A scheduling control method, a scheduling control apparatus, an electronic device, and a storage medium

Abstract

The application provides a scheduling control method, a scheduling control device, electronic equipment and a storage medium, and relates to the technical field of intelligent warehouse management. The method comprises the following steps: generating a plurality of groups of job basis sequences in full permutation form based on Monte Carlo random sampling; for each group of job basis sequences, generating an initial job sequence of each straddle carrier corresponding to the group of job basis sequences, and permuting the terminal tasks of each initial job sequence; taking the rigid demand cycle of the power generation side as a step, performing discrete event simulation on each job sequence after permutation, and screening under the constraint condition of meeting the power generation side single-cycle consumption throughout the process to obtain a candidate job sequence meeting the constraint condition; selecting a sequence meeting a preset optimization index from the candidate job sequence as a final scheduling sequence, and controlling each straddle carrier to perform a job based on the final scheduling sequence. Through the application, the final scheduling sequence is generated, multiple straddle carriers are coordinated to perform a job, congestion is avoided, and the work efficiency is improved.

Description

Technical Field

[0001] This application relates to the field of intelligent warehouse management technology, and more specifically, to a scheduling control method, a scheduling control device, an electronic device, and a storage medium. Background Technology

[0002] Currently, gravity energy storage systems store and retrieve electrical energy by moving heavy objects. Their operation requires coordinating the logistics between the discrete supply side (gantry crane loading and unloading) and the continuous consumption side (generator sets). The supply side has a randomness due to the discrete distribution of operation time caused by different loading and unloading depths; while the demand side requires continuous and stable material supply at a fixed pace, and the system's buffer capacity is limited, making it impossible to smooth out fluctuations through large-scale stockpiling.

[0003] Faced with the aforementioned contradictions, existing scheduling methods often perform poorly. Work plans generated using fixed patterns (such as zigzag or simple incremental sequences) inherently exhibit periodicity, easily leading to "phase resonance" among multiple gantry cranes operating in parallel, causing periodic congestion and supply interruptions. While simple random scheduling strategies can avoid periodicity, they lack effective control over system operating boundaries (especially in the final stages), failing to address the "tail-end supply interruption" problem caused by gantry cranes gradually leaving the site. More critically, existing methods struggle to effectively prevent gantry cranes from being forced to stop and wait due to full buffer zones (i.e., "blockage"), resulting in low utilization of expensive lifting equipment and the system's overall throughput failing to reach its design limits. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a scheduling control method, scheduling control device, electronic device and storage medium to overcome at least one of the above-mentioned defects.

[0005] In a first aspect, embodiments of this application provide a scheduling control method, the method comprising: performing random sampling in the task permutation solution space based on Monte Carlo random sampling to generate multiple sets of operation basic sequences in permutation form; for each set of operation basic sequences, applying a cyclic shift rule to generate an initial operation sequence for each gantry crane corresponding to the set of operation basic sequences, and permuting the end task of each initial operation sequence; performing discrete event simulation on each permuted initial operation sequence with the rigid demand cycle of the power generation side as the step size, and filtering with the constraint of meeting the single-cycle consumption of the power generation side throughout the process as the constraint condition to obtain candidate operation sequences that meet the constraint condition; selecting the sequence that meets the preset optimization index from the candidate operation sequences as the final scheduling sequence, and controlling each gantry crane to perform operations based on the final scheduling sequence.

[0006] In one optional embodiment of this application, multiple sets of random permutation job base sequences are generated in the following manner: the sequence length of the permutation is determined according to the total number of tasks defined in the system model; multiple sets of random permutations that conform to the sequence length are generated using a pseudo-random number generator; and each generated random permutation is used as a job base sequence.

[0007] In one optional embodiment of this application, the initial operation sequence for each gantry crane in the entire field is generated in the following manner: based on a predetermined cyclic shift rule, and based on the operation base sequence and the number of each gantry crane, the task time corresponding to each task in the operation base sequence of that gantry crane is determined; based on the determined task time of each task in the operation base sequence of each gantry crane, the initial operation sequence corresponding to that gantry crane is generated.

[0008] In one optional embodiment of this application, the initial operation sequence for each gantry crane is obtained using the following formula:

[0009] Where Si is the initial operation sequence of the i-th gantry crane, Si(k) represents the time taken for the k-th task in the initial operation sequence of the i-th gantry crane, Sbase represents the basic operation sequence, which is the basic operation order obtained by Monte Carlo random sampling and filtering, k is the task execution sequence number, ranging from 1 to Ntasks, which represents the position of the current task in the sequence, i is the gantry crane number, which is used to determine the phase offset of the gantry crane relative to the basic operation sequence, Ntasks represents the total number of tasks per layer for a single gantry crane, which is the number of tasks that a single gantry crane needs to execute in each layer, (k+i-2) represents the index calculation value based on the gantry crane number and the task sequence number, and mod represents the modulo operation.

[0010] In one optional embodiment of this application, the end tasks of each initial operation sequence are replaced in the following manner: for each gantry crane's initial operation sequence, the last set of tasks responsible for that operation sequence is identified; from all the last set of tasks, a predetermined number of tasks with the shortest execution time are locked; and the locked tasks are swapped to the end of the initial operation sequence.

[0011] In an optional embodiment of this application, candidate job sequences that satisfy the constraints are obtained by: checking the real-time inventory of the buffer at each moment corresponding to the rigid demand cycle in the discrete event simulation; if at any moment the real-time inventory of the buffer is detected to be lower than the single-cycle consumption of the power generation side, the corresponding initial job sequence is eliminated; if at all moments the real-time inventory of the buffer is detected to be not lower than the single-cycle consumption of the power generation side, the initial job sequence is retained as a candidate job sequence.

[0012] In one optional embodiment of this application, the preset optimization indicators include the instantaneous maximum concurrency of the logistics system and the standard deviation of the periodic throughput; The final scheduling sequence is obtained in the following manner: For each candidate job sequence that meets the constraints, the instantaneous maximum concurrency and the standard deviation of the periodic throughput of the logistics system for that candidate job sequence are determined based on the results of its discrete event simulation. Among all candidate job sequences, select one or more candidate job sequences with the smallest instantaneous maximum concurrency of the logistics system; If only one candidate job sequence is obtained after filtering, it will be used as the final scheduling sequence. If multiple candidate job sequences are obtained after screening, the sequence with the smallest periodic throughput standard deviation is selected from the multiple candidate job sequences as the final scheduling sequence.

[0013] Secondly, embodiments of this application also provide a scheduling and control device, the device comprising: a basic task sequence generation module, used to generate multiple sets of basic task sequences in permutation form by randomly sampling within the task permutation solution space based on Monte Carlo random sampling; an initial task sequence replacement module, used to generate an initial task sequence for each gantry crane corresponding to each set of basic task sequences by applying a cyclic shift rule, and replacing the end task of each initial task sequence; a candidate task sequence obtaining module, used to perform discrete event simulation on each replaced initial task sequence with the rigid demand cycle of the power generation side as the step size, and to filter by using the constraint of meeting the single-cycle consumption of the power generation side throughout the process, to obtain candidate task sequences that meet the constraint conditions; and a gantry crane operation module, used to select a sequence that meets a preset optimization index from the candidate task sequences as the final scheduling sequence, and control each gantry crane to perform operations based on the final scheduling sequence.

[0014] Thirdly, embodiments of this application also provide an electronic device, including: a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and when the machine-readable instructions are executed by the processor, the steps of the method described above are performed.

[0015] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the method described above.

[0016] The scheduling control method, scheduling control device, electronic equipment, and storage medium provided in this application are based on Monte Carlo random sampling. Random sampling is performed within the task permutation solution space to generate multiple sets of basic operation sequences in permutation form. For each set of basic operation sequences, a cyclic shift rule is applied to generate an initial operation sequence for each gantry crane corresponding to that set of basic operation sequences, and the end tasks of each initial operation sequence are permuted. Discrete event simulation is performed on each permuted operation sequence using the rigid demand cycle of the power generation side as the step size, and candidate operation sequences that meet the constraint condition of satisfying the single-cycle consumption of the power generation side throughout the entire process are selected. From the candidate operation sequences, the sequence that meets the preset optimization index is selected as the final scheduling sequence, and each gantry crane is controlled to perform operations based on the final scheduling sequence. Through this application, a final scheduling sequence is generated, multiple gantry cranes are coordinated to perform operations, congestion is avoided, and work efficiency is improved.

[0017] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart of the scheduling and control method provided in the embodiments of this application; Figure 2 A flowchart illustrating the generation of multiple sets of random permutations of the basic sequence of tasks provided in this application embodiment; Figure 3 A flowchart illustrating the generation of the initial operation sequence for each overhead crane in the field, provided in an embodiment of this application; Figure 4 A flowchart illustrating the replacement of the end-of-sequence tasks in each initial job sequence, provided in an embodiment of this application; Figure 5 A flowchart illustrating the process of obtaining a candidate job sequence that satisfies the constraints, as provided in an embodiment of this application. Figure 6 This is a flowchart illustrating the process of obtaining the final scheduling sequence provided in an embodiment of this application; Figure 7 This is a schematic diagram illustrating the system inventory level and cycle throughput under the optimized sequence provided in an embodiment of this application. Figure 8This is a schematic diagram illustrating the change of the operating state of the overhead crane over time under the same optimized sequence, as provided in an embodiment of this application. Figure 9 This is a schematic diagram of the scheduling and control device provided in the embodiments of this application; Figure 10 The present application provides a schematic diagram of the structure of an electronic device. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0021] First, the applicable application scenarios of this application will be introduced. This application can be applied to the field of intelligent warehouse management technology.

[0022] Research has shown that gravity energy storage systems store and release electrical energy by mechanically moving solid weights. Its core operational characteristic is a severe nonlinear coupling between the supply and demand sides: 1) Strong randomness and discreteness on the supply side: When gantry cranes retrieve items from storage units at different depths, the time consumed in a single operation is significantly discretely distributed (e.g., from 1×t to N×t), and phase interference is easily generated when multiple gantry cranes operate in parallel.

[0023] 2) Demand rigidity and continuity: Generator sets require a continuous supply of heavy blocks at a strict fixed pace (tdemand). Any interruption in supply at any time will lead to a drop in power generation or shutdown.

[0024] 3) Limited buffer capacity: Due to physical space constraints, the system buffer capacity is small and cannot be used to smooth out long-term production capacity fluctuations through large-scale stockpiling.

[0025] Existing scheduling techniques often struggle to address these contradictions. Traditional mathematically constructed sequences (such as Zig-Zag sequences or simple increasing sequences) inherently possess periodicity, leading to "phase resonance" in gantry cranes and causing periodic capacity congestion and depletion. Meanwhile, stochastic scheduling lacking boundary control cannot cope with capacity decay during the system's final stages, resulting in "tail-end supply disruptions." More importantly, existing methods fail to effectively solve the problem of gantry cranes being forced to stop due to full buffer zones (i.e., "blockage"), resulting in low utilization of expensive gantry crane equipment and the system's overall throughput failing to reach its design limits.

[0026] Based on this, embodiments of this application provide a scheduling control method, a scheduling control device, an electronic device, and a storage medium, which construct a "Monte Carlo random sampling" system. Circular shift decoupling Terminal replacement The "hysteresis control" collaborative scheduling framework solves the problem of mismatch between the discrete randomness of the supply side and the continuous rigidity of the demand side in gravity energy storage systems, which can easily lead to phase resonance and tail-end supply interruption, ensuring zero supply interruption throughout the entire life cycle and maximizing equipment utilization.

[0027] Furthermore, this application provides a gravity energy storage scheduling and control method based on an industrial control computer (such as a PLC, Programmable Logic Controller, or DCS, Distributed Control System), the complete implementation process of which includes an offline optimization stage and an online execution stage.

[0028] The "final scheduling sequence" and its control logic generated in this application can be used by PLC to control the coordinated operation of one or more overhead cranes, or it can be integrated into the DCS of the entire station as an upper-level optimization instruction for unified management and monitoring.

[0029] Please see Figure 1 , Figure 1 A flowchart illustrating the scheduling and control method provided in an embodiment of this application. Figure 1 As shown in the embodiments of this application, the scheduling control method includes: S101. Based on Monte Carlo random sampling, random sampling is performed in the task permutation solution space to generate multiple sets of basic job sequences in permutation form.

[0030] First, system modeling is completed based on the physical design parameters of the gravity energy storage power station. This includes defining the number of gantry cranes (a), the storage depth per column (b), the total number of layers (c), the gantry crane foundation operation time unit (t0), the rigid demand cycle on the power generation side (t1), and the single-cycle consumption M on the power generation side.

[0031] In the embodiments of this application, the number of gantry cranes 'a' refers to the total number of gantry cranes arranged in parallel in the gravity energy storage system, that is, the number of lifting devices that can simultaneously perform the operation of picking up and placing objects, which determines the parallel operation capability of the system.

[0032] The single-column storage depth b represents the total number of storage locations (i.e., the number of depth cells) along the access direction of the gantry crane in a single-column storage cell. This parameter defines the number of independent tasks that a single gantry crane needs to complete.

[0033] The total number of layers, c, represents the number of layers in the vertical direction of the storage structure. Together with the number of columns and the depth, it forms a three-dimensional storage matrix, which determines the total energy storage capacity of the system.

[0034] The gantry crane basic operation time unit t0 refers to the standard time required for the gantry crane to perform one benchmark pick-and-place operation (such as storing or retrieving a block at the shallowest position), and it is the benchmark unit for calculating all operation time.

[0035] The rigid demand cycle t1 on the power generation side refers to the fixed time interval during which the heavy blocks must be supplied on time and continuously when the generator unit is continuously generating electricity. It is a rigid time constraint that the dispatching must meet.

[0036] The single-cycle consumption M on the power generation side refers to the number of heavy blocks that the generator set must consume to maintain rated power output within each rigid demand cycle t1 on the power generation side. It is a key threshold for determining whether the system will experience a power outage.

[0037] The supply side (inside the warehouse) is determined by a, b, c, t0. Its characteristics are that the operation time varies (the operation time increases linearly with the storage and retrieval depth, ranging from the basic operation time t0 to the longest operation time b×t0), and multiple machines (a) are prone to mutual interference.

[0038] The demand side (shipping window) is determined by t1 and M, and its characteristic is that the demand must be timely and sufficient.

[0039] The above parameters are specifically defined by the engineer implementing this application based on the actual physical design and operational requirements of the target gravity energy storage power station.

[0040] This step fundamentally abandons the construction methods of traditional fixed patterns (such as Zig-Zag sequences and simple increasing sequences), avoiding the risk of "phase resonance" caused by the inherent periodicity of the sequence from the source. Through large-scale random sampling, a large number of candidate schemes with both diversity and randomness are provided, laying the foundation for finding high-quality "non-resonance" sequences in a huge solution space.

[0041] This application addresses the enormous (N!) solution space of gantry crane operation sequences by employing the Monte Carlo method for global exploration. This method leverages its ergodicity to perform high-throughput uniform random sampling within the solution space, thereby enabling the discovery of feasible solution regions with excellent aperiodic properties with a very high probability.

[0042] First, a massive (e.g., tens of thousands) number of random permutations are generated to form the initial sample pool of basic job sequences. Then, a discrete event simulation kernel is used to perform a complete system lifecycle simulation for each sequence in the sample pool.

[0043] The simulation process uses "meeting inventory demand throughout the entire process (i.e., inventory is not negative)" as a rigid constraint for feasibility screening. Based on this, the core optimization objective is "minimizing the instantaneous maximum concurrency of the logistics system," selecting the optimal sequence from all feasible solutions. The "non-resonant basic sequence" obtained through this process naturally possesses highly discretized task distribution characteristics in the time domain, thus fundamentally eliminating the risk of in-phase interference and congestion when multiple overhead cranes are operating.

[0044] In the scheduling and control of gravity energy storage systems, each of the above physical design parameters is not an isolated physical quantity. They constitute a stringent, strongly coupled constraint through the global energy conservation principle and the microscopic time-domain continuity requirement.

[0045] The feasible solution space of the system is constrained by the following rigid boundary: Global energy conservation boundary (rigid bottom line) For a system to achieve uninterrupted supply throughout its entire lifecycle, its macro-average supply rate must be greater than or equal to its macro-average consumption rate.

[0046] The theoretical total time Tlayer for a single gantry crane to complete one layer (depth b) of work is calculated as the sum of an arithmetic sequence:

[0047] Where Tlayer is the theoretical total time for a single gantry crane to complete one layer (depth b) of work, k is the task execution sequence number, t0 is the time unit for the basic operation of the gantry crane, and b is the single-column storage depth.

[0048] All a overhead cranes operate in parallel, producing a×b blocks within the Tlayer. Therefore, the average theoretical time interval for producing a single heavy block in the storage area is... for:

[0049] in, Tlayer is the average theoretical time interval for generating a single heavy block in the storage area, Tlayer is the theoretical total time for a single gantry crane to complete one layer (depth b) of work, b is the storage depth of a single column, a is the number of columns of the gantry crane, and t0 is the time unit for the basic operation of the gantry crane.

[0050] Meanwhile, the average time interval between consuming a single heavy block on the power generation side for:

[0051] in, t1 represents the average time interval during which a single heavy block is consumed on the power generation side, t1 represents the rigid demand cycle on the power generation side, and M represents the consumption amount on the power generation side per cycle.

[0052] Numerical substitution and surplus rate calculation: In actual engineering, the average time interval between consuming a single heavy block on the power generation side is considered. Greater than the average theoretical time interval for the generation of a single heavy block in the storage area At the same time, the values ​​of the two should be made as close as possible. This ensures that the theoretical time surplus of the system is low while maintaining sufficient transport capacity, thereby minimizing the wasted time of the gantry crane.

[0053] The theoretical safety margin of a system is defined as follows:

[0054] Where η is the theoretical time surplus rate of the system. The average time interval for consuming a single weight on the power generation side. The average theoretical time interval for generating a single heavy block in the storage area.

[0055] Substitute the specific parameters into this embodiment: The average theoretical time interval for a single heavy block to be generated in the storage area: =8.2×(23+1)÷(2×23)=196.8÷46≈4.278 seconds / time The average time interval between the consumption of a single weight block on the power generation side: =27.86÷6≈4.643 seconds / piece The theoretical time margin of the system (Safety Margin):

[0056] Difficulty analysis conclusion: Extremely tight balance.

[0057] Calculations show that the system is in an "extremely tight balance" state, with only 7.8% time resource tolerance. This means that if the scheduling algorithm causes the average blocking or invalid waiting time of the crane group to exceed 7.8%, the system will inevitably experience a global energy deficit, leading to a final power outage. This explains why simple random scheduling or full-load start-up strategies fail.

[0058] Please refer to Figure 2 , Figure 2 The flowchart for generating multiple sets of random permutations of the basic sequence of jobs provided in the embodiments of this application is as follows: Figure 2 As shown, multiple sets of random permutations of the basic job sequence are generated in the following way: S201. Determine the sequence length of all permutations based on the total number of tasks defined in the system model.

[0059] This step, based on the physical design of the gravity energy storage system, clarifies the scale of the tasks that need to be scheduled.

[0060] During the system parametric modeling phase, physical parameters have been defined: a single-column storage depth of b (e.g., b=23). This parameter b directly determines the number of independent storage locations that a single gantry crane needs to access; in this application, it is defined as the total number of tasks.

[0061] Therefore, the length of the permutation sequence is determined to be equal to the total number of tasks, b. This means that each complete job sequence to be generated will be a permutation containing b elements, each element representing a specific task (such as pick-and-place jobs corresponding to different depths). Determining this length is the basis for all subsequent combinatorial space calculations and random sampling.

[0062] S202. Using a pseudo-random number generator, generate multiple sets of random permutations that conform to the sequence length.

[0063] After determining the sequence length b, a pseudo-random number generator (PRNG) is used as a tool to generate random job sequences on a large scale.

[0064] PRNG is a computational method that uses deterministic algorithms to generate approximate random sequences. It is widely used in computer science for simulation and sampling.

[0065] The algorithm works as follows: It executes the algorithm N times (e.g., N ≥ 2000). In each iteration, PRNG generates a random sequence of integers from 1 to b. This sequence must be a permutation, meaning each integer from 1 to b appears exactly once. This can be achieved using existing techniques such as the Fisher-Yates shuffle algorithm, which efficiently and uniformly shuffles an ordered array (e.g., [1, 2, ..., b]). Ultimately, this step produces N random permutations of length b, each representing a possible order of operations.

[0066] This step abandons traditional mathematical construction methods (such as incremental or Zig-Zag) and instead explores extensively in a huge solution space (b!) through high-throughput random sampling. The large number of random sequence samples (Srand) generated provides a rich candidate pool for subsequent screening, fundamentally avoiding the periodicity defects that may be inherent in the use of deterministic rules.

[0067] S203. Randomly arrange each generated group as a basic sequence for a job.

[0068] Each independent random permutation (i.e., an array of length b) generated in the previous step is formatted and encapsulated as a basic sequence for the job.

[0069] In the system's data representation, this sequence is typically stored as a vector or list, such as Srand[j]=[x1,x2,...,xb], where j represents the j-th sample, and x1 to xb are random permutations of 1 to b. This job base sequence will be considered as a complete, prototype scheduling scheme to be evaluated in subsequent processes, or a candidate job baseline sequence (Sbase).

[0070] S102. For each set of basic operation sequences, apply the cyclic shift rule to generate the initial operation sequence for each gantry crane corresponding to the set of basic operation sequences, and replace the end task of each initial operation sequence.

[0071] This step involves two core operations: topology unrolling and boundary correction.

[0072] First, a topology expansion is performed, using the basic operation sequence generated in the previous step as the baseline sequence. Then, based on the Latin square matrix topology logic, a unique initial operation sequence is generated for each of the *a* gantry cranes in the field. This is achieved through a predetermined Latin square matrix cyclic shift rule. Next, boundary correction (tail permutation) is performed: for each gantry crane's newly generated initial operation sequence, the algorithm scans all tasks in the last layer (layer c) it is responsible for, identifying a predetermined number (e.g., m=3) of tasks with the shortest execution time. Then, these short-duration tasks are forcibly swapped to the very end of the gantry crane's operation sequence, thus constructing a "finishing sprint zone".

[0073] This application addresses the issue of the number of gantry cranes still in operation gradually decreasing during the final stage of system operation. If the original sequence is followed, the total capacity will decrease in a stepwise manner, posing a risk of failing to meet the rigid demand at the end. The principle of this mechanism is to restructure the tail end of the operation sequence, storing the inherent "high-frequency operation capability" of the gantry cranes as "potential energy" and releasing it at the end.

[0074] This mechanism executes automatically after generating the initial job sequences for each gantry crane. The algorithm first parses the job sequence for each gantry crane, identifying the set of tasks it is responsible for located at the bottom level of the storage structure. Then, it evaluates the estimated time consumption of all tasks in this set (typically proportional to access depth) and locks a predetermined number of task units with the shortest consumption time (e.g., equivalent to 1-3 basic time units). Finally, these locked short-time tasks are retrieved from their original positions and sequentially replaced at the end of the gantry crane's job sequence.

[0075] The sequence of tasks is changed only at the last level; it remains unchanged at the earlier levels. The initial task sequence index is 1-Ntasks, and can only contain information for one level.

[0076] The algorithm automatically scans the end of each crane task queue, identifies the few task units with the shortest execution time, and forcibly replaces them with the last stage of the work sequence.

[0077] The "high-frequency operation capability" is reserved for the end of the system's life cycle. When the system enters the final stage, the remaining gantry cranes automatically switch to "sprint mode," using extremely high single-machine turnover rate to compensate for the loss of fewer online units, ensuring that the last batch of heavy blocks can also be delivered on time.

[0078] Through this replacement, each gantry crane will continuously execute a batch of tasks that can be completed as quickly as possible during the final stage of its operational life. When the system enters its final phase, the gantry cranes that have not yet been decommissioned will automatically enter a "high-frequency sprint" mode. By increasing the turnover rate of the remaining individual machines, the total capacity loss caused by the reduction in the number of parallel operating devices is compensated, thereby ensuring that the capacity output can meet the continuous rigid demand of the power generation side throughout the entire cycle from system startup to shutdown, eliminating the possibility of tail-end power outages.

[0079] Please refer to Figure 3 , Figure 3 The flowchart for generating the initial operation sequence for each gantry crane in the entire field, as provided in the embodiments of this application, is as follows: Figure 3 As shown, the initial operation sequence for each gantry crane in the entire field is generated in the following way: S301. Based on the predetermined cyclic shift rules, and based on the basic operation sequence and the number of each gantry crane, determine the task time corresponding to each task in the basic operation sequence of the gantry crane.

[0080] To address the spatial interference problem of multiple gantry cranes operating in parallel, a Latin square topology is used for phase allocation. This topology ensures that the task difficulty (duration) of all gantry cranes in the field is spatially uniformly distributed within any microscopic time window. This allows the overall instantaneous output rate of the system to constantly converge to the average design rate, eliminating macroscopic capacity fluctuations.

[0081] According to the cyclic shift rule, the initial operation sequence for each gantry crane is obtained using the following formula:

[0082] Among them, S i S is the initial operation sequence for the i-th gantry crane. i (k) represents the time taken for the k-th task in the initial operation sequence of the i-th gantry crane, S base This represents the basic sequence of operations, which is the basic order of operations obtained through Monte Carlo random sampling and filtering. k is the task execution sequence number, ranging from 1 to Ntasks, representing the position of the current task in the sequence. i is the gantry crane number, used to determine the phase offset of the gantry crane relative to the basic sequence of operations. Ntasks represents the total number of tasks per layer for a single gantry crane, i.e., the number of tasks that a single gantry crane needs to execute at each layer. (k+i-2) represents the index calculation value based on the gantry crane number and task sequence number, used to achieve phase misalignment. mod represents the modulo operation, used to achieve the cyclic index of the sequence.

[0083] Example, S i (k) represents the task identifier corresponding to the kth task in the initial operation sequence of the i-th gantry crane. The task time can be determined based on this identifier.

[0084] This formula is based on the calculation logic where the index starts from 1; if a system with an index starting from 0 is used, the formula can be adjusted accordingly to (k+i) mod Ntasks.

[0085] Preferably, Si represents the initial operation sequence of the i-th gantry crane, which is a vector composed of the time consumption of all tasks of that gantry crane (such as t0, etc.); Si(k) represents the specific time consumption of the k-th task in the sequence. The variable k is the task execution sequence number, and i is the physical number of the gantry crane. Sbase is the basic operation sequence obtained through Monte Carlo global optimization, serving as the template for generating all gantry crane operation sequences. Ntasks is the total number of tasks, equal to the storage depth b, and also serves as the radix for modular arithmetic with circular indexing. The calculation formula is (k+i... 2) The core of mod Ntasks+1 is to convert the 1-based index to a 0-based index by subtracting 2 for modulo operation, and then convert the result back to a 1-based physical index by adding 1. This formula as a whole implements Sbase-based cyclic shift, generating a phase-shifted operation sequence for each gantry crane, thereby distributing the workload in the time dimension and effectively avoiding congestion and resonance phenomena caused by multi-machine collaborative operation.

[0086] In implementation, the algorithm iterates through every overhead crane in the field (numbered i from 1 to the total number of columns a). For the i-th overhead crane, its corresponding basic operation sequence is used as the base sequence Sbase for this calculation.

[0087] Then, for each task in the gantry crane operation sequence, the sequence number k (k ranges from 1 to the total number of tasks Ntasks, i.e., storage depth b) is used to calculate an index value using the formula above. The mod in the formula is a standard modulo operation, ensuring that the calculation result cycles between 1 and Ntasks. Finally, the calculated value of Si(k) is the task time that the k-th task of the gantry crane (i) should execute (corresponding to which task in the baseline sequence Sbase).

[0088] This step, through defined mathematical rules, transforms a single basic sequence of operations into a set of interconnected but phase-shifted gantry crane-specific sequences. This ensures that at any given moment, the tasks handled by different numbered gantry cranes (in terms of their duration) are staggered within the baseline sequence, thus spatially deconstructing interference from multi-machine parallel operations.

[0089] S302. Based on the time consumption of each task in the determined basic operation sequence of each gantry crane, generate the initial operation sequence corresponding to that gantry crane.

[0090] Here, the discretely calculated task times are combined sequentially according to the order of k from 1 to Ntasks to form a complete and ordered task list.

[0091] This list is defined as the initial operation sequence Si for the i-th gantry crane. In terms of data structure, it is usually represented as a vector or array: Si = [Si(1), Si(2), ..., Si(Ntasks)]. At this point, each of the a gantry cranes in the field has its own independent initial operation sequence based on the same reference but with different phases.

[0092] In one embodiment of this application, a simplified example is given below to intuitively demonstrate the process by which the formula generates the initial job sequence: The assumptions are as follows: number of overhead cranes a=3 (numbered i=1,2,3); total number of tasks Ntasks=4; basic sequence Sbase=[A,B,C,D], where each letter represents a specific task (e.g., A=1t, B=23t, etc.).

[0093] Calculation process: 1) For the first overhead crane 1 (i=1): its task index has no phase offset and directly corresponds to the basic sequence order.

[0094] k=1:(1+1 2) mod 4+1=0+1=1→Sbase(1)=A k=2:(2+1 2) mod 4+1=1+1=2→Sbase(2)=B k=3:(3+1 2) mod 4+1=2+1=3→Sbase(3)=C k=4:(4+1 2) mod 4+1=3+1=4→Sbase(4)=D The resulting sequence S1 = [A, B, C, D] 2) For the second gantry crane (i=2): its task index is cyclically shifted left by 1 bit relative to the basic job sequence.

[0095] k=1:(1+2 2) mod 4+1=1+1=1→Sbase(2)=B k=2:(2+2 2) mod 4+1=2+1=3→Sbase(3)=C k=3:(3+2 2) mod 4+1=3+1=4→Sbase(4)=D k=4:(4+2 2) mod 4+1=4 mod 4+1=0+1=1-Sbase(1)=A The resulting sequence S2 = [B, C, D, A] 3) For the 3rd gantry crane (i=3): its task index is cyclically shifted left by 2 bits relative to the basic job sequence.

[0096] k=1:(1+3 2) mod 4+1=2+1=3-Sbase(3)=C k=2:(2+3 2) mod 4+1=3+1=4-Sbase(4)=D k=3:(3+3 2) mod 4+1=4 mod 4+1=0+1=1-Sbase(1)=A k=4:(4+3 2) mod 4+1=5 mod 4+1=1+1=2-Sbase(2)=B The resulting sequence S3=[C,D,A,B] The generated overall scheduling table (Latin square matrix effect) is shown in Table 1 below: Table 1:

[0097] The table above shows that at any given time (each row of the table), the tasks handled by the three overhead cranes are {A,B,C}, {B,C,D}, etc., which achieves spatial decoupling of task difficulty and avoids logistics congestion caused by all overhead cranes simultaneously handling the most time-consuming task (such as handling B at the same time).

[0098] As can be seen from the above examples, based on the same base sequence Sbase, the cyclic shift rule can generate initial operation sequences with phase misalignment for gantry cranes with different numbers, thereby realizing the discrete distribution of operation tasks in time sequence and laying the foundation for avoiding multi-machine congestion and phase resonance in the future.

[0099] Please refer to Figure 4 , Figure 4 The flowchart for replacing the end tasks of each initial job sequence provided in the embodiments of this application is as follows: Figure 4 As shown, the tasks at the end of each initial job sequence are replaced in the following way: S401. For each gantry crane's initial work sequence, identify the last set of tasks that the work sequence is responsible for.

[0100] Here, the task at the very end of the system storage structure in each gantry crane's work plan can be located.

[0101] During the system parametric modeling phase, the total number of storage layers, c, has been defined. The algorithm traverses the initial operation sequence of every gantry crane in the entire field (this sequence is generated by step S302, where the time consumption of each task corresponds to a specific storage location). For any initial operation sequence, the algorithm needs to parse the physical location (layer, depth) corresponding to the time consumption of each task.

[0102] Specifically, it will filter out all tasks whose corresponding layer number is equal to the total number of layers, c. The set of these tasks constitutes the "last layer task set" for the gantry crane. For example, if c=6, this set contains all tasks for which the gantry crane needs to perform pick-and-place operations from the 6th layer (the deepest or highest layer, depending on the design).

[0103] This step identifies a subset of critical tasks directly relevant to the final stage (closing phase) of system operation. Since the gantry crane will complete its work layer by layer and then leave the site during the later stages of system operation, the completion status of the tasks on the last layer directly determines the sustainability of production capacity in the final phase. Identifying this set provides a clear operational target for subsequent targeted optimization of closing performance.

[0104] S402. From all the last-level task sets, lock the predetermined number of tasks with the shortest execution time.

[0105] After identifying the last layer of tasks for each gantry crane, this step takes different amounts of time depending on the location of the operation, which is determined by the "base time unit t0" and the "depth factor".

[0106] The algorithm needs to evaluate the estimated time for each task in the set (e.g., the task time implies depth information, where depth d corresponds to a time of d×t0). Then, all tasks are sorted in ascending order based on their time.

[0107] Here, depth d is a task-level variable. It refers to the specific row number (depth position) within a single column where a particular task is located. The value of d ranges from 1 to b.

[0108] In this embodiment, the parameter "predetermined quantity" is m (e.g., m=3). The algorithm selects the m tasks with the shortest execution time from the sorted list. These selected tasks are the "short-time tasks" that the crane can complete fastest on the last layer, such as task units corresponding to 1t, 2t, and 3t operation times.

[0109] Here, by proactively identifying and marking the tasks that each gantry crane can complete most easily (fastest) in the final stage, the system can maintain a high output rate by executing these high-frequency tasks during the final stage when the number of gantry cranes decreases.

[0110] S403. Move the locked task to the end of the initial job sequence.

[0111] Here, this application directly modifies the structure of the job sequence.

[0112] For each gantry crane, the algorithm has locked a specific set of short-duration tasks through step S402. Now, the algorithm needs to find the current position of these locked tasks in the initial job sequence of the gantry crane. The task that was originally at the end of the sequence will be swapped to the original position of the locked task with the shortest duration.

[0113] This can be done by manipulating list or array elements. After this operation is completed, the initial job sequence of the gantry crane is updated, and its end portion forms a "finishing sprint zone" consisting of the fastest tasks.

[0114] The initial job sequence is the object to be modified, which is the sequence generated in step S302 and identified in step S401; the updated job sequence is the new sequence formed after the locked tasks in the original sequence are moved to the end.

[0115] This application ensures that each gantry crane, after completing most of its tasks, performs the fastest batch of operations last. When the system enters the final stage, the remaining gantry cranes will automatically enter "sprint mode," using extremely high single-machine turnover rate to compensate for the total capacity loss due to the withdrawal of some gantry cranes, thereby eliminating the "final stage slowdown effect" and providing a key guarantee for achieving "zero supply disruption" at the end of the process under stringent constraints.

[0116] S103. Using the rigid demand cycle of the power generation side as the step size, perform discrete event simulation on each initial operation sequence after replacement, and use the constraint of meeting the single-cycle consumption of the power generation side throughout the process as the screening condition to obtain candidate operation sequences that meet the constraint conditions.

[0117] Traditional control strategies often stop feeding only when the buffer is nearly full, which can easily lead to the gantry crane passively waiting due to lack of space. The principle of this application is to set an optimal start-up threshold (Thopt) significantly lower than the physical upper limit of the buffer, so that the system actively maintains a "non-saturated flow state" for most of its operating time. This state fundamentally eliminates "blockage" events, transforming potential equipment idling time into effective material handling capacity.

[0118] The optimal start-up threshold Thopt is optimized based on the single-cycle consumption M on the power generation side. Its empirical relationship can be expressed as Thopt≈Q×M, where Q is an optimization coefficient determined based on the system buffer capacity and the crane operation speed (e.g., Q=2). This threshold is lower than the maximum buffer capacity.

[0119] For further details, please refer to Figure 5 , Figure 5is a flowchart for obtaining a candidate job sequence that satisfies the constraint conditions provided in the embodiments of the present application, as Figure 5 shown, the candidate job sequence satisfying the constraint conditions is obtained in the following manner: S501. Check the real-time inventory of the buffer at each moment corresponding to the rigid demand period in the discrete event simulation.

[0120] Here, the demand response moment Tk is periodically generated by the simulation clock according to t1, and the real-time buffer inventory I(Tk) is dynamically calculated during the simulation process. Its initial value is usually 0, which increases as the crane completes the transportation event, and decreases as the material taking event of the generator set (taking M pieces in each cycle) proceeds.

[0121] The advancement of the simulation strictly follows the rigid demand period t1 on the power generation side defined by the system parameters.

[0122] Every time the simulation clock advances by t1, it reaches a demand response moment Tk=k×t1 (k is an integer incrementing starting from 1). At each such critical moment Tk, the simulation engine pauses and performs a state check to query a critical state variable: the real-time buffer inventory I(Tk). I(Tk) is dynamically calculated and maintained during the simulation process, and it records the actual number of heavy blocks stored in the buffer at the instant of Tk. This value reflects the net difference between the cumulative output on the supply side (crane transportation) and the cumulative consumption on the demand side (generator set material taking) up to the current moment.

[0123] S502. If it is detected at any moment that the real-time inventory of the buffer is lower than the single-cycle consumption of the power generation side, the corresponding initial job sequence is eliminated.

[0124] After obtaining the real-time buffer inventory I(Tk) at each demand response moment Tk, the algorithm immediately compares it with another key system parameter, the single-cycle consumption M of the power generation side.

[0125] The criterion is: I(Tk)≥M.

[0126] If at any Tk moment, the situation of I(Tk)<M is detected, it means that under this scheduling scheme, the generator set cannot obtain a sufficient number of heavy blocks in the current cycle, that is, a "supply interruption" occurs. Once a supply interruption event is detected, regardless of how the job sequence performs at other moments, the simulation will be terminated early (or marked as failed), and the initial job sequence scheme corresponding to this sequence (and the derived full-site scheduling scheme) will be immediately eliminated and will not participate in any subsequent evaluations.

[0127] This ensures that none of the selected scheduling schemes ever trigger a power outage during its entire lifecycle simulation. This fundamentally guarantees that the schemes meet the minimum requirement of "continuous and stable operation," completely eliminating ineffective or high-risk schemes that could lead to a drop in power generation or shutdown.

[0128] S503. If, at all times, the initial job sequence is found to have a real-time inventory in the buffer that is not lower than the single-cycle consumption of the power generation side, it is retained as a candidate job sequence.

[0129] For a given initial task sequence scheme, if the algorithm traverses all demand response times T1, T2, ..., Tend throughout the entire discrete event simulation cycle (from the start to the completion of all tasks), and checks at each time point that the condition I(Tk)≥M is satisfied, then the scheme has successfully passed the rigid constraint test.

[0130] The simulation engine will fully record the final performance data of the solution (such as the inventory curve) and mark its status as "feasible". All job sequence solutions marked as "feasible" will be collected into a set, which is the candidate job sequence set. These are all qualified alternative solutions that meet the rigid requirement of uninterrupted supply and have completed the full simulation, thus allowing the calculation of subsequent optimization indicators (such as the instantaneous maximum concurrency Pmax of the logistics system and the standard deviation of periodic throughput σ).

[0131] S104. Select the sequence that meets the preset optimization index from the candidate operation sequence as the final scheduling sequence, and control each overhead crane to perform operations based on the final scheduling sequence.

[0132] In this step, after screening all candidate job sequences that meet the rigid constraint (uninterrupted supply) through simulation, the process enters the "optimal solution" stage. The preset optimization indicators are the instantaneous maximum concurrency Pmax and the standard deviation of periodic throughput σ of the logistics system calculated in the previous step, with the goal of finding the most stable logistics solution.

[0133] Here, the preset optimization indicators include the instantaneous maximum concurrency and the standard deviation of the periodic throughput of the logistics system.

[0134] Please refer to Figure 6 , Figure 6 The flowchart for obtaining the final scheduling sequence provided in the embodiments of this application is as follows: Figure 6 As shown, the final scheduling sequence is obtained in the following way: S601. For each candidate job sequence that meets the constraints, determine the instantaneous maximum concurrency and the standard deviation of the periodic throughput of the logistics system for that candidate job sequence based on the results of its discrete event simulation.

[0135] For each candidate sequence in the set, the algorithm calls its complete discrete event simulation result data.

[0136] First, calculate the instantaneous maximum concurrency Pmax of the logistics system. The simulation records the total number of heavy blocks successfully transported to the buffer by all gantry cranes within each generation-side rigid demand cycle t1 (for example, 5 heavy blocks are transported in the first generation-side rigid demand cycle t1, 6 heavy blocks are transported in the second generation-side rigid demand cycle, and so on).

[0137] Pmax is the maximum value of this transportation volume across all t1 periods, reflecting the peak pressure the logistics system needs to withstand. Next, the standard deviation σ of the periodic throughput is calculated. Treating the transportation volume within each t1 period as a sample sequence, the standard deviation σ of these sample values ​​is calculated. σ measures the degree of fluctuation in output across periods; a smaller value indicates more stable logistics.

[0138] S602. Among all candidate job sequences, select one or more candidate job sequences with the smallest instantaneous maximum concurrency of the logistics system.

[0139] First, the peak pressure is minimized. The algorithm iterates through the performance evaluation results of all candidate job sequences and reads their Pmax values. By comparison, the minimum Pmax value is found. Then, all candidate job sequences with Pmax values ​​equal to this minimum value are filtered out to form a better subset. This is because there may be multiple different scheduling schemes whose simulated maximum instantaneous concurrency is exactly the same and is all minimum.

[0140] S603. If only one candidate job sequence is obtained after filtering, then it shall be used as the final scheduling sequence.

[0141] S604. If multiple candidate job sequences are obtained after screening, the sequence with the smallest standard deviation of periodic throughput is selected from the multiple candidate job sequences as the final scheduling sequence.

[0142] If there is only one candidate job sequence in the subset, then it is both the one with the minimum peak pressure and the only optimal solution under the current circumstances, and is therefore directly selected as the final scheduling sequence (Spt).

[0143] If the subset contains multiple candidate job sequences (meaning they are tied for first place in minimizing peak pressure), a secondary optimization objective is needed to determine the winner. In this case, the algorithm reads another performance metric for these tied sequences—the standard deviation of cycle throughput, σ. By comparison, the sequence with the smallest σ value is selected and designated as the final scheduling sequence (Spt). This achieves the complete logic of "selecting the sequence with the smallest Pmax and the smallest σ as the final execution sequence Spt".

[0144] In this way, this application, through a two-tiered optimization criterion with clear priorities (Pmax first, then σ), ensures that even when multiple options are tied on the primary objective, the option with the most stable operation (minimal fluctuation) is automatically selected as the final execution strategy. This makes the final solution not only optimal in dealing with peak pressure but also the best choice in providing continuous and stable logistics output, thereby maximizing the "smoothing of logistics fluctuations".

[0145] In one embodiment of this application, to verify the effectiveness of the scheduling control method, simulation verification was performed based on the following system parameters: the gantry crane array size is 23 columns × 23 storage depth × 6 layers, with a total of 3174 schedulable heavy blocks; the time system parameters are set to a basic operation time unit t0 = 8.2 seconds, corresponding to a maximum operation time of 23t0 = 188.6 seconds; the power generation side demand parameters are a rigid demand cycle t1 = 27.86 seconds, and a single cycle consumption M = 6 units; the system constraint condition is set to a maximum buffer capacity of 24 units.

[0146] The present application will be described in detail below with reference to specific exemplary parameters, but the scope of protection of the present application is not limited to these specific parameters and can be adjusted according to actual needs.

[0147] A basic sequence sample obtained by optimization using the method described in this application: [6 5 19 23 7 9 8 10 22 3 16 14 18 2 11 4 13 17 20 1 15 21 12] The simulation results are as follows Figure 7 As shown. Please refer to [the image / reference]. Figure 7 , Figure 7 This is a schematic diagram illustrating the system inventory level and cycle throughput under the optimized sequence, as provided in an embodiment of this application.

[0148] like Figure 7 As shown, the horizontal axis of the upper chart represents the cycle number, indicating the time step number divided according to the rigid demand cycle t1 of the power generation side during the simulation process; the vertical axis of the upper chart (inventory level chart) represents the real-time inventory level of the buffer, indicating the actual number of heavy blocks stored in the buffer at the corresponding cycle number; Demand is the demand line / consumption line, indicating the number of heavy blocks that the power generation side must consume in each cycle, and its value is determined by M; Max is the maximum capacity of the buffer, that is, the upper limit of the heavy blocks that the buffer can hold. The figure shows the actual change curve of the buffer inventory, the demand line representing the single-cycle consumption M of the power generation side, and the upper limit of the maximum capacity of the buffer (24 in this example).

[0149] like Figure 7As shown, the horizontal axis of the lower chart (throughput graph) also represents the cycle number, and the vertical axis represents the number of heavy blocks (unit: units) successfully transported by the gantry crane to the buffer zone within each generation-side rigid demand cycle t1. Target is the target throughput, which usually corresponds to the generation-side single-cycle consumption M. The graph shows the actual change curve of cycle throughput and the target throughput line (value is M, 6 units in this example). According to the statistical results in the graph below, the average cycle throughput is 5.98 units, the maximum is 12 units, the minimum is 1 unit, and the standard deviation is 2.05, effectively reflecting the instantaneous peak pressure and cycle fluctuation of logistics.

[0150] Simulation results show that under this optimized sequence scheduling, the real-time inventory level in the buffer remains above the demand line throughout the entire process, with no instances of inventory falling below demand. This indicates that the system achieves zero supply interruption throughout the entire process, meeting the rigid constraint requirements. Meanwhile, although the throughput in each cycle exhibits normal fluctuations, its standard deviation (STD=2.05) remains at a low level, indicating good stability in logistics output. The instantaneous maximum throughput (Max=12) is effectively controlled, demonstrating the significant effect of this method in suppressing logistics peaks and avoiding system congestion. These simulation results intuitively verify that the scheduling method proposed in this application can optimize logistics balance while ensuring continuous and stable material supply, thereby improving the overall operating efficiency of the system.

[0151] Please refer to Figure 8 , Figure 8 This is a schematic diagram illustrating the change in the operating state of the overhead crane over time under the same optimized sequence, as provided in an embodiment of this application. Figure 8 As shown, the horizontal axis represents the simulation time in seconds (s), indicating the actual operation time from system startup; the vertical axis represents the crane number, and the color of the horizontal line segment corresponding to different crane numbers indicates their operation status. Among them, the black line segment indicates that the corresponding crane is in the running operation state and is performing pick-up, drop-off or movement operations; the white line segment indicates that the corresponding crane is in the stopped / blocked state, including waiting due to buffer fullness, resource contention or task intermittent.

[0152] Depend on Figure 8 As can be seen, during the simulation period of approximately 14,000 seconds, the operational states of multiple gantry cranes were evenly and interspersed along the time axis, without any prolonged or large-scale concentrated congestion (white segment). This state distribution clearly demonstrates that the scheduling sequence generated by the method in this application effectively avoids operational conflicts and phase overlaps between multiple gantry cranes, enabling the gantry crane group to maintain an efficient and coherent collaborative operation rhythm. Consequently, the invalid waiting time caused by buffer fullness or equipment contention is significantly reduced, thereby improving the overall utilization rate of the gantry crane equipment and the actual throughput efficiency of the system.

[0153] To verify the effectiveness of the method in this application, the following two schemes were set up for simulation comparison: Experimental group: using the scheduling control method described in this application; Control group: using the traditional fixed step size job sequence (such as Z-shaped traversal) scheduling method.

[0154] In the simulation comparison analysis, the experimental group and the control group were set up. After a full-process simulation (approximately 14,800 seconds), the experimental group refers to the scheme using the scheduling control method proposed in this application, namely the complete scheduling process based on Monte Carlo random optimization, cyclic shift generation, end-task replacement, and hysteresis control strategy. The control group refers to existing technologies or traditional scheduling methods, such as fixed job sequences (e.g., Z-shaped or sequential traversal), simple random scheduling, or baseline scheduling strategies without end-task optimization and threshold control.

[0155] Therefore, the data shown in Table 2 compares the key performance indicators of the method of this application (experimental group) with a certain traditional or comparative method (control group) under the full life cycle simulation, highlighting the significant improvement of this application in terms of supply disruption rate, logistics stability, and equipment utilization. The performance of this solution is shown in Table 2: Table 2:

[0156] Based on the system parameters and simulation verification of the foregoing embodiments, the scheduling control method proposed in this application exhibits significant advantages in the following aspects: Significant optimization of logistics load: The aperiodic sequence obtained through Monte Carlo random optimization effectively eliminates the phase resonance phenomenon of multiple overhead cranes operating, stabilizing the instantaneous peak logistics pressure of the system within 12 heavy blocks. This improvement fundamentally avoids scheduling congestion and pressure surges caused by a large number of heavy blocks arriving at the buffer zone in a short period of time, providing a reliable guarantee for a smooth logistics connection from the storage area to the power generation area.

[0157] Zero-disruption supply guarantee during the final stage: Compared to the control group that did not adopt tail optimization (which experienced a supply interruption in the last 200 seconds due to the gantry crane performing a slow task, resulting in output lower than consumption), this application uses a "tail task replacement" mechanism to force the remaining gantry cranes to switch to the task with the shortest execution time (e.g., 1 ton) at the end of system operation. This strategy allows the gantry cranes to enter a "high-speed sprint" state during the final stage, precisely matching the final demand of the power generation side, and perfectly achieving continuous and stable material supply throughout the entire process, until the very last moment.

[0158] System operating efficiency is effectively improved: The dynamic hysteresis control strategy integrated in this application optimizes the buffer feeding trigger threshold to 12 (approximately half of the maximum capacity of 24). Simulations show that, compared to the control group using the full capacity threshold (24), this strategy almost completely avoids the gantry crane entering a blocking waiting state due to a "full buffer." Thus, the system converts all potential time surplus into inventory safety margin, significantly improving the actual utilization rate of the gantry crane and the overall system throughput efficiency.

[0159] The scheduling control method, scheduling control device, electronic equipment, and storage medium provided in this application generate diverse basic operation sequences through Monte Carlo random sampling to avoid periodic congestion at the source; they use cyclic shift rules to generate phase-displaced initial operation sequences for each gantry crane, effectively decoupling multi-machine parallel interference; by replacing the end tasks, a "final sprint zone" is constructed to completely eliminate the risk of tail-end supply interruption; discrete event simulation and constraint screening are then performed with the rigid demand cycle as the step size to ensure that the continuous material supply requirements of the power generation side are met throughout the process; finally, the optimal scheduling sequence is selected based on the dual optimization indicators of the instantaneous maximum concurrency and the standard deviation of the periodic throughput of the logistics system, thereby minimizing the peak pressure of logistics and stabilizing operation.

[0160] Compared with existing technologies where fixed sequences are prone to resonance and random scheduling lacks tail control, this application uses "random sampling" to... Circular shift Terminal replacement Hysteresis simulation The collaborative scheduling framework of "multi-objective optimization" improves the utilization rate of gantry cranes and the overall throughput efficiency of the system while meeting the rigid constraint of zero power outage, and achieves efficient and stable matching between discrete random supply and continuous rigid demand in the gravity energy storage system. It aims to maximize the effective operating efficiency of the gantry crane group and achieve continuous energy supply throughout its entire life cycle while meeting the rigid time window constraint.

[0161] This application belongs to the field of control for large-scale physical energy storage systems. Addressing the spatiotemporal mismatch between the discrete variable-cycle input on the supply side and the continuous rigid output on the demand side of gravity energy storage systems, this application proposes a spatiotemporal collaborative control architecture based on "discrete phase decoupling + boundary potential energy compensation." By constructing a Monte Carlo global stochastic optimization model, non-resonant operation sequences with both low autocorrelation and high throughput stability are selected from the vast solution space. Combined with a Latin square cyclic topology, spatial orthogonality of the gantry crane operation phases is achieved. A tail-end greedy permutation mechanism and a dynamic hysteresis threshold strategy are introduced to effectively eliminate the "capacity reduction" effect at the system's end stage and the "blocking internal friction" during operation. Simulation results show that this method, without increasing redundant equipment, taps the ultimate throughput of the physical system, achieves high-reliability operation with zero supply interruptions throughout its entire lifecycle, and significantly smooths out logistics fluctuations.

[0162] This application achieves extremely high equipment utilization by eliminating overhead crane blockage and operational resonance, ensuring stable material supply from start to finish and overcoming the problem of easy supply interruption in the final stage. Under the premise of meeting the material supply requirements, the peak pressure of logistics can be significantly reduced (by about 50%), thereby reducing the performance requirements of downstream equipment and saving hardware investment and long-term maintenance costs. The scheduling core is based on time parameters rather than specific physical specifications, so it can be quickly adapted to gravity energy storage power stations of different sizes and power, and has strong versatility.

[0163] The technical contributions of this application are not limited to the specific application scenario of gravity energy storage. In essence, it constructs a general scheduling framework and methodology to solve the problem of spatiotemporal coordination and matching between "multi-source discrete random supply" and "rigid continuous flow demand".

[0164] The framework integrates "Monte Carlo global stochastic optimization" Phase cyclic shift decoupling The core architecture of "boundary task substitution compensation" possesses excellent cross-domain generalization capabilities and can be widely applied to any complex logistics or resource scheduling system with finite buffer constraints and high concurrency, discrete event-driven characteristics. Typical potential application scenarios include, but are not limited to: High-end discrete manufacturing and flexible assembly system: Applicable to flexible manufacturing systems, it coordinates multiple automated guided vehicles to deliver materials to automated assembly lines on time, so as to resolve the contradiction between the random fluctuations of the production cycle of upstream processes and the rigid and continuous demand of downstream final assembly lines.

[0165] Automated container terminals and smart hub logistics: Used to coordinate container transfer operations between quay cranes and yard cranes in automated terminals, eliminating operational interference and congestion between equipment through sequence optimization, and improving the overall port throughput efficiency.

[0166] Traffic scheduling in high-performance computing and communication networks: In the fields of computing networks or data communication, it can be used to smoothly schedule data packets or computing tasks under limited buffer queues, avoid buffer overflows, congestion or packet loss caused by sudden traffic, and ensure the stability of service quality.

[0167] In summary, this application provides a general control paradigm for various physical flow (such as materials and containers), information flow (such as data packets), or energy flow (such as heavy blocks) systems facing the challenge of matching "discrete random supply" with "continuous rigid demand," which avoids resonance from the source, ensures supply throughout the entire life cycle, and strives for stable operation. It has broad engineering application prospects and promotional value.

[0168] Based on the same inventive concept, this application also provides a scheduling control device corresponding to the scheduling control method. Since the principle of the device in this application is similar to the scheduling control method described above in this application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0169] Please see Figure 9 , Figure 9 This is a schematic diagram of the scheduling and control device provided in an embodiment of this application. Figure 9 As shown, the scheduling control device 900 includes: The task base sequence generation module 901 is used to generate multiple sets of task base sequences in the form of permutations by performing random sampling in the task permutation solution space based on Monte Carlo random sampling. The initial operation sequence replacement module 902 is used to apply a cyclic shift rule to each group of basic operation sequences to generate an initial operation sequence for each gantry crane corresponding to that group of basic operation sequences, and to replace the end task of each initial operation sequence. The candidate job sequence acquisition module 903 is used to perform discrete event simulation on each of the replaced initial job sequences with the rigid demand cycle of the power generation side as the step size, and to filter them with the constraint of meeting the single-cycle consumption of the power generation side throughout the process, so as to obtain the candidate job sequence that meets the constraint. The overhead crane operation module 904 is used to select a sequence that meets preset optimization indicators from the candidate operation sequences as the final scheduling sequence, and control each overhead crane to perform operations based on the final scheduling sequence.

[0170] Please see Figure 10 , Figure 10 This is a schematic diagram of the structure of the electronic device provided in an embodiment of this application. Figure 10 As shown, the electronic device 1000 includes a processor 1010, a memory 1020, and a bus 1030.

[0171] The memory 1020 stores machine-readable instructions executable by the processor 1010. When the electronic device 1000 is running, the processor 1010 communicates with the memory 1020 via the bus 1030. When the machine-readable instructions are executed by the processor 1010, they can perform the operations described above. Figure 1 The specific implementation of the scheduling control method in the illustrated method embodiment can be found in the method embodiment, and will not be repeated here.

[0172] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described actions. Figure 1 The specific implementation of the scheduling control method in the illustrated method embodiment can be found in the method embodiment, and will not be repeated here.

[0173] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0174] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0175] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0176] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0177] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (RM), random access memory (RAM), magnetic disks, or optical disks.

[0178] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A scheduling and control method, characterized in that, include: Based on Monte Carlo random sampling, random sampling is performed in the task permutation solution space to generate multiple sets of basic job sequences in permutation form. For each set of basic operation sequences, apply the cyclic shift rule to generate the initial operation sequence for each gantry crane corresponding to the set of basic operation sequences, and replace the end task of each initial operation sequence. Using the rigid demand cycle of the power generation side as the step size, discrete event simulations are performed on each of the replaced initial operation sequences, and candidate operation sequences that meet the constraints of satisfying the single-cycle consumption of the power generation side throughout the process are selected. From the candidate operation sequences, a sequence that meets the preset optimization index is selected as the final scheduling sequence, and each overhead crane is controlled to perform operations based on the final scheduling sequence; The initial operation sequence for each overhead crane in the entire field is generated in the following way: Based on the predetermined cyclic shift rules, and based on the basic operation sequence and the number of each gantry crane, the task time corresponding to each task in the basic operation sequence of the gantry crane is determined; and based on the determined task time in the basic operation sequence of each gantry crane, the initial operation sequence corresponding to the gantry crane is generated. The initial operation sequence for each gantry crane is obtained using the following formula: Where Si is the initial operation sequence of the i-th gantry crane, Si(k) represents the time taken for the k-th task in the initial operation sequence of the i-th gantry crane, Sbase represents the basic operation sequence, which is the basic operation order obtained by Monte Carlo random sampling and filtering, k is the task execution sequence number, which ranges from 1 to Ntasks, representing the position of the current task in the sequence, i is the gantry crane number, which is used to determine the phase offset of the gantry crane relative to the basic operation sequence, Ntasks represents the total number of tasks per layer for a single gantry crane, which is the number of tasks that a single gantry crane needs to execute in each layer, (k+i-2) represents the index calculation value based on the gantry crane number and the task sequence number, and mod represents the modulo operation; The tasks at the end of each initial job sequence are replaced using the following method: For each gantry crane's initial work sequence, identify the last set of tasks that the work sequence is responsible for; From all the last-level task sets, lock in the predetermined number of tasks with the shortest execution time; The locked task is swapped to the very end of the initial job sequence.

2. The method according to claim 1, characterized in that, Generate multiple sets of random permutations of the basic sequence of the job using the following method: Determine the sequence length of all permutations based on the total number of tasks defined in the system model; Using a pseudo-random number generator, multiple sets of random permutations conforming to the specified sequence length are generated; Each generated group is randomly arranged and used as a base sequence for a job.

3. The method according to claim 1, characterized in that, The candidate job sequence that satisfies the constraints is obtained in the following way: At each moment corresponding to the rigid demand cycle in the discrete event simulation, check the real-time inventory of the buffer. If at any of the aforementioned times, it is detected that the real-time inventory of the buffer is lower than the single-cycle consumption of the power generation side, then the corresponding initial operation sequence is eliminated. If, at all the aforementioned times, the initial job sequence is detected to have a real-time inventory in the buffer that is not lower than the single-cycle consumption of the power generation side, it is retained as a candidate job sequence.

4. The method according to claim 3, characterized in that, The preset optimization indicators include the instantaneous maximum concurrency of the logistics system and the standard deviation of the periodic throughput; The final scheduling sequence is obtained in the following manner: For each candidate job sequence that meets the constraints, the instantaneous maximum concurrency and the standard deviation of the periodic throughput of the logistics system for that candidate job sequence are determined based on the results of its discrete event simulation. Among all candidate job sequences, select one or more candidate job sequences with the smallest instantaneous maximum concurrency of the logistics system; If only one candidate job sequence is obtained after filtering, it will be used as the final scheduling sequence. If multiple candidate job sequences are obtained after screening, the sequence with the smallest periodic throughput standard deviation is selected from the multiple candidate job sequences as the final scheduling sequence.

5. A scheduling and control device, characterized in that, include: The task base sequence generation module is used to generate multiple sets of task base sequences in the form of permutations by performing random sampling in the task permutation solution space based on Monte Carlo random sampling. The initial work sequence replacement module is used to apply a cyclic shift rule to each set of basic work sequences to generate an initial work sequence for each gantry crane corresponding to that set of basic work sequences, and to replace the end tasks of each initial work sequence. Specifically, the initial work sequence for each gantry crane in the entire field is generated as follows: based on a predetermined cyclic shift rule, and based on the basic work sequence and the number of each gantry crane, the task time corresponding to each task in the basic work sequence of that gantry crane is determined; based on the determined task time of each task in the basic work sequence of each gantry crane, the initial work sequence corresponding to that gantry crane is generated. The initial operation sequence for each gantry crane is obtained using the following formula: Where Si is the initial operation sequence of the i-th gantry crane, Si(k) represents the time taken for the k-th task in the initial operation sequence of the i-th gantry crane, Sbase represents the basic operation sequence, which is the basic operation order obtained by Monte Carlo random sampling and filtering, k is the task execution sequence number, which ranges from 1 to Ntasks, representing the position of the current task in the sequence, i is the gantry crane number, which is used to determine the phase offset of the gantry crane relative to the basic operation sequence, Ntasks represents the total number of tasks per layer for a single gantry crane, which is the number of tasks that a single gantry crane needs to execute in each layer, (k+i-2) represents the index calculation value based on the gantry crane number and the task sequence number, and mod represents the modulo operation; The tasks at the end of each initial work sequence are replaced in the following way: For each gantry crane's initial work sequence, identify the last set of tasks that the work sequence is responsible for; from all the last set of tasks, lock the predetermined number of tasks with the shortest execution time; and swap the locked tasks to the end of the initial work sequence. The candidate job sequence acquisition module is used to perform discrete event simulation on each of the replaced initial job sequences with the rigid demand cycle of the power generation side as the step size, and to filter them with the constraint of meeting the single-cycle consumption of the power generation side throughout the process, so as to obtain the candidate job sequence that meets the constraint. The overhead crane operation module is used to select a sequence that meets preset optimization indicators from the candidate operation sequences as the final scheduling sequence, and control each overhead crane to perform operations based on the final scheduling sequence.

6. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is in operation, the processor communicates with the memory via the bus, and the processor executes the machine-readable instructions to perform the steps of the method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the method as described in any one of claims 1 to 4.

Images