Multi-station task optimization scheduling method for automobile oil seal automatic detection production line

Abstract

This invention relates to the field of oil seal production line scheduling technology, specifically to a multi-station task optimization scheduling method for automated automotive oil seal testing production lines. The method includes: acquiring a set of tasks to be tested containing multiple oil seals; sorting the set of tasks to be tested using an improved ant colony algorithm to obtain an optimal task sequence; generating and issuing production line control commands based on the optimal task sequence to control the production line to perform testing operations according to the optimal task sequence. In other words, the solution of this invention can quickly and accurately obtain the optimal scheduling task sequence for automated automotive oil seal testing production lines, solving the problem of frequent changeover losses during mixed-line testing of automotive oil seals.

Description

Technical Field

[0001] This invention relates to the field of automotive oil seal production line scheduling technology. More specifically, this invention relates to a multi-station task optimization scheduling method for automated automotive oil seal testing production lines. Background Technology

[0002] As a critical sealing component, automotive oil seals involve multiple stages in their production process, including dimensional measurement, appearance defect detection (such as missing material or burrs), and spring assembly inspection. To meet the market demand for diverse varieties and small batches, modern automated testing production lines typically adopt a mixed-line production mode, that is, continuously processing orders for oil seals of different models (different inner diameters, outer diameters, and lip shapes) on the same production line.

[0003] To reduce downtime losses caused by equipment adjustments due to product model switching (such as changing tooling, adjusting camera focal length, or loading new algorithms), the Ant Colony Optimization (ACO) algorithm is typically used to optimize the production task sequence.

[0004] However, existing ant colony algorithms have the following significant limitations when applied to this type of scheduling problem: The blindness in the initialization phase leads to slow convergence and wasted computing power. Traditional ant colony algorithms typically set the pheromone concentration on each path to an equal value at the initial moment. This means that when the "ants" construct the initial solution, they completely ignore the objective differences in manufacturing processes between different oil seals (for example, model A differs from model B by only 1mm in inner diameter, while model B differs from model C by 20mm). The algorithm must initially perform a large number of random walks, gradually accumulating pheromones to distinguish between superior and inferior paths through lengthy iterations.

[0005] The aforementioned cold start characteristics result in excessively long computation time when facing real-time production line scheduling requirements. Furthermore, it is easy to consume a large amount of computing resources in the initial search without quickly locking in a better feasible solution, making it difficult to meet the high response requirements of the production line to the cycle time. Summary of the Invention

[0006] The purpose of this invention is to propose a multi-station task optimization scheduling method for automated automotive oil seal testing production lines, in order to solve the problem that existing algorithms cannot quickly lock in a better feasible scheduling solution and cannot meet the high response requirements of the testing production line for cycle time; to this end, this invention provides a solution in one aspect.

[0007] The multi-station task optimization scheduling method for an automated automotive oil seal testing production line provided by this invention includes: Obtain a set of tasks to be tested that contains multiple oil seals; An improved ant colony algorithm is used to sort the set of tasks to be detected to obtain the optimal task sequence; production line control instructions are generated and issued according to the optimal task sequence to control the production line to perform detection operations according to the optimal task sequence. The improved ant colony algorithm includes a positive correlation between the state transition probability between any two tasks and the corresponding pheromone concentration, traditional heuristic information, and pre-acquired prior guiding potential. The prior guiding potential energy is negatively correlated with the pre-acquired mechanical adjustment damping coefficient, the mechanical adjustment damping coefficient is positively correlated with the logarithm of the dispersion index, and is positively correlated with the exponential term including the dispersion index and the set threshold; the dispersion index characterizes the difference between the process attribute characteristics of any two oil seals.

[0008] The above scheme transforms the dispersion index between any two oil seals into a mechanical adjustment damping coefficient, and further constructs a priori guiding potential energy to directly intervene in the state transition probability of the ant colony algorithm. This enables the algorithm to avoid high-loss paths caused by excessive physical differences during the first iteration, and significantly reduces the number of iterations required for convergence in the automated mixed-line detection scenario. This greatly improves the search efficiency in the early stage of the algorithm, thereby quickly outputting the optimal scheduling sequence and significantly reducing the mechanical adjustment losses and downtime caused by equipment replacement.

[0009] Optionally, the state transition probability for: ; in, , , These are the influencing factors that control pheromones, distance, and prior guiding potential energy, respectively. For path The state transition probability at the t-th iteration. Number the ants. For path The pheromone concentration at the t-th iteration , Paths ,path The distance at the t-th iteration. , Paths ,path The prior guiding potential energy, For ants The set of next tasks that can be selected. is the sequence number of the task to be tested in the next task set, where i and j are the sequence numbers of the oil seal corresponding to the task to be tested.

[0010] The above scheme introduces prior guiding potential energy into the state transition probability, so that in the early stage of iteration without any historical pheromone accumulation, the oil seal detection task nodes with similar process attributes and low switching costs can be given a higher selection probability through mathematical mechanisms. This overcomes the blindness of production line scheduling during cold start and meets the application requirements for millisecond-level response in sudden order insertion scenarios.

[0011] Optionally, the process attribute features include numerical features and discrete features; the numerical features include at least one of the inner diameter of the oil seal, the outer diameter of the oil seal, and the height of the oil seal; the discrete features are type identifiers; the type identifiers are the types of oil seals identified by a pre-inspection system or a visual recognition algorithm.

[0012] The above scheme can comprehensively and accurately depict the objective differences in physical size and structural shape of different types of oil seals, providing a reliable data dimension basis for subsequent accurate quantitative calculation of mechanical changeover resistance of production line equipment, and avoiding the defect of subsequent scheduling being divorced from the underlying actual physical characteristics.

[0013] Optionally, the dispersion index is obtained by fusing the squared difference between the numerical features of two oil sealing tasks with the judgment result of whether the discrete features are the same.

[0014] The dispersion index in the above scheme can comprehensively and quantitatively evaluate the degree of difference between any two oil seal inspection tasks in terms of overall process characteristics. Thus, under complex mixed-line production conditions, it provides a highly quantitative benchmark for accurately evaluating the mold change or parameter fine-tuning range required for production line mechanisms such as servo motors.

[0015] Optionally, the mechanically adjustable damping coefficient for: ; in, Let be the dispersion index between the detection tasks of the i-th oil seal and the detection tasks of the j-th oil seal; To set a threshold; To control the normalized positive number of the step smoothness, ln() is the logarithmic function, and e is the natural constant.

[0016] The mechanical adjustment damping coefficient of the above scheme, through the comprehensive nonlinear physical laws of the marginal decrease in basic adjustment time of the production line mechanical structure during task switching and the triggering of large-scale mold-changing actions (i.e., dead zone effect) when the difference exceeds a certain threshold, makes the final scheduling scheme not only the shortest in theoretical calculation time, but also ensures smoothness in actual physical action, effectively reducing the violent acceleration and deceleration reciprocating motion of the servo mechanism and extending the service life of the production line equipment.

[0017] Optionally, the prior guiding potential energy for: ; in, The mechanical adjustment damping coefficient between the test tasks of the i-th oil seal and the test tasks of the j-th oil seal; The gain is constant. To prevent the default value from having a denominator of zero.

[0018] The prior guiding potential energy in the above scheme makes the recommended potential energy smaller for task paths with larger mechanical adjustment ranges on the production line. It successfully maps the physical change resistance of the underlying equipment into a negative penalty mechanism at the algorithm decision level, ensuring the stability of scheduling guidance and physical executability in the real production line.

[0019] Optionally, before calculating the dispersion index, the data of each feature in the numerical feature is normalized.

[0020] Optionally, the initial steps of the ant colony algorithm include: Set the pheromone concentration to a uniform initial value, and randomly place a set number of ants on different starting oil seal task nodes.

[0021] Optionally, the production line control command includes at least one of the following: Oil seal feeding sequence based on optimal task sequence; Target position parameters of servo motors at each testing station during task switching; Loading instructions used to switch detection algorithms.

[0022] Optionally, the ant colony algorithm further includes: After each ant completes a traversal, the globally optimal task sequence is recorded, and the pheromone concentration on each path is globally updated and locally evaporated. This process continues until the preset maximum number of iterations is reached or the globally optimal task sequence remains unchanged within a specified number of consecutive generations.

[0023] The beneficial effects of this invention are as follows: The present invention extracts the size and type characteristics of oil seals to construct a mechanical adjustment damping coefficient that reflects the nonlinear time consumption and dead zone effect of equipment changeover, and converts it into prior guiding potential energy to directly intervene in the initial state transition probability of the ant colony algorithm. This breaks the limitation of blind search in the early stage of the traditional ant colony algorithm, greatly reduces the number of iterations, and realizes millisecond-level dynamic production scheduling. It not only shortens the theoretical cycle time of the production line, but also makes the physical servo changeover action smooth, effectively reduces downtime losses and extends equipment life, and solves the problem of frequent changeover losses during mixed-line detection of automotive oil seals. Attached Figure Description

[0024] Figure 1 This schematically illustrates the steps of the multi-station task optimization scheduling method for the automated automotive oil seal testing production line in this embodiment. Figure 2 The illustration shows a thermodynamic diagram of the prior guiding potential energy in this embodiment. Detailed Implementation

[0025] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0026] Taking an automated automotive oil seal testing production line as an example, this paper introduces the multi-station task optimization scheduling method for the automated automotive oil seal testing production line of the present invention.

[0027] Specifically, such as Figure 1 As shown, the multi-station task optimization scheduling method for the automated automotive oil seal testing production line in this embodiment includes the following steps: Step S1: Obtain a set of tasks to be tested that includes multiple oil seals.

[0028] In this embodiment, the task set to be tested includes several tasks with different oil seals to be tested, in order to meet the needs of the mixed-line production mode of the automated testing production line. Specifically, the set of tasks to be tested for the oil seals currently to be scheduled for production is obtained through the production line manufacturing execution system. , where n is the oil seal number. This is the task to be tested for the nth oil seal.

[0029] It should be noted that the serial number of the task to be tested corresponds one-to-one with the serial number of the oil seal.

[0030] The aforementioned tasks to be tested specifically include: process attribute characteristics and corresponding execution logic.

[0031] The process attribute features of each task in the task set to be tested include numerical features and discrete features.

[0032] Numerical features include at least the inner diameter of the oil seal. Outer diameter of oil seal and the height of the oil seal At least one of them. Discrete features serve as type identifiers.

[0033] The type identification mentioned above is automatically obtained through a pre-inspection system or visual recognition algorithm. Specifically, before the oil seal enters the formal inspection station, an industrial camera at the loading station takes a full-view picture of the oil seal, and a pre-trained deep learning model or feature extraction algorithm is used to identify the structural features of the oil seal; the structural features of the oil seal include skeleton type, lip structure type, or return flow direction.

[0034] Among them, the skeleton type includes internal skeleton, external skeleton, or assembled skeleton; the lip structure type includes single lip, double lip, or structure with secondary lip; the flow pattern direction includes left-handed, right-handed, or bidirectional flow pattern.

[0035] To eliminate the impact of differences in dimensions on subsequent calculations, the numerical features of the process attributes are normalized before calculating the similarity of process attributes. Specifically, the maximum-minimum normalization method is used to map numerical features such as inner diameter, outer diameter, and height to... Within the range.

[0036] Each task to be inspected corresponds to a unique set of execution logic on the production line, including the coordinates of the robotic arm grasping, the brightness of the light source of the vision camera, and the path of the detection algorithm model.

[0037] Step S2: Obtain the prior guiding potential energy.

[0038] In this embodiment, the process of obtaining the prior guiding potential energy is as follows: Step S21: Calculate the dispersion index based on process attribute characteristics.

[0039] In this embodiment, the dispersion index is obtained by fusing the squared difference between the numerical features of two oil sealing tasks with the judgment result of whether the discrete features are the same.

[0040] When the process attributes include the inner diameter of the oil seal Outer diameter of oil seal and type identifier At that time, the dispersion index is: ; in, The baseline weights for each attribute, , Let be the inner diameters of the i-th oil seal and the j-th oil seal, respectively. , Let be the outer diameters of the i-th oil seal and the j-th oil seal, respectively. , These are the type identifiers for the i-th and j-th oil seals, respectively. For discrete data, the step function, when and If the type identifiers are different, take 1; otherwise, take 0.

[0041] The sum of the three benchmark weights is 1. Specifically, the values ​​of the three benchmark weights can be determined based on the actual situation.

[0042] In the above embodiments, the weighted fusion method can comprehensively quantify the overall differences between the two oil seals in terms of physical size and structural shape.

[0043] Step S22: Calculate the mechanical adjustment damping coefficient.

[0044] When switching production lines between different models, the greater the difference in attributes, the longer the stroke required for the servo motors and pneumatic components, resulting in greater adjustment resistance. Furthermore, this resistance does not increase linearly; it often exhibits a starting threshold and a non-linear stroke duration. To quantify the mechanical adjustment resistance of the equipment caused by task switching, a formula for calculating the mechanical adjustment damping coefficient is constructed.

[0045] Specifically, the mechanical adjustment damping coefficient is: ; in, For the first The task to be tested for the oil seal is related to the first... The mechanical adjustment damping coefficient between the oil seals to be tested. It is a logarithmic function. For the first The task to be tested for the oil seal is related to the first... The dispersion index between the oil seals to be tested. It is a natural constant. To set a threshold, 0 is a normalized positive number that controls the smoothness of the step transition.

[0046] In the formula, the logarithmic term represents the physical law that as the difference increases, the basic adjustment time increases but the marginal effect decreases; the exponential term in the latter part simulates the dead zone effect of the mechanical structure, that is, when the dispersion index exceeds the set threshold, it may trigger additional mechanical reset or large-scale mold change action, resulting in a step increase in the damping coefficient.

[0047] As can be seen from the above formula, when the dispersion index As the value gradually increases, the first half of the growth is gradual because the logarithmic function is monotonically increasing; however, when the dispersion index... Exceeding the set threshold When the exponential term increases rapidly, the overall mechanical damping coefficient exhibits a step-like, sharp increase. This reflects that in actual production lines, when the difference between the two oil seals is small, only slight adjustments are needed, while when the difference is too large, scenarios requiring costly tooling replacements or significant equipment adjustments are necessary.

[0048] Set threshold The range of values ​​is Setting the threshold too low can cause the algorithm to become overly sensitive and miss some globally optimal solutions; setting it too high will fail to effectively reflect changeover resistance on the production line. Therefore, setting a threshold... Control as Within the range, it can accurately distinguish between the critical points that require tooling changes and those that only require parameter fine-tuning.

[0049] In this embodiment, a threshold is set. The value is 0.6.

[0050] Step S23: Calculate the prior guiding potential energy.

[0051] To incorporate the physical damping coefficient into the decision-making logic of the ant colony algorithm, it needs to be transformed into a positive guiding probability. The smaller the damping, the lower the switching loss, and the stronger the ants' tendency to choose that path should be. Therefore, in order to provide effective heuristic guidance in the early stages of the ant colony algorithm and guide the search direction to avoid high-loss paths, a prior guiding potential energy formula needs to be constructed.

[0052] Specifically, the prior guiding potential energy is: ; in, Let be the prior guiding potential energy for the detection tasks of the i-th oil seal and the j-th oil seal. For constant gain, To prevent the default value from having a denominator of zero.

[0053] The preset value is 10. -6 .

[0054] From this formula, we can see that the a priori guided potential energy With mechanical adjustment damping coefficient They exhibit an inverse correlation. When the switching damping coefficient between two tasks increases, the denominator of the formula increases, and the final output prior guiding potential energy decreases accordingly. This reflects the inverse correlation between the production line switching tasks. Switch to task The greater the required mechanical adjustment range, the lower the similarity of its inherent process attributes, and the smaller the prior guiding potential energy given to the path, thus the algorithm tends not to select the path during planning.

[0055] constant gain The preferred value is When this value is too small, the prior potential is too weak to provide guidance; when this value is too large, it will mask the true role of conventional heuristic information. Therefore, in this embodiment, it is controlled to be [value missing]. This allows the prior guiding potential energy of the output to be within a reasonable range that matches the order of magnitude of traditional pheromones.

[0056] The aforementioned prior guiding potential directly quantifies the ability of the detection task to be performed solely based on process similarity in the absence of any historical pheromone accumulation. As a task to be tested The recommendation level of subsequent nodes.

[0057] like Figure 2 As shown, it is a specific heat map of a priori guiding potential energy. In the figure, the different colors correspond to different coupling situations of the two oil seals to be tested. Yellow represents a high value, which proves that the two oil seals to be tested are the most significant strong correlation.

[0058] Step S3: Introduce the prior guiding potential into the traditional ant colony algorithm to obtain an improved ant colony algorithm, and use the improved ant colony algorithm to sort the set of tasks to be detected to obtain the optimal task sequence.

[0059] Among them, the traditional ant colony algorithm is a heuristic search algorithm that simulates the foraging behavior of ants. It is a probabilistic algorithm used to find the optimal path. Its core principle is that ants release pheromones on the path, and other ants tend to choose the path with a high pheromone concentration, forming a positive feedback mechanism, thereby gradually finding the optimal path.

[0060] In this embodiment, prior guiding potential is introduced into the traditional ant colony algorithm to obtain an improved ant colony algorithm.

[0061] Specifically, the process of sorting the set of tasks to be detected using the improved ant colony algorithm to obtain the optimal task sequence is as follows: Step 31, initialize parameters: number of ants m, maximum number of iterations, pheromone heuristic factor, expected heuristic factor, and pheromone evaporation factor.

[0062] Specifically, set the iteration counter g=0 and the maximum number of iterations GN.

[0063] Step S32: Encode each task to be tested in the set of multiple oil seals to be tested, and randomly generate m ant individuals as the initial population.

[0064] Specifically, a set number of ants are randomly placed on the task nodes to be tested at different starting oil seals.

[0065] Step S33: Calculate the pheromone linked list on the path number of each group of ants, and select the next traversal point using the roulette wheel method according to the transition probability, until all traversal points have been selected, and calculate the objective function value of each task sequence.

[0066] The objective function value is the total time required to complete all oil seal tasks in the current task sequence.

[0067] In this embodiment, the state transition rules of the ant colony algorithm are redefined. In the early stages of the algorithm iteration, ants move from the task to be detected... Select the next task to be tested. transition probability Instead of relying solely on empty pheromones or simple distance reciprocals, it introduces a priori guiding potential energy.

[0068] Specifically, the state transition probability is: ; in, , , These are the influencing factors that control pheromones, distance, and prior guiding potential energy, respectively. For path The state transition probability at the t-th iteration. Number the ants. For path The pheromone concentration at the t-th iteration , Paths ,path The distance at the t-th iteration. , Paths ,path The prior guiding potential energy, For ants The set of next tasks that can be selected. This is the sequence number of the task to be tested in the next task set.

[0069] The pheromone influence factor mentioned above reflects the importance of ants accumulating experience in pathfinding. Its typical value range is [1, 5]. If the value is too large, the algorithm is prone to premature convergence to a local optimum. The distance influence factor reflects the immediate impact of task switching time on path selection. When the production line has high cycle time requirements, it should be appropriately increased. This makes the algorithm more inclined to select nodes with shorter switching times. The influence factor of the prior guiding potential is used to control the strength of the intervention of the prior guiding heuristic potential on the initial search.

[0070] Specifically, during the system initialization phase, the influencing factors are calibrated using a preset set of typical oil seal inspection task samples and an offset length search method. For example, a fixed... and The value is changed in steps of 0.1 within the range of [0.5,3]. Record the number of iterations required for the algorithm to converge; select the set of parameters that minimizes the number of iterations and the total switching time as the default influencing factor in the production environment.

[0071] As shown in the above formula, the numerator is the product of pheromone concentration, heuristic information, and prior guiding potential. When the path formed by the i-th and j-th detection tasks... Corresponding prior guiding potential energy When the product term increases, the overall product term increases, thus increasing the state transition probability of that path being chosen. The corresponding increase reflects that, in the process of finding the optimal task sequence, introducing prior guiding potential can directly intervene in the calculation of transition probabilities, even in the early stages of iteration when the pheromone concentration of all paths is... Under equal circumstances, due to The difference in process parameters allows ants to prioritize task nodes with similar process attributes and low switching costs with a high probability, effectively overcoming the slow convergence speed caused by the blind initialization of traditional ant colony algorithms and improving the algorithm's solution efficiency.

[0072] Step S34: Update the pheromone linked list on the path between each point in the current iteration, and iterate multiple times until the stopping condition is met, and output the optimal task sequence.

[0073] The pheromone concentration in the pheromone chain is updated as follows: ; ; ; in, For the first The pheromone concentration is updated in the next iteration. The pheromone concentration that needs to be updated after the t-th iteration. This represents the pheromone increment on the current iteration path (i,j). Let m be the amount of pheromone left by the k-th ant on the path (i,j) in the current iteration, m be the number of ants, and c be the pheromone evaporation coefficient. For positive integers, Let be the path length of the k-th ant in the current iteration.

[0074] After each ant completes one traversal, calculate the total switching time of the current task sequence. Record the globally optimal task sequence and perform global updates and local evaporation operations on the pheromone concentration on each path until the preset maximum number of iterations is reached or the globally optimal task sequence remains unchanged within a specified number of consecutive generations.

[0075] The preferred pheromone volatility coefficient When the pheromone volatility coefficient is 100%, At the same time, it can effectively balance global exploration and local development capabilities, avoiding the algorithm from getting stuck in local optima.

[0076] Step S4: Generate and issue production line control instructions based on the optimal task sequence to control the production line to perform detection operations according to the optimal task sequence.

[0077] In this embodiment, after receiving the optimal task sequence, the production line control system generates corresponding production line control instructions in sequence.

[0078] The production line control instructions include oil seal loading sequence instructions based on the optimal task sequence, servo motor target position parameter instructions for each inspection station during task switching, and loading instructions for switching inspection algorithms.

[0079] In this embodiment, by sending production line control commands to the production line PLC and industrial control computer, the control transmission mechanism and the detection unit work together to enable the automated detection production line to perform assembly line processing according to the planned low switching loss sequence.

[0080] In this embodiment, by transforming the scheduling algorithm into a low-level motion control instruction that includes the oil seal feeding sequence, servo motor change target position parameters, and corresponding detection algorithm loading, seamless collaboration from high-level production scheduling planning to low-level production line manufacturing execution systems (such as PLCs and industrial control computers) is achieved, ensuring that the planned low-loss sequence can be strictly and automatically executed by the automated production line.

[0081] The solution of this invention improves the traditional ant colony algorithm by introducing prior guiding potential energy, which can effectively solve the problem of slow convergence speed of the algorithm in the cold start state, reduce the equipment adjustment downtime losses caused by product model switching in the production line, and thus meet the high response requirements of modern automated testing production lines for cycle time.

[0082] In the description of this specification, "multiple" means at least two, such as two, three or more, etc., unless otherwise expressly and specifically defined.

[0083] While various embodiments of the invention have been shown and described in this specification, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention.

Claims

1. A multi-station task optimization scheduling method for an automated automotive oil seal testing production line, characterized in that, include: Obtain a set of tasks to be tested that contains multiple oil seals; An improved ant colony algorithm is used to sort the set of tasks to be detected to obtain the optimal task sequence; production line control instructions are generated and issued according to the optimal task sequence to control the production line to perform detection operations according to the optimal task sequence. The improved ant colony algorithm includes a positive correlation between the state transition probability between any two tasks and the corresponding pheromone concentration, traditional heuristic information, and pre-acquired prior guiding potential. The prior guiding potential energy is negatively correlated with the pre-acquired mechanical adjustment damping coefficient, the mechanical adjustment damping coefficient is positively correlated with the logarithm of the dispersion index, and is positively correlated with the exponential term including the dispersion index and the set threshold; the dispersion index characterizes the difference between the process attribute characteristics of any two oil seals.

2. The multi-station task optimization scheduling method for an automated automotive oil seal testing production line according to claim 1, characterized in that, The state transition probability for: ； in, , , These are the influencing factors that control pheromones, distance, and prior guiding potential energy, respectively. For path The state transition probability at the t-th iteration. Number the ants. For path The pheromone concentration at the t-th iteration , Paths ,path The distance at the t-th iteration. , Paths ,path The prior guiding potential energy, For ants The set of next tasks that can be selected. is the sequence number of the task to be tested in the next task set, where i and j are the sequence numbers of the oil seal corresponding to the task to be tested.

3. The multi-station task optimization scheduling method for an automated automotive oil seal testing production line according to claim 1, characterized in that, The process attribute features include numerical features and discrete features; the numerical features include at least one of the inner diameter of the oil seal, the outer diameter of the oil seal, and the height of the oil seal; the discrete features are type identifiers; the type identifiers are the types of oil seals identified by a pre-inspection system or a visual recognition algorithm.

4. The multi-station task optimization scheduling method for an automated automotive oil seal testing production line according to claim 2, characterized in that, The dispersion index is obtained by fusing the squared difference between the numerical features of two oil sealing tasks with the judgment results of whether the discrete features are the same.

5. The multi-station task optimization scheduling method for an automated automotive oil seal testing production line according to claim 1, characterized in that, The mechanically adjustable damping coefficient for: ； in, Let be the dispersion index between the detection tasks of the i-th oil seal and the detection tasks of the j-th oil seal; To set a threshold; To control the normalized positive number of the step smoothness, ln() is the logarithmic function, and e is the natural constant.

6. The multi-station task optimization scheduling method for an automated automotive oil seal testing production line according to claim 1, characterized in that, The prior guiding potential energy for: ； in, The mechanical adjustment damping coefficient between the test tasks of the i-th oil seal and the test tasks of the j-th oil seal; The gain is constant. To prevent the default value from having a denominator of zero.

7. The multi-station task optimization scheduling method for an automated automotive oil seal testing production line according to claim 1, characterized in that, Before calculating the dispersion index, the normalization process is also performed on the feature data of each feature in the numerical feature.

8. The multi-station task optimization scheduling method for an automated automotive oil seal testing production line according to claim 1, characterized in that, The initial steps of the ant colony algorithm include: Set the pheromone concentration to a uniform initial value, and randomly place a set number of ants on different starting oil seal task nodes.

9. The multi-station task optimization scheduling method for an automated automotive oil seal testing production line according to claim 1, characterized in that, The production line control instructions include at least one of the following: Oil seal feeding sequence based on optimal task sequence; Target position parameters of servo motors at each testing station during task switching; Loading instructions used to switch detection algorithms.

10. The multi-station task optimization scheduling method for an automated automotive oil seal testing production line according to claim 1, characterized in that, The ant colony algorithm also includes: After each ant completes one traversal, the globally optimal task sequence is recorded, and the pheromone concentration on each path is globally updated and locally evaporated until the preset maximum number of iterations is reached or the globally optimal task sequence remains unchanged within a specified number of consecutive generations.

Images