Container terminal cost analysis system based on activity-based costing and big data analysis

Abstract

The present application relates to the technical field of port operation cost accounting, in particular to a container terminal cost analysis system based on operation cost method and big data analysis, comprising: an operation data acquisition preprocessing unit; an operation cost hierarchical collection accounting unit, which carries out cost collection on standardized operation data set based on an improved operation flow topology cross consumption coefficient incremental iteration algorithm, constructs a resource-operation-cost object three-level dynamic mapping relationship; an accounting result storage and compliance verification unit; a terminal cost classification summary and information display unit. The present application relies on real-time consumption data of operation, adopts adaptive learning rate and normalized incremental iteration to realize dynamic tracking of allocation coefficient; at the same time, based on operation deviation absolute value and weight configuration, it completes the accurate attribution of idle consumption, solves the problems of insufficient adaptability of coefficient and idle suspension.

Description

Technical Field

[0001] This invention relates to the field of port operation cost accounting technology, and more specifically, to a container terminal cost analysis system based on activity-based costing and big data analysis. Background Technology

[0002] As a core node in the global supply chain, port container terminals are undergoing automation and intelligent upgrades, generating massive amounts of heterogeneous data from multiple sources through their internal operating systems, equipment control systems, and energy management systems. Against this backdrop, refined cost control has become a rigid requirement for terminal operation and management. Activity-based costing (ABC), as an accounting system that accurately maps resource consumption to operations and cost objects, provides the theoretical foundation and practical framework for lean cost analysis at the single container and single operation line level in the port industry. Current mainstream technical solutions generally rely on a hardware and software collaborative architecture consisting of terminal IoT sensor networks, production management systems, and enterprise resource planning systems to complete the collection of operational data and cost allocation calculations.

[0003] One existing, widely implemented solution is an activity-based costing system (ABC) based on static cost driver rates. This system pre-sets resource consumption coefficients for each activity based on historical periodic data, and at the end of the accounting period, it aggregates the total cost of the resource pool and allocates it to each activity and cost object according to a fixed ratio. Its advantages include a clear model structure, controllable implementation costs, and suitability for scenarios with relatively small business fluctuations. Another solution is a dynamic allocation system based on real-time activity metering. This system collects real-time resource consumption data through energy and material metering devices deployed on key equipment and allocates costs based on physical rules or engineering experience formulas. Its advantages include high accuracy in calculating direct energy consumption costs and the ability to meet monthly or ten-day cost analysis needs.

[0004] The aforementioned technical solutions have revealed common industry challenges in practice, namely, the difficulty in simultaneously achieving dynamic adaptability of cost allocation coefficients and accurate attribution of idle losses. Specifically, this manifests in two problems: First, it is difficult to achieve an effective balance between static benchmarks and real-time operating conditions for cost allocation coefficients. Static solutions cannot respond to the slow drift in resource consumption patterns caused by factors such as adjustments in operational processes, equipment aging, and environmental changes, resulting in a decrease in allocation accuracy over time. While real-time metering solutions can capture some fluctuations in direct consumption, the lack of a complete data loop for allocation coefficient calibration stems from the difficulty in comprehensively installing independent metering terminals for indirect resources. Second, there is a lack of accurate causal attribution mechanisms for idle consumption during equipment downtime. Waiting times caused by fluctuations in operational plans are difficult to accurately trace back to specific operational stages based on plan deviation responsibility, leading to some idle costs becoming unallocated overhead costs, or forcing the adoption of simple average allocation, reducing the attribution accuracy of single-container cost accounting. These problems result in coarse-grained cost accounting and difficulty in tracing abnormal consumption during project implementation, limiting the terminal's data support capabilities for operational process optimization and pricing decisions. In view of this, we propose a container terminal cost analysis system based on activity-based costing and big data analysis. Summary of the Invention

[0005] The purpose of this invention is to provide a container terminal cost analysis system based on activity-based costing and big data analysis, so as to solve the problems of insufficient dynamic adaptability of cost allocation coefficient and lack of causal attribution of empty-load consumption mentioned in the background art.

[0006] To address the aforementioned technical problems, the present invention aims to provide a container terminal cost analysis system based on activity-based costing and big data analysis, comprising: The operation data acquisition and preprocessing unit collects raw operation data from all operations at the terminal, performs basic normalization processing on the raw operation data, and outputs a standardized operation dataset. The activity cost hierarchical aggregation and accounting unit performs cost aggregation on the standardized activity dataset based on the improved activity flow topology cross consumption coefficient incremental iterative algorithm, constructs a three-level dynamic mapping relationship of resource-activity-cost object, completes the accurate splitting and allocation calculation of activity cross consumption and topology gap idle consumption, and generates the activity-level initial cost accounting result. The calculation result storage and compliance verification unit receives the initial cost calculation results at the job level, completes the persistent storage of data and basic compliance verification, and outputs compliant cost data. The terminal cost classification and information display unit performs routine classification and summary of compliant cost data, and completes the visualization presentation of cost information according to the terminal business dimension.

[0007] As a further improvement to this technical solution, the basic normalization processing of the job data acquisition preprocessing unit specifically includes: The system performs data deduplication, time-series alignment, field format normalization, and invalid data removal operations on the collected multi-source raw operational data of the dock, generating a standardized operational dataset with unified fields and time-series matching.

[0008] As a further improvement to this technical solution, the hierarchical cost aggregation and accounting unit includes a workflow topology construction module, a cross-consumption coefficient matrix management module, an incremental iteration calculation module, a topology gap idle consumption processing module, and a three-level cost allocation execution module, wherein: The workflow topology construction module constructs a directed acyclic graph topology of the dock workflow based on a standardized workflow dataset, completing the basic mapping between atomic operation links and resource consumption flow. The cross-consumption coefficient matrix management module initializes and dynamically maintains the cross-consumption coefficient matrix, storing the proportion of various resources consumed by the corresponding atomic operation links at each time. The incremental iterative operation module listens for the completion trigger event of the atomic job and performs incremental update operation on the cross-consumption coefficient matrix based on the actual resource consumption data of the job. The topology gap idle consumption processing module identifies the time gap between adjacent atomic operation links of the same device and completes the accurate mapping of gap idle consumption data to the corresponding operation links; The three-level cost allocation execution module completes the layer-by-layer allocation calculation of resource costs to operational links and cost objects based on the updated cross-consumption coefficient matrix.

[0009] As a further improvement to this technical solution, the workflow topology construction module includes a job process decomposition submodule, a topology node generation submodule, and a topology edge generation submodule, wherein: The operation process decomposition submodule parses the standardized operation dataset and extracts the complete container operation process, breaking it down into indivisible atomic operation links; The topology node generation submodule maps each atomic operation step to a directed acyclic graph topology node and binds the node with a unique device identifier, operation start and end timestamps, and operation quantity parameters. The topology edge generation submodule matches the corresponding consumption relationship between terminal production resources and atomic operation links. It constructs unidirectional topology directed edges with production resources as the starting point and atomic operation links as the ending point, so that each directed edge uniquely corresponds to a type of terminal consumption resource, thereby representing the exclusive consumption support relationship of resources for operation links.

[0010] As a further improvement to this technical solution, the cross-consumption coefficient matrix management module includes a matrix dimension determination submodule, a matrix initialization submodule, and a matrix dynamic maintenance submodule, wherein: The matrix dimension determination submodule retrieves the directed acyclic graph topology of the dock operation flow generated by the operation flow topology construction module, and uses the total number of the consumed resource types as the number of matrix rows and the total number of nodes corresponding to the atomic operation links as the number of matrix columns to determine the row and column dimensions of the dynamic cross-consumption coefficient matrix. The matrix initialization submodule retrieves historical operational consumption statistics of the dock, calculates the average consumption ratio of various resources allocated to each type of atomic operation link, and uses it as the initial value of the matrix. The matrix dynamic maintenance submodule receives incremental update instructions, updates the matrix element values ​​in real time, and synchronously stores the latest version of the matrix.

[0011] As a further improvement to this technical solution, the incremental update operation process of the incremental iteration operation module includes the following steps: S23.1 Real-time capture of event trigger signals indicating the completion of a single atomic operation step; S23.2 Collect the actual resource consumption data of this atomic operation link, calculate the actual value of unit operation resource consumption, and at the same time calculate the predicted value of unit operation resource consumption of this atomic operation link based on the current cross-consumption coefficient matrix. S23.3 Calculate the sum of squares of the deviations between the predicted and actual unit operation resource consumption values, and solve for the gradient of the loss function; S23.4. An adaptive learning rate that dynamically decreases with the number of completed tasks is used to perform incremental corrections on the cross-consumption coefficient matrix element by element. S23.5 Perform proportional normalization on the cross-consumption coefficient matrix after incremental correction so that the sum of the cross-consumption coefficients of all atomic operation links corresponding to the same type of resource is 1. S23.6. Synchronize the normalized cross-consumption coefficient matrix to the matrix dynamic maintenance submodule to complete this iteration update.

[0012] As a further improvement to this technical solution, the process of accurately mapping the gap idle consumption of the topology gap idle consumption processing module includes the following steps: S24.1. Based on the start and end timestamps of the jobs bound to the topology nodes, extract the time interval between two adjacent atomic job segments on the same device; S24.2 When the time interval is greater than the standard operation switching time of the dock equipment, it is determined to be an effective idle gap, and the total idle consumption generated by the equipment within the gap is collected. S24.3 Calculate the deviation between the actual execution time and the planned execution time of the previous atomic operation step. And the deviation between the actual execution time and the planned execution time of the next atomic operation. ; S24.4 According to the cause-and-effect attribution rule of work plan deviation, the total idle consumption is allocated to the two atomic work links before and after in proportion. The allocation ratio is positively correlated with the absolute value of the corresponding deviation value. The allocation ratio of different working conditions such as early work and delayed work can be distinguished by preset weight. S24.5. Add the allocated idle consumption to the consumption of the corresponding atomic operation link to obtain the total consumption of the atomic operation link. S24.6 Synchronize the total consumption data of the atomic operation to the incremental iterative calculation module, triggering the correction of the corresponding cross-consumption coefficient matrix elements.

[0013] As a further improvement to this technical solution, the three-level cost allocation execution module includes a resource-level allocation submodule, an activity-level aggregation submodule, and a cost object-level summary submodule, wherein: The resource layer allocation submodule retrieves the final cross-consumption coefficient matrix after incremental iteration and idle loss correction, and allocates the total cost of various resources to the corresponding atomic operation links. The task layer aggregation submodule summarizes the self-consumption and allocated idle consumption of each atomic task step to generate the total cost of the atomic task steps. The cost object layer aggregation submodule aggregates the costs of each atomic operation into the corresponding container cost object, constructs a three-level dynamic mapping relationship of resource-operation-cost object, and generates the initial cost accounting result at the operation level.

[0014] As a further improvement to this technical solution, the persistent storage of data and basic compliance verification of the accounting result storage and compliance verification unit specifically includes: First, perform basic compliance verification on the received job-level initial cost accounting results, distinguish between qualified and unqualified data, and retain the exception log of unqualified data separately. Perform structured data entry on the verified initial cost accounting results at the job level, complete the persistent storage of all data, and output the compliant cost data after verification.

[0015] As a further improvement to this technical solution, the classification, summarization, and visualization presentation of the terminal cost classification, summarization, and information display unit specifically includes: The received compliant cost data is categorized and summarized by at least one of the following: resource type, operation type, equipment type, and container type, to obtain a categorized and summarized cost dataset. The cost dataset, after being categorized and summarized, is parsed and rendered according to at least one of the following terminal business dimensions: single container cost, operation line cost, equipment cost, and shift cost, to achieve a visual presentation of cost information.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention constructs a loss function gradient by combining the real-time actual consumption value of the operation with the predicted value calculated by the current cross-consumption coefficient matrix through an incremental iterative calculation module. It also uses an adaptive learning rate that dynamically decays with the number of completed operations to perform element-by-element incremental correction and proportional re-normalization of the coefficient matrix. This enables the allocation coefficient to continuously track the resource consumption pattern drift caused by the gradual change of the operation process and the degradation of equipment performance. This solves the problems of static allocation rate fixation and lack of closed-loop real-time metering calibration data, and ensures the stability of cost allocation accuracy under long-term operation. 2. This invention relies on the topology gap idle consumption processing module, which uses the start and end timestamps of the work bound to the topology nodes to extract the time interval between adjacent work steps. Based on the absolute value of the work plan deviation, and by using preset weights to distinguish the allocation ratio of early and delayed work conditions, the total consumption within the effective idle gap is accurately allocated to the corresponding work steps according to the responsibility ratio. This triggers the correction of the cross-consumption coefficient matrix elements, fundamentally solving the accounting distortion problem caused by the suspension or equal allocation of idle costs. It provides reliable data support for the accurate attribution of single-box costs and the optimization of work processes. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall system framework of the present invention; The meanings of the labels in the diagram are as follows: 1. Operation data acquisition and preprocessing unit; 2. Hierarchical cost collection and accounting unit for activity-based costs; 21. Workflow topology construction module; 22. Cross-consumption coefficient matrix management module; 23. Incremental iterative calculation module; 24. Topology gap idle consumption processing module; 25. Three-level cost allocation execution module; 3. Calculation result storage and compliance verification unit; 4. Terminal cost classification summary and information display unit. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] like Figure 1As shown, this embodiment provides a container terminal cost analysis system based on activity-based costing and big data analytics, applied to a fully automated, specialized container terminal operation scenario. This terminal is equipped with mainstream core operating equipment such as automated quay cranes, rail-mounted yard cranes, and yard trucks. It also features a terminal operation scheduling system, equipment energy consumption monitoring system, equipment PLC control system, enterprise resource management system, and IoT smart metering terminals. The entire operation process generates multi-source heterogeneous production data that is sequential, diverse in type, and covers all operational procedures. The system adopts a layered deployment architecture of edge acquisition + central computing + compliant storage + visualization applications. Both hardware and software are selected from port industrial-grade products, and deployment is completed using a private server cluster at the terminal, meeting industry management requirements such as port intranet isolation, data security, operation tracking, and audit traceability.

[0020] The system's overall business process strictly follows a closed-loop operational logic of data preprocessing—refined cost accounting—compliance verification and storage—multi-dimensional visualization. Using activity-based costing (ABC) as its core accounting theory, it integrates technologies such as big data workflow topology modeling, incremental iteration of cross-consumption coefficient matrices, and causal attribution of operational plan deviations. It achieves layered decomposition, dynamic updating, and precise allocation of full-link cost analysis for effective operational resource consumption and equipment idle time consumption across all processes in terminal loading / unloading, storage, and in-yard transfer. This embodiment is entirely based on real terminal operating conditions, historical consumption statistics, and equipment runtime sequence characteristics for practical application. It does not use idealized simulation parameters or fictitious operating conditions. All calculation rules are consistent with port cost accounting standards, equipment energy consumption metering standards, and on-site operation scheduling logic. Data flow between functional units is smooth, and functional connections are tight. It effectively solves existing problems in traditional accounting schemes, such as the gradual decline in the accuracy of static cost allocation, the difficulty in accurately attributing equipment idle time losses, and the inability to reasonably write off suspended costs.

[0021] The container terminal cost analysis system based on activity-based costing and big data analytics used in this embodiment includes: an activity data acquisition and preprocessing unit 1; an activity cost hierarchical aggregation and accounting unit 2; an accounting result storage and compliance verification unit 3; and a terminal cost classification, summary, and information display unit 4, wherein: Operation data acquisition and preprocessing unit 1 collects raw operation data from all terminal operations and performs basic normalization processing on the raw operation data, outputting a standardized operation dataset; specifically, the basic normalization processing includes: The system performs data deduplication, time-series alignment, field format normalization, and invalid data removal operations on the collected multi-source raw operational data of the dock, generating a standardized operational dataset with unified fields and time-series matching.

[0022] Specifically, the operation data acquisition and preprocessing unit 1 is deployed on the terminal's industrial intranet server, adopting a lightweight architecture combining edge gateways and application services. Its logical links connect to existing business modules such as the terminal's TOS scheduling system, equipment energy consumption metering terminals, and equipment PLC control systems. It comprehensively collects raw operation data generated from all processes of container loading and unloading, yard storage, and in-yard transfer. The collection scope includes core information such as operation ID, equipment ID, operation time parameters, operation volume, various resource consumption, and scheduling instructions. The operation data acquisition and preprocessing unit 1 is configured with multi-protocol data access methods and a breakpoint resume mechanism, completely preserving the raw data collected on-site without tampering with or correcting basic information such as metering values ​​and timestamps, ensuring the authenticity and integrity of the raw data.

[0023] Furthermore, after data acquisition is completed, the operation data acquisition preprocessing unit 1 performs basic standardization work sequentially according to preset rules. The operation data acquisition preprocessing unit 1 uses the operation ID, equipment ID, and timestamp as joint judgment conditions to perform data deduplication, removing duplicate entries generated by network retransmissions and multi-channel parallel acquisition; the operation data acquisition preprocessing unit 1 uses the terminal's global time synchronization server as a unified benchmark to complete time sequence alignment, correcting clock deviations between various business subsystems; the operation data acquisition preprocessing unit 1 unifies the format, data type, and measurement units of all data fields, completing field format normalization; the operation data acquisition preprocessing unit 1 identifies and removes invalid data such as missing key fields, operation time sequence logic errors, and abnormal resource consumption values.

[0024] Finally, after all the standardization procedures are completed, the operation data acquisition and preprocessing unit 1 generates a standardized operation dataset with unified fields and time sequence matching, and pushes the standardized operation dataset to the downstream operation cost hierarchical collection and accounting unit 2 through the intranet dedicated data interface.

[0025] The hierarchical cost aggregation and accounting unit 2 for operation costs aggregates costs on a standardized operation dataset based on an improved incremental iterative algorithm for the cross-consumption coefficient of the operation flow topology. It constructs a three-level dynamic mapping relationship between resources, operations, and cost objects, completes the accurate splitting and allocation calculation of cross-consumption of operations and idle consumption of topology gaps, and generates initial cost accounting results at the operation level. In this embodiment, the hierarchical cost aggregation and accounting unit 2 includes a workflow topology construction module 21, a cross-consumption coefficient matrix management module 22, an incremental iteration calculation module 23, a topology gap idle consumption processing module 24, and a three-level cost allocation execution module 25. These five modules operate in a hierarchical manner, using the workflow topology model as the spatial carrier, the cross-consumption coefficient matrix as the allocation rule, and incremental iteration and idle attribution algorithms as dynamic correction methods, thereby upgrading static activity cost accounting to dynamic adaptive lean accounting. Wherein: The workflow topology construction module 21 constructs a directed acyclic graph topology of the dock workflow based on a standardized workflow dataset, completing the basic mapping between atomic work steps and resource consumption flow. The workflow topology construction module 21 includes a work process decomposition submodule, a topology node generation submodule, and a topology edge generation submodule, and the specific implementation is as follows: The work process decomposition submodule parses the standardized work dataset and extracts the complete container work process, breaking it down into indivisible atomic work links; Specifically, the work process breakdown sub-module receives the standardized work dataset output by the work data acquisition and preprocessing unit 1, and, based on the standardized work process regulations of the container terminal, refines the entire complex work process of quay crane loading and unloading, yard crane stacking and retrieval, in-yard truck transfer, container alignment, and container lashing.

[0026] Traditional port cost accounting generally adopts a coarse-grained aggregation method of "whole operation line, whole shift," treating multi-process coupled operations as a single accounting object. This has inherent defects such as overlapping consumption of cross-processes, inability to isolate idle periods, and inability to locate cost responsibility. The core improvement of the operation process decomposition sub-module in this embodiment is: using "independent equipment operation, independent resource consumption cycle, and independent time interval" as the sole decomposition criteria, coupled composite operations are decomposed layer by layer into the smallest indivisible, independently executed, independently statistically resource-consuming, and single-function operation units, defined as atomic operation links.

[0027] The dismantling process strictly adheres to engineering constraints: the same piece of equipment corresponds to only one atomic operation within the same time interval; processes with different technological purposes, resource consumption types, and operational functions must be broken down into independent atomic operation processes; complex operations spanning multiple time periods and processes must be segmented according to time boundaries to prevent the overlapping and aggregation of multiple processes. Through these dismantling rules, the uniform minimization of the granularity of terminal operation accounting is achieved, providing a unique and standardized basic accounting unit for subsequent cross-consumption decoupling, accurate identification of idle gaps, and dynamic coefficient iterative correction.

[0028] The topology node generation submodule maps each atomic operation step to a directed acyclic graph topology node and binds the node with a unique device identifier, operation start and end timestamps, and operation quantity parameters; Specifically, the topology node generation submodule takes all atomic operation links output by the operation process decomposition submodule as the modeling object, and uniquely maps each independent atomic operation link to an independent topology node in the directed acyclic graph of the terminal operation flow, constructing a structured, parameterized, and time-sequential set of topology nodes, thereby realizing the digitization and modeling of all atomic operations.

[0029] Meanwhile, in this embodiment, the mathematical expression for the set of topological nodes is defined as follows: ; This represents the set of all topological nodes in the directed acyclic graph of the job flow. Indicates the first The topology nodes corresponding to each atomic operation step; Indicates the first A unique identifier for the operating equipment bound to each topology node; Indicates the first The actual start timestamp of each atomic operation step; Indicates the first The actual end timestamp of each atomic operation step; Indicates the first The workload corresponding to each atomic operation step; This represents the total number of all valid atomic jobs identified by the system within the current statistical period.

[0030] Furthermore, the topology node generation submodule enforces a mandatory parameter binding mechanism for each topology node. All parameters are derived from standardized work datasets, without manual fitting or subjective correction, ensuring that node attributes are completely consistent with the actual on-site work conditions. Simultaneously, all topology nodes are automatically sorted according to their chronological order, preserving the temporal continuity of equipment operations and the logical sequence of processes, giving the topology model inherent chronological traceability capabilities. This parameterized node design completely solves the modeling defects of traditional static work models, which lack chronological attributes, equipment binding, and workload correlation. It provides accurate and reusable node data sources for subsequent extraction of gaps between adjacent equipment operations, calculation of idle time, verification of work plan deviations, and attribution of idle cost responsibility.

[0031] The topology edge generation submodule matches the corresponding consumption relationship between terminal production resources and atomic operation links. It constructs unidirectional topology directed edges with production resources as the starting point and atomic operation links as the ending point, so that each directed edge uniquely corresponds to a type of terminal consumption resource, thereby representing the exclusive consumption support relationship of resources for operation links.

[0032] Specifically, the topology edge generation submodule matches the corresponding support relationships between various production resources at the dock and each atomic operation link based on the resource consumption associations recorded in the standardized operation dataset, constructs a set of directed edges in the directed acyclic graph of the operation flow, and establishes a one-way mapping relationship from production resources to atomic operation links.

[0033] In this embodiment, the mathematical definition of the topological directed edge set of the directed acyclic graph of the job flow is as follows: ; In the formula, Represents the set of all topological directed edges in the directed acyclic graph of the job flow; Indicates the first Class production resources point to the first One-way directed edges of a topological node; Indicates the first Port-like production resources; , These are the production resource type count subscript and the total number of production resource types included in the accounting scope, respectively.

[0034] Meanwhile, the topology edge generation submodule strictly adheres to a unique mapping rule: a single directed edge corresponds to only one type of production resource and a specific consumption relationship with one atomic operation; the same type of production resource can provide support for multiple different atomic operation processes simultaneously through multiple independent directed edges, adapting to the operational characteristics of shared consumption of port public resources by multiple devices; a single directed edge cannot be bound to multiple types of resources, nor can it establish mapping relationships across topology nodes. This modeling method transforms complex scenarios of multiple resources and multiple operations with overlapping consumption into a rule-based topology network structure.

[0035] Understandably, traditional cost models only establish fuzzy many-to-many relationships between resources and operations, failing to achieve a quantitative breakdown of cross-consumption. This topology model completes the decoupling preprocessing of cross-consumption at the modeling level, laying a rigorous mathematical foundation for subsequent cross-consumption coefficient matrix initialization, incremental iterative dynamic correction, and accurate allocation of hierarchical costs.

[0036] The cross-consumption coefficient matrix management module 22 initializes and dynamically maintains the cross-consumption coefficient matrix, storing the proportion of various resources consumed by their corresponding atomic operation stages at each time point. This module is responsible for initializing and dynamically maintaining the cross-consumption coefficient matrix, which stores the proportion of various resources consumed by their corresponding atomic operation stages at each time point. It is the core data carrier for accurately allocating resource costs to operation stages. The cross-consumption coefficient matrix management module 22 includes a matrix dimension determination submodule, a matrix initialization submodule, and a matrix dynamic maintenance submodule. These three submodules work collaboratively according to the business process, as detailed below: The matrix dimension determination submodule retrieves the directed acyclic graph topology of the dock operation flow generated by the operation flow topology construction module 21, and uses the total number of resource consumption types as the number of matrix rows and the total number of nodes corresponding to atomic operation links as the number of matrix columns to determine the row and column dimensions of the dynamic cross-consumption coefficient matrix. Specifically, the matrix dimension determination submodule actively retrieves the directed acyclic graph topology of the dock operation flow generated by the operation flow topology construction module 21, and extracts two types of core statistical values ​​from the topology model: First, the total number of all resource types consumed within the system's accounting scope; Second, the total number of topology nodes corresponding to all atomic operation stages.

[0037] The matrix dimension determination submodule uses the total number of resource consumption types as the number of rows in the cross-consumption coefficient matrix and the total number of nodes corresponding to atomic operation stages as the number of columns in the cross-consumption coefficient matrix to determine the complete row and column dimensions of the dynamic cross-consumption coefficient matrix.

[0038] Furthermore, based on the previous definition of the topology model, let the total number of resource types be... The total number of atomic job nodes is The mathematical expression for the cross-consumption coefficient matrix is ​​as follows: ; In the formula, Represents the cross-consumption coefficient matrix; Represents the first element in the matrix Line 1 Column matrix elements; For resource type counting subscripts; This is the index for counting atomic job nodes; The row number represents the total number of resource types consumed by the dock. The column number represents the total number of topological nodes in the atomic operation.

[0039] Finally, the matrix dimensions are determined entirely by the workflow topology model, ensuring that the matrix rows and columns correspond one-to-one with resources and atomic operation links. This structurally achieves a precise match between the resource consumption ratio and the accounting object, avoiding allocation logic errors caused by dimension mismatch.

[0040] The matrix initialization submodule retrieves historical operational consumption statistics for the dock, calculates the average consumption percentage of various resources allocated to each type of atomic operation, and uses it as the initial value of the matrix. Specifically, the matrix initialization submodule retrieves historical operational consumption statistics for the terminal over the past 30 days. The statistical criteria, resource classification standards, and atomic operation division rules of this historical data are consistent with the current topology model. The matrix initialization submodule categorizes and summarizes data according to resource type and atomic operation stage, calculates the average consumption percentage of each resource type allocated to each atomic operation stage, and assigns this average consumption percentage as the initial element of the matrix, thus completing the initial cross-consumption coefficient matrix. The complete calculation process for its construction is as follows: First, summarize historical consumption data by day and by dimension. Let the total number of days in the historical data statistical period be (…). ), iterate through the daily data within the statistical period, and calculate the first number of each day. Japan and China, No. Class resources in the Consumption of each atomic job node and the Total daily consumption of all types of resources ,in The formula for calculating the total daily resource consumption is: ; In the formula, Indicates the first day Total consumption of this type of resource; Indicates the first day Class resources in the The consumption of each atomic job node.

[0041] Then, calculate the daily resource consumption percentage. For each day's data, solve for the... Class of resources allocated to the first Daily consumption percentage of each atomic job node The calculation formula is: ; In the formula, Indicates the first day Class of resources for the first The consumption percentage of each atomic work node; this parameter is a dimensionless proportion.

[0042] Next, calculate the average consumption percentage over the period. Calculate the arithmetic mean of the daily consumption percentages over 30 consecutive days to obtain the [number of days]. Class of resources allocated to the first The average consumption percentage of each atomic operation step is the initial element of the cross-consumption coefficient matrix. The calculation formula is: ; In the formula, Indicates the first Class of resources corresponds to the first Average consumption percentage of each atomic job node; Represents the initial cross-consumption coefficient matrix The Middle Line 1 The elements of the column.

[0043] The fourth step is normalization verification and correction. To meet the business rule of full resource allocation, the sum of the consumption percentages of all atomic job nodes corresponding to the same type of resource must equal 1, i.e., the constraint: ; Due to historical floating-point calculation errors, if there is a slight deviation in the summation result of a single row, the module will make a minor correction to all elements in that row proportionally, ultimately strictly satisfying the normalization constraint. After all elements have been calculated and verified, the average consumption percentage is assigned to the corresponding positions in the matrix one by one, generating a complete initial cross-consumption coefficient matrix. After initialization, the initial matrix data is temporarily cached, awaiting management by the dynamic maintenance submodule.

[0044] The matrix dynamic maintenance submodule receives incremental update commands, updates the matrix element values ​​in real time, and synchronously stores the latest version of the matrix.

[0045] Specifically, the matrix dynamic maintenance submodule monitors the incremental update commands issued by the downstream incremental iterative calculation module 23 in real time. Upon receiving a valid update command, the matrix dynamic maintenance submodule updates the element values ​​at the corresponding positions of the cross-consumption coefficient matrix point by point according to the element correction data carried in the command. After the value update is completed, the latest version of the cross-consumption coefficient matrix is ​​stored synchronously, while historical version matrices are retained for data traceability and review.

[0046] During system operation, the cross-consumption coefficient matrix always uses the latest version in cost allocation calculations, and historical versions of the matrix are archived and saved in accordance with the terminal data management specifications. The matrix dynamic maintenance submodule ensures the real-time performance, integrity, and traceability of matrix data throughout the process, adapting to the dynamic adjustment needs of consumption ratios caused by continuous changes in terminal operating conditions.

[0047] Understandably, the cross-consumption coefficient matrix management module 22 relies on the output of the upstream workflow topology construction module 21 to define the matrix dimensions. Based on the terminal's real historical operating data for the past 30 days, it obtains the average consumption ratio through a complete calculation process of hierarchical aggregation, daily percentage calculation, multi-day average solution, and normalization verification. This completes the matrix initialization, and in conjunction with the downstream calculation module, it achieves dynamic matrix updates and persistent storage. The cross-consumption coefficient matrix managed by the module establishes a quantitative consumption correlation between "resources and atomic operation links," inheriting the consumption patterns of historical operations while continuously and adaptively adjusting to changes in on-site operating conditions.

[0048] The incremental iterative calculation module 23 listens for the completion trigger event of atomic operation stages and performs incremental update calculations on the cross-consumption coefficient matrix based on the actual resource consumption data of the operation. Specifically, the incremental iterative calculation module 23 is the core calculation unit for realizing the dynamic adaptive update of coefficients in this system. The incremental iterative calculation module 23 listens for the completion trigger event of atomic operation stages in real time. Whenever a single atomic operation stage is completed, it retrieves the actual resource consumption data corresponding to that operation and performs incremental update calculations on the cross-consumption coefficient matrix in combination with a preset algorithm, so that the predicted consumption value continuously approaches the actual consumption value, thereby adapting to changes in resource consumption patterns caused by adjustments in port operation processes, equipment aging, and environmental changes. The incremental update calculation of the incremental iterative calculation module 23 is strictly executed in sequence according to the six steps from S23.1 to S23.6, and the specific implementation method is as follows: S23.1 Real-time capture of event trigger signals indicating the completion of a single atomic operation step; In this step, the incremental iterative calculation module 23 connects to the terminal industrial event bus and continuously monitors the running status of all atomic operation stages. When any atomic operation stage is completed, the system automatically generates a job completion event message and pushes it outward. The incremental iterative calculation module 23 captures this trigger signal in real time and simultaneously extracts the associated data such as the topology node number, equipment identifier, actual resource consumption, and workload corresponding to the currently completed atomic operation, as the basic data for this round of incremental iterative calculation. The system adopts a single-job trigger, single-iteration operation mode, with each iteration calculation independent of the others, which can effectively avoid data conflict problems caused by multi-task concurrency.

[0049] S23.2 Collect the actual resource consumption data of this atomic operation link, calculate the actual value of unit operation resource consumption, and at the same time calculate the predicted value of unit operation resource consumption of this atomic operation link based on the current cross-consumption coefficient matrix. In this step, the incremental iterative calculation module 23 extracts the actual resource consumption data measured on-site for this atomic operation and calculates the actual value of resource consumption per unit operation; at the same time, it reads the currently effective cross-consumption coefficient matrix and calculates the predicted value of resource consumption per unit operation in combination with the cumulative resource consumption in the current statistical period.

[0050] The formula for calculating the actual value of resource consumption per unit of work is as follows: ; The formula for calculating the predicted unit resource consumption is as follows: ; In the formula, Indicates the first Class of resources corresponds to the first The actual unit resource consumption of each atomic job node; Indicates the first Class of resources corresponds to the first Predicted unit resource consumption of each atomic job node; This indicates that before the completion of this assignment, the [number]th [item]... The cumulative total consumption of this type of resource within the current statistical period. Unit operations are based on standard container units (TEU). and The dimensions remain consistent.

[0051] S23.3 Calculate the sum of squares of the deviations between the predicted and actual unit operation resource consumption values, and solve for the gradient of the loss function; In this step, combining the actual and predicted values ​​obtained in the previous step, the incremental iterative calculation module constructs a loss function using the sum of squared deviations to quantify the degree of deviation between the two sets of data. Based on this, the partial derivatives of the loss function with respect to the matrix elements are calculated to obtain the gradient of the loss function. The gradient result will be used to determine the direction and magnitude of the coefficient correction.

[0052] The loss function calculation formula for a single resource type and a single atomic job node is as follows: ; In the formula, This represents the loss function value corresponding to a single type of resource and a single atomic job node.

[0053] Taking the partial derivative of the loss function, we obtain the formula for calculating the gradient of the loss function: ; In the formula, For the loss function on matrix elements The gradient. This refers to the [number]th [item] in the current statistical period before the completion of this task. The cumulative total consumption of the resource class. The sign of the gradient indicates the direction of correction of the coefficient, and the absolute value of the gradient indicates the initial correction magnitude of the coefficient.

[0054] S23.4. An adaptive learning rate that dynamically decreases with the number of completed tasks is used to perform incremental corrections on the cross-consumption coefficient matrix element by element. In this step, to balance the need for rapid deviation correction in the early stages of iteration with the need for stable model operation in the later stages, the incremental iteration calculation module 23 introduces an adaptive learning rate that dynamically decays with the number of completed jobs. Combined with the gradient values ​​obtained in the previous step, incremental corrections are performed on the corresponding elements in the cross-consumption coefficient matrix one by one. Let the total number of atomic jobs completed after system startup be... The system's preset initial learning rate is The preset attenuation coefficient is In the context of cost accounting at port container terminals, The typical value range is 0.01 to 0.1. The typical value range is 0.001 to 0.01, and the optimal parameter can be obtained by using a grid search method to optimize the historical operation data of the past 30 days.

[0055] The formula for calculating the adaptive learning rate is as follows: ; In the formula, This is the adaptive learning rate for the current iteration round. It increases with the cumulative number of jobs. As the learning rate increases, it gradually decreases, achieving the effect of quickly correcting deviations in the early stages of iteration and making small fine adjustments to stabilize the model in the later stages of iteration.

[0056] Substituting the adaptive learning rate and gradient into the calculation, we obtain the formula for adjusting the matrix element increment: ; In the formula, These are temporary matrix elements that have undergone incremental correction but have not been normalized. This iteration only corrects all matrix column elements corresponding to the current atomic job node; the remaining matrix elements retain their original values.

[0057] S23.5 Perform proportional normalization on the cross-consumption coefficient matrix after incremental correction so that the sum of the cross-consumption coefficients of all atomic operation links corresponding to the same type of resource is 1. In this step, after the matrix elements undergo incremental correction, the original constraint that "the sum of the cross-consumption coefficients of all atomic operation steps corresponding to the same type of resource is 1" will be broken. In order to comply with the business rule of full allocation of resource costs, the incremental iterative calculation module 23 performs proportional renormalization processing on the temporary matrix row by row according to the resource category.

[0058] First calculate the... The sum of all temporary matrix elements in a row is calculated using the following formula: ; In the formula, Indicates the first The summation of all elements of the temporary matrix.

[0059] The matrix elements are scaled proportionally based on the sum of the elements in each row to obtain the final normalized matrix elements. The calculation formula is as follows: ; In the formula, These are the final matrix elements after normalization.

[0060] The cross-consumption factor must meet the following requirements. The constraints. If the temporary matrix elements after incremental correction... If a value is less than 0 or greater than 1, it is first truncated to the boundary value of 0 or 1, and then the proportional normalization process is re-executed for all elements in that row.

[0061] After normalization, the matrix strictly satisfies the following constraints: ; This step can effectively avoid problems such as unallocated or duplicated resource allocation, and ensure that the cost allocation logic is compliant.

[0062] S23.6. Synchronize the normalized cross-consumption coefficient matrix to the matrix dynamic maintenance submodule to complete this iteration update.

[0063] In this step, after the matrix completes all processing steps of incremental correction and proportional re-normalization, the incremental iterative operation module 23 synchronizes the latest version of the cross-consumption coefficient matrix to the matrix dynamic maintenance submodule of the cross-consumption coefficient matrix management module 22 through an internal dedicated data interface. Upon receiving the data, the matrix dynamic maintenance submodule immediately updates the currently running matrix version and archives historical version data for subsequent data traceability and operational review. At this point, the incremental iterative update process corresponding to a single atomic job is complete, and the incremental iterative operation module 23 automatically resumes event listening mode, continuously waiting for the completion trigger signal of the next atomic job stage.

[0064] Understandably, the incremental iterative calculation module 23 is driven by atomic operation completion events, constructs a loss function and gradient based on real-time operation consumption data, and completes incremental coefficient correction by combining a dynamically decaying adaptive learning rate, while maintaining coefficient constraint rules through row normalization. This incremental iterative calculation module 23 can continuously optimize the allocation coefficients based on changes in dock equipment conditions, operational processes, and the environment, solving the problem of traditional static coefficients gradually decreasing matching accuracy as runtime increases. The incremental iterative calculation module 23 connects upwards to the front-end operation data and topology model, and downwards to the matrix dynamic maintenance submodule to push updated results; it is the core computational link for the entire system to achieve dynamic lean cost accounting.

[0065] The topology gap idle consumption processing module 24 identifies the time gap between adjacent atomic operation segments of the same device, and completes the accurate mapping of gap idle consumption data to the corresponding operation segments. Based on the timing and device association information of the topology nodes, the topology gap idle consumption processing module 24 identifies the time gap between adjacent atomic operation segments of the same device, distinguishes between normal operation switching time and effective idle time, and completes the accurate mapping and reasonable allocation of gap idle consumption data to the corresponding atomic operation segments. The process of the topology gap idle consumption processing module 24 performing the accurate mapping of gap idle consumption is carried out sequentially according to S24.1 to S24.6, and the specific implementation method is as follows: S24.1. Based on the start and end timestamps of the jobs bound to the topology nodes, extract the time interval between two adjacent atomic job segments on the same device; In this step, the topology gap idle consumption processing module 24 first relies on the job start and end timestamps and unique device identifiers already bound in the topology nodes to group all atomic job nodes according to their device numbers, and then sorts them according to the job occurrence sequence within the same device group. For two adjacent atomic job nodes after sorting, the time interval between them is calculated.

[0066] Assuming that on the same device, the previous atomic job node is the... Each node corresponds to a job's actual completion timestamp. The next adjacent atomic job node is the first Each node corresponds to a job's actual start timestamp. The formula for calculating the time interval between adjacent tasks is as follows: ; In the formula, This represents the time interval between two adjacent atomic operation stages on the same device, measured in milliseconds (ms). This calculation method allows for the precise extraction of all interval durations between continuous operations of a single device, providing a timing basis for subsequent no-load condition determination.

[0067] S24.2 When the time interval is greater than the standard operation switching time of the dock equipment, it is determined to be an effective idle gap, and the total idle consumption generated by the equipment within the gap is collected. In this step, container terminal equipment has a fixed standard operation switchover time, which is the inherent time required for the equipment to complete routine connection actions such as workstation adjustment and container alignment. The topology gap idle consumption processing module 24 retrieves the terminal's preset equipment standard operation switchover time as a judgment threshold, and compares the time interval obtained in the previous step with this threshold to distinguish between normal switchover gaps and effective idle gaps.

[0068] Assume the standard operation switchover time for dock equipment is This parameter is calibrated separately for different equipment types such as quay cranes, rail-mounted cranes, and in-yard trucks, and is a constant fixed on-site. Among them, automated quay cranes are typical... The typical time is 5-10 seconds for a rail-mounted yard crane. The typical time for a container truck in the field is 3-8 seconds. The time is 2-5 seconds, which can be determined by statistically averaging the switching times of normal operations over the past 30 days. When the following conditions are met... When the effective idle time interval is determined, the topology gap idle consumption processing module 24 connects to various resource metering terminals on site to collect the total idle consumption of various resources generated by the equipment during this time interval.

[0069] Record No. The total consumption of this type of resource within the effective idle gap is Its dimensions are consistent with the actual consumption of similar resources mentioned above, and this data serves as the original basis for subsequent allocation of idle consumption.

[0070] S24.3 Calculate the deviation between the actual execution time and the planned execution time of the previous atomic operation step. And the deviation between the actual execution time and the planned execution time of the next atomic operation. ; In this step, after the effective idle gap is determined, the topology gap idle consumption processing module 24 counts the planned execution time and actual execution time of two adjacent atomic operation links respectively, calculates the time deviation value between the two, and uses this as the core basis for idle consumption attribution and allocation.

[0071] For the previous atomic operation, let its planned execution time be... The actual execution time is The time deviation value is denoted as For the latter atomic operation, let its planned execution time be... The actual execution time is The time deviation value is denoted as The formulas for calculating the two sets of deviation values ​​are as follows: ; ; In the formula, , All units are milliseconds (ms). A deviation value greater than 0 indicates that the actual execution time of the task exceeds the planned time, i.e., the task is delayed; a deviation value less than 0 indicates that the actual execution time of the task is shorter than the planned time, i.e., the task is completed ahead of schedule.

[0072] S24.4 According to the cause-and-effect attribution rule of work plan deviation, the total idle consumption is allocated to the two atomic work links before and after in proportion. The allocation ratio is positively correlated with the absolute value of the corresponding deviation value. The allocation ratio of different working conditions such as early work and delayed work can be distinguished by preset weight. In this step, the topology gap idle consumption processing module 24 strictly follows the work plan deviation causal attribution rule to carry out idle consumption allocation, and firstly, the rule is fully disclosed: First, the basis for rule formulation. Idle gaps exceeding the standard switchover time between adjacent atomic operations on the same equipment are mainly caused by deviations in the planned execution time of preceding and subsequent operations, as well as delays in scheduling system instructions. Therefore, this rule prioritizes the planned execution deviation between two adjacent operations as the core basis for dividing responsibility for idle consumption, following the core causal principle that abnormal operations are the primary cause of idle gaps and that consumption proportions are divided according to the degree of impact. For idle gaps caused by instruction delays confirmed by the scheduling system logs, the corresponding idle consumption will be included in the terminal's common cost pool and allocated to all operational stages according to the proportion of the total workload in the current period. Second, the standard for classifying operating conditions. Two categories of abnormal operating conditions are distinguished by the positive or negative value of the time deviation: Defined as a work delay, this condition will slow down the overall work pace and directly cause subsequent equipment to wait for idleness; Defined as early operation, this condition causes the equipment to complete the task ahead of schedule and enter a standby no-load state. This means the operation was carried out exactly as planned, with no deviation from the plan, and the operator does not bear any corresponding idle consumption. Third, differentiated weighting rules. In line with container terminal operation performance evaluation and cost attribution management requirements, fixed weighting coefficients are set for two types of abnormal operating conditions: defining the weight of operational delays. Priority weighting of assignments Both types of weights are system-predefined equipment dimension constants, which can be configured individually according to different equipment types such as quay cranes, yard cranes, and container trucks; because operational delays can easily lead to chain-like waiting and empty loads with a wider impact, the terminal's unified configuration rule is as follows: This reflects the differences in the impact of different abnormal operating conditions on the no-load clearance; Fourth, weight matching rules. For two adjacent atomic tasks, a corresponding weight is assigned to each task based on the sign of their respective time deviation values: a task is assigned a weight if it is delayed. If the task is completed ahead of schedule, the weight will be matched. If there is no deviation in the operation, the corresponding weight value is 0, and it does not participate in the idle consumption allocation.

[0073] Under the constraints of the above rules, the topology gap idle consumption processing module 24 first calculates the absolute values ​​of the two sets of deviation values. , Weights are assigned to the preceding and following operations based on the working conditions. , Multiply the absolute value of the deviation by its corresponding weight to obtain the comprehensive allocation base for a single operation. Add the comprehensive allocation bases for all operations to obtain the total base. The calculation formula is as follows: ; This represents the total base amount allocated to the two atomic operation steps before and after the previous one, and is a dimensionless value.

[0074] Using the total base amount for comprehensive allocation as a reference, the topology gap idle consumption processing module 24 calculates the idle consumption allocation ratio for the two atomic operation stages before and after the operation, respectively. The allocation ratio is proportional to the comprehensive allocation base amount for the operation. ; ; In the formula, , These represent the idle consumption allocation ratios for the preceding and subsequent atomic operation stages, respectively. Both are dimensionless coefficients and naturally satisfy... .

[0075] Finally, the topology gap idle consumption processing module 24, combined with the allocation ratio, calculates the first [number]th [unit] of the idle consumption allocated to the two work stages. Resource idle consumption: ; ; In the formula, , These are the first and second atomic operation stages allocated to each other, respectively. Resource idle consumption.

[0076] The following example illustrates how the rule is implemented: If the preceding work is delayed but the subsequent work is performed normally, then... , All idle time consumption is borne by the preceding task; if the preceding task is completed ahead of schedule and the following task is delayed, weights are assigned accordingly. and Consumption is divided according to the proportion of the comprehensive base, which is in line with the logic of cause and effect attribution.

[0077] S24.5. Add the allocated idle consumption to the consumption of the corresponding atomic operation link to obtain the total consumption of the atomic operation link. In this step, after the idle consumption allocation is completed, the topology gap idle consumption processing module 24 adds the allocated idle consumption amount to the resource consumption amount of the corresponding atomic operation link, and integrates them to obtain the total operation consumption including effective operation consumption and allocated idle consumption.

[0078] Following the previous definition, let the first... The atomic operation step for the first The self-consumption of this type of resource is After the idle consumption is amortized and distributed, the first step of this operation... Total consumption of class resources The calculation formula is: ; Similarly, the total resource consumption of each subsequent adjacent atomic task is calculated using the same logic. The integrated total consumption data fully reflects the total resource consumption corresponding to a single atomic task.

[0079] S24.6. Synchronize the total consumption data of the atomic operation to the incremental iterative calculation module 23, triggering the correction of the corresponding cross-consumption coefficient matrix elements.

[0080] In this step, after the total consumption calculation of all atomic operation stages is completed, the topology gap idle consumption processing module 24 synchronously pushes the summarized total consumption data to the incremental iterative calculation module 23 through the system's internal data interface. After receiving the data, the incremental iterative calculation module 23, according to the predetermined algorithm flow, triggers incremental correction of the cross-consumption coefficient matrix elements associated with the corresponding atomic operation stages, realizing dynamic optimization of the allocation coefficient based on the full consumption data. At this point, the accurate mapping and allocation process of a single topology gap idle consumption is completed, and the topology gap idle consumption processing module 24 continues to enter the gap detection and processing process of the next group of adjacent atomic operations.

[0081] Understandably, the Topology Gap Idle Consumption Processing Module 24, relying on the temporal and equipment association attributes of the workflow topology model, achieves automatic identification of equipment operation gaps and accurate determination of effective idle conditions. The Topology Gap Idle Consumption Processing Module 24 fully implements the causal attribution rules for operation plan deviations, configuring differentiated weights based on different abnormal conditions such as operation ahead of schedule and operation delay, and completing the proportional allocation of idle consumption according to causal relationships and the degree of impact, perfectly matching the actual needs of container terminal operation management, cost assessment, and responsibility attribution. The Topology Gap Idle Consumption Processing Module 24 incorporates idle consumption into the total consumption statistics of atomic operations, making up for the shortcomings of the traditional accounting model, such as the ambiguity of idle consumption attribution and the inability to quantify and allocate it; at the same time, it outputs complete total operation consumption data to the downstream incremental iterative calculation module 23, providing a comprehensive and accurate data source for the continuous iterative optimization of the cross-consumption coefficient matrix, and is an important intermediate module connecting topology modeling, consumption statistics, and dynamic coefficient updates.

[0082] The three-level cost allocation execution module 25, based on the updated cross-consumption coefficient matrix, completes the layer-by-layer allocation calculation of resource costs to operational processes and cost objects, thereby constructing a complete cost accounting chain. The three-level cost allocation execution module 25 includes a resource-level allocation submodule, an operational-level aggregation submodule, and a cost object-level summary submodule. These three submodules operate collaboratively in the order of resource allocation and aggregation at each level. The specific implementation method is as follows: The resource layer allocation submodule retrieves the final cross-consumption coefficient matrix after incremental iteration and idle loss correction, and allocates the total cost of various resources to the corresponding atomic operation links. Specifically, the resource allocation submodule retrieves the final cross-consumption coefficient matrix after incremental iteration and no-load loss correction, and combines it with the total cost of various resources in the current period of the terminal to allocate the total cost of each type of resource to the corresponding atomic operation link according to the consumption ratio.

[0083] Let the total number of resource types included in the accounting be The total number of topology nodes corresponding to the atomic operation is , No. The total current cost of this type of resource is The unit is a monetary unit of measurement; These are the elements of the final cross-consumption coefficient matrix after normalization. The data collected by the current period's data acquisition and preprocessing unit 1 The total consumption of each resource category is calculated by multiplying the corresponding unit price of the resource. The unit price is obtained from the port enterprise resource management system (ERP) and updated monthly to match the resource consumption statistical cycle. Combining the characteristics of the cross-consumption coefficient in representing the proportion of resource consumption, the [further details are needed]. Resource allocation to the first The formula for calculating the resource cost of an atomic operation is as follows: ; In the formula, Indicates the first Resource allocation to the first The cost of each atomic operation step.

[0084] The resource layer cost allocation submodule traverses all resource types and atomic operation nodes, performing cost allocation calculations row by row and column by column. The calculation process strictly follows the constraint that the sum of the coefficients of all atomic operation links corresponding to the same type of resource is equal to 1, ensuring that the total cost of all types of resources can be fully allocated without any cost omissions or duplicate allocations.

[0085] The task-level aggregation submodule summarizes the self-consumption and allocated idle consumption of each atomic task step to generate the total cost of the atomic task steps; Specifically, after the resource layer allocation calculation is completed, the job layer aggregation submodule summarizes the cost corresponding to the self-consumption of each atomic job step and the cost corresponding to the allocated idle consumption, and integrates them to generate the total cost of a single atomic job step.

[0086] For the For each atomic task, the allocated costs of all resource types are summed, and the idle consumption costs allocated to that atomic task are also added to obtain the total cost of the atomic task. The calculation formula is as follows: ; In the formula, Indicates the first The total cost of each atomic operation step; Indicates the first The total cost corresponds to the idle consumption allocated to each atomic operation step.

[0087] The task-level aggregation submodule takes atomic task links as independent accounting units, integrates direct task consumption costs and indirect idle cost allocation, and fully restores the full-dimensional cost expenditures corresponding to a single atomic task, providing accurate task-level cost data for subsequent top-level cost aggregation.

[0088] The cost object layer aggregation submodule aggregates the costs of each atomic operation into the corresponding container cost object, constructs a three-level dynamic mapping relationship between resources, operations, and cost objects, and generates the initial cost accounting results at the operation level.

[0089] Specifically, after the work-level aggregation is completed, the cost object-level aggregation submodule aggregates the costs of each atomic work step to the corresponding container cost object according to the system's preset association relationship, and simultaneously constructs a three-level dynamic mapping relationship of resource-work-cost object, and finally generates the initial cost accounting result at the work level.

[0090] Each atomic operation at the terminal is pre-assigned to a unique container cost object. Let the k-th container be an independent cost object. The total cost of all atomic operations associated with that container is summed to obtain the total cost for that single container, calculated using the following formula: ; In the formula, Indicates the first The total cost corresponding to each container cost object; Indicates binding to the first The set of all atomic operation nodes for each container.

[0091] The cost object layer aggregation submodule, while completing cost aggregation, records the correlation and correspondence between resource types, atomic operation stages, and container cost objects, forming a traceable and queryable three-level dynamic mapping relationship between resources, operations, and cost objects. After all aggregation operations are completed, the cost object layer aggregation submodule outputs complete operation-level initial cost accounting results, including details of resource allocation, costs of atomic operations, costs of containers, and three-level mapping correlation data.

[0092] Understandably, the three-level cost allocation execution module 25, relying on the dynamically updated cross-consumption coefficient matrix, sequentially completes resource-level cost allocation, operation-level cost aggregation, and cost object-level cost summarization, strictly adhering to the resource-operation-cost object hierarchy for layer-by-layer calculation. This module 25 receives the coefficient matrix and operation consumption data output from the upstream module and outputs standardized, traceable cost accounting results, fully realizing the system's three-level dynamic cost mapping design goal. Compared to traditional single-dimensional cost statistics methods, the three-level cost allocation execution module 25 can not only break down the cost distribution of various resources across different operations and containers, but also achieve reverse tracing to pinpoint the resource consumption and operation composition corresponding to the cost of a single container, comprehensively meeting the actual business needs of container terminals for refined cost accounting, cost analysis, and control.

[0093] The accounting result storage and compliance verification unit 3 receives the initial cost accounting results at the job level, completes persistent data storage and basic compliance verification, and outputs compliant cost data. Specifically, as a backend auxiliary unit of the system, the accounting result storage and compliance verification unit 3 is mainly responsible for receiving the initial cost accounting results at the job level, sequentially completing basic compliance verification, retention of abnormal data logs, persistent storage of qualified data, and finally outputting compliant cost data. The specific implementation method of the accounting result storage and compliance verification unit 3 is as follows: First, perform basic compliance verification on the received job-level initial cost accounting results, distinguish between qualified and unqualified data, and retain the exception log of unqualified data separately. Perform structured data entry on the verified initial cost accounting results at the job level, complete the persistent storage of all data, and output the compliant cost data after verification.

[0094] Specifically, the accounting result storage and compliance verification unit 3 first performs basic compliance verification operations on the received initial cost accounting results at the operation level. It sets verification dimensions based on the general specifications for cost accounting in container terminals, mainly including three categories of basic verification content: data integrity verification, cost value reasonableness verification, and verification of the validity of the three-level mapping relationship between resources, operations, and cost objects. The accounting result storage and compliance verification unit 3 completes item-by-item judgment according to preset verification rules, thereby distinguishing between qualified and unqualified data.

[0095] Furthermore, for initial cost accounting results at the work level that are deemed unqualified for verification, the accounting result storage and compliance verification unit 3 generates an exception log separately and manages it. The exception log fully records key information such as the exception data number, exception type, data reception time, and original data summary, which is used for subsequent manual verification, problem localization, and business review. Unqualified data is not stored in the database.

[0096] Furthermore, for the initial cost accounting results at the operation level that are deemed to have passed verification, the accounting result storage and compliance verification unit 3 performs a structured data entry operation. The accounting result storage and compliance verification unit 3 completes data regularization according to the predefined data table structure of the system. The data table fields correspond one-to-one with the three-level accounting system of resources, atomic operations, and container cost objects, completing the persistent retention of all qualified data. Structured data entry and persistent storage in the industrial database are well-known technologies in the industry, and will not be elaborated upon in this embodiment.

[0097] Finally, after completing the persistent storage of data, the accounting result storage and compliance verification unit 3 will output the verified compliant cost data to downstream business processes such as terminal cost report generation, operational data analysis, and cost control.

[0098] Terminal Cost Classification and Information Display Unit 4 performs routine classification and summarization of compliant cost data and visualizes cost information according to terminal business dimensions. Specifically, as a system terminal application unit, Terminal Cost Classification and Information Display Unit 4 mainly connects to upstream compliant cost data, sequentially completing multi-dimensional classification and summarization, business-dimensional analysis and rendering, and visualization, providing intuitive data support for terminal operation management, cost analysis, and business scheduling. The specific implementation method of Terminal Cost Classification and Information Display Unit 4 is as follows: The received compliant cost data is categorized and summarized by at least one of the following: resource type, operation type, equipment type, and container type, to obtain a categorized and summarized cost dataset. The cost dataset, after being categorized and summarized, is parsed and rendered according to at least one of the following terminal business dimensions: single container cost, operation line cost, equipment cost, and shift cost, to achieve a visual presentation of cost information.

[0099] Specifically, the terminal cost classification and information display unit 4 first classifies and summarizes the received compliant cost data. Combining the actual management needs of the terminal, it selects at least one dimension from resource type, operation type, equipment type, and container type for data collection and statistics, and finally obtains the classified and summarized cost dataset.

[0100] When summarizing by resource type, the terminal cost classification summary and information display unit 4 uses resource category as the grouping basis to collect the same resource cost scattered under each atomic operation and each container object, and to count the total consumption cost and allocation details of each type of resource. When summarizing by operation type, the terminal cost classification and information display unit 4 categorizes atomic operation links with the same attributes and calculates the overall cost corresponding to different operation categories; when summarizing by equipment type, the terminal cost classification and information display unit 4 summarizes all costs generated by equipment of the same model and category based on the equipment affiliation. When summarizing by container type, Terminal Cost Classification and Information Display Unit 4 classifies containers according to attributes such as container specifications and uses, and completes the cost collection corresponding to different container types.

[0101] The above classification and summarization logic strictly follows the unified rules for resource division, operation division, equipment identification, and container binding mentioned above, ensuring consistency in the summary data.

[0102] Based on this, after completing the classification and summarization and generating the corresponding cost dataset, the terminal cost classification and summarization and information display unit 4 selects at least one terminal business dimension from single container cost, operation line cost, equipment cost, and team cost to parse and render the classified and summarized cost dataset to achieve a visual presentation of cost information.

[0103] When analyzing and rendering from the perspective of single container cost, the terminal cost classification summary and information display unit 4 extracts the full-chain cost data corresponding to a single container, and intuitively displays the comprehensive cost of a single container and the composition of various costs. When analyzing and rendering from the perspective of work line cost, the terminal cost classification summary and information display unit 4 integrates data according to the established work line of the terminal to present the overall cost distribution of the entire work line; When analyzing and rendering from the perspective of equipment cost, the Terminal Cost Classification Summary and Information Display Unit 4 focuses on a single or similar piece of equipment, displaying the various costs generated by the equipment during its operation throughout the entire time period. When analyzing and rendering from the perspective of team cost, the dock cost classification summary and information display unit 4 combines the team division rules of the operators to collect the costs generated by all operations within the jurisdiction of the corresponding team.

[0104] In addition, data parsing, chart rendering, and front-end interface display are all well-known technologies in the industry, and will not be elaborated on in this embodiment.

[0105] Those skilled in the art will understand that the process of implementing all or part of the steps of the above embodiments can be carried out by hardware or by a program instructing the relevant hardware.

[0106] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A container terminal cost analysis system based on activity-based costing and big data analytics, characterized in that: include: Operation data acquisition and preprocessing unit (1) collects raw operation data of the entire terminal business, performs basic normalization processing on the raw operation data, and outputs a standardized operation dataset; The hierarchical cost collection and accounting unit (2) performs cost collection on the standardized operation dataset based on the improved operation flow topology cross consumption coefficient incremental iteration algorithm, constructs a three-level dynamic mapping relationship of resource-operation-cost object, completes the accurate splitting and allocation calculation of operation cross consumption and topology gap idle consumption, and generates operation-level initial cost accounting results. The accounting result storage and compliance verification unit (3) receives the initial cost accounting result at the job level, completes the data persistence storage and basic compliance verification, and outputs the compliant cost data. Terminal cost classification and information display unit (4) performs routine classification and summary of compliant cost data and completes the visualization of cost information according to the terminal business dimension.

2. The container terminal cost analysis system based on activity-based costing and big data analysis according to claim 1, characterized in that, The basic normalization process of the operation data acquisition and preprocessing unit (1) specifically includes: The system performs data deduplication, time-series alignment, field format normalization, and invalid data removal operations on the collected multi-source raw operational data of the dock, generating a standardized operational dataset with unified fields and time-series matching.

3. The container terminal cost analysis system based on activity-based costing and big data analysis according to claim 1, characterized in that, The hierarchical cost aggregation and accounting unit (2) includes a workflow topology construction module (21), a cross-consumption coefficient matrix management module (22), an incremental iteration calculation module (23), a topology gap idle consumption processing module (24), and a three-level cost allocation execution module (25), wherein: The operation flow topology construction module (21) constructs a directed acyclic graph topology of the dock operation flow based on the standardized operation dataset, and completes the basic mapping between atomic operation links and resource consumption flow direction; The cross-consumption coefficient matrix management module (22) initializes and dynamically maintains the cross-consumption coefficient matrix, storing the proportion of various resources consumed by the corresponding atomic operation links at each time. The incremental iteration operation module (23) listens for the completion trigger event of the atomic operation and performs incremental update operation on the cross-consumption coefficient matrix based on the actual resource consumption data of the operation. The topology gap idle consumption processing module (24) identifies the time gap between adjacent atomic operation links of the same device and completes the accurate mapping between gap idle consumption data and corresponding operation links; The three-level cost allocation execution module (25) completes the layer-by-layer allocation calculation of resource costs to operation links and cost objects based on the updated cross-consumption coefficient matrix.

4. The container terminal cost analysis system based on activity-based costing and big data analysis according to claim 3, characterized in that, The workflow topology construction module (21) includes a workflow process splitting submodule, a topology node generation submodule, and a topology edge generation submodule, wherein: The operation process decomposition submodule parses the standardized operation dataset and extracts the complete container operation process, breaking it down into indivisible atomic operation links; The topology node generation submodule maps each atomic operation step to a directed acyclic graph topology node and binds the node with a unique device identifier, operation start and end timestamps, and operation quantity parameters. The topology edge generation submodule matches the corresponding consumption relationship between terminal production resources and atomic operation links. It constructs unidirectional topology directed edges with production resources as the starting point and atomic operation links as the ending point, so that each directed edge uniquely corresponds to a type of terminal consumption resource, thereby representing the exclusive consumption support relationship of resources for operation links.

5. The container terminal cost analysis system based on activity-based costing and big data analysis according to claim 3, characterized in that, The cross-consumption coefficient matrix management module (22) includes a matrix dimension determination submodule, a matrix initialization submodule, and a matrix dynamic maintenance submodule, wherein: The matrix dimension determination submodule retrieves the directed acyclic graph topology of the dock operation flow generated by the operation flow topology construction module (21), uses the total number of the consumed resource types as the number of matrix rows, and uses the total number of nodes corresponding to the atomic operation links as the number of matrix columns to determine the row and column dimensions of the dynamic cross-consumption coefficient matrix. The matrix initialization submodule retrieves historical operational consumption statistics of the dock, calculates the average consumption ratio of various resources allocated to each type of atomic operation link, and uses it as the initial value of the matrix. The matrix dynamic maintenance submodule receives incremental update instructions, updates the matrix element values ​​in real time, and synchronously stores the latest version of the matrix.

6. The container terminal cost analysis system based on activity-based costing and big data analysis according to claim 3, characterized in that, The incremental update operation of the incremental iteration operation module (23) includes the following steps: S23.1 Real-time capture of event trigger signals indicating the completion of a single atomic operation step; S23.2 Collect the actual resource consumption data of this atomic operation link, calculate the actual value of unit operation resource consumption, and at the same time calculate the predicted value of unit operation resource consumption of this atomic operation link based on the current cross-consumption coefficient matrix. S23.3 Calculate the sum of squares of the deviations between the predicted and actual unit operation resource consumption values, and solve for the gradient of the loss function; S23.

4. An adaptive learning rate that dynamically decreases with the number of completed tasks is used to perform incremental corrections on the cross-consumption coefficient matrix element by element. S23.5 Perform proportional normalization on the cross-consumption coefficient matrix after incremental correction so that the sum of the cross-consumption coefficients of all atomic operation links corresponding to the same type of resource is 1. S23.

6. Synchronize the normalized cross-consumption coefficient matrix to the matrix dynamic maintenance submodule to complete this iteration update.

7. The container terminal cost analysis system based on activity-based costing and big data analysis according to claim 3, characterized in that, The process of accurately mapping the gap idle consumption of the topology gap idle consumption processing module (24) includes the following steps: S24.

1. Based on the start and end timestamps of the jobs bound to the topology nodes, extract the time interval between two adjacent atomic job segments on the same device; S24.2 When the time interval is greater than the standard operation switching time of the dock equipment, it is determined to be an effective idle gap, and the total idle consumption generated by the equipment within the gap is collected. S24.3 Calculate the deviation between the actual execution time and the planned execution time of the previous atomic operation step. And the deviation between the actual execution time and the planned execution time of the next atomic operation step. ; S24.4 According to the cause-and-effect attribution rule of work plan deviation, the total idle consumption is allocated to the two atomic work links before and after in proportion. The allocation ratio is positively correlated with the absolute value of the corresponding deviation value. The allocation ratio of different working conditions such as early work and delayed work can be distinguished by preset weight. S24.

5. Add the allocated idle consumption to the consumption of the corresponding atomic operation link to obtain the total consumption of the atomic operation link. S24.

6. Synchronize the total consumption data of the atomic operation to the incremental iterative operation module (23) to trigger the correction of the corresponding cross-consumption coefficient matrix elements.

8. The container terminal cost analysis system based on activity-based costing and big data analysis according to claim 3, characterized in that, The three-level cost allocation execution module (25) includes a resource-level allocation submodule, an operation-level aggregation submodule, and a cost object-level summary submodule, wherein: The resource layer allocation submodule retrieves the final cross-consumption coefficient matrix after incremental iteration and idle loss correction, and allocates the total cost of various resources to the corresponding atomic operation links. The task layer aggregation submodule summarizes the self-consumption and allocated idle consumption of each atomic task step to generate the total cost of the atomic task steps. The cost object layer aggregation submodule aggregates the costs of each atomic operation into the corresponding container cost object, constructs a three-level dynamic mapping relationship of resource-operation-cost object, and generates the initial cost accounting result at the operation level.

9. The container terminal cost analysis system based on activity-based costing and big data analysis according to claim 1, characterized in that, The data persistence storage and basic compliance verification of the accounting result storage and compliance verification unit (3) specifically include: First, perform basic compliance verification on the received job-level initial cost accounting results, distinguish between qualified and unqualified data, and retain the exception log of unqualified data separately. Perform structured data entry on the verified initial cost accounting results at the job level, complete the persistent storage of all data, and output the compliant cost data after verification.

10. The container terminal cost analysis system based on activity-based costing and big data analysis according to claim 1, characterized in that, The classification, summarization, and visualization of the terminal cost classification, summarization, and information display unit (4) specifically includes: The received compliant cost data is categorized and summarized by at least one of the following: resource type, operation type, equipment type, and container type, to obtain a categorized and summarized cost dataset. The cost dataset, after being categorized and summarized, is parsed and rendered according to at least one of the following terminal business dimensions: single container cost, operation line cost, equipment cost, and shift cost, to achieve a visual presentation of cost information.

Images