Multi-objective optimization driven coal preparation plant product loading intelligent scheduling method and system
By constructing a loading node diagram for coal preparation plants, calculating the concentration and fluctuation correlation of flow using historical flow data, and dynamically adjusting the propagation weight, the problem of multi-objective optimization in loading scheduling of coal preparation plants is solved, achieving more efficient scheduling and energy utilization.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- YULIN SHENHUA ENERGY CO LTD
- Filing Date
- 2026-03-04
- Publication Date
- 2026-06-05
AI Technical Summary
Existing coal preparation plant product loading scheduling methods typically focus on a single optimization objective, which cannot effectively cope with factors such as fluctuations in loading demand, limited track resources, conflicting work sequences, and difficulties in equipment coordination. This leads to increased path congestion and equipment wear and tear, and fails to achieve multi-objective optimization.
By constructing a loading node graph structure, calculating the traffic concentration, fluctuation correlation, and conflict risk index using historical traffic data, dynamically adjusting the propagation weight of the scheduling path, and employing the Dijkstra algorithm for multi-objective optimization for path planning, real-time scheduling optimization is achieved.
It improved overall scheduling efficiency, reduced conflicts and waiting time, maximized energy utilization efficiency, and enhanced the level of intelligent scheduling.
Smart Images
Figure CN122155245A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of industrial scheduling optimization, specifically to a multi-objective optimization-driven intelligent scheduling method and system for loading coal preparation plant products. Background Technology
[0002] In the field of coal preparation plant product loading and scheduling, traditional scheduling methods typically focus on a single optimization objective, such as simply pursuing the shortest transportation distance or the shortest operation time. However, loading and scheduling in actual production environments is a complex systems engineering project, and its optimization objectives are inherently multi-objective, often contradictory or restrictive. For example, improving loading and unloading efficiency may conflict with reducing equipment energy consumption, and shortening the single transportation time may conflict with balancing the overall system load. Therefore, the concept of multi-objective optimization-driven approaches has significant application value in this field. It emphasizes that the decision-making of the scheduling system should be driven by multiple optimization objectives, rather than relying on the control logic of a single objective, and requires the optimization algorithm to effectively weigh and coordinate these objectives to find the optimal or suboptimal solution set that meets the actual needs.
[0003] In practical applications, the loading points, transportation route intersections, and destinations of coal preparation plants are often abstracted as nodes in a graph structure, and the transportation routes connecting these nodes are abstracted as edges, thus transforming the scheduling problem into a path planning problem on the graph. Dijkstra's algorithm, as a classic single-source shortest path algorithm, is widely used for basic path optimization in such scenarios. The basic steps of this algorithm include: initializing node distances, iteratively selecting the nearest unvisited node, updating the distances of its adjacent nodes, until the shortest path for all nodes is determined.
[0004] However, applying the standard Dijkstra algorithm directly to dynamic loading scheduling in coal preparation plants encounters significant problems. Due to factors such as fluctuating loading demand, limited track resources, conflicting work sequences, and difficulties in equipment coordination, the actual travel costs (such as time and energy consumption) of transportation routes are not static but dynamically changing. For example, when real-time loading demand surges on a certain scheduling route, using static weights as the "shortest path" will lead to congestion on that route, reducing overall loading and unloading efficiency and increasing equipment wear and tear. In this case, the route weights (or costs) need to be dynamically adjusted according to real-time conditions, a capability that traditional static algorithms lack. Summary of the Invention
[0005] This invention provides a multi-objective optimization-driven intelligent scheduling method and system for loading coal preparation plant products, in order to solve existing problems.
[0006] The multi-objective optimization-driven intelligent scheduling method for coal preparation plant product loading of the present invention adopts the following technical solution:
[0007] One embodiment of the present invention provides a multi-objective optimization-driven intelligent scheduling method for product loading in coal preparation plants, the method comprising the following steps:
[0008] Construct a loading node diagram structure for the coal preparation plant, and for each scheduling path in the loading node diagram structure, obtain its historical flow data corresponding to each time interval;
[0009] Based on historical traffic data, the traffic concentration of each scheduling path is calculated in each time interval. The traffic concentration is used to characterize the possibility of loading conflicts occurring on the scheduling path in the corresponding time interval.
[0010] For each scheduling path, the traffic concentration in each time interval is combined with the traffic concentration in the adjacent time interval to obtain the traffic concentration sequence of the scheduling path in that time interval.
[0011] Calculate the sequence similarity between the traffic concentration sequences of adjacent scheduling paths in the same time interval to obtain the correlation degree of traffic fluctuation of each scheduling path in each time interval;
[0012] The system conflict risk index for the target time interval is calculated based on the sum of the correlation of traffic fluctuations of all scheduling paths in the target time interval and the average of the traffic concentration.
[0013] Based on the historical traffic data and system conflict risk index of each scheduling path in the target time interval, the real-time conflict contribution of each scheduling path is calculated.
[0014] Based on the real-time conflict contribution of each scheduling path, calculate its corresponding real-time propagation weight, update the propagation weight of each scheduling path in the loading node graph structure to the real-time propagation weight of the corresponding scheduling path, and use the updated propagation weight to calculate and output the optimal loading scheduling path through the path planning algorithm.
[0015] Optionally, the historical traffic data corresponding to each time interval includes traffic values for at least two scheduling dates, and the scheduling dates for the different time intervals are the same. The step of calculating the traffic concentration for each scheduling path in each time interval based on the historical traffic data specifically includes:
[0016] For the a-th scheduling path, obtain the sum of the traffic values for the b-th scheduling date in each time interval to get the total daily load intensity of the path on the b-th scheduling date for the a-th scheduling path;
[0017] The ratio of the flow value on the b-th scheduling date in the t-th time interval to the total daily load intensity of the path is determined as the interval load ratio coefficient of the b-th scheduling path in the t-th time interval on the b-th scheduling date.
[0018] The average value of the interval load ratio coefficient of all scheduling dates of the a-th scheduling path in the t-th time interval is calculated to obtain the traffic concentration of the a-th scheduling path in the t-th time interval.
[0019] Optionally, for each scheduling path, the traffic concentration in each time interval is combined with the traffic concentration in adjacent time intervals to obtain the traffic concentration sequence of that scheduling path in that time interval, specifically including:
[0020] For the a-th scheduling path, the traffic concentration in the (t-1)-th, t-th, and t+1-th time intervals is combined according to time sequence to obtain the traffic concentration sequence of the a-th scheduling path in the t-th time interval.
[0021] Optionally, the sequence similarity between the traffic concentration sequences of adjacent scheduling paths in the same time interval is calculated to obtain the correlation of traffic fluctuations of each scheduling path in each time interval, specifically including:
[0022] Calculate the Pearson correlation coefficient between the traffic concentration sequence of the a-th scheduling path in the t-th time interval and the traffic concentration sequence of the (a+1)-th scheduling path in the t-th time interval. Use the Pearson correlation coefficient as the sequence similarity to obtain the traffic fluctuation correlation degree of the a-th scheduling path in the t-th time interval.
[0023] Optionally, based on the sum of the traffic fluctuation correlation of all scheduling paths in the target time interval and the average traffic concentration, a system conflict risk index for the target time interval is calculated, specifically including:
[0024] The average traffic fluctuation correlation of all scheduling paths in the target time interval is calculated to obtain the mean traffic fluctuation correlation of all scheduling paths in the target time interval.
[0025] Normalize the traffic concentration of all scheduling paths in the target time interval, and then average the normalized values to obtain the average traffic concentration of all scheduling paths in the target time interval.
[0026] The system conflict risk index for the target time interval is determined by multiplying the sum of the traffic fluctuation correlation of all scheduling paths in the target time interval by the average traffic concentration.
[0027] Optionally, based on the historical traffic data and system conflict risk index of each scheduling path within the target time interval, the real-time conflict contribution of each scheduling path is calculated, specifically including:
[0028] Obtain the traffic value of the a-th scheduling path in the target time interval from the historical traffic data and normalize it to obtain the normalized traffic value of the a-th scheduling path in the target time interval.
[0029] The ratio of the normalized traffic value of the a-th scheduling path in the target time interval to the system conflict risk index of the target time interval is determined as the real-time conflict contribution of the a-th scheduling path.
[0030] Optionally, based on the real-time conflict contribution of each scheduling path, its corresponding real-time propagation weight is calculated, specifically including:
[0031] The reciprocal of the real-time conflict contribution of the a-th scheduling path is used to obtain the initial real-time propagation weight of the a-th scheduling path.
[0032] The initial real-time propagation weight of each scheduling path is obtained and normalized to obtain the real-time propagation weight of each scheduling path. The sum of the real-time propagation weights of all scheduling paths is 1.
[0033] Optionally, the path planning algorithm is the shortest path algorithm.
[0034] Optionally, the shortest path algorithm is Dijkstra's algorithm, and the real-time propagation weights correspond to the edge weights in Dijkstra's algorithm.
[0035] This invention proposes a multi-objective optimization-driven intelligent scheduling system for loading coal preparation plant products, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of the multi-objective optimization-driven intelligent scheduling method for loading coal preparation plant products.
[0036] The beneficial effects of the technical solution of the present invention are:
[0037] In this embodiment of the invention, by analyzing historical traffic data and real-time status awareness, a traffic concentration quantity reflecting the possibility of scheduling conflicts and a traffic fluctuation correlation degree characterizing the collaborative status between paths are constructed. Then, a dynamic system conflict risk index and real-time conflict contribution are calculated, and finally, the real-time propagation weights of each scheduling path in the optimization algorithm are dynamically adjusted accordingly. This method realizes the transformation of scheduling strategies from static to dynamic, and from single-objective to multi-objective collaboration, aiming to optimize overall scheduling efficiency, reduce conflicts and waiting, and ultimately maximize energy utilization efficiency and improve the level of intelligent scheduling. Attached Figure Description
[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0039] Figure 1 A flowchart of a multi-objective optimization-driven intelligent scheduling method for loading coal preparation plant products according to an embodiment of the present invention;
[0040] Figure 2 This is a structural diagram of a multi-objective optimization-driven intelligent scheduling system for loading coal preparation plant products, provided in one embodiment of the present invention. Detailed Implementation
[0041] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the multi-objective optimization-driven intelligent scheduling method for loading coal preparation plant products according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.
[0042] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0043] The following description, in conjunction with the accompanying drawings, details the specific scheme of the multi-objective optimization-driven intelligent scheduling method for loading coal preparation plant products provided by this invention.
[0044] This invention provides a multi-objective optimization-driven intelligent scheduling method and system for loading coal preparation plant products. Please refer to [link / reference]. Figure 1 The diagram illustrates a flowchart of a multi-objective optimization-driven intelligent scheduling method for loading coal preparation plant products, according to an embodiment of the present invention. The method includes the following steps:
[0045] S101. Construct the loading node diagram structure of the coal preparation plant, and for each scheduling path in the loading node diagram structure, obtain its historical flow data corresponding to each time interval.
[0046] For example, the process of constructing the loading node diagram structure of a coal preparation plant is as follows:
[0047] Key physical operation points within the coal preparation plant are abstracted as nodes in a graph structure. These nodes include product loading stations, track scales, key turnouts, marshalling yards, and entrances / exits to storage areas. The actual feasible transportation routes connecting these nodes are abstracted as edges in the graph, with each edge corresponding to a scheduling path. An initial weight is preset for each edge, which can be determined based on at least one of the following factors: physical distance of the path, baseline travel time, or designed transportation capacity. Based on the definitions of nodes and edges, a weighted graph G=(V, E, W) is constructed, where V is the set of nodes, E is the set of edges, and W is the set of initial weights for each edge. This weighted graph represents the topological connection relationship of the coal preparation plant's scheduling network, where the scheduling path is the edge in the loading node graph structure, and the optimal loading path is the complete path from the starting point to the destination.
[0048] Based on the established loading node graph structure, all possible valid scheduling paths in the graph can be calculated using path planning algorithms (such as Dijkstra's algorithm), which are connected sequences from any loading start point to the destination. Static parameters are recorded for each scheduling path, including path ID, start point, end point, sequence of nodes along the route, theoretical transport time, and designed transport capacity, forming a basic path information database.
[0049] Historical traffic data from multiple complete operating cycles was extracted from databases such as the coal preparation plant's production management system and train dispatch logs. For each dispatch route, the data was organized into fixed time intervals (e.g., every 15 minutes), forming historical traffic data containing different dispatch dates.
[0050] Optionally, the length of the time interval here can be set according to actual needs. There are no specific restrictions here, and it can be changed in real time according to the actual transportation needs of the coal preparation plant.
[0051] By deploying sensors (such as track occupancy sensors and RFID readers) at key nodes, GPS devices, or interfacing with the real-time dispatch system, the system continuously acquires information on the number of trains passing through each dispatch path, their load capacity, real-time location, and task status.
[0052] Historical data and real-time data streams are aligned, cleaned, and integrated according to scheduling paths and time dimensions to construct a unified structured data source, namely historical traffic data. This ensures that for any scheduling path and any time interval, its historical traffic sequence and current real-time traffic observations can be obtained. Data preprocessing includes noise reduction, missing value imputation, and normalization operations to ensure the accuracy of subsequent analysis.
[0053] Historical traffic data can be in the following form: For the a-th scheduling path, its real-time traffic observation value in the t-th time interval is denoted as... Over a time interval x, a historical sequence of traffic values spanning multiple scheduling dates can be denoted as... .
[0054] S102. Based on historical traffic data, calculate the traffic concentration of each scheduling path in each time interval. The traffic concentration is used to characterize the possibility of loading conflicts occurring on the scheduling path in the corresponding time interval.
[0055] In this embodiment, the historical traffic data corresponding to each time interval includes traffic values for at least two scheduling dates, and the scheduling dates for different time intervals are the same. Based on the historical traffic data, the traffic concentration of each scheduling path in each time interval is calculated, specifically including:
[0056] For the a-th scheduling path, obtain the sum of the traffic values for the b-th scheduling date in each time interval to get the total daily load intensity of the path on the b-th scheduling date for the a-th scheduling path;
[0057] The ratio of the flow value on the b-th scheduling date in the t-th time interval to the total daily load intensity of the path is determined as the interval load ratio coefficient of the b-th scheduling path in the t-th time interval on the b-th scheduling date.
[0058] The average value of the interval load ratio coefficient of all scheduling dates of the a-th scheduling path in the t-th time interval is calculated to obtain the traffic concentration of the a-th scheduling path in the t-th time interval.
[0059] For example, in a real production environment, the core challenge faced by coal preparation plants in loading scheduling lies in the significant inconsistency in loading flow across different scheduling paths at different time periods. This inconsistency directly leads to problems such as conflicting work sequences and difficulties in equipment coordination, resulting in decreased loading and unloading efficiency, path congestion, and even system blockage. Simply relying on static path planning algorithms cannot adapt to this dynamically changing and complex scenario.
[0060] To address the aforementioned conflicts, this embodiment starts with historical data analysis, calculating traffic concentration to understand the load variation patterns of a single path across different time intervals. However, understanding the patterns of a single path alone is insufficient to address systemic conflicts. This is because when multiple paths within the system are under high load simultaneously, any local adjustment to a single path is unlikely to effectively alleviate overall congestion.
[0061] For example, the loading operations at a coal preparation plant can be analyzed using a calendar day as a basic analysis cycle (scheduling date). For any given scheduling path:
[0062] Within the same day, the traffic volume of this path may vary across different time intervals. For example, traffic volume during morning and evening peak hours may be significantly higher than during off-peak hours. Within the same time interval, the traffic volume of this path may also differ on different scheduled dates (i.e., different days) due to factors such as production plans and market fluctuations. Traffic concentration refers to a situation where, within a specific time interval, the traffic volume of the path remains consistently higher than in other time intervals of the day, and this "peak" characteristic exhibits statistical stability across different dates.
[0063] Based on the above analysis, the traffic concentration of the scheduling path a in the t-th time interval is... The calculation formula can be:
[0064]
[0065] in, This represents the traffic concentration of the a-th scheduling path in the t-th time interval; This represents the historical traffic value of the b-th scheduling date of the a-th scheduling path in the t-th time interval; This represents the total number of time intervals within a day. For example, if each interval is 15 minutes, then T = 96. This represents the b-th scheduling date. This indicates the total number of days in the scheduling period, which is the total number of days in the historical data. This represents the total daily load intensity of the 'a'-th scheduling path on the 'b'-th scheduling date, where 'j' is the index of the time interval. This represents the load ratio coefficient of the traffic of the a-th scheduling path on the b-th scheduling date and the t-th time interval to the total traffic of that path on that day. It quantifies the concentration of traffic in a specific time interval within a single day. j is the index of the time interval, from 1 to T.
[0066] In the formula, the value of traffic concentration ranges from 0 to 1, directly reflecting the degree of concentration of historical traffic for that path within a given time interval. The closer the value is to 1, the more it indicates that, based on historical data, the traffic for that path consistently accounts for a very high proportion of the total daily traffic in the t-th time interval. In other words, this time interval is a clear and stable peak period, and based on historical patterns, the likelihood of scheduling conflicts occurring at this time is also higher. Conversely, the closer the value is to 0, the more it indicates that the historical traffic share for that time interval is very low, and the path is relatively idle.
[0067] Optionally, during actual scheduling operations, the system needs to make dynamic decisions based on real-time data. For the current real-time scheduling date (taking December 12, 2025 as an example), when calculating its real-time conflict contribution, it is necessary to obtain the traffic data for the current time interval t and its subsequent time intervals. However, at the point in time for real-time calculation (such as the 5th time interval of the day), there is actually no observed data for future time intervals (the 6th to the 96th time intervals).
[0068] Based on this, this embodiment provides an optional data supplementation strategy to ensure the integrity of the real-time scheduling date traffic data sequence, as follows:
[0069] Taking December 12, 2025 (the real-time scheduling date) as an example, assume that we are currently in the t-th time interval of that day (e.g., the 5th time interval). At this time, the system has obtained the real-time traffic observation values for the 1st to t-th time intervals of that day, but there is actually no data for the t+1 to T-th time intervals (e.g., the 6th to 96th time intervals).
[0070] For the (t+1)th to (T)th time interval (i.e., future periods after the current time), the historical traffic values for the corresponding time interval are supplemented using the most recently fully recorded scheduling date (e.g., the previous day, December 11, 2025). For example:
[0071] The 6th time interval of the real-time scheduling date (December 12, 2025) uses the traffic value of the 6th time interval of the historical scheduling date (December 11, 2025);
[0072] The 7th time interval of the real-time scheduling date uses the traffic value of the 7th time interval of the historical scheduling date;
[0073] This process continues until all T time intervals are completed.
[0074] The formula for the supplementary strategy can be shown below:
[0075] set up To schedule the observations for the current time interval t of the current date in real time, Given the historical value of the j-th time interval of the previous scheduling date, then the final traffic value of the j-th time interval of the real-time scheduling date. It can be defined as:
[0076]
[0077] in, .
[0078] This strategy can construct a complete real-time scheduling date traffic sequence without affecting the timeliness of real-time decision-making. Supplementing with data from the most recent historical dates considers both the continuity of production scheduling and the relative stability of the data pattern, thus ensuring the rationality and reliability of subsequent calculations of parameters such as real-time conflict contribution.
[0079] S103. For each scheduling path, combine the traffic concentration in each time interval with the traffic concentration in the adjacent time interval to obtain the traffic concentration sequence of the scheduling path in that time interval.
[0080] In this embodiment, for each scheduling path, the traffic concentration in each time interval is combined with the traffic concentration in adjacent time intervals to obtain the traffic concentration sequence of the scheduling path in that time interval, specifically including:
[0081] For the a-th scheduling path, the traffic concentration in the (t-1)-th, t-th, and t+1-th time intervals is combined according to time sequence to obtain the traffic concentration sequence of the a-th scheduling path in the t-th time interval.
[0082] For example, to achieve quantitative analysis of the synergistic effect of traffic fluctuations between paths, this embodiment first needs to construct a contextual representation of the traffic concentration of each path in the time dimension. Furthermore, in this embodiment, the scheduling paths are sorted according to their path length.
[0083] Specifically, for the a-th scheduling path, the traffic concentration sequence for its t-th time interval is constructed as follows:
[0084] Collect the concentrated flow values of the path at three consecutive time intervals: t-1, t, and t+1.
[0085] Arrange these three concentrated flow values in chronological order to form an ordered sequence containing three elements, represented as:
[0086]
[0087] in, This represents the sequence of concentrated traffic volume for the a-th scheduling path in the t-th time interval. These represent the traffic concentrations calculated in advance for the path during the three time intervals of t-1, t, and t+1, respectively.
[0088] By constructing the above-mentioned traffic concentration sequence, not only are the load characteristics of the path recorded in the current time interval, but also... It also captured its state in the previous moment. The load state and its load characteristics at the next moment. .
[0089] Optionally, when constructing the traffic concentration volume sequence, for the time intervals in special boundary cases (i.e., the first and last time intervals of each day), specific processing methods need to be adopted to ensure the integrity of the data sequence and the consistency of calculations.
[0090] The specific processing rules are as follows:
[0091] For the first time interval (t = 1): When t is the first time interval, there is no previous time interval t - 1. At this time, the traffic concentration volume value of the current time interval is used to fill the front and back positions of the sequence to form a complete traffic concentration volume sequence:
[0092]
[0093] Among them, represents the traffic concentration volume sequence of the a-th scheduling path in the first time interval. The first and second elements of the sequence are both filled with (the traffic concentration volume of the current time interval), and the third element is filled with the traffic concentration volume of the next time interval .
[0094] For the last time interval (t = T): When t is the last time interval, there is no subsequent time interval t + 1. At this time, the traffic concentration volume value of the current time interval is used to fill the front and back positions of the sequence to form a complete traffic concentration volume sequence:
[0095]
[0096] Among them, represents the traffic concentration volume sequence of the a-th scheduling path in the T-th time interval. The first element of the sequence is filled with the traffic concentration volume of the previous time interval , and the second and third elements are both filled with (the traffic concentration volume of the current time interval).
[0097] For the intermediate time intervals (1 < t < T): For the time intervals that are neither the first nor the last, construct the traffic concentration volume sequence containing three consecutive time intervals in the conventional way:
[0098]
[0099] By employing the aforementioned boundary processing strategy, it is ensured that a traffic concentration sequence of length 3 can be generated for all time intervals (including boundary time intervals). Through appropriate numerical padding, the complete structure of the sequence is maintained while fully considering the special characteristics of the boundary time intervals. This approach ensures that all time intervals have the same dimensional input characteristics when subsequently calculating the correlation of traffic fluctuations, thereby guaranteeing consistency and comparability in the calculations. Particularly at the beginning and end of each day, this padding strategy effectively avoids system anomalies caused by missing data, while maintaining the continuity and smoothness of the time series, thus enhancing the time-dimensional analysis capabilities of the entire scheduling system.
[0100] S104. Calculate the sequence similarity between the traffic concentration sequences of adjacent scheduling paths in the same time interval to obtain the traffic fluctuation correlation of each scheduling path in each time interval.
[0101] In this embodiment, the sequence similarity between the traffic concentration sequences of adjacent scheduling paths in the same time interval is calculated to obtain the traffic fluctuation correlation of each scheduling path in each time interval, specifically including:
[0102] Calculate the Pearson correlation coefficient between the traffic concentration sequence of the a-th scheduling path in the t-th time interval and the traffic concentration sequence of the (a+1)-th scheduling path in the t-th time interval. Use the Pearson correlation coefficient as the sequence similarity to obtain the traffic fluctuation correlation degree of the a-th scheduling path in the t-th time interval.
[0103] For example, after obtaining the traffic concentration sequence for each scheduling path, it is necessary to quantify the similarity of traffic fluctuation patterns between different paths. The similarity calculation is as follows:
[0104] For adjacent paths a-th and a+1-th, the formula for calculating the similarity between their traffic concentration sequences in the same target time interval t (i.e., the correlation of traffic fluctuations of the a-th scheduling path in the t-th time interval) can be:
[0105]
[0106] in, This represents the similarity value between path a and path (a+1). This represents the sequence of concentrated traffic flow for path a in the t-th time interval. This represents the sequence of concentrated traffic volume for the (a+1)th path in the t-th time interval. This indicates the calculation of the Pearson correlation coefficient between two sequences.
[0107] In the formula, The value range of is [-1, 1]. The closer the value is to 1, the more consistent the trends of the traffic concentration series changes between the two paths are, meaning the traffic fluctuation patterns of the two paths are highly similar; when... The closer the value is to -1, the more opposite the trends of the flow concentration sequences of the two paths are; when A value close to 0 indicates that the traffic fluctuation patterns of the two paths are independent and have no significant correlation. This is achieved by calculating the correlation between all adjacent paths. This value provides a metric system that reflects the degree of coordination in traffic fluctuations between paths throughout the entire scheduling network.
[0108] S105. Based on the sum of the correlation of traffic fluctuations of all scheduling paths in the target time interval and the average of the traffic concentration, calculate the system conflict risk index for the target time interval.
[0109] In this embodiment, the system conflict risk index for the target time interval is calculated based on the sum of the traffic fluctuation correlation of all scheduling paths in the target time interval and the average traffic concentration, specifically including:
[0110] The average traffic fluctuation correlation of all scheduling paths in the target time interval is calculated to obtain the mean traffic fluctuation correlation of all scheduling paths in the target time interval.
[0111] Normalize the traffic concentration of all scheduling paths in the target time interval, and then average the normalized values to obtain the average traffic concentration of all scheduling paths in the target time interval.
[0112] The system conflict risk index for the target time interval is determined by multiplying the sum of the traffic fluctuation correlation of all scheduling paths in the target time interval by the average traffic concentration.
[0113] For example, based on the obtained correlation of traffic fluctuations along each path, the system conflict risk index is calculated by comprehensively evaluating the overall traffic concentration and inter-path coordination of the system.
[0114] In this embodiment, the t-th time interval is regarded as the target time interval, and the formula for calculating the system conflict risk index of the target time interval can be:
[0115]
[0116] in, This represents the system conflict risk index in the t-th time interval. This represents the average traffic concentration across all paths in the t-th time interval. This indicates the total number of scheduling paths. This represents the similarity value between path a and path (a+1). This represents the sum of the correlation of traffic fluctuations across all scheduled paths within the target time interval.
[0117] In the formula, Before the calculation, the original traffic concentration of all paths in the t-th time interval needs to be normalized. The normalization method can be the min-max normalization method, which is an existing technology and will not be elaborated on here. It integrates information from two dimensions: the concentration of traffic on a single path and the coordination of traffic across multiple paths.
[0118] S106. Based on the historical traffic data and system conflict risk index of each scheduling path in the target time interval, calculate the real-time conflict contribution of each scheduling path.
[0119] In this embodiment, based on the historical traffic data and system conflict risk index of each scheduling path within the target time interval, the real-time conflict contribution of each scheduling path is calculated, specifically including:
[0120] Obtain the traffic value of the a-th scheduling path in the target time interval from the historical traffic data and normalize it to obtain the normalized traffic value of the a-th scheduling path in the target time interval.
[0121] The ratio of the normalized traffic value of the a-th scheduling path in the target time interval to the system conflict risk index of the target time interval is determined as the real-time conflict contribution of the a-th scheduling path.
[0122] For example, the production operations of coal preparation plants have significant periodicity and event-based characteristics, which directly affect the spatiotemporal distribution of loading flow:
[0123] For example, 8:00-10:00 and 16:00-18:00 are typically peak transportation periods, corresponding to the concentrated arrival and departure times of transport vehicles. Operational arrangements such as production line handovers and shift changes further exacerbate traffic density during peak hours. Planned activities such as equipment maintenance and repairs reduce the capacity of certain routes during specific periods. Traffic flow is closely related to coal supply and shipping cycles; significant peak flow occurs during raw material arrivals and product shipments. Changes in the pace of coal mining and processing directly affect fluctuations in transportation demand. Unpredictable events such as sudden transportation tasks and temporary scheduling arrangements can disrupt regular traffic flow patterns.
[0124] Based on the above production characteristics, the calculation of real-time conflict contribution needs to comprehensively consider two dimensions: historical patterns and real-time status.
[0125] Taking the t-th time interval as the target time interval, the formula for calculating the real-time conflict contribution of the a-th scheduling path is:
[0126]
[0127] in, This represents the real-time conflict contribution of the a-th scheduling path in the current t-th time interval.
[0128] The normalized traffic value of the scheduling path a in the t-th time interval. It is calculated based on the actual monitored traffic data of the t-th time interval, reflecting the actual load status of the a-th scheduling path in the t-th time interval.
[0129] For the a-th scheduling path: Let represent the normalized traffic value of the 'a'-th scheduling path in the 't'-th time interval. This represents the system conflict risk index in the t-th time interval. When... When the value is close to 1, it indicates that, based on historical patterns, the system as a whole is in a state of high conflict risk during the current time interval; when When the value is close to 1, it indicates that the real-time traffic of the a-th scheduling path in the t-th time interval is close to saturation. Therefore, when When the value is close to 1, it means that the a-th scheduling path is facing dual pressure in the current t-th time interval: the system as a whole is in a high-conflict-risk environment and the path itself has reached a high-load state.
[0130] When the system is experiencing high real-time traffic and low system risk ( The value is relatively large. When the value is small, at this time A significant increase in the value indicates that the current path is experiencing high real-time traffic, while other paths in the system are relatively idle. In this case, it is necessary to reduce the Dijkstra propagation weight of this path to avoid over-allocating scheduling tasks to high-load paths and to guide tasks to idle paths. A higher value indicates that the path conflict risk exceeds the system average, and its selection probability should be appropriately reduced. When the system is experiencing low real-time traffic and high system risk (…),… The value is small. When the value is relatively large, at this time The value decreases significantly, indicating that the real-time traffic on the current path is low, but the overall system is in a high-collision state. In this case, it is necessary to appropriately increase the Dijkstra propagation weight of this path to effectively utilize the traffic offloading capacity of the low-load path when the overall system is under pressure. A lower value indicates that the path is relatively idle, and its scheduling priority should be appropriately increased.
[0131] S107. Based on the real-time conflict contribution of each scheduling path, calculate its corresponding real-time propagation weight, and update the propagation weight of each scheduling path in the loading node graph structure to the real-time propagation weight of the corresponding scheduling path. Using the updated propagation weight, calculate and output the optimal loading scheduling path through the path planning algorithm.
[0132] In this embodiment, the real-time propagation weight is calculated based on the real-time conflict contribution of each scheduling path, specifically including:
[0133] The reciprocal of the real-time conflict contribution of the a-th scheduling path is used to obtain the initial real-time propagation weight of the a-th scheduling path.
[0134] The initial real-time propagation weight of each scheduling path is obtained and normalized to obtain the real-time propagation weight of each scheduling path. The sum of the real-time propagation weights of all scheduling paths is 1.
[0135] The path planning algorithm in this embodiment is the shortest path algorithm, which is Dijkstra's algorithm. The real-time propagation weights correspond to the edge weights or costs in Dijkstra's algorithm.
[0136] For example, taking the t-th time interval as the target time interval, the formula for calculating the Dijkstra real-time propagation weight of the a-th path in the t-th time interval based on the real-time conflict contribution can be:
[0137]
[0138] in, This represents the Dijkstra real-time propagation weight of the a-th scheduling path in the t-th time interval. This represents the real-time conflict contribution of the a-th scheduling path in the current t-th time interval. This represents the normalization function.
[0139] The specific steps for performing dynamic weight adjustment and obtaining the optimal scheduling path allocation result are as follows:
[0140] In the t-th time interval, the original fixed propagation weights of each scheduling path in the coal preparation plant loading node graph structure are updated to the real-time propagation weights calculated for that path. When selecting the optimal scheduling path, Dijkstra's algorithm changes the weight of each path (edge) from its original baseline value (e.g., 1) to a dynamic real-time propagation weight. Based on this updated weight set, the algorithm calculates the minimum cost path from the starting point to the ending point, thus obtaining the optimal scheduling path allocation result for the current time node. This process is repeated in each time interval. In each new time interval t (e.g., every 15 minutes), the system recalculates based on the latest real-time and historical data. And then update It outputs a new optimal scheduling path allocation result, thereby achieving continuous dynamic optimization of loading scheduling.
[0141] In the above embodiments, the propagation weight value and the real-time conflict contribution are negatively correlated: the larger the conflict coefficient value, the higher the real-time conflict contribution of the path, and the smaller the calculated propagation weight value. Consequently, the probability of this path being selected as the optimal path in Dijkstra's algorithm also decreases. This effectively prevents high-conflict paths from being frequently scheduled, guiding traffic flow to relatively idle paths, achieving the intelligent scheduling goal of dynamically balancing system load and avoiding local congestion.
[0142] In summary, this invention, through historical traffic data analysis and real-time status awareness, constructs a traffic concentration quantity reflecting the possibility of scheduling conflicts and a traffic fluctuation correlation degree characterizing the collaborative state between paths. This allows for the calculation of a dynamic system conflict risk index and a real-time conflict contribution, ultimately dynamically adjusting the real-time propagation weights of each scheduling path in an optimization algorithm (such as Dijkstra's algorithm). This method realizes a transformation in scheduling strategies from static to dynamic, and from single-objective to multi-objective collaboration, aiming to optimize overall scheduling efficiency, reduce conflicts and waiting times, and ultimately maximize energy utilization efficiency and improve the level of intelligent scheduling.
[0143] This invention also proposes a multi-objective optimization-driven intelligent scheduling system for loading coal preparation plant products. Please refer to [link / reference]. Figure 2 The diagram shows a structural diagram of a multi-objective optimization-driven intelligent scheduling system for loading coal preparation plant products, provided by an embodiment of the present invention. The system includes: a data acquisition module 101, a data processing module 102, and a path output module 103.
[0144] The data acquisition module 101 is used to construct the loading node diagram structure of the coal preparation plant and to acquire the historical flow data corresponding to each scheduling path in the loading node diagram structure in each time interval.
[0145] Data processing module 102 is used to calculate the traffic concentration of each scheduling path in each time interval based on historical traffic data, wherein the traffic concentration is used to characterize the possibility of loading conflict occurring on the scheduling path in the corresponding time interval.
[0146] For each scheduling path, the traffic concentration in each time interval is combined with the traffic concentration in the adjacent time interval to obtain the traffic concentration sequence of the scheduling path in that time interval.
[0147] Calculate the sequence similarity between the traffic concentration sequences of adjacent scheduling paths in the same time interval to obtain the correlation degree of traffic fluctuation of each scheduling path in each time interval;
[0148] The system conflict risk index for the target time interval is calculated based on the sum of the correlation of traffic fluctuations of all scheduling paths in the target time interval and the average of the traffic concentration.
[0149] Based on the historical traffic data and system conflict risk index of each scheduling path in the target time interval, the real-time conflict contribution of each scheduling path is calculated.
[0150] The path output module 103 is used to calculate the real-time propagation weight of each scheduling path based on the real-time conflict contribution, update the propagation weight of each scheduling path in the loading node graph structure to the real-time propagation weight of the corresponding scheduling path, and use the updated propagation weight to calculate and output the optimal loading scheduling path through the path planning algorithm.
[0151] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer equipment can be divided into different functional modules to complete all or part of the functions described above. In addition, the multi-objective optimization-driven intelligent scheduling system for loading coal preparation plant products and the multi-objective optimization-driven intelligent scheduling method for loading coal preparation plant products provided in the above embodiments belong to the same concept. The specific implementation process is detailed in the method embodiments and will not be repeated here.
[0152] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0153] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.
[0154] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A multi-objective optimization-driven intelligent scheduling method for product loading in coal preparation plants, characterized in that, include: Construct a loading node diagram structure for the coal preparation plant, and for each scheduling path in the loading node diagram structure, obtain its historical flow data corresponding to each time interval; Based on historical traffic data, the traffic concentration of each scheduling path is calculated in each time interval. The traffic concentration is used to characterize the possibility of loading conflicts occurring on the scheduling path in the corresponding time interval. For each scheduling path, the traffic concentration in each time interval is combined with the traffic concentration in the adjacent time interval to obtain the traffic concentration sequence of the scheduling path in that time interval. Calculate the sequence similarity between the traffic concentration sequences of adjacent scheduling paths in the same time interval to obtain the correlation degree of traffic fluctuation of each scheduling path in each time interval; The system conflict risk index for the target time interval is calculated based on the sum of the correlation of traffic fluctuations of all scheduling paths in the target time interval and the average of the traffic concentration. Based on the historical traffic data and system conflict risk index of each scheduling path in the target time interval, the real-time conflict contribution of each scheduling path is calculated. Based on the real-time conflict contribution of each scheduling path, calculate its corresponding real-time propagation weight, update the propagation weight of each scheduling path in the loading node graph structure to the real-time propagation weight of the corresponding scheduling path, and use the updated propagation weight to calculate and output the optimal loading scheduling path through the path planning algorithm.
2. The intelligent scheduling method for loading coal preparation plant products driven by multi-objective optimization according to claim 1, characterized in that, The historical traffic data corresponding to each time interval includes traffic values for at least two scheduling dates, and the scheduling dates for different time intervals are the same. The calculation of the traffic concentration for each scheduling path in each time interval based on the historical traffic data specifically includes: For the a-th scheduling path, obtain the sum of the traffic values for the b-th scheduling date in each time interval to get the total daily load intensity of the path on the b-th scheduling date for the a-th scheduling path; The ratio of the flow value on the b-th scheduling date in the t-th time interval to the total daily load intensity of the path is determined as the interval load ratio coefficient of the b-th scheduling path in the t-th time interval on the b-th scheduling date. The average value of the interval load ratio coefficient of all scheduling dates of the a-th scheduling path in the t-th time interval is calculated to obtain the traffic concentration of the a-th scheduling path in the t-th time interval.
3. The intelligent scheduling method for loading coal preparation plant products driven by multi-objective optimization according to claim 1, characterized in that, For each scheduling path, the traffic concentration in each time interval is combined with the traffic concentration in adjacent time intervals to obtain the traffic concentration sequence of that scheduling path in that time interval, specifically including: For the a-th scheduling path, the traffic concentration in the (t-1)-th, t-th, and t+1-th time intervals is combined according to time sequence to obtain the traffic concentration sequence of the a-th scheduling path in the t-th time interval.
4. The intelligent scheduling method for loading coal preparation plant products driven by multi-objective optimization according to claim 1, characterized in that, The calculation of the sequence similarity between the traffic concentration sequences of adjacent scheduling paths in the same time interval, to obtain the traffic fluctuation correlation of each scheduling path in each time interval, specifically includes: Calculate the Pearson correlation coefficient between the traffic concentration sequence of the a-th scheduling path in the t-th time interval and the traffic concentration sequence of the (a+1)-th scheduling path in the t-th time interval. Use the Pearson correlation coefficient as the sequence similarity to obtain the traffic fluctuation correlation degree of the a-th scheduling path in the t-th time interval.
5. The intelligent scheduling method for loading coal preparation plant products driven by multi-objective optimization according to claim 1, characterized in that, The calculation of the system conflict risk index for the target time interval, based on the sum of the correlation of traffic fluctuations across all scheduling paths within the target time interval and the average of the traffic concentration, specifically includes: The average traffic fluctuation correlation of all scheduling paths in the target time interval is calculated to obtain the mean traffic fluctuation correlation of all scheduling paths in the target time interval. Normalize the traffic concentration of all scheduling paths in the target time interval, and then average the normalized values to obtain the average traffic concentration of all scheduling paths in the target time interval. The system conflict risk index for the target time interval is determined by multiplying the sum of the traffic fluctuation correlation of all scheduling paths in the target time interval by the average traffic concentration.
6. The intelligent scheduling method for loading coal preparation plant products driven by multi-objective optimization according to claim 1, characterized in that, The calculation of the real-time conflict contribution of each scheduling path based on historical traffic data and system conflict risk index within the target time interval specifically includes: Obtain the traffic value of the a-th scheduling path in the target time interval from the historical traffic data and normalize it to obtain the normalized traffic value of the a-th scheduling path in the target time interval. The ratio of the normalized traffic value of the a-th scheduling path in the target time interval to the system conflict risk index of the target time interval is determined as the real-time conflict contribution of the a-th scheduling path.
7. The intelligent scheduling method for loading coal preparation plant products driven by multi-objective optimization according to claim 1, characterized in that, The calculation of the real-time propagation weight based on the real-time conflict contribution of each scheduling path specifically includes: The reciprocal of the real-time conflict contribution of the a-th scheduling path is used to obtain the initial real-time propagation weight of the a-th scheduling path. The initial real-time propagation weight of each scheduling path is obtained and normalized to obtain the real-time propagation weight of each scheduling path. The sum of the real-time propagation weights of all scheduling paths is 1.
8. The intelligent scheduling method for loading coal preparation plant products driven by multi-objective optimization according to claim 1, characterized in that, The path planning algorithm is the shortest path algorithm.
9. The intelligent scheduling method for loading coal preparation plant products driven by multi-objective optimization according to claim 8, characterized in that, The shortest path algorithm is Dijkstra's algorithm, and the real-time propagation weights correspond to the edge weights in Dijkstra's algorithm.
10. A multi-objective optimization-driven intelligent scheduling system for loading coal preparation plant products, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent scheduling method for loading coal preparation plant products driven by multi-objective optimization as described in any one of claims 1-9.