p-value. The term with the largest p-value that is greater than the significance threshold is removed. This process is repeated until all remaining terms are significant (i.e., all p-values ​​are less than the significance threshold).

[0093] In the aforementioned process of eliminating insignificant terms, the hierarchical principle must also be followed, that is, when a quadratic term / interaction term of a certain variable is retained, its corresponding linear term is also retained.

[0094] For example, after the aforementioned simplification, the second-order response surface regression model for the single-bundle cycle time and binding success rate obtained in one embodiment of the present invention is as follows:

[0095] ;

[0096] .

[0097] In this invention, in S2, the established second-order response surface regression model of the single-bundle cycle time and binding success rate can be tested through analysis of variance to determine its significance and the influence of each factor and interaction term. Specifically:

[0098] The F-test was used to calculate the p-values ​​of each term and the p-value of the lack-of-fit term in the second-order response surface regression model of the single-bundle cycle time and the success rate of the bundle established by S2.

[0099] Set a significance threshold, such as 0.05. Only when the p-values ​​of all items are less than the significance threshold and the p-values ​​of the lack-of-fit items are greater than the significance threshold, that is, the second-order response surface regression model of single-bundle cycle time and binding success rate established by S2 is deemed to meet the requirements. Otherwise, it is necessary to return to S2 and re-establish the second-order response surface regression model of single-bundle cycle time and binding success rate.

[0100] In this invention, if the p-values ​​for each term are all less than the significance threshold, it indicates that the second-order response surface regression model for single-bundle cycle time and binding success rate established by S2 is significant, indicating that there is a reliable regression relationship between the response variable and the optimization variable. If the p-values ​​for the lack-of-fit terms are greater than the significance threshold, it indicates that the lack-of-fit terms are not significant, indicating that the second-order response surface regression model for single-bundle cycle time and binding success rate established by S2 does not have obvious underfitting problems, its fit to experimental data is good, it can be used to analyze the influence of each variable on the single-bundle cycle time and binding success rate, and it can be used as a surrogate model for subsequent parameter optimization.

[0101] In this invention, in S3, the binding success rate predicted by the second-order response surface regression model of S2 is used to construct constraints, and the predicted single binding cycle time is used as the optimization objective, as follows:

[0102] ;

[0103] ;

[0104] Where X=[x1, x 2, x 3, x4],x1, x 2, x 3, x4 represents the encoded values ​​of the impact parameters of the single-bundle cycle time and binding success rate obtained by S1; T(X) and S(X) represent the single-bundle cycle time and binding success rate predicted by the second-order response surface regression model of the single-bundle cycle time and binding success rate established by S2; S min The minimum binding success rate is used as a constraint on the binding success rate; for example, considering that the success rate may decrease in actual operation, in order to achieve the goal of a 97% binding success rate, S min 98% is acceptable.

[0105] In this invention, in step S3, based on the encoding of S1, the various influencing parameters obtained in S1 are used as variables to construct a particle swarm optimization model and perform global optimization, as follows:

[0106] Each variable can be randomly initialized within its encoding range, and the corresponding velocity vector can be randomly initialized within a set range. The maximum number of iterations is set, and a nonlinear adaptive decreasing inertial weight and a time-varying asynchronously adjusted learning factor are used to perform global optimization.

[0107] This invention employs a nonlinear adaptive decreasing inertia weight, which decreases at a slower rate in the early stages of the search to maintain sufficient global exploration capability, and accelerates the decreasing rate in the mid-to-late stages to promote fine-grained local convergence. Compared to a linear decreasing strategy, it retains a higher level of inertia at the end of the iteration, helping to reduce the risk of the algorithm getting trapped in local optima too early. Furthermore, this invention uses a time-varying asynchronously adjusted learning factor, which allows the early stages of the search to focus on individual exploration to maintain population diversity, and the later stages to focus on group collaboration to accelerate convergence to the global optimum neighborhood.

[0108] In one embodiment of the present invention, the population size (i.e., the total number of particles) of the particle swarm optimization algorithm is set to 50, the particle velocity range can be set to [-0.15, 0.15], and the maximum number of iterations is set to 120. Each particle is a four-dimensional vector, i.e., [x1, x2, x3, x4], corresponding to the encoded values ​​of four influencing parameters A, B, C, and D. The encoded values ​​of each influencing parameter are randomly initialized within their respective encoding ranges, thus obtaining 50 particles. In the subsequent optimization process, the particle velocity is the change range of the encoded value of each influencing parameter in each particle, which can be randomly initialized within the aforementioned range.

[0109] Therefore, we can conclude that:

[0110] The particle velocity update formula is:

[0111] ;

[0112] The position update formula is:

[0113] ;

[0114] in, , , c1 and c2 are the particle velocities at the kth and k+1th iterations, respectively, representing the change in the encoded value of a certain influencing parameter; c1 and c2 are the individual learning factor and the group learning factor, respectively; r1 and r2 are random numbers in the interval [0,1]. This represents the historical optimal position of the i-th particle after the k-th iteration. This represents the globally optimal position of the population after the k-th iteration. , These are the particle positions at the k-th and k+1-th iterations, respectively.

[0115] In one embodiment of the present invention, the nonlinear adaptively decreasing inertia weight w(k) is specifically as follows:

[0116] w(k)=w min +(w max −w min )·(1−k / K) m ;

[0117] wherein, w max and w min are respectively the upper limit and the lower limit of the inertial weight, which can take empirical values to ensure stable convergence. For example, they can be 0.9 and 0.4 respectively; k is the current iteration number, K is the maximum iteration number, and (1−k / K) is the remaining iteration proportion; when m=1, the inertial weight decreases linearly; when m>1, the inertial weight decreases rapidly in the early stage and gradually slows down in the later stage, so that the inertial weight is reduced earlier, which is beneficial to enhancing the local exploitation capability of the algorithm; when 0<m<1, the inertial weight decreases slowly in the early stage and rapidly in the later stage, which is beneficial to maintaining a strong global exploration capability in the early stage and accelerating convergence in the later stage. Therefore, 0.8 can be selected in this embodiment.

[0118] In an embodiment of the present invention, time-varying asynchronously adjusted learning factors are adopted, which are specifically as follows:

[0119] The individual learning factor c1(k)=2.5−2.0×k / K;

[0120] The social learning factor c2(k)=0.5+2.0×k / K.

[0121] In the present invention, in S3, performing global optimization may adopt a feasibility-first tournament selection strategy, which is specifically as follows:

[0122] For each particle, according to its four-dimensional encoding position, the predicted bundling success rate and single bundling cycle time in the second-order response surface regression model of the single bundling cycle time and the bundling success rate established in S2 can be obtained;

[0123] When comparing two particles:

[0124] If both are feasible, that is, satisfying the constraints, the one with the smaller objective function value T is better;

[0125] If one is feasible and the other is infeasible, the feasible particle is better;

[0126] If both are infeasible, the one with the smaller constraint violation degree is better, wherein the constraint violation degree is defined as g(X)=max(0, S min −S(X)), that is, the difference between the success rate and the minimum bundling success rate, and the larger the difference, the more serious the violation degree;

[0127] In this step, each particle carries a set of four-dimensional encoded variables (corresponding to the encoded values ​​of the four influencing parameters A, B, C, and D), and iteratively searches within the solution space. In each iteration, each particle independently records its historical best position according to the aforementioned feasibility-first tournament selection strategy, denoted as the individual historical best position p. best Simultaneously, from the individual historical bests of all particles, the combination of the four influence parameters corresponding to the particle with the best overall performance is selected according to a tournament selection strategy based on feasibility priority, and this combination is taken as the swarm global optimum g of the current population. best It should be noted that g best It must be a complete combination of the encoded values ​​of the four influencing parameters of a real particle, rather than a combination of the independent optimal values ​​of each dimension. This is because there is an interactive coupling effect between the four optimization variables, and piecing them together independently will destroy the intrinsic relationship between the variables.

[0128] When the iteration satisfies the convergence condition (reaching the maximum number of iterations or multiple consecutive iterations g), best After (no longer updating), the current g best The encoded values ​​of the four influencing parameters are combined and output as the global optimal solution.

[0129] In one embodiment of the present invention, based on the above optimization, the final optimal combination of encoded variables is A=+0.453, B=+0.300, C=+0.600, and D=−0.280. According to the aforementioned encoded value x and actual value X... real Converting the conversion relationship into actual process parameters, we can obtain: the rotational speed A of the cutting mechanism's drive unit when driving the blade to cut is 95.4 rpm; the rotational speed of the clamping mechanism's drive unit when driving the rotating lifting arm to rotate upward is 75.5 rpm; the synchronization time difference between the upward rotation of the clamping mechanism's drive unit and the translation of the support claw mechanism is 0.32 s; and the rotational speed of the support claw mechanism's drive unit when driving the expansion component to expand is 351.9 rpm.

[0130] The present invention may also include a verification step, which involves writing the actual control parameters obtained in S4 into the control unit of the bundled platform and conducting three rounds of independent verification tests, 100 times per round, with the test conditions being consistent with the aforementioned prototype test.

[0131] A total of 300 binding cycles were performed across three rounds. The measured average binding cycle time T = 11.92 s, and the measured average binding success rate S = 97.7%, meeting the target requirement of a binding success rate of not less than 97%. Compared with the prediction values ​​of the response surface model at the optimal parameters (binding cycle time T = 12.01 s, binding success rate S = 98%), the prediction error of the binding cycle time T is 0.75%, and the absolute deviation of the binding success rate S is 0.3 percentage points, both of which are below the engineering acceptable threshold of 1%. This verifies the effectiveness and reliability of the multi-step binding timing control method for swimming crabs of the present invention.

[0132] The present invention also provides a multi-step binding timing control system for swimming crabs based on the aforementioned multi-step binding timing control method, comprising:

[0133] Bundled platform;

[0134] The parameter encoding module is used to obtain the parameters affecting the single binding cycle time and binding success rate, including the cutting parameters of the cutting mechanism, the lifting parameters of the clamping mechanism, the connection sequence of the lifting of the clamping mechanism and the translation of the support claw mechanism, and the expansion parameters of the support claw mechanism, and encodes them according to the corresponding value range as low, medium and high levels;

[0135] The prototype testing module is used to conduct prototype testing using a four-factor central composite design method based on the aforementioned coding, thereby establishing second-order response surface regression models for single-bundle cycle time and bundle success rate.

[0136] The optimization model building module is used to construct constraints based on the binding success rate predicted by the second-order response surface regression model, with the predicted single binding cycle time as the optimization objective. Combined with the encoding of the parameter encoding module, the combination of all influencing parameters is used as particles to construct a particle swarm optimization model and perform global optimization.

[0137] The result acquisition module is used to obtain the optimal combination of encoded variables based on global optimization, convert it into actual control parameters, and input it into the control unit of the bundled platform.

[0138] The present invention also provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the aforementioned multi-step binding timing control method for swimming crabs.

[0139] This invention uses key parameters of specific work steps as optimization variables and establishes a data-driven optimization framework based on response surface regression model and adaptive constrained particle swarm optimization algorithm. By establishing a second-order response surface regression model with single-bundle cycle time and binding success rate, the invention replaces expensive point-by-point physical experiments to evaluate the merits of candidate parameter combinations. This enables the particle swarm optimization algorithm to complete the global search of the four-dimensional parameter space with only a few dozen prototype experiments. It also minimizes the single-bundle cycle time while ensuring that the binding success rate is not lower than a set threshold, thus balancing optimization accuracy and experimental cost.

[0140] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of the invention is limited to these examples; within the framework of the invention, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of the different aspects of the embodiments of the invention as described above, which are not provided in detail for the sake of brevity.

[0141] The embodiments of this invention are intended to cover all such substitutions, modifications, and variations falling within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of this invention should be included within the protection scope of this invention.

Claims

1. A method for controlling the timing of multi-step binding of swimming crabs, applied to a binding platform consisting of a rubber tube conveying mechanism, a cutting mechanism, a clamping mechanism, a support claw mechanism, and a material ejection mechanism arranged sequentially along a straight line in multiple steps, characterized in that... include: S1. Obtain the parameters affecting the single binding cycle time and binding success rate, including the cutting parameters of the cutting mechanism, the lifting parameters of the clamping mechanism, the connection sequence of the lifting of the clamping mechanism and the translation of the support claw mechanism, and the expansion parameters of the support claw mechanism, and encode them according to the corresponding value range as low, medium and high levels; S2. Based on the coding of S1, a prototype test was carried out using a four-factor central composite design method, thereby establishing second-order response surface regression models for single-bundle cycle time and bundle success rate. S3. The binding success rate predicted by the second-order response surface regression model of S2 is used to construct constraints. The predicted single binding cycle time is used as the optimization objective. Based on the encoding of S1, the combination of all the influence parameters obtained by S1 is used as particles to construct a particle swarm optimization model and perform global optimization. S4. Based on the global optimization in S3, obtain the optimal combination of encoded variables and convert it into actual control parameters.

2. The multi-step binding timing control method for swimming crabs according to claim 1, characterized in that, In S1, the value range of each influencing parameter is determined based on the rated parameters of the drive unit and the safety boundary obtained from the pre-test. The corresponding low, medium and high level values ​​are obtained from the boundary values ​​and center values ​​of the corresponding value range, and then encoded accordingly.

3. The multi-step binding timing control method for swimming crabs according to claim 2, characterized in that, In S2, a prototype test is conducted using a four-factor central composite design method based on the coding of S1, comprising three parts: (1) 2 4 The full factorial design includes 16 points, each containing a boundary value of its corresponding range for each influencing parameter. (2) Each influencing parameter is located at a coding level of ±α, where α is within the coding value range, and the other influencing parameters are kept at a coding level of 0, for a total of 8 points; (3) Repeat the experiment several times with all influencing parameters at the coding level of 0; This results in a corresponding number of prototype tests.

4. The multi-step binding timing control method for swimming crabs according to claim 1, characterized in that, In S2, for the second-order response surface regression models for the single-bundle cycle time and the success rate of the bundle, respectively, least squares regression is used and insignificant terms are removed, thereby obtaining the final second-order response surface regression models for the single-bundle cycle time and the success rate of the bundle.

5. The multi-step binding timing control method for swimming crabs according to claim 1, characterized in that, In S3, the binding success rate predicted by the second-order response surface regression model in S2 is used to construct constraints, and the predicted single binding cycle time is used as the optimization objective, as follows: ； ； Where X = [x1, x2, x3, x4], x1, x2, x3, x4 are the encoded values ​​of the influence parameters of the single-bundle cycle time and the binding success rate obtained by S1, respectively; T(X) and S(X) are the single-bundle cycle time and the binding success rate predicted by the second-order response surface regression model of the single-bundle cycle time and the binding success rate established by S2, respectively; S min This represents the lowest possible success rate for bundling.

6. The multi-step binding timing control method for swimming crabs according to claim 1, characterized in that, In S3, using the influencing parameters obtained in S1 as variables, a particle swarm optimization model is constructed and global optimization is performed, as follows: Each influencing parameter is randomly initialized within its encoding range, thereby obtaining several particles; Set the optimal velocity range and maximum number of iterations for the particles; The variation range of each influencing parameter is randomly initialized within the set optimization speed range of the particle, and global optimization is performed by using nonlinear adaptive decreasing inertial weights and time-varying asynchronously adjusted learning factors.

7. The multi-step binding timing control method for swimming crabs according to claim 6, characterized in that, When comparing two particles If both satisfy the constraints, the one with the smaller optimization objective is preferred; If one is feasible while the other is not, the feasible particle is preferred. If neither is feasible, the one with the lower degree of constraint violation is preferred; wherein the degree of constraint violation is obtained by the difference between the success rate and the lowest binding success rate.

8. The multi-step binding timing control method for swimming crabs according to claim 1, characterized in that, It also includes a verification step, in which the actual control parameters obtained in S4 are written into the control unit of the bundled platform, and several rounds of independent verification tests are carried out, with the test conditions being consistent with those of the prototype test.

9. A multi-step binding timing control system for swimming crabs based on the timing control method for multi-step binding of swimming crabs according to any one of claims 1-8, characterized in that, include: Bundled platform; The parameter encoding module is used to obtain the parameters affecting the single binding cycle time and binding success rate, including the cutting parameters of the cutting mechanism, the lifting parameters of the clamping mechanism, the connection sequence of the lifting of the clamping mechanism and the translation of the support claw mechanism, and the expansion parameters of the support claw mechanism, and encodes them according to the corresponding value range as low, medium and high levels; The prototype testing module is used to conduct prototype testing using a four-factor central composite design method based on the aforementioned coding, thereby establishing second-order response surface regression models for single-bundle cycle time and bundle success rate. The optimization model building module is used to construct constraints based on the binding success rate predicted by the second-order response surface regression model, with the predicted single binding cycle time as the optimization objective. Combined with the encoding of the parameter encoding module, the combination of all influencing parameters is used as particles to construct a particle swarm optimization model and perform global optimization. The result acquisition module is used to obtain the optimal combination of encoded variables based on global optimization, convert it into actual control parameters, and input them into the control unit of the bundled platform.

10. A computer-readable medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the multi-step binding timing control method for swimming crabs as described in claims 1-8.

Images