Operating decision system for cost control based on multi-dimensional big data analysis

Abstract

The present application relates to the technical field of port operation management, in particular to an operating decision system for cost control based on multi-dimensional big data analysis, comprising: a cost-related data collection and indexing unit; a port cost driving factor extraction and correlation unit; a multi-scenario cost control simulation operation unit; a cost control operating decision generation and verification unit; an execution feedback and deviation comparison unit. The present application builds a port cost driving factor atlas covering the load-cost correspondence and the distribution of high-load delay consumption, adopts multi-scenario simulation operation and execution deviation closed-loop calibration of resource domain load time series superposition followed by nonlinear cost mapping, solves the problem of difficult pre-implementation of multi-scenario cost accurate calculation for multiple operating schemes to be implemented, realizes accurate prediction of port multi-class shared resource cost, and provides quantitative support for cost control operating decisions.

Description

Technical Field

[0001] This invention relates to the field of port operation and management technology, and more specifically, to an operational decision-making system for cost control based on multi-dimensional big data analysis. Background Technology

[0002] As a core hub in the integrated transportation system, ports involve various shared resources such as loading and unloading equipment, berths and storage yards, and operational personnel. Cost control is one of the core aspects of port operation and management. With the continuous increase in the scale and complexity of port operations, technologies for cost analysis and control based on operational data have gradually become an important research direction in the field of port operation optimization.

[0003] In the prior art, such as the prior art document with application number CN202512028736.0, a port cost tracing method based on multi-source data fusion is disclosed, including: S1, collecting multi-source data of port operations and performing standardization processing; S2, performing time and spatial alignment; S3, generating composite identifiers for operation units and constructing operation unit models to form operation unit datasets; S4, performing evidence modeling and fusion through Dirichlet evidence distribution to generate fusion results with confidence and conflict levels; S5, generating power curves from the energy consumption data in the fusion results and integrating to obtain energy consumption; processing personnel work hour records to obtain work hours; performing text extraction on financial voucher records to generate cost information; S6, identifying cost driving factors and using an improved Aumann–Shapley method to generate operation unit allocation value sets; S7, calculating the total cost of operation units and outputting cost tracing results. This invention significantly improves the accuracy and interpretability of port operation cost tracing. For example, prior art document CN202510131369.4 discloses a smart energy optimization control method and system for ports. This method involves collecting energy data from the target port area and obtaining historical energy consumption datasets; constructing a nonlinear regression model based on the historical energy consumption datasets to predict future energy consumption values; constructing confidence intervals based on the historical energy consumption datasets; comparing future energy consumption values ​​with the confidence intervals to determine abnormal energy consumption values; calculating optimization coefficients for each data point of the abnormal energy consumption values; determining the abnormal points corresponding to the abnormal energy consumption values ​​and the degree of abnormality of the abnormal points based on the optimization coefficients; and performing optimization control on the abnormal points based on the degree of abnormality.

[0004] While both of the aforementioned technical solutions have their own applicable scenarios and technical advantages, they still suffer from the following technical shortcomings: They cannot accurately calculate costs in advance for multiple operational plans to be implemented, taking into account the load variation characteristics of various shared resources in the port. For example, the port cost tracing method based on multi-source data fusion disclosed in CN202512028736.0 is a post-event cost accounting approach, only capable of allocating cost-driving factors and calculating costs based on already occurred operational data. It cannot perform pre-implementation cost prediction and multi-scenario comparison for operational plans to be implemented, and cannot provide prior data support for cost control decisions. Furthermore, the port smart energy optimization control scheme disclosed in CN202510131369.4 only focuses on energy consumption prediction and anomaly control from a single energy dimension, failing to cover the cost response patterns of various shared resources in the port, and not considering the impact of resource load superposition effects on costs when multiple operational measures work together. Therefore, it is difficult to support accurate comprehensive cost calculation under a combination of multiple measures. In view of this, we propose an operational decision-making system for cost control based on multi-dimensional big data analysis. Summary of the Invention

[0005] The purpose of this invention is to provide an operational decision-making system for cost control based on multi-dimensional big data analysis, in order to solve the problem mentioned in the background art of being unable to accurately calculate costs in advance for multiple scenarios for various operational plans to be implemented, combined with the load change characteristics of various shared resources in the port.

[0006] To address the aforementioned technical problems, the present invention aims to provide an operational decision-making system for cost control based on multi-dimensional big data analysis, comprising: The cost-related data collection and indexing unit receives heterogeneous operational data output from various port production business systems, and uses data integration and classification indexing techniques to standardize and collect cost-related data such as production operations, equipment maintenance, energy consumption, and manpower scheduling to obtain a structured port cost dataset. The port cost driver feature extraction and association unit receives the structured port cost dataset and uses correlation mapping and statistical feature extraction techniques to extract the corresponding correlation between shared resource load data and cost data, the cost change pattern corresponding to changes in shared resource load, the conversion relationship between operating measures and resource load, and the distribution of high resource load delay consumption, thereby generating a port cost driver feature map. The multi-scenario cost control simulation calculation unit receives the port cost driver characteristic map and externally input operational decision parameters to be evaluated. It uses resource load time series superposition and nonlinear cost response mapping techniques to first perform time series integration and superposition calculation on the current load of port shared resources and the delayed load triggered by high load in the resource domain to obtain the total load rate. Then, it generates time series costs based on the load-cost nonlinear response relationship mapping of shared resources. After load contribution decomposition processing, it obtains the multi-scenario cost control prediction results. The cost control and operation decision generation and verification unit receives the multi-scheme cost control prediction results, and uses decision matching and closed-loop verification technology to screen and adjust the parameters of the alternative operation schemes under cost constraints to obtain executable port cost control and operation decision instructions. The execution feedback and deviation comparison unit receives the port cost control operation decision instruction and the port's actual operating cost data. It uses data comparison technology to compare the actual cost and expected cost after the decision is executed, obtains the cost deviation feedback result, and sends it back to the port cost driver feature extraction and correlation unit for closed-loop calibration of the feature parameters of the port cost driver feature map.

[0007] As a further improvement to this technical solution, the cost-related data collection and indexing unit includes a data access module, a data cleaning and standardization module, a classification index construction module, and a structured dataset output module, wherein: The data access module connects to various production and business systems of the port and receives heterogeneous operational data such as production operations, equipment maintenance, energy consumption, and manpower scheduling. The data cleaning and standardization module performs format unification, missing value filling, and outlier removal on the received heterogeneous operational data, thereby completing the standardized collection of cost-related data. The classification index construction module classifies and labels the standardized cost association data according to the cost association dimension, and constructs the corresponding data index. The structured dataset output module encapsulates the cost-related data carrying the index and outputs a structured port cost dataset.

[0008] As a further improvement to this technical solution, the port cost driver feature extraction and correlation unit includes a correlation mapping construction module, a load change law statistics module, a measure load conversion construction module, a delay consumption distribution extraction module, and a feature map generation module, wherein: The association mapping construction module establishes a corresponding association between shared resource load data and cost data based on the structured port cost dataset using association mapping technology. The load change pattern statistics module uses statistical feature extraction techniques to extract the cost change patterns corresponding to the load changes of shared resources, based on the structured port cost dataset. The measure load conversion construction module extracts the correspondence between the operational measure parameters and the shared resource load increment based on the structured port cost dataset, and constructs the operational measure-resource load conversion relationship; The delay consumption distribution extraction module extracts delay consumption data under high load conditions of shared resources based on the structured port cost dataset, and generates a high load delay consumption distribution of resources. The feature map generation module generates a port cost driver feature map based on corresponding relationships, cost change patterns, operational measures-resource load conversion relationships, and resource high load delay consumption distribution.

[0009] As a further improvement to this technical solution, the multi-scenario cost control simulation calculation unit includes a resource load time-series profiling module, a resource load time-series overlay module, a nonlinear cost response mapping module, and a load contribution decomposition module, wherein: The resource load time series profiling module uses the port cost driver feature map as a basis to convert the externally input operational decision parameters to be evaluated into the current load increment of the corresponding port shared resources, and matches and generates the delayed load increment corresponding to the high resource load caused by the current load increment, and integrates them according to the time period to obtain the load time series sequence of the shared resources corresponding to each scheme. The resource load time series overlay module receives the load time series output by the resource load time series profiling module, takes the physical resource quantity as the overlay object, and overlays the shared resource load corresponding to each measure in the same scheme to the base load according to the time period to obtain the time series of the total load rate of shared resources, and completes the load aggregation of multiple measures in the resource input domain. The nonlinear cost response mapping module receives the total load rate time series output by the resource load time series superposition module. Based on the load-cost nonlinear response relationship of shared resources, it performs cost mapping calculation on the superimposed total load periodically to obtain the time series cost data of each scheme. The calculation logic is to aggregate resources first and then map costs. The load contribution decomposition module receives the total load rate time series and the load time series of each measure from the resource load time series superposition module, and combines the time series cost data from the nonlinear cost response mapping module to decompose the contribution of each measure to the total load and nonlinear cost increment, and integrates them to obtain the multi-scheme cost control prediction results.

[0010] As a further improvement to this technical solution, the process by which the resource load time-series profiling module generates a load time-series sequence includes the following steps: S31.1 Based on the operational measures-resource load conversion relationship in the port cost driver characteristic map, the externally input operational decision parameters to be evaluated are converted into the current load increment of the corresponding port shared resources; S31.2 Based on the distribution of high-load delay consumption of resources in the port cost driver feature map, match and generate the port shared resource delay load increment for each subsequent period corresponding to the current load increment. S31.3 Integrate the current load increment and the delayed load increment according to the time period to generate the load time sequence of the port shared resources corresponding to each scheme.

[0011] As a further improvement to this technical solution, the process by which the resource load time series overlay module obtains the time series of the total load rate of shared port resources includes the following steps: S32.1 Receive the port shared resource load time sequence corresponding to each measure in the same scheme output by the resource load time sequence profiling module, and superimpose the load increment of each measure according to the time period to obtain the port shared resource measure superimposed total load increment time sequence. S32.2. The time series of the total load increment of the port shared resources measures is superimposed with the time series of the port shared resources baseline load of the corresponding period to obtain the time series of the total load of the port shared resources. S32.3 Calculate the time series ratio of the total load of port shared resources to the rated capacity of the corresponding port shared resources on a periodic basis to obtain the time series ratio of the total load rate of port shared resources; where the first... Total load factor of port shared resources during the period The calculation formula is: ; In the formula, For the first The baseline load of shared port resources during the period; For the first The operational measures in the first The port shared resource load increment corresponding to the period includes the current period load increment component and the delayed load increment component; This refers to the rated capacity of shared resources for the corresponding port.

[0012] As a further improvement to this technical solution, the nonlinear cost response mapping module includes an equipment load energy consumption mapping submodule, a berth and yard load congestion mapping submodule, a manpower load working hour mapping submodule, and a time-series cost aggregation submodule, wherein: The equipment load energy consumption mapping submodule receives the total load rate time series of port shared resources for equipment, and calculates the corresponding cost of equipment energy consumption for each period based on the load-energy consumption physical operation characteristics of port loading and unloading equipment through a piecewise nonlinear response function. The berth yard load congestion mapping submodule receives the total load rate time series of port shared resources such as berth yards, and calculates the cost corresponding to congestion delay in each period based on the physical operation characteristics of load-congestion queuing of port berth yards through piecewise nonlinear response functions. The manpower load-hour mapping submodule receives the total load rate time series of manpower-related port shared resources, and calculates the corresponding cost of manpower hours for each period based on the physical operation characteristics of port operation manpower load-hour efficiency through a piecewise nonlinear response function. The time-series cost aggregation submodule aggregates the periodic cost data output by the equipment load energy consumption mapping submodule, the berth and yard load congestion mapping submodule, and the manpower load working hour mapping submodule to obtain the time-series cost data for each scheme.

[0013] As a further improvement to this technical solution, the process by which the load contribution decomposition module obtains the multi-scheme cost control prediction results includes the following steps: S34.1 Receive the port shared resource total load rate time series and port shared resource load time series corresponding to each measure output by the resource load time series overlay module, calculate the proportion of the load increment corresponding to each operation measure in the total load of port shared resources on a cycle-by-cycle basis, and obtain the load contribution proportion time series of each measure. S34.2. Combining the load rate segment threshold of the piecewise nonlinear response function, locate the time period during which the total load rate of the port's shared resources crosses the segment threshold, identify the operational measures that play a dominant role in the crossing of the load rate segment threshold, and obtain nonlinear cost increment attribution information. S34.3 Integrate the time-series cost data output by the nonlinear cost response mapping module, the time-series load contribution ratio of each measure, and the nonlinear cost increment attribution information to obtain the multi-scheme cost control prediction results.

[0014] As a further improvement to this technical solution, the cost control and operational decision generation and verification unit includes a solution cost constraint screening module and a decision closed-loop verification and adjustment module, wherein: The cost constraint screening module uses decision matching technology to select alternative business plans that meet the cost control objectives under cost constraints, based on the cost control prediction results of the multiple plans. The decision-making closed-loop verification and adjustment module uses closed-loop verification technology to verify the execution adaptability of alternative operating plans that meet the cost control objectives, based on the alternative operating plans output by the scheme cost constraint screening module. It also adjusts and optimizes the parameters of the alternative operating plans to obtain executable port cost control operating decision instructions.

[0015] As a further improvement to this technical solution, the execution feedback and deviation comparison unit includes a cost deviation comparison calculation module and a deviation feedback return calibration module, wherein: The cost deviation comparison and calculation module receives the port cost control and operation decision-making instructions and the actual operating cost data of the port. It uses data comparison technology to compare the actual cost after the decision is executed with the corresponding expected cost on a cycle basis to obtain the cost deviation feedback result. The deviation feedback and calibration module, based on the cost deviation feedback results output by the cost deviation comparison and calculation module, sends the cost deviation data back to the port cost driver feature extraction and correlation unit for feature parameter calibration of the port cost driver feature map.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention constructs a port cost driver feature map that covers the load-cost correspondence and the distribution of high load delay consumption. It employs multi-scenario simulation calculations and execution deviation closed-loop calibration by first superimposing the resource domain load time series and then nonlinear cost mapping. This solves the problem of difficulty in accurately calculating the costs of multiple scenarios for various operational plans to be implemented in advance. It enables accurate prediction of the costs of multiple types of shared resources in the port and provides quantitative support for cost control and operational decision-making. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall system framework structure of the present invention; The meanings of the labels in the diagram are as follows: 1. Cost-related data collection and indexing unit; 11. Data access module; 12. Data cleaning and standardization module; 13. Classification index construction module; 14. Structured dataset output module; 2. Port cost driver feature extraction and correlation unit; 21. Correlation mapping construction module; 22. Load change law statistics module; 23. Measures load conversion construction module; 24. Delay consumption distribution extraction module; 25. Feature map generation module; 3. Multi-scenario cost control simulation calculation unit; 31. Resource load time series profiling module; 32. Resource load time series overlay module; 33. Nonlinear cost response mapping module; 34. Load contribution decomposition module; 4. Cost control and operational decision generation and verification unit; 41. Solution cost constraint screening module; 42. Decision closed-loop verification and adjustment module; 5. Execution feedback and deviation comparison unit; 51. Cost deviation comparison calculation module; 52. Deviation feedback and calibration module. 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 1 As shown, this embodiment provides an operational decision-making system for cost control based on multi-dimensional big data analysis, including: Cost-related data collection and indexing unit 1 receives heterogeneous operational data output from various port production business systems. Using data integration and classification indexing techniques, it standardizes and collects cost-related data related to production operations, equipment maintenance, energy consumption, and manpower scheduling to obtain a structured port cost dataset. This includes a data access module 11, a data cleaning and standardization module 12, a classification index construction module 13, and a structured dataset output module 14. The data access module 11 connects to various production business systems of the port to receive heterogeneous operational data such as production operations, equipment maintenance, energy consumption, and manpower scheduling. Specifically, the data access module 11 connects to the container terminal operating system (TOS), equipment management system (EAM), energy metering and management system (EMS), and human resources management system (HRMS) through standardized data interfaces.

[0020] The system receives production operation data such as vessel berthing and departure plans, loading and unloading operation instructions, yard operation records, and equipment scheduling tasks from the Container Terminal Operating System (TOS); equipment operation and maintenance data such as equipment operating hours, start-up and shutdown records, fault repair work orders, and maintenance plans from the Equipment Management System (EAM); energy consumption data such as equipment electricity consumption, fuel consumption, and pipeline energy consumption from the Energy Metering and Management System (EMS); and human resource scheduling data such as personnel schedules, attendance records, work hour statistics, and team task assignments from the Human Resources Management System (HRMS). The data output formats of these systems include relational database records, text log files, XML / JSON interface messages, and spreadsheets.

[0021] Finally, the data access module 11 establishes an independent connection channel for each data source, acquires incremental data according to the set collection cycle, and adds a source system identifier to each raw data record to form a raw data buffer set with source tags.

[0022] The data cleaning and standardization module 12 performs format unification, missing value filling and outlier removal on the received heterogeneous operational data, and completes the standardized collection of cost-related data. The format standardization process includes: unifying the timestamp field to the format "YYYY-MM-DD HH:MM:SS"; measuring equipment runtime in hours; measuring electricity consumption in kilowatt-hours and fuel consumption in liters; measuring man-hours in person-hours; and measuring workload in standard containers or natural tons. For basic fields using different coding rules, such as resource number, berth number, and shift number, a unified coding mapping table is established for conversion.

[0023] Missing value filling is handled according to the field type, with corresponding strategies applied: for time-series numerical fields, such as hourly energy consumption readings of equipment, linear interpolation is used to fill the missing values ​​at the preceding and following time points; for categorical fields, such as job type and resource type, the most frequently occurring value is used to fill the missing values; records that cannot be reasonably filled using the above strategies are marked and their original missing state is retained.

[0024] Outlier removal is performed based on business rules and statistical rules: constraints are set such as the equipment running time must not exceed the corresponding natural time of the cycle, energy consumption must not be negative, and the workload of a single quay crane per unit time must not exceed the rated efficiency limit. Records that violate the constraints are removed. At the same time, records in the same category of data that deviate from the mean by more than three times the standard deviation are identified and removed. The removed records are entered into the abnormal data log for future reference.

[0025] The classification index construction module 13 classifies and labels the standardized cost association data according to the cost association dimension, and constructs the corresponding data index. Specifically, the classification index construction module 13 classifies and labels standardized cost-related data according to four dimensions: resource type, time period, cost type, and business process. Among them: Resource types are categorized into equipment resources, berth and yard resources, and human resources. Equipment resources include quay cranes, yard cranes, container trucks, reach stackers, and forklifts. Berth and yard resources include berths, front yards, and rear warehouses. Human resources include quay loaders, yard drivers, and tally clerks. The time cycle dimension is divided into six levels: annual, monthly, weekly, daily, shift, and hourly. Costs are categorized into five types: energy consumption costs, equipment maintenance costs, congestion and delay costs, labor and hour costs, and other costs. The business processes are divided into five categories: ship loading and unloading operations, horizontal transportation operations, yard storage and retrieval operations, gate collection and distribution operations, and auxiliary operations.

[0026] Furthermore, the classification index construction module 13 matches and determines classification tags under four dimensions for each standardized data record based on its source system, resource identifier, and business document number, adds four-element classification index tags, and constructs a four-level index structure of "resource type → time period → cost type → business process", which supports the retrieval of data subsets by any combination of dimensions.

[0027] The structured dataset output module 14 encapsulates the cost-related data carrying indexes to output a structured port cost dataset. Specifically, the structured dataset output module 14 encapsulates the cost-related data carrying four-dimensional classification index labels into a unified data record format. Each record contains an index field area and a data field area: the index field area records the classification label values ​​of four dimensions: resource type, time period, cost type, and business process; the data field area records the resource number, timestamp, load characteristics including equipment operating hours, berth occupancy time, manpower attendance hours, and quantitative fields such as workload, energy consumption, and cost incurred. The structured port cost dataset is organized and stored in the form of partitioned data tables, with the resource type dimension and cost type dimension as the primary partition key and the time period dimension as the sub-partition key. This dataset is output to the port cost driver feature extraction and association unit 2 through a standardized data interface for use in shared resource load-cost association mapping, load change pattern statistics, construction of operational measures-resource load conversion relationship, and extraction of resource high load delay consumption distribution.

[0028] Port cost driver feature extraction and association unit 2 receives a structured port cost dataset and uses correlation mapping and statistical feature extraction techniques to extract the corresponding correlation between shared resource load data and cost data, the cost change pattern corresponding to changes in shared resource load, the conversion relationship between operational measures and resource load, and the distribution of resource high load delay consumption, generating a port cost driver feature map. Port cost driver feature extraction and association unit 2 includes a correlation mapping construction module 21, a load change pattern statistics module 22, a measure load conversion construction module 23, a delay consumption distribution extraction module 24, and a feature map generation module 25.

[0029] In this embodiment, shared resource load data refers to quantitative data reflecting the degree of occupation or use of port shared resources within a specific time period. Equipment resources are measured in operating hours, berth and yard resources in occupancy duration, and manpower resources in attendance hours. Shared resource load increment refers to the increase in shared resource load data relative to the baseline load or the load of the previous period. This can be directly caused by operational measures or by delays in historical high-load states. Delayed consumption data refers to the additional shared resource load increment generated in subsequent time periods after a shared resource has been in a high-load state for a certain time period. High-load delayed consumption distribution refers to the distribution of delayed load increments arranged by subsequent period numbers after statistically processing the delayed consumption data corresponding to multiple high-load periods of the same type of shared resource.

[0030] The association mapping construction module 21 establishes a corresponding association between shared resource load data and cost data based on the structured port cost dataset and using association mapping technology. Specifically, the association mapping construction module 21 reads cost association data records one by one from the structured port cost dataset, and establishes a correspondence between the shared resource load data and cost incurred within the same time period, using the shared resource identifier and timestamp as the association key. The shared resource identifier includes equipment resource number, berth and yard resource number, and human resource number. The cost incurred includes energy consumption cost, equipment maintenance cost, congestion delay cost, human labor hour cost, and other costs related to the corresponding shared resource. In cases where multiple records of different cost types exist for the same shared resource within the same time period, mapping rows are established separately for each cost type. Each row contains the shared resource identifier, time period, shared resource load data, and cost incurred for that cost type. The association mapping construction module 21 organizes the above correspondence into a load-cost mapping table, which is output as the corresponding association between shared resource load data and cost data. The load-cost mapping table is stored separately according to shared resource type for use by the load change pattern statistics module 22 and the feature map generation module 25.

[0031] The load change pattern statistics module 22 is based on the structured port cost dataset and uses statistical feature extraction techniques to extract the cost change pattern corresponding to the load change of shared resources. Specifically, the load change pattern statistics module 22 groups records in the load-cost mapping table according to load rate, based on shared resource type and cost type. The load rate is the ratio of the shared resource load data to the rated capacity of the shared resource. The lower bound of each load rate range is denoted as The upper boundary is recorded as The average cost of all records within the interval is denoted as . The load change pattern statistics module 22 uses the median value of the load rate interval as the independent variable, and... Using the load as the dependent variable, a discrete load-cost sequence is obtained. The load change pattern statistics module 22 further calculates the difference in the average cost between adjacent load rate intervals, which is used as the cost change corresponding to the load change of the shared resource.

[0032] The aforementioned discrete load-cost correspondence sequence and the cost change in adjacent intervals together constitute the cost change law corresponding to the load change of shared resources, reflecting the nonlinear change characteristics of the cost corresponding to the change of shared resource load rate from low to high, and are used by the nonlinear cost response mapping module 33 in the multi-scenario cost control simulation calculation unit 3.

[0033] The measure load conversion module 23 extracts the correspondence between operational measure parameters and shared resource load increments based on the structured port cost dataset, and constructs the operational measure-resource load conversion relationship; Specifically, the measure load conversion construction module 23 filters data records before and after the implementation of operational measures from the structured port cost dataset. These operational measure implementation records originate from the operation plan adjustment records and port production scheduling records in the Container Terminal Operating System (TOS), including measure type, measure parameter values, and implementation time. After being collected by the data access module 11, these records are incorporated into the structured port cost dataset. The measure load conversion construction module 23 associates the measure parameters with the shared resource load data for the corresponding time period based on the implementation time of the operational measures. It establishes a correspondence between the two, using the change in operational measure parameters as the independent variable and the increase in shared resource load as the dependent variable. Operational measure parameters include increasing the number of work lines, adjusting shift duration, extending equipment operating time, and increasing temporary personnel input. The measure load conversion construction module 23 statistically analyzes historical corresponding data of the same type of operational measures acting on the same type of shared resources, calculates the average increase in shared resource load corresponding to a unit change in operational measure parameters, and uses this average as the operational measure-resource load conversion coefficient. For the first... Type of operational measures apply to the first In the case of shared resources, the conversion relationship between operational measures and resource load is expressed as follows: ; In the formula, For the first The first type of business measures caused Incremental load on shared resources For the first Changes in parameters of similar operational measures For the first Type of business measures for the first Load conversion factor for shared resources. For situations where operational parameters have discrete values, such as increasing the number of production lines, Take a non-negative integer value; for operational measure parameters that are continuously variable, such as extending equipment operating time, In hours.

[0034] The measure load conversion construction module 23 organizes the conversion coefficients of each business measure-resource load according to the shared resource type and business measure type into a conversion coefficient table, which is used as the output of the business measure-resource load conversion relationship for use by the resource load time sequence profiling module 31 in the multi-scenario cost control simulation calculation unit 3.

[0035] The delay consumption distribution extraction module 24 extracts delay consumption data under high load conditions of shared resources based on the structured port cost dataset, and generates a delay consumption distribution under high load conditions. Specifically, the delay consumption distribution extraction module 24 performs periodic load rate judgment on each shared resource in the structured port cost dataset. When a certain shared resource in the period... When the load rate of a cycle exceeds the high load threshold corresponding to the shared resource, the cycle is identified as a high load cycle. The high load threshold is determined based on the type of shared resource and historical operating data. For equipment resources, it is a preset percentage of the rated capacity of the equipment; for berth and yard resources, it is a preset percentage of the rated capacity of the berth and yard; and for manpower resources, it is a preset percentage of the rated manpower hours. For each high load cycle, the delay consumption distribution extraction module 24 extracts the continuous delay consumption distribution after that cycle. Latency consumption data of the same shared resource within a period, i.e., the data after the high load period. The additional shared resource load increase caused by high load conditions within a cycle. Take 1 to . The value is determined based on the type of shared resource and historical data.

[0036] Meanwhile, the delay consumption distribution extraction module 24 statistically analyzes all high-load cycles of similar shared resources and calculates the number of cycles after the high-load cycle. The average latency consumption data for each period is denoted as . Generate a high-load latency consumption distribution of resources: ; In the formula, For high-load latency consumption distribution of resources, The first after a high load cycle The average delay load increment over each period, with dimensions consistent with the shared resource load data. (This refers to the load data for equipment-related resources.) Based on operating hours, berth and yard resources Human resources, in terms of time spent Calculated by attendance hours; The statistical period number is the value range. Used to index the first [period] after a high-load cycle. Each cycle. The distribution of high-load latency consumption of resources is stored separately according to the type of shared resource, and is used by the resource load time-series profiling module 31 in the multi-scenario cost control simulation calculation unit 3 for incremental matching and generation of latency load.

[0037] The feature map generation module 25 generates a port cost driver feature map based on corresponding relationships, cost change patterns, operational measures-resource load conversion relationships, and resource high load delay consumption distribution. The feature map generation module 25 is also responsible for receiving calibration data packets from the execution feedback and deviation comparison unit 5, adjusting the feature parameters according to the cost deviation data in the calibration data packets, and regenerating the calibrated port cost driver feature map.

[0038] Specifically, the feature map generation module 25 organizes the load-cost mapping table output by the association mapping construction module 21, the cost change pattern corresponding to load change output by the load change pattern statistics module 22, the conversion coefficient table output by the measure load conversion construction module 23, and the resource high load delay consumption distribution output by the delay consumption distribution extraction module 24 into a map data structure according to the shared resource type. At the same time, the feature map generation module 25 extracts the rated capacity attribute data of each shared resource from the structured port cost dataset, and statistically obtains the baseline load time series of each shared resource based on the historical normal operating cycle load data in the structured port cost dataset, and incorporates them into the map data structure.

[0039] Furthermore, the port cost driver characteristic map uses shared resource types as the first-level nodes and feature categories as the second-level nodes. Feature categories include load-cost correlation mapping, load change cost patterns, operational measures-resource load conversion relationships, resource high-load delay consumption distribution, baseline load time series, and resource rated capacity. Each second-level node corresponds to a data table, containing a load-cost mapping table, a load-cost correspondence sequence, a conversion factor table, a resource high-load delay consumption distribution sequence, a baseline load time series sequence, and a rated capacity data table for that shared resource type. Finally, the port cost driver characteristic map is output to the multi-scenario cost control simulation calculation unit 3 through a standardized data interface.

[0040] Multi-scenario cost control simulation calculation unit 3 receives the port cost driver characteristic map and the external input operation decision parameters to be evaluated. It adopts the resource load time series superposition and nonlinear cost response mapping technology. First, it performs time series integration and superposition calculation on the current load of the port shared resources corresponding to each scheme and the delayed load triggered by high load in the resource domain to obtain the total load rate. Then, it generates time series cost based on the load-cost nonlinear response relationship mapping of shared resources. After load contribution decomposition processing, it obtains the multi-scheme cost control prediction results. Specifically, the multi-scenario cost control simulation unit 3 receives the port cost driver feature map output by the port cost driver feature extraction and correlation unit 2, as well as the externally input operational decision parameters to be evaluated. The externally input operational decision parameters refer to the parameterized expression of specific measures given by port management personnel for alternative operational plans, including the type, quantity, or duration of the measures, and the effective period of the measures. Internally, the multi-scenario cost control simulation unit 3 is serially connected in the order of resource load time-series profiling module 31, resource load time-series overlay module 32, nonlinear cost response mapping module 33, and load contribution decomposition module 34, with the output of the previous module serving as the input of the next. Specifically, the resource load time-series profiling module 31 and the resource load time-series overlay module 32 complete the time-series integration and overlay of load in the resource domain; the nonlinear cost response mapping module 33 completes the mapping from total load rate to time-series cost in the cost domain; and the load contribution decomposition module 34 completes the attribution of nonlinear cost increment measures in the contribution domain.

[0041] In this embodiment, the shared resource load rate is calculated by aggregation according to the shared resource type, and the time period granularity is consistent with the period granularity of the benchmark load time series in the port cost driver feature map. All parameters involved in this embodiment, including load conversion coefficient, high load threshold, resource high load delay consumption distribution, benchmark load time series, slope parameter and intercept parameter in piecewise nonlinear response function, are read from the corresponding data table in the port cost driver feature map, and their calibration process is completed in port cost driver feature extraction and association unit 2.

[0042] Furthermore, unlike the conventional approach of calculating the load of each operational measure independently, mapping costs separately, and then summing them, this embodiment first uses physical resource quantity as the overlay object in the resource load time-series overlay module 32. It then uniformly overlays the current load increment and delayed load increment caused by each measure within the same scheme onto the baseline load, obtaining the true total load rate of shared resources under the combined effect of various measures, before proceeding to the cost mapping stage. The reason for this order is that the cost response of port shared resources exhibits non-linear characteristics when the load rate exceeds the corresponding threshold. If the cost of each individual measure is calculated before load aggregation, the non-linear segment in the cost mapping function cannot be triggered, leading to a systematic underestimation of costs under high load conditions. Simultaneously, this embodiment integrates the current load increment and the delayed load increment triggered by high load according to time periods in the resource load time-series profiling module 31, ensuring that cost prediction covers the entire time range affected by delays and consumption during the implementation period of the measure, rather than only reflecting the static load during the implementation period. Based on this, the load contribution decomposition module 34 addresses the issue that nonlinear costs cannot be simply allocated proportionally. By combining the load rate segmentation threshold of the piecewise nonlinear response function, it identifies the operational measures that play a dominant role in crossing the threshold, attributing the nonlinear cost increment to specific measures, and providing an operational basis for subsequent scheme selection and parameter adjustment.

[0043] In this embodiment, the load increments caused by various operational measures are superimposed linearly. This method is suitable for situations where the shared resource load rate does not exceed the rated capacity by a large margin and there is no significant mutual influence between measures. The current load increment refers to the increase in shared resource load directly caused by the operational measures within the current time period; the delayed load increment refers to the increase in shared resource load in subsequent periods triggered by the shared resources being in a high-load state in the current or historical period; the load time sequence refers to the shared resource load increment data corresponding to various measures within the same scheme, arranged by time period, including the current component and the delayed component. The delayed load increment does not participate in the triggering judgment of high-load states, and the delayed load increment does not recursively trigger new delayed loads.

[0044] The resource load time series profiling module 31, based on the port cost driver characteristic map, converts the externally input operational decision parameters to be evaluated into the current period load increment of the corresponding port shared resources, and matches and generates the delayed load increment corresponding to the high resource load caused by the current load increment. It integrates the load time series sequence of the shared resources corresponding to each scheme according to the time period. The process of generating the load time series sequence includes the following steps: S31.1 Based on the conversion relationship between operating measures and resource load in the port cost driver characteristic map, the externally input operating decision parameters to be evaluated are converted into the current load increment of the corresponding port shared resources. In S31.1, the resource load time-series profiling module 31 reads the operational measure-resource load conversion relationship from the port cost driver characteristic map. This conversion relationship is stored in the form of a conversion coefficient table, recording the load conversion coefficients of each operational measure type to each shared resource type. For each operational measure in the externally input operational decision parameters to be evaluated, the resource load time-series profiling module 31 determines the operational measure type and shared resource type corresponding to the measure, reads the corresponding load conversion coefficient from the conversion coefficient table, and calculates the load conversion coefficient according to the following relationship. The operational measures in the first Current load increment for the period: ; In the formula, For the first The load conversion factor of the operational measure on the shared resources it affects is read from the conversion factor table in the port cost driver characteristic map according to the measure type and the type of shared resources it affects, and is consistent with the corresponding item in the conversion factor table output by the measure load conversion construction module 23. For the first The operational measures in the first The change in the periodic parameter indicates a positive value for increased load and a negative value for decreased load. For cases where the measure parameter has discrete values, such as increasing the number of production lines, Take the integer value; for cases where the measure parameter is a continuous value, such as adjusting shift duration, In hours. The dimensions of the load data are consistent with those of the shared resources being affected: equipment resources are measured in operating hours, berth and yard resources in occupancy time, and human resources in attendance hours. When an operational measure applies to multiple shared resources simultaneously, the corresponding load conversion factor is read for each shared resource, and the current period load increment is calculated.

[0045] S31.2 Based on the distribution of high-load delay consumption of resources in the port cost driver feature map, match and generate the port shared resource delay load increment for each subsequent period corresponding to the current load increment. In S31.2, the resource load time-series profiling module 31 reads the resource high load delay consumption distribution from the port cost driver feature map. The resource high load delay consumption distribution is presented as a distribution sequence. Stored in the form of , where The first after a high load cycle Average delay load increment over each period This is the arithmetic mean of latency consumption data for all high-load cycles of similar shared resources. This represents the number of statistical periods for the delayed consumption distribution. Take 1 to ,Right now From 1 to The statistical period number variable, for The upper limit of the value, for The mean delay load increment corresponding to the time period represents the mean delay load increment of the last period within the statistical range.

[0046] In this embodiment, the resource load time-series profiling module 31 first determines the first... The operational measures in the first Does the current period's load increment trigger a high resource load state? The determination method is: [The text abruptly ends here, likely due to an incomplete sentence or missing information.] The baseline load of the periodic shared resources and the first The sum of the current period load increments of all operating measures within the period is used to obtain the result. The estimated total load for the period; calculate the ratio of the estimated total load to the rated capacity of the shared resource to obtain the value of the first period. The projected total load rate for the period; when the projected total load rate exceeds the high load threshold corresponding to the shared resource, the current period's load increment is determined to trigger a high load state for the resource. The high load threshold is obtained from the data table corresponding to the port cost driver characteristic map, with equipment resources, berth and yard resources, and manpower resources each having their own high load threshold. When the projected total load rate equals the high load threshold, delayed load generation is not triggered.

[0047] When it is determined that the current period's load increment triggers a high resource load state, the resource load time sequence profiling module 31 generates the first [database name] according to the following relationship. The operational measures in the first Periodic triggering, in the first The incremental delay load generated by the period : ; In the formula, For the first The operational measures in the first The periodic delay load trigger ratio factor is determined by the following formula: ; In the formula, For the first The sum of current load increments from all operating measures that result in a positive current load increment within the period. Traversing the first All operational measures that result in a positive increase in current load during the period are included; negative measures are not included in the summation of the denominator. This is a dimensionless ratio. When... At this time, no delay load increment is generated. The dimensions are consistent with those of the corresponding shared resource load data.

[0048] When it is determined that the current load increment does not trigger a high resource load state, the resource load timing profile module 31 does not generate a delayed load increment, i.e. .

[0049] S31.3 Integrate the current load increment and the delayed load increment according to the time period to generate the load time sequence of the port shared resources corresponding to each scheme.

[0050] In S31.3, the resource load time-series profiling module 31 performs a process for the first... The operational measures are integrated based on the load increment within each time period. The operational measures in the first Periodic load increment It is obtained by superimposing the current load increment component and the delayed load increment component: ; In the formula, For the first The operational measures in the first Periodic triggering, in the first The incremental delay load generated by the cycle, i.e., the previous cycle The high load state of the cycle is currently in the first cycle. The incremental component of delayed load formed periodically; For the first Before the cycle Triggered within the current cycle and continued until the next cycle. The sum of all latency load increments for each period. Resource load time-series profiling module 31 will display the latency load increments for each period. Arranged in chronological order, we get the [number]. The load time sequence corresponding to each operational measure. The load time sequences of each operational measure within the same plan constitute the shared resource load time sequence set corresponding to the plan, and are output to the resource load time sequence overlay module 32.

[0051] The resource load time series overlay module 32 receives the load time series output by the resource load time series profiling module 31. Taking physical resource quantity as the overlay object, it overlays the shared resource loads corresponding to various measures within the same scheme onto the baseline load according to the time period to obtain the total shared resource load rate time series. The process of obtaining the port shared resource total load rate time series includes the following steps: S32.1 Receive the port shared resource load time sequence corresponding to each measure in the same scheme output by the resource load time sequence profiling module 31, and superimpose the load increment of each measure according to the time period to obtain the port shared resource measure superimposed total load increment time sequence. In S32.1, the resource load timing overlay module 32 overlays all operational measures acting on the same shared resource within the same scheme in the first... The load increments of each cycle are added together to obtain the first cycle. Periodic measures combined with total load increments: ; In the formula, For the first The measures for sharing resources during the cycle are superimposed on the total load increment; Iterate through all operational measures that apply to the shared resource within the same scheme.

[0052] S32.2. The time series of the total load increment of the port shared resources measures is superimposed with the time series of the port shared resources baseline load of the corresponding period to obtain the time series of the total load of the port shared resources. In S32.2, the resource load time series overlay module 32 reads the baseline load time series of the shared resource from the port cost driver feature map. The baseline load refers to the basic load of the shared resource under normal operating conditions without overlaying the load increment of the operational measures to be evaluated. The baseline load time series is obtained by the port cost driver feature extraction and correlation unit 2 based on the structured port cost dataset. The resource load time series overlay module 32 will then... The periodic measures are superimposed on the total load increment and the shared resource. The baseline load of each cycle is added together to obtain the first cycle. Total load per cycle: ; In the formula, For the first The total load on the shared resources during the period; For the first The baseline load of the shared resource during the period.

[0053] S32.3 The total load time series of port shared resources is calculated by periodically comparing it with the rated capacity of the corresponding port shared resources to obtain the total load rate time series of port shared resources. Specifically, the resource load time series overlay module 32 reads the rated capacity of the shared resource from the port cost driver characteristic map. Rated capacity refers to the maximum load capacity that the shared resource can bear within a safe or reasonable operating range. For equipment resources, it is calculated based on the rated operating hours; for berth and yard resources, it is calculated based on the rated occupancy time; and for human resources, it is calculated based on the rated attendance hours. The total load factor for the cycle is calculated using the following formula: ; In the formula, For the first The baseline load of shared port resources during the period; For the first The operational measures in the first The port shared resource load increment corresponding to the period includes the current period load increment component and the delayed load increment component; This refers to the rated capacity of shared resources for the corresponding port.

[0054] The nonlinear cost response mapping module 33 receives the total load rate time series output by the resource load time series superposition module 32. Based on the load-cost nonlinear response relationship of shared resources, it performs cost mapping calculations on the superimposed total load periodically to obtain the time series cost data of each scheme. The calculation logic adopts a first resource aggregation, then cost mapping approach. Specifically, the nonlinear cost response mapping module 33 receives the total load rate time series of each shared resource output by the resource load time series superposition module 32, distributes it according to the shared resource type to the equipment load energy consumption mapping submodule, the berth yard load congestion mapping submodule, and the manpower load working hour mapping submodule for processing, and then the time series cost aggregation submodule aggregates the outputs of the three submodules. Since the cost response of shared resources exhibits different characteristics in the low load rate range and the high load rate range, this embodiment sets piecewise nonlinear response functions for the three types of shared resources—equipment, berth yard, and manpower—to match the cost mapping results with the physical operating characteristics of each type of resource, rather than using a uniform function for indiscriminate mapping. When the total load rate exceeds 1, the above piecewise functions are still calculated according to their respective high load range function forms to reflect the additional costs under overload conditions.

[0055] The equipment load energy consumption mapping submodule receives the total load rate time series of port shared resources for equipment, and calculates the corresponding cost of equipment energy consumption for each period based on the load-energy consumption physical operation characteristics of port loading and unloading equipment through a piecewise nonlinear response function. Specifically, the equipment load energy consumption mapping submodule reads the cost change patterns corresponding to load changes of shared equipment resources from the port cost driver characteristic map, and determines the piecewise nonlinear response function of the shared equipment resources based on these patterns. This piecewise nonlinear response function uses the total load rate as the independent variable and the cost corresponding to equipment energy consumption as the dependent variable, and includes at least one load rate segmentation threshold, denoted as... . No. The cost corresponding to the energy consumption of periodic equipment is calculated using the following formula: ; In the formula, This is the first slope parameter for shared resources of the device class in the low load rate range. This is the second slope parameter for shared resources of the device class within the high load rate range, and ; The intercept parameter of the piecewise function for shared resources of device class satisfies This ensures the piecewise function remains continuous at the threshold. The parameters mentioned above are determined by the cost variation pattern corresponding to load changes in shared resources across device classes, reflecting the physical characteristic of accelerated energy consumption increases under high load operation.

[0056] The berth yard load congestion mapping submodule receives the total load rate time series of port shared resources such as berth yards and calculates the cost corresponding to congestion delay in each period based on the physical operation characteristics of load-congestion queuing of port berth yards through piecewise nonlinear response functions. Specifically, the berth yard load congestion mapping submodule reads the cost change patterns corresponding to load changes of shared resources like berth yards from the port cost driver characteristic map, and determines the piecewise nonlinear response function of these shared resources based on these patterns. This piecewise nonlinear response function uses the total load rate as the independent variable and the cost corresponding to congestion delays as the dependent variable, and includes at least one load rate segmentation threshold, denoted as... . No. The cost of cyclical congestion delays is calculated using the following formula: ; In the formula, This refers to the nonlinear growth parameters of shared resources such as berths and yards within high load ranges. These parameters are determined by the cost change patterns corresponding to load variations in shared resources like berths and yards, reflecting the physical characteristic that delay costs accelerate after exceeding congestion thresholds.

[0057] The manpower load-hour mapping submodule receives the total load rate time series of manpower-related port shared resources, and calculates the corresponding cost of manpower hours for each period based on the physical operation characteristics of port operation manpower load-hour efficiency through a piecewise nonlinear response function. Specifically, the manpower load-hour mapping submodule reads the cost change patterns corresponding to load changes in shared manpower resources from the port cost driver characteristic map, and determines the piecewise nonlinear response function of shared manpower resources based on these patterns. This piecewise nonlinear response function uses the total load rate as the independent variable and the cost corresponding to manpower hours as the dependent variable, and includes at least one load rate segmentation threshold, denoted as... . No. The cost corresponding to the periodic labor hours is calculated using the following formula: ; In the formula, This represents the first slope parameter for shared human resources within the low load rate range. This is the second slope parameter for shared human resources within the high load rate range, and ; The intercept parameter of the piecewise function for shared human resources satisfies... This ensures the piecewise function remains continuous at the threshold. The parameters mentioned above are determined by the cost variation patterns corresponding to changes in the load of shared human resources, reflecting the physical characteristics of decreased efficiency due to overtime pay and fatigue under high load conditions.

[0058] The time-series cost aggregation submodule aggregates the periodic cost data output by the equipment load energy consumption mapping submodule, the berth and yard load congestion mapping submodule, and the manpower load working hour mapping submodule to obtain the time-series cost data for each scheme.

[0059] Specifically, the time-series cost aggregation submodule adds up the costs corresponding to equipment energy consumption (output from the equipment load energy consumption mapping submodule), congestion delays (output from the berth and yard load congestion mapping submodule), and manpower hours (output from the manpower load hour mapping submodule) periodically according to the time period to obtain the cost of the solution in the [number]th time period. The time-series cost of the cycle: ; In the formula, For the first The total cost forecast for this scheme is calculated separately for equipment energy consumption cost, congestion delay cost, and labor cost, and then summed. Equipment energy consumption cost and labor cost are the full-scope costs, while congestion delay cost is the incremental cost incurred after exceeding the congestion threshold. The time-series cost aggregation submodule performs the above aggregation calculation periodically for each time period to obtain the time-series cost data for this scheme, which is then output to the load contribution decomposition module 34.

[0060] The load contribution decomposition module 34 receives the total load rate time series and the load time series of each measure from the resource load time series superposition module 32. Combined with the time series cost data from the nonlinear cost response mapping module 33, it decomposes the contribution of each measure to the total load and the nonlinear cost increment, integrating them to obtain the multi-scheme cost control prediction results. Specifically, the load contribution decomposition module 34 receives the total load rate time series output by the resource load time series superposition module 32, the load time series sequences corresponding to each operating measure, and the time series cost data output by the nonlinear cost response mapping module 33. It then performs processing steps S34.1 to S34.3 for each alternative operating scheme. Since the piecewise nonlinear response function causes a jump in cost increment when the load rate crosses the piecewise threshold, and this jump is caused by the combined effect of multiple measures, this module identifies the dominant measure that crosses the threshold and attributes the nonlinear cost increment to specific measures, solving the problem that aggregated costs are calculable but not interpretable.

[0061] The process by which the load contribution decomposition module 34 obtains the multi-scheme cost control prediction results includes the following steps: S34.1 Receive the port shared resource total load rate time sequence and the port shared resource load time sequence corresponding to each measure output by the resource load time sequence overlay module 32, calculate the proportion of the load increment corresponding to each operation measure in the total load of the port shared resources on a cycle-by-cycle basis, and obtain the load contribution proportion time sequence of each measure. In S34.1, the load contribution decomposition module 34 decomposes the load contribution into the first load. Within the cycle The percentage of the load increment corresponding to each operational measure in the total load of the shared resources is calculated using the following formula: ; In the formula, For the first The operational measures in the first The load contribution percentage for each period is expressed as a percentage. This percentage uses the total load as the denominator, which includes the baseline load. Therefore, the sum of the load contribution percentages of each measure within the same period is not equal to 100%. The load contribution decomposition module 34 calculates the load contribution percentage of each operational measure for each time period, obtaining the time sequence of the load contribution percentages of each measure.

[0062] S34.2. Combining the load rate segment threshold of the piecewise nonlinear response function, locate the time period during which the total load rate of the port's shared resources crosses the segment threshold, identify the operational measures that play a dominant role in the crossing of the load rate segment threshold, and obtain nonlinear cost increment attribution information. In S34.2, the load contribution decomposition module 34 obtains the load rate segment threshold of the piecewise nonlinear response function corresponding to each shared resource from the nonlinear cost response mapping module 33. The load contribution decomposition module 34 compares the total load rate with the load rate segment threshold periodically. When the... The total load factor of the period is lower than a certain load factor segment threshold. When the total load rate of a period exceeds the threshold of that load rate segment, the first period is determined to be higher. The cycle is a cross-cycle.

[0063] For each cycle, the load contribution decomposition module 34 first determines the measures to be added to the total load increment. Is it greater than 0? When this span is caused by a change in the baseline load timing, no attribution record is generated. At that time, the load contribution decomposition module 34 identifies the operating measures that play a dominant role in crossing the load rate segment threshold. The identification rule is as follows: compare the load increment of each operating measure within the crossing period, and identify the operating measure with the largest load increment as the dominant measure. When multiple operating measures have the same largest load increment, all of them are identified as dominant measures and jointly bear the attribution of the nonlinear cost increment corresponding to the crossing period. The load contribution decomposition module 34 records the crossing period number, the crossed load rate segment threshold, the dominant measure identifier, the load increment corresponding to the dominant measure, and the nonlinear cost increment amount corresponding to the crossing period. This amount is the difference between the actual time-series cost of the crossing period and the theoretical cost calculated according to the corresponding lower segment cost function if the total load rate had not crossed the segment threshold, forming nonlinear cost increment attribution information. When no threshold crossing occurs in a certain period, no attribution record is generated.

[0064] S34.3 Integrate the time-series cost data output by the nonlinear cost response mapping module 33, the time-series and nonlinear cost increment attribution information of the load contribution ratio of each measure, and obtain the multi-scheme cost control prediction results.

[0065] In S34.3, the load contribution decomposition module 34 organizes and integrates the time-series cost data output by the nonlinear cost response mapping module 33, the time-series load contribution ratio of each measure calculated in S34.1, and the nonlinear cost increment attribution information generated in S34.2, according to the scheme number and shared resource type, into a multi-scheme cost control prediction result. The multi-scheme cost control prediction result includes: the time-series cost data of each scheme, the time-series load contribution ratio of each operational measure within each scheme, and the nonlinear cost increment attribution information of each scheme. The multi-scheme cost control prediction result is output to the cost control operational decision generation and verification unit 4 through a standardized data interface for use in scheme cost constraint screening and decision closed-loop verification and adjustment.

[0066] The cost control and operational decision generation and verification unit 4 receives multi-scheme cost control prediction results and uses decision matching and closed-loop verification techniques to screen and adjust parameters of candidate operational schemes under cost constraints, resulting in executable port cost control and operational decision instructions. The multi-scheme cost control prediction results include time-series cost data for each scheme, the time-series load contribution ratio of each operational measure within each scheme, and the attribution information for nonlinear cost increments for each scheme. The cost control and operational decision generation and verification unit 4 includes a scheme cost constraint screening module 41 and a decision closed-loop verification and adjustment module 42, wherein: The scheme cost constraint screening module 41 uses decision matching technology to screen out alternative business schemes that meet the cost control objectives based on the multi-scheme cost control prediction results and under cost constraints. Specifically, the scheme cost constraint screening module 41 extracts the time-series cost data of each scheme from the multi-scheme cost control prediction results, and adds up the time-series costs of the scheme in each time period to obtain the total cost of the scheme within the evaluation period. The scheme cost constraint screening module 41 receives externally input cost control target parameters, including a cost ceiling and a cost control priority. The cost ceiling refers to the maximum allowed total cost of the scheme within the evaluation period. The cost control priority is used to rank the alternative business schemes that meet the cost constraints, with higher-priority schemes being output as alternatives first. The scheme cost constraint screening module 41 compares the total cost of each scheme with the cost ceiling: when the total cost of a scheme does not exceed the cost ceiling, the scheme is marked as a feasible alternative that meets the cost constraints; when the total cost of a scheme exceeds the cost ceiling, the scheme is eliminated.

[0067] When the total cost of all alternative operating plans exceeds the cost ceiling, the plan cost constraint screening module 41 selects the plan with the closest total cost to the cost ceiling as the alternative plan and marks it as the plan to be adjusted, which is then handed over to the decision closed-loop verification and adjustment module 42 for parameter adjustment. When multiple plans have the same total cost and none of them exceed the cost ceiling, the plan cost constraint screening module 41 selects the plan with the fewest leading measures as the priority alternative plan based on the number of leading measures corresponding to the nonlinear cost increment attribution information in each plan, and outputs the remaining plans as alternative plans as well. For alternative operating plans that meet the cost constraints, the plan cost constraint screening module 41 sorts the alternative plans according to the cost control priority, with the plan with higher priority listed first. The plan cost constraint screening module 41 outputs the screened alternative operating plans and their corresponding time-series cost data, load contribution ratio time-series data, and nonlinear cost increment attribution information to the decision closed-loop verification and adjustment module 42.

[0068] The decision-making closed-loop verification and adjustment module 42 uses closed-loop verification technology to verify the execution adaptability of alternative operating plans that meet the cost control objectives, based on the alternative operating plans output by the scheme cost constraint screening module 41. The module also adjusts and optimizes the parameters of the alternative operating plans to obtain executable port cost control operating decision instructions.

[0069] Specifically, the decision-making closed-loop verification and adjustment module 42 receives the alternative operating schemes and their corresponding time-series cost data, load contribution ratio time-series and nonlinear cost increment attribution information output by the scheme cost constraint screening module 41, and performs adaptability verification on each alternative operating scheme.

[0070] The adaptability verification process comprises two parts: measure parameter range verification and shared resource availability verification. Measure parameter range verification checks whether the parameter values ​​of each operational measure in the alternative operating plan are within the allowable range stipulated by port operation management regulations, including the upper limit of the number of work lines, the upper limit of equipment operating time, and the upper limit of personnel input. Shared resource availability verification checks whether the total load rate of each shared resource in the alternative operating plan exceeds the maximum allowable load rate for that shared resource in any given time period. This maximum load rate is obtained from the corresponding data table in the port cost driver characteristic graph. If the total load rate of any shared resource in the alternative operating plan exceeds the maximum allowable load rate in any given time period, the adaptability verification for that plan is deemed to have failed.

[0071] When the alternative business plan passes the adaptability verification, the decision closed-loop verification and adjustment module 42 takes the measure type, measure parameters, and measure effective period in the alternative business plan as executable content and generates port cost control business decision instructions.

[0072] When a candidate business plan fails the adaptability verification, the decision-making closed-loop verification and adjustment module 42 adjusts and optimizes the parameters of the candidate business plan. The adjustment and optimization method is as follows: locate the shared resources that failed the verification and their corresponding overload time periods, read the dominant measure identifier corresponding to the time period from the nonlinear cost increment attribution information, and reduce the parameter value of the dominant measure by a set step size; the reduced parameter value is resubmitted into the multi-scenario cost control simulation calculation unit 3 for simulation calculation to obtain the updated multi-plan cost control prediction results; the decision-making closed-loop verification and adjustment module 42 performs adaptability verification on the updated prediction results again until the verification passes or the preset maximum number of adjustments is reached. If the adaptation verification still fails after reaching the preset maximum number of adjustments, the candidate business plan is removed.

[0073] Finally, the decision-making closed-loop verification and adjustment module 42 takes the alternative operating plan that has passed the adaptability verification as the decision result and generates a port cost control operating decision instruction. The port cost control operating decision instruction includes the plan number, measure type, measure parameters, measure effective period, expected total cost of the plan, and expected time-series cost data. The port cost control operating decision instruction is output to the execution feedback and deviation comparison unit 5 through a standardized data interface, and simultaneously distributed to various port production business systems for execution.

[0074] The execution feedback and deviation comparison unit 5 receives port cost control operation decision instructions and actual port operating cost data. Using data comparison technology, it compares the actual cost with the expected cost after the decision is implemented, obtaining cost deviation feedback results, which are then transmitted back to the port cost driver feature extraction and association unit 2 for closed-loop calibration of the feature parameters of the port cost driver feature map. The port cost control operation decision instructions include the scheme number, measure type, measure parameters, measure effective period, total expected cost of the scheme, and expected time-series cost data. The actual port operating cost data is collected and output by the multi-dimensional port cost association data collection and indexing unit 1, including the actual cost incurred by each shared resource in each time period. The execution feedback and deviation comparison unit 5 includes a cost deviation comparison calculation module 51 and a deviation feedback feedback calibration module 52, wherein: The cost deviation comparison and calculation module 51 receives port cost control and operation decision-making instructions and actual port operating cost data. It uses data comparison technology to compare the actual cost after the decision is implemented with the corresponding expected cost on a cycle basis to obtain cost deviation feedback results. Specifically, the cost deviation comparison and calculation module 51 extracts the expected time-series cost data of the plan from the port cost control operation decision-making instructions, and extracts the actual cost incurred in each time period corresponding to the plan from the port's actual operating cost data. The cost deviation comparison and calculation module 51 compares the actual cost with the expected cost one by one according to the time period, subtracting the expected cost of the corresponding period from the actual cost incurred in each period to obtain the cost deviation for that period. The cost deviation comparison and calculation module 51 also divides the cost deviation of each period by the expected cost of the corresponding period to obtain the cost deviation rate for that period. When the expected cost of a certain period is zero, the cost deviation rate for that period uses the absolute value of the cost deviation as a substitute measure.

[0075] Finally, the cost deviation comparison and calculation module 51 organizes the cost deviations and cost deviation rates for each period into cost deviation feedback results according to the shared resource type and time period. The cost deviation feedback results include: scheme number, time period, shared resource type, expected cost, actual cost, cost deviation, and cost deviation rate. The cost deviation comparison and calculation module 51 outputs the cost deviation feedback results to the deviation feedback feedback calibration module 52.

[0076] Based on the cost deviation feedback results output by the cost deviation comparison and calculation module 51, the deviation feedback and calibration module 52 transmits the cost deviation data back to the port cost driver feature extraction and correlation unit 2 for the calibration of the feature parameters of the port cost driver feature map.

[0077] Specifically, the deviation feedback calibration module 52 determines the calibration trigger condition based on the cost deviation rate in the cost deviation feedback result. When the cost deviation rate of a certain shared resource type exceeds the preset deviation threshold for multiple consecutive time periods, the deviation feedback calibration module 52 marks the shared resource type as a shared resource to be calibrated and packages all cost deviation data corresponding to the shared resource to be calibrated into a calibration data package. The preset deviation threshold is set according to the port cost management accuracy requirements.

[0078] The deviation feedback calibration module 52 transmits the calibration data package back to the port cost driver feature extraction and correlation unit 2 via a standardized data interface. Upon receiving the calibration data package, the port cost driver feature extraction and correlation unit 2 incorporates the actual load data and actual cost data corresponding to each period in the calibration data package into the historical statistical sample of this shared resource type. Based on the updated historical statistical sample, it re-executes the statistics of operational measures-resource load conversion coefficients, high-load delay consumption distribution, and load-cost change patterns. The results of this re-statistical analysis are used to adjust the characteristic parameters corresponding to this shared resource type in the port cost driver feature map, including the operational measures-resource load conversion coefficient, the average delay load increment in the high-load delay consumption distribution, and the slope and intercept parameters in the piecewise nonlinear response function. The calibrated port cost driver feature map is then re-output by the port cost driver feature extraction and correlation unit 2 to the multi-scenario cost control simulation calculation unit 3 for subsequent scheme simulation calculations.

[0079] Furthermore, when the cost deviation rate of each shared resource type in the cost deviation feedback results does not exceed the preset deviation threshold in multiple consecutive time periods, the deviation feedback feedback calibration module 52 determines that the characteristic parameters of the port cost driver characteristic map do not need to be calibrated, and only archives the cost deviation feedback results for future reference.

[0080] 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.

[0081] 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. An operational decision-making system for cost control based on multi-dimensional big data analysis, characterized in that: include: Cost-related data collection and indexing unit (1) receives heterogeneous operational data output by various production business systems of the port, and uses data integration and classification indexing technology to standardize and collect cost-related data such as production operations, equipment maintenance, energy consumption, and manpower scheduling to obtain a structured port cost dataset. Port cost driver feature extraction and association unit (2): The port cost driver feature extraction and association unit (2) receives the structured port cost dataset, and uses association mapping and statistical feature extraction techniques to extract the corresponding relationship between shared resource load data and cost data, the cost change law corresponding to the change of shared resource load, the conversion relationship between operation measures and resource load, and the distribution of high resource load delay consumption, and generates a port cost driver feature map. The multi-scenario cost control simulation calculation unit (3) receives the port cost driver characteristic map and the external input operation decision parameters to be evaluated. It adopts the resource load time series superposition and nonlinear cost response mapping technology. First, it performs time series integration and superposition calculation on the current load of the port shared resources corresponding to each scheme and the delayed load triggered by high load in the resource domain to obtain the total load rate. Then, it generates time series cost based on the load-cost nonlinear response relationship mapping of shared resources. After load contribution decomposition processing, it obtains the multi-scheme cost control prediction results. Cost control operation decision generation and verification unit (4) receives the multi-scheme cost control prediction results, and uses decision matching and closed-loop verification technology to screen and adjust the parameters of the alternative operation schemes under cost constraints to obtain executable port cost control operation decision instructions; The execution feedback and deviation comparison unit (5) receives the port cost control operation decision instruction and the actual port operating cost data. It uses data comparison technology to compare the actual cost and expected cost after the decision is executed, obtains the cost deviation feedback result, and sends it back to the port cost driver feature extraction and association unit (2) for closed-loop calibration of the feature parameters of the port cost driver feature map.

2. The business decision-making system for cost control based on multi-dimensional big data analysis according to claim 1, characterized in that, The cost-related data collection and indexing unit (1) includes a data access module (11), a data cleaning and standardization module (12), a classification index construction module (13), and a structured dataset output module (14), wherein: The data access module (11) connects to various production business systems of the port and receives heterogeneous operational data such as production operations, equipment maintenance, energy consumption, and manpower scheduling. The data cleaning and standardization module (12) performs format unification, missing value filling and outlier removal on the received heterogeneous operation data, and completes the standardized collection of cost-related data. The classification index construction module (13) classifies and labels the standardized cost association data according to the cost association dimension, and constructs the corresponding data index. The structured dataset output module (14) encapsulates the cost association data carrying the index and outputs a structured port cost dataset.

3. The business decision-making system for cost control based on multi-dimensional big data analysis according to claim 1, characterized in that, The port cost driver feature extraction and correlation unit (2) includes a correlation mapping construction module (21), a load change law statistics module (22), a measure load conversion construction module (23), a delay consumption distribution extraction module (24), and a feature map generation module (25), wherein: The association mapping construction module (21) establishes a corresponding association between shared resource load data and cost data based on the structured port cost dataset using association mapping technology. The load change pattern statistics module (22) extracts the cost change pattern corresponding to the load change of shared resources based on the structured port cost dataset and using statistical feature extraction techniques. The measure load conversion construction module (23) extracts the correspondence between the operating measure parameters and the shared resource load increment based on the structured port cost dataset, and constructs the operating measure-resource load conversion relationship; The delay consumption distribution extraction module (24) extracts delay consumption data under high load conditions of shared resources based on the structured port cost dataset, and generates a high load delay consumption distribution of resources; The feature map generation module (25) generates a port cost driver feature map based on the corresponding correlation, cost change law, operation measures-resource load conversion relationship and resource high load delay consumption distribution.

4. The business decision-making system for cost control based on multi-dimensional big data analysis according to claim 1, characterized in that, The multi-scenario cost control simulation unit (3) includes a resource load time series profiling module (31), a resource load time series overlay module (32), a nonlinear cost response mapping module (33), and a load contribution decomposition module (34), wherein: The resource load time series profiling module (31) uses the port cost driver feature map as a basis to convert the externally input operational decision parameters to be evaluated into the current load increment of the corresponding port shared resources, and matches and generates the delayed load increment corresponding to the high resource load caused by the current load increment, and integrates them according to the time period to obtain the load time series sequence of the shared resources corresponding to each scheme. The resource load time sequence overlay module (32) receives the load time sequence output by the resource load time sequence portrait module (31), takes the physical resource quantity as the overlay object, and overlays the shared resource load corresponding to each measure in the same scheme to the benchmark load according to the time period to obtain the shared resource total load rate time sequence, and completes the load aggregation of multiple measures in the resource input domain; The nonlinear cost response mapping module (33) receives the total load rate time sequence output by the resource load time sequence superposition module (32), and performs cost mapping calculation on the superimposed total load periodically based on the load-cost nonlinear response relationship of the shared resources to obtain the time sequence cost data of each scheme. The calculation logic of first resource aggregation and then cost mapping is adopted. The load contribution decomposition module (34) receives the total load rate time series and the load time series of each measure from the resource load time series superposition module (32), and combines the time series cost data from the nonlinear cost response mapping module (33) to decompose the contribution of each measure to the total load and nonlinear cost increment, and integrates to obtain the multi-scheme cost control prediction results.

5. The business decision-making system for cost control based on multi-dimensional big data analysis according to claim 4, characterized in that, The process by which the resource load time-series profiling module (31) generates a load time-series sequence includes the following steps: S31.1 Based on the operational measures-resource load conversion relationship in the port cost driver characteristic map, the externally input operational decision parameters to be evaluated are converted into the current load increment of the corresponding port shared resources; S31.2 Based on the distribution of high-load delay consumption of resources in the port cost driver feature map, match and generate the port shared resource delay load increment for each subsequent period corresponding to the current load increment. S31.3 Integrate the current load increment and the delayed load increment according to the time period to generate the load time sequence of the port shared resources corresponding to each scheme.

6. The business decision-making system for cost control based on multi-dimensional big data analysis according to claim 4, characterized in that, The process by which the resource load time series overlay module (32) obtains the time series of the total load rate of the port shared resources includes the following steps: S32.1 Receive the port shared resource load time sequence corresponding to each measure in the same scheme output by the resource load time sequence portrait module (31), and superimpose the load increment of each measure according to the time period to obtain the port shared resource measure superimposed total load increment time sequence. S32.

2. The time series of the total load increment of the port shared resources measures is superimposed with the time series of the port shared resources baseline load of the corresponding period to obtain the time series of the total load of the port shared resources. S32.3 Calculate the time series ratio of the total load of port shared resources to the rated capacity of the corresponding port shared resources on a periodic basis to obtain the time series ratio of the total load rate of port shared resources; where the first... Total load factor of port shared resources during the period The calculation formula is: ； In the formula, For the first The baseline load of shared port resources during the period; For the first The operational measures in the first The port shared resource load increment corresponding to the period includes the current period load increment component and the delayed load increment component; This refers to the rated capacity of shared resources for the corresponding port.

7. The business decision-making system for cost control based on multi-dimensional big data analysis according to claim 4, characterized in that, The nonlinear cost response mapping module (33) includes an equipment load energy consumption mapping submodule, a berth and yard load congestion mapping submodule, a manpower load working hour mapping submodule, and a time-series cost aggregation submodule, wherein: The equipment load energy consumption mapping submodule receives the total load rate time series of port shared resources for equipment, and calculates the corresponding cost of equipment energy consumption for each period based on the load-energy consumption physical operation characteristics of port loading and unloading equipment through a piecewise nonlinear response function. The berth yard load congestion mapping submodule receives the total load rate time series of port shared resources such as berth yards, and calculates the cost corresponding to congestion delay in each period based on the physical operation characteristics of load-congestion queuing of port berth yards through piecewise nonlinear response functions. The manpower load-hour mapping submodule receives the total load rate time series of manpower-related port shared resources, and calculates the corresponding cost of manpower hours for each period based on the physical operation characteristics of port operation manpower load-hour efficiency through a piecewise nonlinear response function. The time-series cost aggregation submodule aggregates the periodic cost data output by the equipment load energy consumption mapping submodule, the berth and yard load congestion mapping submodule, and the manpower load working hour mapping submodule to obtain the time-series cost data for each scheme.

8. The business decision-making system for cost control based on multi-dimensional big data analysis according to claim 4, characterized in that, The process by which the load contribution decomposition module (34) obtains the multi-scheme cost control prediction results includes the following steps: S34.1 Receive the port shared resource total load rate time sequence and port shared resource load time sequence corresponding to each measure output by the resource load time sequence overlay module (32), calculate the proportion of the load increment corresponding to each operation measure in the total load of port shared resources on a cycle-by-cycle basis, and obtain the load contribution proportion time sequence of each measure. S34.

2. Combining the load rate segment threshold of the piecewise nonlinear response function, locate the time period during which the total load rate of the port's shared resources crosses the segment threshold, identify the operational measures that play a dominant role in the crossing of the load rate segment threshold, and obtain nonlinear cost increment attribution information. S34.3 Integrate the time-series cost data output by the nonlinear cost response mapping module (33), the time-series load contribution ratio of each measure, and the nonlinear cost increment attribution information to obtain the multi-scheme cost control prediction results.

9. The business decision-making system for cost control based on multi-dimensional big data analysis according to claim 1, characterized in that, The cost control and operational decision generation and verification unit (4) includes a scheme cost constraint screening module (41) and a decision closed-loop verification and adjustment module (42), wherein: The scheme cost constraint screening module (41) uses decision matching technology to screen out alternative business schemes that meet the cost control objectives under cost constraints, based on the multi-scheme cost control prediction results. The decision-making closed-loop verification and adjustment module (42) verifies the execution adaptability of the alternative operating schemes that meet the cost control objectives based on the alternative operating schemes output by the scheme cost constraint screening module (41), and adjusts and optimizes the parameters of the alternative operating schemes to obtain executable port cost control operating decision instructions.

10. The business decision-making system for cost control based on multi-dimensional big data analysis according to claim 1, characterized in that, The execution feedback and deviation comparison unit (5) includes a cost deviation comparison calculation module (51) and a deviation feedback return calibration module (52), wherein: The cost deviation comparison calculation module (51) receives the port cost control operation decision instruction and the actual port operating cost data, and uses data comparison technology to compare the actual cost after the decision is executed with the corresponding expected cost on a cycle basis to obtain the cost deviation feedback result. The deviation feedback and calibration module (52) transmits the cost deviation data back to the port cost driver feature extraction and association unit (2) based on the cost deviation feedback result output by the cost deviation comparison calculation module (51), for the calibration of the feature parameters of the port cost driver feature map.

Images