An engineering cost whole life cycle management method and system

Abstract

The application relates to the technical field of information management of engineering projects, and discloses a project cost whole life cycle management method and system, which comprises the following steps: establishing a structured information body describing a source event, and based on shared resource objective operation facts recorded by a mobile terminal, performing dynamic cost collection, and then starting from the source event, chain tracing and quantifying direct costs caused by the source event and indirect costs transmitted by the source event. The application structurally binds the financial attribute of the cost and the cause-effect logic at the data source, so that the cost data itself becomes a lossless information carrier bearing a complete cause-effect chain in project evolution, avoids information breakage caused by a traditional accounting method, and realizes accurate quantification and tracing of all-round financial influence caused by a single source event.

Description

Technical Field

[0001] This invention relates to a method and system for full life-cycle management of engineering costs, belonging to the field of information management technology for engineering projects. Background Technology

[0002] In the field of full life-cycle management of engineering costs, project management information platforms, as centralized data processing hubs, enable dynamic monitoring of project financial status by linking actual costs with pre-set budget items. This has become a widely adopted and fundamental approach in the industry for effectively controlling project investment.

[0003] However, for complex engineering projects with long construction cycles and numerous uncertainties, the aforementioned management method, while maintaining the accuracy of daily financial accounting, reveals an inherent structural limitation. When an initial partial change in a project, such as adjustments to the basic plan due to changes in geological conditions, has its impact transmitted through a series of intermediate links such as project delays and seasonal changes, ultimately manifesting as various indirect costs in different stages and cost items in the later stages of the project. When recording these subsequent costs, the existing information platform's data structure only supports linking expenses to direct construction tasks or financial items. This leads to a break in the connection between the financial information of costs and their logical causes. This information loss occurs at the source of data generation and is a common problem in existing technologies.

[0004] To compensate for this information gap, intuitive improvement approaches include establishing more refined accounting categories or attempting to reconstruct causal relationships using data analysis. The former only provides a more detailed breakdown at the result level, without addressing the transmission chain between cause and effect. The latter, due to the lack of initial key information, inevitably involves speculation and uncertainty in any deduction. These approaches fail to address the root of the problem: an inherent contradiction in the design of existing cost recording methods. The requirements for efficiency and convenience in the accounting process are incompatible with the requirements for the integrity of deep causal relationships in project management. Specifically, existing technologies have the following shortcomings: 1. Cost data loses its deep causal attributes at the source of recording, resulting in incomplete information; 2. The logical transmission path between indirect costs and the initial triggering event cannot be systematically recorded and traced; 3. The true and comprehensive financial impact of a single change event cannot be accurately quantified and assessed. Therefore, the technical problem to be solved by this invention is how to construct a cost recording mechanism that can structurally bind the financial information of costs to its inherent causal logic at the source of data generation, thereby fundamentally avoiding the loss of integrity of project information during the entire life cycle. Summary of the Invention

[0005] This invention provides a method and system for full life-cycle management of engineering costs. Its main purpose is to solve the problem that in the existing technology, the inherent causal chain between indirect costs and their root-causing events is severed during the conventional financial accounting process, which leads to the damage to the integrity of project management information.

[0006] To achieve the above objectives, this invention provides a method for full life-cycle management of engineering costs, comprising:

[0007] Establish and continuously update a structured information body describing source events. The specific steps are as follows: when a potential source event whose nature or scope of impact is not fully clear is discovered, placeholder registration is performed, and the status and direct cost estimate of the source event are updated by selecting the stored standardized evolution operators. For the composite costs generated by shared resources dynamically shared by multiple response measures in the project, dynamic cost collection based on objective operational facts is performed. The dynamic cost collection is specifically as follows: through a mobile terminal application, the operation time period of the shared resources serving each response measure within a specific billing cycle is recorded in real time to form an operation log.

[0008] Based on the work logs, the cost contribution ratio of shared resources to each response measure is calculated. Based on the cost contribution ratio, the single composite cost generated by the shared resources in the service cycle is broken down and aggregated to multiple response measures that it serves and are associated with the source event described by the structured information body. Based on the aggregated cost data, starting from the source event described by the structured information body, the direct costs caused by the source event and the indirect costs transmitted through the shared resources are traced and quantified in a chain.

[0009] Preferably, the step of calculating the cost contribution ratio of shared resources to each response measure based on the work log is further specified as follows: obtaining the working time period of shared resources under each response measure, obtaining the efficiency coefficient corresponding to each response measure, multiplying the working time period of shared resources under each response measure with the corresponding efficiency coefficient to obtain the weighted working time, and determining the cost contribution ratio based on the proportion of the weighted working time of each response measure to the total weighted working time.

[0010] Preferably, the step of obtaining the efficiency coefficient corresponding to each response measure is as follows: by using heterogeneous sensors deployed on shared resources to collect real-time operating parameters that characterize the current operating status, obtain the current task stage information of the response measure, select a benchmark operating parameter from the parameter set stored corresponding to the task stage information based on the current task stage information of the response measure, and dynamically generate the efficiency coefficient by comparing the real-time operating parameters with the selected benchmark operating parameters.

[0011] Preferably, the step of acquiring real-time operating parameters characterizing the current operating status of heterogeneous sensors deployed on shared resources specifically involves: performing data fusion processing on the multi-dimensional real-time operating parameters acquired in parallel by the heterogeneous sensors to generate fused operating parameters; performing anomaly detection on the fused operating parameters to identify and remove abnormal data; when anomalies are detected in the fused operating parameters, correcting or supplementing the abnormal data according to the stored anomaly handling rules; and using the processed operating parameters as the final real-time operating parameters.

[0012] Preferably, the step of performing anomaly detection on the fused operating parameters is achieved by generating a dynamic threshold range that varies with operating conditions. Specifically, generating the dynamic threshold range involves: obtaining historical operating parameters of the shared resource under various historical operating conditions; for each operating condition, calculating the mean and standard deviation of the corresponding historical operating parameters, and generating a corresponding dynamic threshold range for each operating condition. This dynamic threshold range is determined as follows: calculating the difference between the historical mean and the product of a confidence coefficient and the historical standard deviation as the lower limit, and calculating the sum of the historical mean and the product as the upper limit; comparing the fused operating parameters with the dynamic threshold range under the current operating condition, and identifying fused operating parameters exceeding the range as abnormal data.

[0013] Preferably, when an anomaly is detected in the fused operating parameters, the steps for correcting or supplementing the abnormal data according to the stored anomaly handling rules are as follows: identify the type of abnormal data; call the corresponding correction or supplementation algorithm model according to the type of abnormal data; comprehensively evaluate the confidence level of the abnormal data; use the confidence level to adjust the strength of the correction or supplementation algorithm model; execute the correction or supplementation algorithm model after strength adjustment; and generate the correction or supplementation result.

[0014] Preferably, the steps for comprehensively evaluating the confidence level of anomalous data are as follows: obtaining the source information of the heterogeneous sensor that generated the anomalous data, obtaining the initial confidence score given by the anomalous detection algorithm for the anomalous data; querying the historical correction effect evaluation records of similar anomalous data, and combining the source information, the initial confidence score and the historical correction effect evaluation records, calculating the final confidence level through a stored weighted model.

[0015] Preferably, the standardized evolution operators include a nature deepening operator for deepening the understanding of the nature of the source event, and an impact scope expansion operator for expanding the scope of influence of the source event. Furthermore, after the steps of chain tracing and quantifying the direct costs caused by the source event and the indirect costs transmitted through shared resources, a cost risk warning based on evolutionary trends is executed. Specifically, the cost risk warning involves: continuously recording the evolution data of all indirect costs associated with the source event described by a structured information body as the project progresses; calculating the rate of change of the evolution data, which is used to characterize the growth trend of indirect costs; and automatically generating and outputting a warning instruction indicating that the growth trend of indirect costs caused by the source event exceeds the risk threshold when the rate of change continuously exceeds a risk threshold calculated based on historical cost data.

[0016] Preferably, the step of using confidence level to adjust the strength of the modified or supplemented algorithm model specifically involves: selecting a corresponding strength adjustment curve from multiple stored strength adjustment curves representing different adjustment strengths based on the confidence level value; determining the strength parameter used to modify or supplement the algorithm model based on the selected strength adjustment curve; and applying the strength parameter.

[0017] A life-cycle management system for engineering cost includes: one or more processors and a memory, on which a computer program is stored. When the computer program is executed by one or more processors, it implements a life-cycle management method for engineering cost.

[0018] Compared with the prior art, the beneficial effects of the present invention are:

[0019] 1. This invention establishes dual markers for events and measures in the cost entry process and combines them with a synchronous storage mechanism for financial and causal attributes. This gives each piece of cost data an indivisible dual identity when it is generated. This approach avoids the information decay and logical breakage caused by the separate recording of financial information and project management information and subsequent correlation in traditional processes. The cost data in the project is no longer just the result of financial accounting. It itself becomes a structured information carrier that carries the complete causal chain in the project evolution process, providing a lossless and reliable data foundation for subsequent traceability analysis.

[0020] 2. This invention further moves the data collection point for cost attribution from the back-end financial entry to the front-end on-site operation stage. By recording the real-time operation logs of shared resources in the project, and dynamically splitting and aggregating single composite costs based on actual working hours, this transforms the traditional management problem of shared resource cost allocation, which relies on experience-based estimation, into a calculation process based on front-end real-time data. This not only allows the allocation results to directly correspond to the actual on-site operations, but also enables these precisely split cost shares to carry their own clear attribution to measures, seamlessly integrating into the causal chain constructed by event and measure markers. This achieves accurate quantification of shared resource costs under the concurrent impact of multiple source events.

[0021] 3. This invention also introduces a dynamic evolution recording mechanism for source events. Through preset structured evolution operators, complex events with progressively evolving nature and gradually expanding influence are continuously and systematically updated, so that the definition of the event can be continuously revised as exploration or construction progresses. This mechanism, combined with the cost causal attribution system, enables all direct and indirect costs to not only be traced back to an initial event, but also to be precisely associated with the specific state of the event at different cognitive stages. Thus, the system no longer presents static cost attribution results, but rather a dynamic cost impact path that reflects the project team's deepening understanding of complex issues throughout the entire process. Attached Figure Description

[0022] Figure 1 This is a flowchart of the source event-driven dynamic cost collection and risk early warning method of the present invention;

[0023] Figure 2 This is a schematic diagram of the dynamic threshold anomaly detection based on fusion parameters according to the present invention;

[0024] Figure 3 This is a data interaction architecture diagram of the core and functional modules of the system of this invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be described in detail below. However, these embodiments are only used to explain the present invention and are not intended to limit the scope of protection of the present invention.

[0026] The present invention discloses a method and system for full life-cycle management of engineering costs, which is configured as follows: a data processing procedure running on the server and operated collaboratively by multiple functional modules. The method aims to establish a mechanism that structurally binds the financial attributes of costs with their causal logic at the source of data. In turn, by quantifying the operational facts of shared resources, it can achieve the tracing and dynamic risk warning of the full-link cost impact caused by a single source event. The method mainly includes three main stages: structured information body management of source events, dynamic cost collection based on objective operational facts, and cost tracing and risk warning based on causal chains.

[0027] In large-scale engineering projects, potential risk events whose nature or scope of impact is not fully clear pose a challenge to project cost control. Traditional text-based records are not conducive to systematic dynamic tracking and quantitative analysis. To address this, the first step of the method in this invention is to establish and continuously update a structured information body describing the source event. This procedure begins with a placeholder registration operation. When a preliminarily discovered potential source event is entered into the system, such as the preliminary discovery of a geological anomaly in the main bridge pier area, the system generates a data object. The initial state of this data object is marked as pending evolution. Its data structure is designed to include a unique event identifier, an initial description, a discovery timestamp, and an initial direct cost estimation field. This field can be set to zero or a preliminary estimate during the placeholder registration stage. As the understanding of the event deepens, for example, a new geological exploration report confirms that the anomaly is a deep fault zone, project managers update the structured information body of the event by calling a standardized evolution operator. Specifically, the operator selects a nature deepening operator, and the system presents a structured data input interface bound to that operator, requiring the input of the geological report number. After submitting information such as newly discovered geological features, the system will append this update as a timestamped and independent version record to an evolutionary history array of the event information body. Based on the newly input deep fault zone characteristics and their impact depth parameters, and according to the built-in cost linkage rules, the system will automatically recalculate and update the direct cost estimation field of the event. This structured update mechanism driven by evolutionary operators allows the cognitive evolution process of complex events to be recorded completely and without loss, providing a dynamically evolving and continuously accurate attribution starting point for subsequent cost tracing. To solidify the technical implementation path of standardized evolutionary operators, the system provides an operator configuration procedure. This procedure allows the technical lead to first define a set of data fields associated with a specific source event type, where each field is assigned a data type, such as text, a preset menu, or a value. The content of the preset menu is directly read from a specified data table in the project material database or geological information database. Subsequently, a cost estimation linkage rule is bound to this operator. This rule is implemented as a script with conditional branches, whose input is the value of the aforementioned data field, and whose output is a definite direct cost increment. For example, for the geological deepening operator of isolated rock formations in shield tunneling construction, its Through formula The calculation shows that, among which The total number of hobs read from the equipment parameter library. The cost of replacing a single tool is obtained from the material procurement database. This is a two-dimensional lookup table function that uses the diameter of the boulder as input by the operator. With rock strength The two field values ​​return a damage probability value that was determined during the device's factory calibration phase. Finally, the configured operator is assigned a unique version identifier and set to an active state before it can be called by the project's front-end application.

[0028] Furthermore, for resources dynamically shared by multiple response measures in a project, such as large lifting equipment or specialized construction teams, how to allocate the compound costs to different cost drivers is a technical problem in project cost management. Therefore, the method of this invention adopts a dynamic cost collection procedure based on objective operational facts. This procedure moves the data collection point for cost attribution from the back-end financial entry to the front-end on-site operation stage. On-site management personnel use an application deployed on a mobile terminal to record the service status of shared resources in real time. When a large floating crane begins service for response measure A: reinforcing the main bridge pier, the management personnel select the resource and measure in the mobile application and trigger a start operation command. This command, along with the current timestamp, is sent to the server. If the floating crane is reassigned to response measure B: hoisting steel box girders within the same billing cycle, the management personnel trigger a task switch command. The system records the end time of the previous task and the start time of the new task, thus forming an operation log accurate to the second, containing resource ID, measure ID, and service time period. When a compound cost related to the floating crane, such as a week's rental fee, is incurred, the system records the cost. When data is entered into the system, the system automatically retrieves all job logs within the billing cycle and calculates the total job time for each response measure served by that resource. To improve the accuracy of data collection, an efficiency coefficient is introduced into the system. To characterize the differences in operational output of resources under different operating conditions, this efficiency coefficient is determined. The system executes a deterministic procedure by collecting real-time operating parameters characterizing its current operating status through heterogeneous sensors deployed on shared resources, such as engine fuel flow meters, GPS positioning modules, and equipment vibration sensors. For example, if the system determines that the current task phase of countermeasure A is underwater deep-layer solidification, it selects a baseline operating parameter corresponding to that task phase from a preset parameter set, such as baseline fuel consumption. It is 20 liters per hour, while the real-time fuel consumption collected by the sensor is... If the efficiency is 25 liters per hour, then the efficiency coefficient for one dimension can be calculated as follows: The system performs data fusion processing on multi-dimensional real-time operating parameters from different sensors. For example, it uses a weighted model to integrate parameters such as fuel consumption, vibration amplitude, and GPS offset to generate a fused operating parameter, which is then compared with the baseline operating parameter to dynamically generate the final efficiency coefficient. When calculating the cost contribution ratio, the system first considers the operation time period under each response measure. Its corresponding efficiency coefficient Perform a product operation to obtain the weighted operation time. Subsequently, the weighted operation time for each response measure was used as a percentage of the total weighted operation time. The proportion is used to determine its cost contribution ratio. Ultimately, the system will determine the cost contribution ratio. , will single composite cost This involves breaking down and aggregating the responses it serves into multiple measures, i.e., aggregating them into measures. Cost share This procedure transforms the problem of sharing resource costs, which relies on empirical estimation, into a calculation process based on real-time front-end data and dynamic efficiency calibration.

[0029] To ensure the reliability of sensor data, the method of this invention also includes a data anomaly detection and correction procedure. After performing data fusion processing on multi-dimensional real-time operating parameters, the system performs anomaly detection. This detection is achieved by generating a dynamic threshold range that changes with operating conditions. Specifically, the system pre-acquires historical operating parameters of shared resources under various historical operating conditions, and for each operating condition, calculates the average value of the corresponding historical operating parameters. with standard deviation Therefore, the system generates a corresponding dynamic threshold range for each operating condition, and the lower limit of this range is determined as follows: The upper limit is set as Among them, the confidence coefficient This is a configurable parameter, calibrated based on selecting a value from the historical dataset that maximizes the recognition rate of known anomalous data points while controlling the false alarm rate of normal data points to a specific percentage, such as below 5%. For example, it can be set to 3. When the fused operating parameters exceed the dynamic threshold range under the current operating conditions, they are identified as anomalous data. When anomalous data is detected, the system calls the corresponding correction or supplementary algorithm model according to the stored anomaly handling rules. For example, when the anomalous data type is identified as a transient spike pulse, the system calls the median filtering algorithm for correction. Before performing the correction, the system comprehensively evaluates the confidence level of the anomalous data. The estimation process integrates source information from heterogeneous sensors, such as high-precision sensors which have higher weight, the initial confidence score given by the anomaly detection algorithm, and historical correction effect evaluation records of similar anomaly data. The final confidence score is calculated through a stored weighted model. This confidence score value is used to adjust the strength of the correction algorithm model. For example, based on the confidence score value, one of the multiple preset strength adjustment curves representing different adjustment strengths is selected, and the strength parameters used to correct the algorithm model, such as the size of the filter window or the order of the interpolation model, are determined based on the selected curve. Finally, the strength parameters are applied to perform the correction, generating the final real-time operating parameters.

[0030] To further ensure the objectivity of the work logs upon which dynamic cost aggregation is based, the system also executes a cross-validation procedure for work facts based on sensor data. This procedure is automatically triggered after each work log segment is generated. The system first retrieves the historical statistical distribution model of operating parameters corresponding to the response measures marked in the log from the stored equipment operating parameter baseline library. This model includes the mean vector and covariance matrix of the multidimensional operating parameters of the response measures in history. Then, the system extracts the actual multidimensional sensor data stream within the time window corresponding to the work log segment and calculates the Mahalanobis distance between the feature vector of the actual data stream and the historical statistical distribution model. This distance, as a statistical measure, is used to quantify the degree of deviation of a new observation point from the distribution center of its corresponding sample set; the system will calculate... The value is compared with a preset verification threshold. The threshold is calculated during the baseline calibration phase using a large amount of historical normal operation data for the same response. The statistical distribution of the values ​​is determined by taking the 95th percentile. Continue to exceed When this happens, the system marks this segment of the work log as pending review and automatically lowers the confidence weight coefficient of its corresponding cost contribution ratio by a fixed percentage, such as 25%. This reduces the impact weight of the suspicious work record in subsequent cost tracing and risk warning calculations.

[0031] Finally, based on the aggregated cost data, the method of this invention can start from the source event described by the structured information body, perform chain tracing, and quantify the direct costs caused by the source event and the indirect costs transmitted through shared resources. When it is necessary to assess the financial impact of a source event, the system first indexes the direct costs recorded by the event itself through database association queries, then indexes all countermeasures associated with the event, and then summarizes all cost shares directly attributed to and dynamically allocated to these countermeasures to form a complete cost impact view. In addition, the method of this invention also includes a cost risk early warning procedure based on evolutionary trends. The system continuously records the evolution data of all indirect costs associated with a certain source event as the project progresses, forming a cost time series, and calculates the rate of change of the evolution data to quantify the growth trend of indirect costs. When the rate of change continuously exceeds a risk threshold statistically derived from historical cost data of similar projects, for example, exceeding the 85th percentile rate of historical data for three consecutive statistical periods, the system will take action. The system automatically generates and outputs an early warning instruction, indicating that the indirect cost growth trend triggered by the source event has entered a high-risk range. This combination of procedures transforms the static and fragmented financial accounting method into a dynamic, traceable, and life-cycle cost management method that reflects the causal chain of project evolution. It should be noted that the early warning instruction is implemented in the system as a structured data packet. This data packet encapsulates the source event identification code that triggered the warning, the current rate of change exceeding the limit, and a query link to the relevant cost details. The early warning instruction is configured to be automatically pushed to a highlighted early warning module on the project management dashboard through an application programming interface. In this module, the indirect cost growth curve associated with the source event is displayed in a visual manner. At the same time, the system sends a message containing the core information of the early warning instruction through an encrypted channel to the mobile terminal of the preset project cost control manager, thereby transforming a quantitatively analyzed risk status into a decision-making entry point that can be directly intervened by management personnel.

[0032] Example 1: In a large-scale cross-sea infrastructure construction project, the project adopted a full life-cycle cost management approach. Midway through the project, two independent and concurrent unexpected situations arose. First, underwater drilling operations at the No. 2 main pier revealed that the seabed geological conditions beneath it were more complex than described in the exploration report, containing an unexpected weak interlayer. This forced a change in the construction plan to a more time-consuming reinforcement operation. Second, the delivery date of the critical precast steel box girder used for the superstructure installation was delayed by four weeks due to a production line malfunction at the supplier, disrupting the original hoisting schedule. Faced with these concurrent challenges, the project management team recorded them according to the system procedures in the initial stage. The first event was recorded as the source event EVT-00 through the placeholder registration procedure. 1. The geological anomalies of piers 1 and 2 were estimated to be the cost of additional reinforcement materials and design changes. The second event was recorded as the source event EVT-002: delay in the supply of steel box girders. To address these two independent events, the project team developed separate countermeasures. For EVT-001, the countermeasure ACT-A was developed: underwater supplementary reinforcement was carried out. For EVT-002, the countermeasure ACT-B was developed: adjusting the steel box girder hoisting sequence and increasing nighttime operations to catch up with the schedule. At this time, a large floating crane was used as a shared resource, serving as both the equipment for underwater reinforcement and the resource for steel box girder hoisting. In order to maximize the utilization of this resource, the on-site dispatcher frequently instructed it to switch between the two tasks, ACT-A and ACT-B, within the same billing cycle.

[0033] Under conventional cost accounting methods, the rental fee for the floating crane, as a single composite cost, becomes difficult to attribute. It is typically allocated proportionally over time or included in a single management cost category. This not only fails to clarify the indirect costs arising from the two source events but also ignores the differences in resource consumption efficiency of different response measures. However, in the method of this embodiment, a structured information management mechanism for source events and a dynamic cost aggregation mechanism based on objective operational facts work synergistically. First, on-site management personnel record all operating time periods of the crane in ACT-A and ACT-B using a mobile terminal application, creating an operation log. When the crane performs underwater reinforcement work in ACT-A, due to the complexity of its working conditions, real-time operating parameters such as engine fuel consumption and vibration amplitude, collected by heterogeneous sensors, deviate from the equipment's baseline operating parameters for that task phase. Based on this, the system dynamically generates an efficiency coefficient. However, when it switches to ACT-B for standard steel box girder hoisting, its operating parameters are close to the benchmark, and the system generates an efficiency coefficient. At the end of the billing cycle, the system calculates the weighted operating time of the two measures based on the original operating time in the operation log and the dynamically generated efficiency coefficient. Based on the proportion of this weighted operating time to the total weighted operating time, the system automatically splits and aggregates the crane's composite rental fee into two different countermeasures, and then links them to the two source events, EVT-001 and EVT-002, respectively.

[0034] This process allows for parallel operational flexibility on-site and financial accountability at the back end. On-site dispatching can allocate resources according to actual needs, while back-end data processing procedures maintain the certainty of cost attribution through quantitative analysis of objective operational facts. Furthermore, this method provides a quantitative approach to measuring change costs. When project decision-makers review cost reports, they no longer see an isolated cost overrun alert for the machinery usage fee item, but clearly see that the source event EVT-001, due to its inefficiency, although occupying less initial time, has a higher share of indirect costs than EVT-002. This data-driven visualization of cost transmission paths allows management to differentiate and assess the financial impact of different risk events. Ultimately, based on this quantified cost impact data, the project management team's decision-making is enriched. For example, they can determine whether to invest resources in EVT-001 to optimize its construction process and thus improve its operational efficiency coefficient. It may be more cost-effective than adding more working time to EVT-002, and the function of project financial data has thus been transformed from a lagging feedback on budget deviations to an analytical basis for resource allocation decisions.

[0035] Example 2: To objectively verify the accuracy of the dynamic cost aggregation method of the present invention, especially when the efficiency of shared resource operations changes dynamically, this verification experiment was conducted. The purpose of the experiment is to quantitatively compare the deviations between two different cost aggregation algorithms and known ground truth values ​​in a controllable and reproducible simulation environment, thereby verifying the introduction of an efficiency coefficient. Its role in improving the accuracy of cost aggregation.

[0036] This experiment was conducted on a hardware-in-the-loop simulation platform. This platform consists of an industrial server running the core algorithm of the management method of this invention, a mobile terminal for inputting work logs, and a simulation server responsible for generating simulated operating condition data. The simulation server uses a physical model of engineering machinery to simulate the operating state of a large floating crane performing two different types of tasks, and generates multi-dimensional simulated sensor data streams in real time, including engine fuel consumption rate, hydraulic system pressure fluctuations, and the vibration spectrum of the boom end. The output interval of this data stream is set to 20ms, which is based on the fact that its value is less than the hydraulic main circuit response time of the simulated crane, which is approximately 200ms, by an order of magnitude. This method is used to capture dynamic details of equipment operation without exceeding the server's data processing load. In the experiment, a 10.0-hour work cycle was set, with a total composite cost of 10,000 cost units for the shared resources. Within this cycle, the resource was allocated to execute two independent responses, Response A and Response B, each serving for 5.0 hours. Response A was designed as a high-difficulty, low-efficiency task with a ground-based efficiency value of 0.7 in the simulation model, while Response B was designed as a standardized, high-efficiency task with a ground-based efficiency value of 1.2. Based on these settings, the total effective workload generated by the shared resource within the cycle was... Therefore, the ground truth value of cost allocation should be: allocation of response A. Unit, Response Measure B Shared The experiment was conducted in two groups: a control group and an experimental group. The control group used a method that allocated costs based solely on time data from the work logs, while the experimental group used the complete method of this invention, which performed dynamic cost aggregation based on the work logs and simulated sensor data generated by the simulation server. The control group's calculation result showed that response A and response B each allocated 5000.0 cost units. This result showed a deviation of +35.7% from the actual ground value of 3684.2 units for response A, and a deviation of -20.8% from the actual ground value of 6315.8 units for response B. In contrast, the experimental group's algorithm module received the simulated sensor data stream and calculated the average dynamic efficiency coefficients of resources after stabilization during the execution of response A and B. and Based on this weighted calculation, the final cost allocation result is that response A is allocated 3717.3 units and response B is allocated 6282.7 units. The deviation rate between this result and the actual ground value is reduced to +0.9% and -0.5%, respectively.

[0037] Experimental data shows that when resource operation efficiency varies, a time-based allocation method can lead to cost mismatch. The dynamic cost aggregation procedure used in the experimental group, however, introduces and dynamically calculates efficiency coefficients. The method quantifies the differences in work output of resources under different working conditions and uses this as a basis to correct costs. As a result, the deviation between the allocation results and the actual ground values ​​is significantly reduced. This experiment verifies that the method can more accurately allocate the composite costs of shared resources to the different coping measures they serve, and provides a more accurate allocation basis for cost accounting in complex scenarios in engineering projects.

[0038] Example 3: This example combines Figures 1 to 3 This describes a method and system for managing the entire lifecycle of engineering costs, such as... Figure 1 As shown, the process begins with the identification of a potential source event, followed by the generation of an initial data object through placeholder registration, and then updating it through a standardized evolution operator to form a dynamically traceable structured information body. Simultaneously, for shared resources and compound costs, dynamic cost aggregation is performed by recording work logs and combining them with dynamic efficiency coefficients generated from heterogeneous sensor data. During this process, the sensor data undergoes data anomaly detection and correction. After aggregation, the aggregated cost data carrying a clear causal chain, along with the structured information body of the source event, is sent to the chain-based tracing and quantification stage to form a cost impact view. Based on this view, cost risk warnings are executed, and finally, a warning instruction is generated when the conditions are met.

[0039] like Figure 2 As shown in the figure, the horizontal axis represents the sampling time, and the vertical axis represents the normalized parameter values. The figure shows two normalized original operating parameter curves, fuel consumption rate and vibration amplitude, and a processed fused parameter curve. An upper limit for anomaly threshold is set. When the fused parameter curve shows peaks at approximately the 28th minute and the 72nd minute and exceeds the upper limit of the threshold, the system can identify it as abnormal data.

[0040] like Figure 3 As shown, the architecture centers on the core of the engineering cost lifecycle management system. It receives input information such as work logs from mobile terminal applications and operating parameters from heterogeneous sensor groups. The system core processes this data and drives the data fusion processing module, anomaly detection and correction module, cost aggregation and calculation module, and chain traceability and risk warning module on the right to work together through clearly defined data flows, namely data fusion, anomaly handling, cost allocation and risk output, thereby forming a complete data-driven engineering cost management system.

[0041] Example 4: In an engineering project using a new type of tunnel boring machine (TBM), the core algorithm module of the present invention needs to be initialized and calibrated before the project starts. Since this type of TBM lacks directly relevant historical operating data, the data fusion model used to quantify operational efficiency, the evolutionary operator rules used for cost estimation linkage, and the cost change rate threshold used for risk warning are all calibrated according to the following procedure to establish the initial data processing baseline for this specific project. First, a data fusion model calibration procedure for this new type of TBM is executed. This procedure is carried out on a ground test bench capable of simulating different geotechnical conditions. The TBM runs continuously for 2.0 hours each under three conditions—soft soil, mixed rock surface, and intact hard rock—according to a preset program. During this period, the system collects multi-dimensional real-time operating parameters at 50ms intervals using its heterogeneous sensors, mainly including cutterhead torque. Hydraulic propulsion system and mud-water loop pressure Simultaneously record the standard earthwork excavation volume per unit time. After collecting all test data, the system performs a multiple linear regression analysis to... As the dependent variable, Establish a linear model with the independent variable. The goal of regression analysis is to solve for a set of weight coefficients. With constant This makes it possible to achieve the following across the entire test dataset: and By minimizing the root mean square error between them, the resulting set of weight coefficients is solidified as the data fusion model parameters specific to this tunnel boring machine under this project.

[0042] Subsequently, project managers configured the standardized evolution operators within the system. For an operator called "Encountering Unforeseen Geological Bodies: Property Deepening," a logical rule engine was configured. Managers set a rule within this engine with the following trigger condition: when the system receives input from the operator that a newly discovered geological feature is a cluster of isolated boulders, and the length parameter of its influence range exceeds 20.0m, the system will automatically call the cost recalculation module with ID CR-007. This cost recalculation module is associated with a calculation formula. ,in, This represents the number of cutterheads within the affected area; this value is directly read from the tunnel boring machine design parameter library. This represents the probability of damage to a single hobbing cutter when encountering this type of geological formation. This value is cited from the technical manual provided by the equipment manufacturer. This represents the cost of replacing a single hobbing cutter. This value is obtained from the project's material procurement database. A qualitative description of an external event is then transformed into a direct cost estimate increment through a rule and formula driven by multi-source data. Finally, regarding the indirect cost growth trend risk threshold in the cost risk early warning procedure, the system adopts a two-stage approach for setting it. At the initial stage of project initiation, the system selects 50 cases from an industry database containing past tunnel engineering cases that are similar to this project in terms of scale and geological type, and calculates the indirect cost growth of these 50 cases in the first 20% of the project execution period. The distribution of the growth rate is used, with the 80th percentile as the initial risk threshold for this project. Simultaneously, the system is configured to automatically trigger an adaptive threshold update after the project has accumulated 60 working days. At that time, the system will recalculate the statistical distribution of the growth rate based entirely on the indirect cost data generated by the project within those 60 days, and use the 85th percentile as the updated risk threshold. Through the above series of procedures, before the system is officially put into operation, its core parameters are given traceable and verifiable setting basis, enabling cost management throughout the project lifecycle to have a data processing foundation based on the calibrated parameters.

[0043] Example 5: In the deployment phase of the method of the present invention, in order to calibrate the dynamic efficiency coefficient The required reference benchmark needs to be determined by a pre-baseline calibration procedure for specific shared resources in the project. This procedure is carried out in a controlled environment. The operator first defines a set of standardized task phases corresponding to the project's work content. Then, the shared resource is operated stably for a preset duration under each standard task phase. During this period, the system continuously records its multi-dimensional real-time operating parameters using a high-frequency sampling method and removes outliers from the collected dataset. Finally, for each standard task phase, the system calculates the statistical average of all real-time operating parameters within its corresponding time period and stores this set of averages as the benchmark operating parameters for that task phase in a lookup library of the system.

[0044] To establish the system's abnormal data processing capabilities, a controlled fault injection test can be added to the baseline calibration procedure. In this test, when the shared resources are running stably in a certain standard task phase, the testers inject a series of typical fault signals into the sensor signal links. These signals are used to simulate instantaneous spike pulses, short-term signal loss, and sensor zero-point drift that may occur in field operations. After receiving these sensor data containing fault characteristics, the system will automatically record their data morphology characteristics. The system administrator will then map and associate specific data morphology characteristics with a corresponding correction or supplementary algorithm model in the anomaly handling rule base based on these recorded characteristics. For example, the data characteristics of short-term signal loss will be associated with a supplementary algorithm model using cubic spline interpolation. In this way, the system establishes an automated data correction mechanism with clear triggering conditions and response logic for different anomaly types.

[0045] Example 6: When applying the method of the present invention to a mixed fleet project containing various types of engineering machinery, in order to enable the abnormal data processing procedure to adapt to the signal characteristics of different devices, an offline optimization process for generating an anomaly correction intensity parameter matrix needs to be executed. This process aims to find an optimal mapping relationship between the anomaly data confidence level and the correction model intensity parameter for each anomaly correction algorithm model in the system. The objective function of the optimization is to minimize the root mean square error between the corrected data and the ground truth value. The optimization process first defines the intensity parameter optimization space of the correction algorithm. Taking a median filtering algorithm for processing instantaneous spike pulse anomalies as an example, its intensity parameter is discretized into three levels, with level one corresponding to the filter. The wave window size is 3, the window size for level 2 is 5, and the window size for level 3 is 7. Subsequently, the system randomly extracts segments from a large amount of clean sensor data obtained from the previous baseline calibration and injects synthetic spike pulse signals with known amplitude and duration at different signal-to-noise ratios in a programmed manner to form a test dataset containing thousands of anomalous samples. Each sample retains its original uncontaminated data as the ground truth value. The system then uses the aforementioned median filtering algorithm and iterates through all three intensity parameter levels to correct each anomalous sample in the test dataset and calculates the root mean square error between each correction result and the corresponding ground truth value as a measure of the correction performance of the intensity parameter on that sample.

[0046] Finally, based on the corrected performance measurement data generated by the above process, the system constructs the final intensity parameter matrix. This matrix uses the confidence interval of the abnormal data as an index and the intensity parameter level determined by optimization as the output. Specifically, the system divides the confidence value range into multiple intervals and, within each interval, statistically identifies the corrected intensity parameter level that achieves the lowest mean root square error, storing this level in the matrix. After calculation, for spike pulses with a confidence level between 0.70 and 0.89, the average correction error generated by the level two filtering algorithm (i.e., a window size of 5) is the smallest. Therefore, in the final generated parameter matrix, this confidence interval is mapped to intensity parameter level two. After this process is completed, when the system detects any spike pulse anomalies during online operation, it can directly query and call the corresponding offline optimized corrected intensity parameter from the matrix based on the calculated confidence level, thereby automating the data correction process and enabling parameter self-adaptation.

[0047] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for full life-cycle management of engineering costs, characterized in that, include: Establish and continuously update a structured information body describing source events. The specific steps are as follows: when a potential source event whose nature or scope of impact is not fully clear is discovered, placeholder registration is performed, and the status and direct cost estimate of the source event are updated by selecting the stored standardized evolution operators. For the composite costs generated by shared resources dynamically shared by multiple response measures in the project, dynamic cost collection based on objective operational facts is performed. The dynamic cost collection is specifically as follows: through a mobile terminal application, the operation time period of the shared resources serving each response measure within a specific billing cycle is recorded in real time to form an operation log. Based on the work logs, the cost contribution ratio of shared resources to each response measure is calculated. Based on the cost contribution ratio, the single composite cost generated by the shared resources in the service cycle is broken down and aggregated to multiple response measures that it serves and are associated with the source event described by the structured information body. Based on the aggregated cost data, starting from the source event described by the structured information body, the direct costs caused by the source event and the indirect costs transmitted through the shared resources are traced and quantified in a chain.

2. The method for full life-cycle management of engineering costs according to claim 1, characterized in that, The steps for calculating the cost contribution ratio of shared resources to each response measure based on the work log are further specified as follows: obtain the working time period of shared resources under each response measure, obtain the efficiency coefficient corresponding to each response measure, multiply the working time period of shared resources under each response measure with the corresponding efficiency coefficient to obtain the weighted working time, and determine the cost contribution ratio based on the proportion of the weighted working time of each response measure to the total weighted working time.

3. The method for full life-cycle management of engineering costs according to claim 2, characterized in that, The steps for obtaining the efficiency coefficient corresponding to each response measure are as follows: real-time operating parameters characterizing the current operating status are collected in real time by heterogeneous sensors deployed on shared resources to obtain the current task stage information of the response measure; based on the current task stage information of the response measure, a baseline operating parameter is selected from the parameter set stored corresponding to the task stage information; and the efficiency coefficient is dynamically generated by comparing the real-time operating parameters with the selected baseline operating parameters.

4. The method for full life-cycle management of engineering costs according to claim 3, characterized in that, The steps for acquiring real-time operating parameters that characterize the current operating status of heterogeneous sensors deployed on shared resources are as follows: data fusion processing is performed on the multi-dimensional real-time operating parameters acquired in parallel by the heterogeneous sensors to generate fused operating parameters; anomaly detection is performed on the fused operating parameters; when anomalies are detected in the fused operating parameters, the abnormal data is corrected or supplemented according to the stored anomaly handling rules, and the processed operating parameters are used as the final real-time operating parameters.

5. The method for full life-cycle management of engineering costs according to claim 4, characterized in that, The step of performing anomaly detection on the fused operating parameters is achieved by generating a dynamic threshold range that varies with operating conditions. Specifically, the step of generating the dynamic threshold range that varies with operating conditions involves: obtaining the historical operating parameters of the shared resources under various historical operating conditions; for each operating condition, calculating the mean and standard deviation of the corresponding historical operating parameters, and generating a corresponding dynamic threshold range for each operating condition. The dynamic threshold range is determined as follows: the difference between the historical mean and the product of a confidence coefficient and the historical standard deviation is used as the lower limit, and the sum of the historical mean and the product is used as the upper limit. The fused operating parameters are compared with the dynamic threshold range under the current operating condition, and the fused operating parameters that exceed the range are identified as abnormal data.

6. The method for full life-cycle management of engineering costs according to claim 4, characterized in that, When an anomaly is detected in the fused operating parameters, the steps to correct or supplement the abnormal data according to the stored anomaly handling rules are as follows: identify the type of abnormal data, call the corresponding correction or supplementation algorithm model according to the type of abnormal data, comprehensively evaluate the confidence level of the abnormal data, use the confidence level to adjust the strength of the correction or supplementation algorithm model, execute the correction or supplementation algorithm model after strength adjustment, and generate the correction or supplementation result.

7. The method for full life-cycle management of engineering costs according to claim 6, characterized in that, The steps for comprehensively evaluating the confidence level of anomalous data are as follows: obtain the source information of the heterogeneous sensors that generated the anomalous data, and obtain the initial confidence score given by the anomaly detection algorithm for the anomalous data; The system queries historical correction effectiveness evaluation records for similar anomaly data, combines source information, initial confidence scores, and historical correction effectiveness evaluation records, and calculates the final confidence level using a stored weighted model.

8. The method for full life-cycle management of engineering costs according to claim 1, characterized in that, The standardized evolution operators include a nature deepening operator for deepening the understanding of the nature of the source event, and an impact expansion operator for expanding the scope of the source event's influence. It also includes, after the steps of chain tracing and quantifying the direct costs caused by the source event and the indirect costs transmitted through shared resources, executing a cost risk warning based on evolutionary trends. Specifically, the cost risk warning involves continuously recording the evolution data of all indirect costs associated with the source event described by a structured information body as the project progresses, calculating the rate of change of the evolution data, and using the rate of change to characterize the growth trend of indirect costs. When the rate of change continuously exceeds a risk threshold calculated based on historical cost data, an early warning instruction is automatically generated and output, indicating that the growth trend of indirect costs caused by the source event exceeds the risk threshold.

9. A method for full life-cycle management of engineering costs according to claim 6, characterized in that, The steps for using confidence level to adjust the strength of the modified or supplemented algorithm model are as follows: based on the confidence level value, select a corresponding strength adjustment curve from multiple stored strength adjustment curves representing different adjustment strengths; based on the selected strength adjustment curve, determine the strength parameter used to modify or supplement the algorithm model, and apply the strength parameter.

10. A project cost lifecycle management system, characterized in that, include: One or more processors and a memory, on which a computer program is stored, wherein when the computer program is executed by one or more processors, it implements a method for full life-cycle management of engineering costs as claimed in any one of claims 1 to 9.

Images