Production line multi-device collaborative scheduling method and system based on particle swarm optimization

Abstract

The invention relates to the technical field of intelligent production line control, in particular to a production line multi-device collaborative scheduling method and system based on a particle swarm algorithm. The method comprises the following steps: acquiring a future speed sequence formed by a plurality of speeds of the leveler in a prediction time window, and taking the future speed sequence as a position vector of particles in a particle swarm optimization algorithm; calculating fitness values of particles through a target function in a particle swarm optimization algorithm; and obtaining a future speed sequence with the maximum fitness value based on a particle swarm optimization algorithm so as to control the operation of the leveler. The situation that raised grains appear on the strip due to frequent shutdown and acceleration and deceleration of the leveler can be reduced.

Description

Technical Field

[0001] This application relates to the field of intelligent production line control technology, and in particular to a method and system for collaborative scheduling of multiple equipment in a production line based on particle swarm optimization algorithm. Background Technology

[0002] After the coil is uncoiled, the material retains a tendency to bend, especially near the core where the curvature is greater. Without leveling, the sheet cannot remain straight, and direct cutting or processing will result in a bent and deformed finished product. Therefore, a leveling machine is usually installed after the uncoiler. One end of the uncoiled material passes through the leveling machine, which releases the internal stress of the steel and levels it. It then passes through a buffer pit, which stores a section of the strip in a suspended state, using space to save time and decoupling the speed difference between the preceding and following equipment. The leveling machine is usually followed by laser cutting equipment, which performs intermittent step-feeding and contour cutting of the strip according to the nesting diagram. Therefore, a complete automated uncoiler-leveling-laser-cutting blanking production line typically includes multiple pieces of equipment such as an uncoiler, leveling machine, and laser cutting machine. During the operation of the production line, multiple equipment systems work together to complete the continuous cutting of materials.

[0003] In related technologies, the speed of the leveling machine is mainly controlled by PID feedback, that is, the speed of the leveling machine is adjusted according to the deviation of the material height in the buffer pit detected by the sensor. However, PID control has a lag. When the feeding speed of the downstream cutting equipment fluctuates drastically, the leveling machine is forced to perform frequent and rapid acceleration and deceleration. This speed oscillation will cause inconsistent plastic deformation process of the plate by the leveling roller system, destroy the steady-state mechanical environment required for the Bauschinger effect, resulting in uneven elimination of residual stress in the plate, producing stopping spots or hidden wavy lines, and ultimately resulting in poor leveling quality of the plate. Summary of the Invention

[0004] To improve the quality of sheet metal leveling by the leveling machine, this application provides a method and system for collaborative scheduling of multiple equipment in a production line based on particle swarm optimization algorithm.

[0005] Firstly, this application provides a multi-device collaborative scheduling method for production lines based on particle swarm optimization, employing the following technical solution: A multi-equipment collaborative scheduling method for production lines based on particle swarm optimization (PSO) algorithm is proposed. This method obtains the future speed sequence composed of multiple speeds of the leveling machine within the prediction time window, and uses the future speed sequence as the position vector of particles in the PSO algorithm. The fitness value of the particles is calculated through the objective function of the PSO algorithm. The future velocity sequence with the highest fitness value is obtained based on the particle swarm optimization algorithm, and the operation of the leveling machine is controlled accordingly. The step of calculating the fitness value of a particle using the objective function in the particle swarm optimization algorithm includes: calculating the material level deviation cost and process quality cost in the objective function of the particle swarm optimization algorithm; the material level deviation cost is positively correlated with the difference between the predicted height of the material in the buffer pit and the set safety height at each moment within the prediction time window; the process quality cost is positively correlated with the local process cost at each moment within the prediction time window; the local process cost is positively correlated with the absolute value of the leveler acceleration, the absolute value of the difference between the leveler real-time speed and the optimal process speed; and the weighted sum of the material level deviation cost and the process quality cost is used as the fitness value of the particle, with the material level deviation cost and the process quality cost being negatively correlated with the fitness value.

[0006] A prediction time window is constructed, and the future velocity sequence of the leveler within the prediction time window is used as the position vector of particles in the particle swarm optimization algorithm. Then, the objective function is solved using the particle swarm optimization algorithm to obtain the future velocity sequence with the maximum fitness value, which is then used to control the speed of the leveler. The objective function includes material level deviation cost and process quality cost, realizing the synergistic optimization of the material level safety in the buffer pit and the stability of the leveling process. Compared with the traditional PID control method based solely on material height feedback, this scheme can detect the downstream consumption change trend in advance, avoiding the leveler from passively generating frequent and violent acceleration and deceleration. Thus, while ensuring that the buffer pit does not touch the bottom or straighten out, the leveler's operating speed is smoother and the acceleration is more controlled, reducing quality defects such as stopping spots and hidden ripples caused by uneven residual stress.

[0007] Optionally, the automated production line also includes a laser cutting machine located behind the buffer pit. In the process of weighted summation of material level deviation cost and process quality cost, the operating state of the laser cutting machine is divided into laser cutting state and feeding state according to the material consumption sequence that reflects the material consumption rate of the laser cutting equipment. The state with high material consumption is the feeding state. For any given moment, in response to the laser cutting state at that moment, the first weighted strategy is used to calculate the weighted sum of the material level deviation cost and the process quality cost; For any given moment, in response to the feeding state at that moment, the second weighted strategy is used to calculate the weighted sum of the material level deviation cost and the process quality cost.

[0008] The laser cutting equipment is differentiated into laser cutting and feeding states based on its material consumption rate. A differentiated weighting strategy is employed in each state to dynamically reflect the actual operating conditions of the production line. In the cutting state, safety constraints on the material height are strengthened, while in the feeding state, process quality constraints are emphasized. This avoids optimization bias caused by fixed weights, enabling the leveling machine to adopt speed response strategies that better meet actual needs under different operating conditions, thus improving the overall adaptability and robustness of the system's scheduling.

[0009] Optionally, the steps for obtaining the material consumption sequence within the future prediction time window include: reading the G-code instructions that are not currently executed on the cutting surface of the laser cutting machine; calculating the total path length contained in the G-code instructions; and obtaining the predicted consumption sequence of the laser cutting machine based on the set cutting speed and feeding speed.

[0010] By analyzing the unexecuted G-code instructions in the laser cutting machine, which are the control codes used for CNC machine tools, the material consumption sequence within the future prediction time window can be obtained in advance.

[0011] Optionally, in the first weighted strategy, the weight corresponding to the material level deviation cost is greater than the weight corresponding to the process quality cost; In the second weighted strategy, the weight corresponding to the material level deviation cost is less than the weight corresponding to the process quality cost.

[0012] Optionally, the steps for obtaining the predicted height of the material in the buffer pit include: for any moment in the prediction time window, obtaining the length of the material entering the pit and the length of the material leaving the pit within the corresponding time step; calculating the difference between the length of the material leaving the pit and the length of the material entering the pit, using the ratio of this difference to a preset buffer pit geometric transformation coefficient as the height change, and using the sum of the height change and the predicted height of the material in the previous moment as the predicted height at the current moment.

[0013] In the buffer pit, as more material is moved into the pit, it causes the material to descend closer to the ground, while removing material causes it to move further away from the ground. By quantifying the length of the material entering and leaving the buffer pit, and combining this with the geometric transformation coefficient of the buffer pit to estimate the change in material height, the prediction and estimation of the material height at each moment in the prediction time window can be achieved.

[0014] Optionally, the steps for obtaining the geometric transformation coefficient of the buffer pit include: recording the lowest point height of different lengths of material in the buffer pit while the machine is stopped, generating a length-height mapping curve; and using the slope of the mapping curve as the geometric transformation coefficient of the buffer pit.

[0015] Optionally, the calculation steps for the material level deviation cost include: for any given moment, calculating the difference between the predicted height of the material in the buffer pit and the set safe height corresponding to each moment within the prediction time window, using this difference as the local deviation cost, and using the sum of the squares of the local deviation costs corresponding to each moment as the material level deviation cost corresponding to the particle.

[0016] By using the sum of squares of the deviations as the cost of the material level deviation, the algorithm becomes more sensitive to larger material level deviations. This forces the control system to prioritize eliminating dangerous large deviations that could cause the material to bottom out or become taut, while maintaining a certain tolerance for small fluctuations within the safe range, thus enhancing the robustness of the control system.

[0017] Optionally, the sum of the squares of the local process costs at each moment within the prediction time window can be used as the process quality cost of the particle.

[0018] By calculating the sum of squares of local process costs, the algorithm can significantly suppress changes in the acceleration of the leveling machine; it guides the leveling machine to run along a trajectory with optimal energy or minimum rate of change, ensuring the smoothness of mechanical operation.

[0019] Optionally, the safety height is one-third of the depth of the buffer pit.

[0020] Setting the safety height to one-third of the depth of the buffer pit avoids the material touching the bottom and prevents the material from becoming taut, while also reserving the maximum buffer margin for the subsequent rapid feeding of the laser cutting machine, thus maximizing the utilization rate of the buffer pit's buffering capacity.

[0021] Secondly, this application provides a multi-equipment collaborative scheduling system for production lines based on particle swarm optimization, employing the following technical solution: A production line multi-device collaborative scheduling system based on particle swarm optimization algorithm includes: a processor and a memory. The memory stores computer program instructions. When the computer program instructions are executed by the processor, the production line multi-device collaborative scheduling method based on particle swarm optimization algorithm described above is implemented.

[0022] The aforementioned multi-device collaborative scheduling method for production lines based on particle swarm optimization is generated into a computer program and stored in memory for loading and execution by a processor. Thus, a system is created based on the memory and processor for convenient use.

[0023] This application has the following technical advantages: During the control process, a future velocity sequence of the leveling machine is generated. This velocity sequence is used as the initial position in the particle swarm optimization algorithm. Subsequently, the particle is iterated through the particle swarm optimization algorithm to obtain the particle with the largest fitness value. The speed of the leveling machine is controlled according to the future velocity sequence of the particle. Compared with traditional PID feedback control, this can reduce the occurrence of rapid acceleration and deceleration and improve the straightening quality of the leveling machine. Attached Figure Description

[0024] Figure 1 This is a flowchart of a method for collaborative scheduling of multiple devices on a production line based on particle swarm optimization algorithm, according to an embodiment of this application.

[0025] Figure 2 This is a flowchart illustrating the steps of calculating the fitness value of a particle using the objective function in the particle swarm optimization algorithm, as described in this application embodiment.

[0026] Figure 3 This is a graph showing the material consumption of a laser cutting machine in the multi-equipment collaborative scheduling method for a production line based on particle swarm optimization in this application embodiment.

[0027] Figure 4 This is a graph showing the changes in the first and second weights in the embodiments of this application. Detailed Implementation

[0028] This application discloses a multi-equipment collaborative scheduling method for production lines based on particle swarm optimization (PSO). It combines the processing progress of the laser cutting machine to predict future material consumption and constructs a multi-objective optimization model that includes material level deviation costs and process quality costs. The PSO algorithm is used to find the optimal leveling machine speed sequence within the prediction time window, and the weights in the objective function are dynamically adjusted according to the current state of the laser cutting machine. This ensures the safety of the material level in the buffer pit while maximizing the leveling process quality and reducing surface defects on the sheet metal.

[0029] Reference Figure 1 The production line multi-equipment collaborative scheduling method based on particle swarm optimization algorithm includes steps S1-S2.

[0030] S1: Obtain the future velocity sequence composed of multiple velocities of the leveling machine within the prediction time window, and use the future velocity sequence as the position vector of the particles in the particle swarm optimization algorithm; calculate the fitness value of the particles through the objective function in the particle swarm optimization algorithm.

[0031] For any given moment, a prediction time window is constructed, and particle positions are randomly generated using a particle swarm optimization algorithm. Each particle position corresponds to a future velocity sequence. The future velocity sequence represents the predicted velocities at each moment within the prediction time window.

[0032] The steps for calculating the fitness value of a particle using the objective function in the particle swarm optimization algorithm include: steps S11-S13.

[0033] S11: Calculate the material level deviation cost in the objective function of the particle swarm optimization algorithm. The material level deviation cost is positively correlated with the difference between the predicted height of the material in the buffer pit and the set safety height at each time point within the prediction time window.

[0034] For any given particle, the local deviation cost is calculated based on the difference between the material height and the safety height at each moment. The sum of the squares of the local deviation costs at each moment is taken as the material level deviation cost of the particle.

[0035] Specifically, the formula for calculating the cost of particle level deviation can be expressed as: In the formula, The particle level deviation cost represents the particle, and the particle swarm optimization algorithm aims to minimize this value. Indicates the time within the prediction time window The height of the conveyor belt; This is the system-set safety height, which in this embodiment is set to one-third of the buffer pit depth. At this position, the strip will neither touch the bottom nor become taut, and there is a greater tolerance for the height of the strip during subsequent laser cutting, allowing the strip height to decrease. On the one hand, the laser cutting machine consumes zero strip material during the cutting process, thus effectively utilizing the material pit; on the other hand, it also pre-stores more strip material for subsequent laser cutting, enabling the laser cutting machine to feed material at a faster speed.

[0036] Indicates time The local deviation cost is expressed in square form. This ensures that errors are correctly accumulated rather than canceled out, and also improves the sensitivity to large level deviations, thus forcing the algorithm to prioritize eliminating large deviations that could lead to bottoming out or tautness.

[0037] For the real-time conveyor height at any point within the prediction time window, it is necessary to extrapolate future changes in the conveyor height based on the current system state. For each time node within the prediction time window... The formula for calculating the height of the conveyor belt can be expressed as: In the formula, Indicates time The height of the conveyor belt; Indicates time The height of the conveyor belt; Indicates the time step between two moments; To optimize the time-series of future velocities generated by the algorithm speed; Indicates the first The expected length of strip consumed by the laser cutting machine within a given time step; is the geometric transformation coefficient of the buffer pit. In this embodiment, the probability of this value being 0 is extremely low, so the case where this value is 0 is not considered here. Alternatively, in other embodiments, the sum of the geometric transformation coefficient of the buffer pit and 0.1 can be calculated as the denominator in the formula.

[0038] For the first moment in the prediction time window, the calculation is based on the currently collected real-time material height, that is, the currently collected real-time material height is used as the formula. part.

[0039] The height of the material in the buffer pit can be measured in real time using a laser sensor or a structured light rangefinder, with a fixed sampling frequency. Due to mechanical vibrations and changes in the reflective surface of the strip in the production environment, the acquired raw strip height signal is often accompanied by high-frequency noise. To obtain smooth and accurate strip height data, this embodiment uses a one-dimensional Kalman filter to process the raw signal. Kalman filtering is an algorithm that uses the state equations of a linear system to optimally estimate the system state using system input and output observation data. In this embodiment, the state variable is defined as the actual strip height, and the observation variable is the sensor reading. Through two stages—prediction update and measurement update—random noise is effectively filtered out. The observation noise variance can be obtained based on sensor calibration data or static sampling in the field.

[0040] about To obtain the material consumption sequence within the future forecast time window, the steps are as follows: Combine... Figure 3 Because laser cutting blanking lines employ static cutting and step-feeding processes, downstream laser cutting machines are typically stationary during the cutting process to improve cutting accuracy. Material consumption exhibits a non-linear characteristic, alternating between static and high-speed feeding pulse consumption. This step involves... System communication establishes an offline progress consumption model. The control system reads the current processing progress of the laser cutting machine and uses a virtual interpreter to pre-read the remaining unexecuted tasks. The system uses code instructions to determine the cutting and feeding times for the laser cutting machine. For example, the system can accumulate all remaining cutting instructions ( The path length is divided by the set cutting speed, then the piercing and idle movement are added. Calculate the remaining time until the next feeding based on the time consumed. This allows us to obtain the amount of material consumed at each moment.

[0041] For any moment within the prediction time window, if that moment falls within the laser cutting processing period, then That is, at this point, the cutting machine consumes 0 sheet metal.

[0042] For any moment within the prediction time window, if that moment coincides with the feeding period of laser cutting, then the value at that moment... In the formula, Indicates the time point within the prediction time window The length of strip material consumed by the laser cutting machine; Indicates the laser cutting machine time The feeding speed of the laser cutting machine is preset by the operator. At this point, the cutting machine completes processing and begins pulling the material.

[0043] This is the geometric factor of the buffer pit. Because the strip within the buffer pit has a catenary or approximately parabolic shape, the lower the pit, the less significant the change in height caused by the same length change. Therefore, It's about the current altitude. nonlinear functions, i.e. This functional relationship can be obtained through static calibration of the buffer pit on-site. Specifically, while the machine is stopped, the lowest point height corresponding to different strip extensions is recorded, and a mapping curve between length and height is fitted. The slope of this curve is... At the calculation time During the process of increasing the height of the conveyor belt, For the mapping curve The slope corresponding to the height.

[0044] S12: Calculate the process quality cost in the objective function of the particle swarm optimization algorithm. The process quality cost is positively correlated with the local process cost at each moment within the prediction time window. The local process cost is positively correlated with the absolute value of the leveler acceleration, the absolute value of the difference between the leveler real-time speed and the optimal process speed.

[0045] In this step, for any moment within the prediction time window, the instantaneous process stability is calculated based on the leveling machine speed at that moment and the sensitivity of the material itself, in order to characterize the current processing quality.

[0046] For any moment within the prediction time window, the formula for calculating the instantaneous process stability can be expressed as:

[0047] In the formula, Indicates time Instantaneous process stability; This represents the absolute value of the current acceleration of the leveling machine; This is the maximum allowable acceleration for the leveling machine; This represents the current real-time speed of the leveling machine; The optimal processing speed for this material can be obtained by consulting the process manual or based on the experience of the staff.

[0048] The exponential decay form quantitatively reflects the combined impact of material sensitivity and equipment dynamic behavior on leveling quality. When acceleration... The larger the value of the exponential term, or the further the leveling machine speed deviates from the optimal process speed, the larger the value of the exponential term, leading to a decrease in instantaneous process stability approaching zero. For highly sensitive materials, changes in leveling machine speed and acceleration are even more sensitive; therefore, the same motion fluctuations can cause greater changes in instantaneous process stability. In subsequent optimization, the algorithm will seek a trajectory with minimal acceleration and a speed close to the optimal process speed.

[0049] Based on the instantaneous process stability at each moment, the process quality cost is calculated. For any given moment, the higher the instantaneous process stability, the better the quality resulting from the current processing parameters, and consequently, the lower the process quality. Therefore, in this embodiment, the result of subtracting the instantaneous process stability from 1 is taken as the local process cost, and the sum of the squares of the local process costs at each moment is taken as the process quality cost of the particle.

[0050] Specifically, for any particle, the formula for calculating its corresponding process quality cost can be expressed as: In the formula, This represents the cost of manufacturing processes for particles; Indicates the time point within the prediction time window Instantaneous process stability; This indicates the length of the prediction time window.

[0051] S13: The weighted sum of the material level deviation cost and the process quality cost is used as the fitness value of the particle. The material level deviation cost and the process quality cost are negatively correlated with the fitness value.

[0052] For any particle, its total objective function It can be represented as: In the formula, This represents the fitness value of a particle. It is the first weight; It is the second weight; This represents the cost of particle level deviation. This represents the cost of manufacturing quality for particles.

[0053] The controller detects the consumption at the start of the current prediction time window. .like or less than the preset trace threshold The current state is determined to be cutting and processing; if The current state is determined to be step feeding.

[0054] Combination Figure 4Regarding the first and second weights, this embodiment constructs two weighting strategies. For any moment within the prediction time window, if that moment is in the laser cutting state, the first weighting strategy is used to calculate the weighted sum of the material level deviation cost and the process quality cost. In this strategy, the first weight is determined as the dominant weight, and the value range of the first weight is... The second weight is an auxiliary weight, and its value range is... During laser cutting, because the laser cutting machine consumes less material, the height of the material in the pit is continuously decreasing. Therefore, setting a higher first weight makes the fitness function more sensitive to material level deviation, preventing the material from hitting the bottom.

[0055] For any moment within the prediction time window, if the moment is in the feeding state, the second weighting strategy is used to calculate the weighted sum of the material level deviation cost and the process quality cost. In this strategy, the second weight is determined as the dominant weight, and the first weight is the weakened weight. The value range of the first weight can be... This allows the fitness function to be sensitive to sudden acceleration changes. In this case, the optimization algorithm allows for larger variations in the conveyor height, instead focusing on finding a smooth curve with the smallest rate of change in acceleration. This ensures that the leveler responds with the most compliant posture during vigorous feeding, avoiding stress oscillations and thus reducing hidden waviness.

[0056] S2: Obtain the future velocity sequence with the largest fitness value based on the particle swarm optimization algorithm to control the operation of the leveling machine.

[0057] Specifically, define a particle position vector , representing the prediction time window The algorithm generates a sequence of future velocities for a set of leveling machines. During the algorithm iteration process, each particle updates its velocity and position based on its individual historical best solution and the global historical best solution. The specific update process is a conventional technique in this field and will not be elaborated here.

[0058] The algorithm initializes a preset number of particles, such as After multiple iterations and convergence (e.g., 50 iterations), the velocity curve with the highest fitness value is selected as the optimal control command. The system executes only the first velocity value of this optimal sequence and repeats the above process at the next time step. The method achieves rolling time-domain control. Through this approach, the leveling machine does not completely stop during laser cutting, but rather smoothly reduces its speed and gently accelerates during feeding, effectively solving the lag and impact problems caused by traditional PID control.

[0059] This application also discloses a multi-device collaborative scheduling system for a production line based on particle swarm optimization, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the multi-device collaborative scheduling method for a production line based on particle swarm optimization according to this application.

[0060] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0061] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A multi-equipment collaborative scheduling method for a production line based on particle swarm optimization, applied to an automated production line comprising a leveling machine and a buffer pit connected sequentially; characterized in that, include: Obtain the future velocity sequence composed of multiple velocities of the leveler within the prediction time window, and use the future velocity sequence as the position vector of particles in the particle swarm optimization algorithm; calculate the fitness value of the particles through the objective function in the particle swarm optimization algorithm; The future velocity sequence with the highest fitness value is obtained based on the particle swarm optimization algorithm, and the operation of the leveling machine is controlled accordingly. The step of calculating the fitness value of a particle using the objective function of the particle swarm optimization algorithm includes: calculating the material level deviation cost in the objective function of the particle swarm optimization algorithm, where the material level deviation cost is positively correlated with the difference between the predicted height of the material in the buffer pit and the set safe height at each moment within the prediction time window; calculating the process quality cost in the objective function of the particle swarm optimization algorithm, where the process quality cost is positively correlated with the local process cost at each moment within the prediction time window; the local process cost is positively correlated with the absolute value of the leveler acceleration, the absolute value of the difference between the leveler real-time speed and the optimal process speed; and using the weighted sum of the material level deviation cost and the process quality cost as the fitness value of the particle, where the material level deviation cost and the process quality cost are negatively correlated with the fitness value.

2. The production line multi-equipment collaborative scheduling method based on particle swarm optimization algorithm according to claim 1, characterized in that, The automated production line also includes a laser cutting machine located behind the buffer pit. In the process of weighted summation of material level deviation cost and process quality cost, it also includes: dividing the operating state of the laser cutting machine into laser cutting state and feeding state according to the material consumption sequence that reflects the material consumption rate of the laser cutting equipment. Among them, the state with high material consumption is the feeding state. For any given moment, in response to the laser cutting state at that moment, the first weighted strategy is used to calculate the weighted sum of the material level deviation cost and the process quality cost; For any given moment, in response to the feeding state at that moment, the second weighted strategy is used to calculate the weighted sum of the material level deviation cost and the process quality cost.

3. The production line multi-equipment collaborative scheduling method based on particle swarm optimization algorithm according to claim 2, characterized in that, The steps for obtaining the material consumption sequence within the future prediction time window include: reading the G-code instructions that are not currently executed on the cutting surface of the laser cutting machine; calculating the total path length contained in the G-code instructions; and obtaining the predicted consumption sequence of the laser cutting machine based on the set cutting speed and feeding speed.

4. The production line multi-equipment collaborative scheduling method based on particle swarm optimization algorithm according to claim 2, characterized in that, In the first weighted strategy, the weight corresponding to the material level deviation cost is greater than the weight corresponding to the process quality cost. In the second weighted strategy, the weight corresponding to the material level deviation cost is less than the weight corresponding to the process quality cost.

5. The production line multi-equipment collaborative scheduling method based on particle swarm optimization algorithm according to claim 1, characterized in that, The steps for obtaining the predicted height of the material in the buffer pit include: for any moment in the prediction time window, obtaining the length of the material entering the pit and the length of the material leaving the pit within the corresponding time step; calculating the difference between the length of the material leaving the pit and the length of the material entering the pit, using the ratio of this difference to the preset geometric transformation coefficient of the buffer pit as the height change, and using the sum of the height change and the predicted height of the material in the previous moment as the predicted height of the current moment.

6. The production line multi-equipment collaborative scheduling method based on particle swarm optimization algorithm according to claim 5, characterized in that, The steps for obtaining the geometric transformation coefficient of the buffer pit include: recording the lowest point height of different lengths of material in the buffer pit while the machine is stopped, generating a length-height mapping curve; and using the slope of the mapping curve as the geometric transformation coefficient of the buffer pit.

7. The production line multi-equipment collaborative scheduling method based on particle swarm optimization algorithm according to claim 1, characterized in that, The calculation steps for the material level deviation cost include: for any given moment, calculating the difference between the predicted height of the material in the buffer pit and the set safe height corresponding to each moment within the prediction time window, using this difference as the local deviation cost, and using the sum of the squares of the local deviation costs corresponding to each moment as the material level deviation cost corresponding to the particle.

8. The production line multi-equipment collaborative scheduling method based on particle swarm optimization algorithm according to claim 1, characterized in that, The sum of the squares of the local process costs at each moment within the prediction time window is used as the process quality cost of the particle.

9. The production line multi-equipment collaborative scheduling method based on particle swarm optimization algorithm according to claim 7, characterized in that, The safe height is one-third of the depth of the buffer pit.

10. A multi-equipment collaborative scheduling system for a production line based on particle swarm optimization, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement the multi-device collaborative scheduling method for a production line based on the particle swarm optimization algorithm according to any one of claims 1-9.

Images