A new energy service target cost management system

Abstract

The application relates to the technical field of cost control, and discloses a new energy service target cost control system, which aims at the core pain points that cost control is separated from power generation and equipment life management, and compliance control is separated from cost control in the new energy power station operation and maintenance scene, takes the maximization of the net income of the whole life cycle of the power station as a core target, mines the strong coupling correlation of four dimensions of cost, power generation, equipment life and compliance through a multi-head attention mechanism, constructs a target cost calculation model with endogenous embedding of double rigid constraints of coupling constraints and compliance constraints, simultaneously runs through the whole process link of target cost decomposition, execution, deviation attribution and iterative optimization with the double constraint conditions, forms a whole life cycle closed-loop control system which can self-learn and self-iterate, and realizes fine, dynamic and whole-process control of the new energy operation and maintenance target cost under the premise of rigidly guaranteeing equipment health, power generation up to standard and compliance execution.

Description

Technical Field

[0001] This invention relates to the field of cost control technology, and in particular to a target cost control system for new energy services. Background Technology

[0002] Against the backdrop of the continuous advancement of the "dual carbon" strategy, my country's photovoltaic and wind power industries have entered a stage of large-scale and high-quality development. The operation and maintenance market of existing power plants has become the core growth pole of the new energy industry. According to official data released by the National Energy Administration, by the end of 2025, the cumulative installed capacity of wind power and photovoltaic power in China exceeded 1.4 billion kilowatts, and the newly installed capacity has ranked first in the world for many consecutive years. The corresponding domestic photovoltaic / wind power plant operation and maintenance service market has exceeded 100 billion yuan. Moreover, as the service life of existing power plants increases, the market space and profit pressure of operation and maintenance services are increasing simultaneously.

[0003] The aforementioned and existing related technologies often suffer from the following shortcomings: 1. Existing technologies for photovoltaic / wind power plants with cross-regional distributed layouts and ultra-long operation and maintenance cycles of 20-25 years all treat operation and maintenance cost control, power generation guarantee, and equipment life cycle health management as independent single-dimensional modules. They fail to address the strong coupling relationship between cost input, marginal gain of power generation, and equipment remaining lifespan degradation at the level of a single station, single equipment, and single operation and maintenance process. They also fail to construct quantitative constraint models adapted to different stages of the power plant's life cycle and embed these coupling constraints into the entire process control chain of target cost calculation, decomposition, execution, and optimization. This results in the target cost calculation and control being completely divorced from the actual operating conditions of the power plant's operation and maintenance, leading to problems such as unplanned equipment downtime, irreversible loss of power generation, and a significant reduction in equipment service life caused by excessive compression of operation and maintenance costs. 2. Existing technologies for photovoltaic / wind power plants with distributed layouts across regions and differentiated operations in multiple scenarios all treat data compliance and ethical control as additional functions separate from the target cost control system. They do not break down the compliance requirements of different regions, different scenarios, and dynamic updates into quantifiable constraint parameters, nor do they embed the cost impact of compliance requirements on single-site operation and maintenance procedures, data collection, personnel scheduling, and storage solutions into the entire process control link of target cost calculation, decomposition, execution, and optimization. As a result, the regional differences and dynamic updates of compliance requirements cannot be transmitted to the cost control system in real time, leading to problems such as uncontrolled growth of compliance costs and serious deviations from target cost execution. Summary of the Invention

[0004] The technical problem to be solved by this invention is that the existing technology has the shortcomings of coupling imbalance and lack of compliance adaptation in the operation and maintenance cost control of photovoltaic and wind power. To this end, we propose a new energy service target cost control system.

[0005] To achieve the above objectives, this application adopts the following technical solution: a new energy service target cost control system, including a multi-source heterogeneous compliance data base module, a power plant full life cycle coupled feature encoding module, a dual-constraint target cost dynamic calculation core engine module, a target cost hierarchical decomposition and responsibility implementation module, a full-process closed-loop control and dynamic adaptation execution module, a multi-dimensional deviation attribution and full-link iterative optimization module, and a hierarchical visualization and performance evaluation closed-loop module.

[0006] The multi-source heterogeneous compliance data foundation module outputs standardized homogeneous compliance data to the power plant full life cycle coupling feature encoding module. The power plant full life cycle coupling feature encoding module generates a coupling feature set based on the standardized homogeneous compliance data and outputs it to the dual-constraint target cost dynamic calculation core engine module. The dual-constraint target cost dynamic calculation core engine module has built-in dual rigid constraints of coupling constraints and compliance constraints, and outputs a target cost benchmark value that meets both constraints. The coupling constraint mechanism is used to characterize the dynamic correlation between cost input, power generation, and equipment life. The compliance constraint mechanism is used to quantify the dynamically changing compliance requirements into calculable constraint parameters. The dual-constraint target cost dynamic calculation core engine module takes maximizing the net benefit of the power plant throughout its entire life cycle as the optimization objective and outputs a target cost benchmark value that simultaneously meets the coupling constraints and compliance constraints. The multi-dimensional deviation attribution and full-link iterative optimization module, based on the deviation between the actual execution data and the target cost benchmark value, locates the root cause of the deviation by backtracking the feature weights generated by the coupling feature encoding module, and feeds back the optimization results to the multi-source heterogeneous compliance data foundation module, the power plant full life cycle coupling feature encoding module, and the dual-constraint target cost dynamic calculation core engine module to achieve full-link self-iteration.

[0007] Preferably, the multi-source heterogeneous compliance data foundation module is used to complete the pre-compliance verification of multi-source data, unified encoding of homogeneous master data, and full lifecycle compliant storage. Specifically, it includes: real-time collection of full-dimensional raw data from the power station through SCADA systems, IoT sensor platforms, operation and maintenance work order systems, financial ERP systems, SRM supply chain systems, and compliance management systems deployed at photovoltaic / wind power station sites. This raw data includes equipment technical parameters reflecting the operating status of the power station equipment, business process data reflecting the execution of power station operation and maintenance, financial accounting data reflecting the cost occurrence of the power station, and compliance rule data reflecting regional compliance requirements. Each data source access port is equipped with a pre-verification gateway to perform compliance pre-verification on the collected raw data. The verification includes verification of the legal authorization of data collection, verification of the minimum necessary principle of the collection scope, and verification of the matching degree of regional compliance requirements. Raw data that does not meet the verification rules is directly intercepted and a compliance risk log is generated. Raw data that passes the verification is subjected to irreversible de-identification processing of sensitive data. The raw data that passes the verification is arranged according to the time sequence of data collection to generate a standardized homogeneous compliance dataset with time sequence relationship. Each data unit in the standardized homogeneous compliance dataset contains corresponding timestamp information and globally unified master data code.

[0008] Preferably, the power plant full lifecycle coupled feature encoding module is used to extract features from four dimensions: cost, power generation, equipment lifespan, and compliance. It employs a multi-head attention mechanism network to learn the intrinsic correlation weights between features of different dimensions, generating a cross-dimensional coupled feature encoding matrix to represent the strong coupling relationship between cost, power generation, and equipment lifespan. The output is a coupled feature set adapted to different stages of the power plant's full lifecycle. Specifically, this includes: filtering standardized, homogeneous compliance datasets, removing redundant parameters, retaining key parameters that directly reflect the power plant's operation and maintenance status, standardizing the key parameters to obtain standardized operation and maintenance full-dimensional parameters; and standardizing the standardized operation and maintenance parameters. The system processes parameters across all dimensions, divides the power plant into stages throughout its entire lifecycle, and completes stage coding. It extracts four fundamental features related to cost control, generating sets of cost features, power generation features, equipment lifespan features, and compliance features. Each feature set has a one-to-one correspondence with key parameters. The four fundamental features are then jointly coded to extract coupling features that reflect the inherent relationships between the multi-dimensional features, generating a set of coupling features containing features from multiple dimensions. This set of coupling features reflects the interconnected changes in cost, power generation, equipment lifespan, and compliance of the photovoltaic / wind power plant throughout its entire operation and maintenance lifecycle.

[0009] Preferably, the standardized O&M full-dimensional parameters are processed, specifically including: dividing the standardized O&M full-dimensional parameters into three life cycle stages—commissioning period, stable period, and aging period—according to the power plant's commissioning years, equipment design life, and actual operating conditions; performing one-hot encoding on each life cycle stage to generate basic features for the life cycle stage; dividing the standardized O&M full-dimensional parameters into multiple parameter time periods according to time series, with each parameter time period containing multiple continuously collected O&M full-dimensional parameters; calculating the mean, variance, and rate of change of the parameters for each parameter time period as basic features of the O&M status; analyzing the variation patterns of the O&M full-dimensional parameters between different parameter time periods, extracting periodic features to reflect the periodic changes in the power plant's O&M status, and generating basic feature sets corresponding to the cost dimension, power generation dimension, equipment life dimension, and compliance dimension; and combining the life cycle stage basic features, O&M status basic features, and periodic features to generate a full set of basic features for the four dimensions.

[0010] Preferably, the four basic features are jointly encoded to extract coupling features that reflect the intrinsic relationships between the multi-dimensional features, generating a coupling feature set. Specifically, this includes: constructing a multi-head attention mechanism network, inputting the set of basic features from the four dimensions into the multi-head attention mechanism network, automatically learning the intrinsic relationship weights between different dimensional features and between different features, and identifying strongly correlated feature pairs that significantly affect target cost control; jointly encoding the identified strongly correlated feature pairs to generate a coupling feature vector, while incorporating the basic features of the life cycle stage into the coupling feature vector, adjusting the feature weights of different life cycle stages to form a coupling feature matrix adapted to different stages of the power plant's entire life cycle; and standardizing and reducing the dimensionality of the coupling feature matrix to generate a coupling feature set that can be directly input into the core engine module for dynamic calculation of dual-constraint target cost. Each coupling feature in the coupling feature set corresponds to a clear dimensional relationship and physical meaning.

[0011] Preferably, the core optimization objective function of the dual-constraint target cost dynamic calculation core engine module is: Where T represents the total number of years in the power plant's entire lifecycle, t represents the t-th year in the entire lifecycle, S represents the total number of lifecycle stages, s represents the s-th lifecycle stage, K represents the total number of operation and maintenance cost items, and k represents the k-th cost item. The core decision variables are the target cost values ​​for year t, stage s, and cost category k. Let be the power generation coupling function, representing the annual power generation of the power plant in year t and stage s, and be a function of operation and maintenance cost input. The benchmark on-grid tariff for photovoltaic / wind power in year t. Here, represents the compliance requirement coefficient, reflecting the dynamic impact of compliance requirements on costs; r is the industry benchmark discount rate. The weight coefficient is the penalty coefficient for coupling constraints. For compliance constraint penalty weighting coefficient, For violations of coupling constraints, the total penalty term is applied. To comply with the total penalty for violating the regulations.

[0012] Preferably, the specific calculation methods for the power generation coupling function, the total penalty term for violation of coupling constraints, and the total penalty term for violation of compliance constraints in the core optimization objective function include: power generation coupling function The calculation formula is: ; in, The rated power generation of the power plant in year t and stage s. This is the baseline power generation loss rate under conditions of no operation and maintenance investment. Let be the power generation gain coefficient for the k-th type of cost input. The minimum baseline operation and maintenance investment is defined as the cost for category k, where M represents the total number of core equipment in the power plant. Let be the health gain coefficient of the m-th device, and be a function of operation and maintenance cost input; the total penalty term for violation of coupling constraints. The calculation formula is: in, The minimum health threshold for equipment in year t and phase s. For the overall health of the power plant equipment, This represents the minimum power generation guarantee threshold for year t and phase s. Let be the total target cost for year t and stage s. The target cost threshold for year t and stage s is given.

[0013] Total Penalty Items for Violation of Compliance Constraints The calculation formula is: Where N represents the total number of compliant clauses. This represents the minimum compliance requirement threshold for the nth compliance clause. To determine the degree of actual compliance satisfaction of the nth compliance clause, The risk level coefficient for the nth compliance clause.

[0014] Preferably, the target cost hierarchy decomposition and responsibility assignment module is used to integrate dual constraints throughout the entire decomposition process, generating a hierarchical budget control ledger that binds responsible entities. Specifically, it includes: receiving the approved and locked target cost baseline value; decomposing the total target cost into core maintenance processes such as daily inspections, preventative maintenance, fault repair, technical upgrades, and compliance testing based on the WBS (Work Breakdown Structure) for maintenance work; ensuring that the cost amount for each process strictly matches the dual rigid constraints of coupling and compliance; and further decomposing the target cost of each maintenance process into labor costs, spare parts costs, etc., based on the CBS (Cost Breakdown Structure) for cost decomposition. The cost items corresponding to outsourcing service costs, travel scheduling costs, compliance costs, technological transformation costs, and management costs are clearly defined, specifying the budget amount, scope of use, and constraint rules for each cost item. The decomposed process costs and cost items are mapped to specific power plants, equipment units, and operation and maintenance grids, and simultaneously bound to the three-level responsibility system of group headquarters-regional company-power plant, clarifying the budget amount, control authority, and performance indicators for each responsible entity. According to the preset ledger format, the decomposed cost amounts, constraint rules, and responsible entities are integrated to generate a hierarchical target cost control ledger, which serves as a rigid basis for closed-loop control of the entire process.

[0015] The preferred modules include a full-process closed-loop control and dynamic adaptation execution module, and a multi-dimensional deviation attribution and full-link iterative optimization module. These modules are used to achieve rigid control, root cause localization, and full-link self-iteration throughout the entire operation and maintenance process. Specifically, they include: receiving the target cost control ledger, constructing a full-process online control link from pre-application to in-process execution to post-verification; verifying the matching degree between the budget and the dual constraints for all operation and maintenance work order initiation, spare parts requisition, procurement application, travel expense reimbursement, and outsourced service settlement actions; automatically intercepting applications without budget or that do not meet the constraints, and automatically triggering a hierarchical approval process for applications exceeding the budget; and collecting dynamic change data on power plant equipment operating conditions, compliance requirements, and weather conditions in real time. When abnormal equipment health, updated compliance requirements, sudden failures, or sudden changes in weather conditions occur, the budget is automatically adjusted without exceeding the total target cost limit or violating the dual constraints. The system dynamically adjusts processes and cross-subjects to update the target cost control ledger; it compares the target cost benchmark value, constraint threshold, and actual execution data to calculate multi-dimensional deviation data; based on the coupled feature coding matrix, it traces back to locate the core causes of deviations through feature weights, distinguishing five types of deviation causes: changes in operating conditions, updates to compliance requirements, model calculation errors, inadequate execution, and sudden changes in the external environment, generating deviation analysis reports and optimization suggestions; the optimization suggestions and parameter adjustment results are fed back to all front-end modules, optimizing the model parameters of the dual-constraint target cost dynamic calculation core engine module for model calculation errors, optimizing the coding logic of the power plant full life cycle coupled feature coding module for feature weight deviations, optimizing the target cost decomposition and control execution rules for execution rule defects, and optimizing the data collection logic of the multi-source heterogeneous compliance data base module for data collection defects, achieving end-to-end self-iteration.

[0016] Preferably, a method for controlling the target cost of new energy services includes the following steps: S1: Acquire multi-source raw data throughout the entire lifecycle of photovoltaic / wind power plant operation and maintenance, complete compliance pre-verification and unified encoding of the same-source master data, and generate a standardized same-source compliance dataset with time-series relationships; S2: Perform feature extraction processing on the standardized same-source compliance dataset, divide the entire lifecycle of the power plant into stages and complete stage encoding, extract basic features of four dimensions: cost, power generation, equipment life, and compliance, construct a cross-dimensional coupled feature encoding matrix, and generate a coupled feature set; S3: Input the coupled feature set into a preset dual-constraint target cost calculation model, with the maximization of the net profit of the power plant throughout its lifecycle as the optimization objective, and measure the target cost under the dual rigid constraints of coupling constraints and compliance constraints. S4: Based on the target cost benchmark value, the dual constraints are decomposed throughout the entire process to complete the hierarchical decomposition of the target cost and the binding of the responsible entities, generating a hierarchical target cost control ledger; S5: Based on the target cost control ledger, rigid control of the entire operation and maintenance process and dynamic adaptation execution under changes in working conditions / compliance are completed, and actual execution data is collected synchronously; S6: Based on the target cost benchmark value and actual execution data, multi-dimensional deviation calculation and root cause attribution are completed, and the optimization results are fed back to all front-end links to achieve full-link self-iteration; S7: Based on the full-link control data, a hierarchical and visualized control system and a multi-dimensional performance evaluation system with dual constraint targets as the core are constructed to achieve a closed loop for the implementation of control targets.

[0017] The technical effects and advantages of this invention are as follows: This invention deeply explores the inherent strong coupling relationship between operation and maintenance cost input, marginal gain of power generation, and equipment remaining lifespan decline at the single station, single equipment, and single operation and maintenance process level through a multi-head attention mechanism. It constructs a quantitative coupling constraint model adapted to different stages of the power plant's entire lifecycle—commissioning, stabilization, and aging. This coupling constraint is then fully embedded as an endogenous rigid condition into the entire process control chain of target cost calculation, decomposition, execution, and optimization. With maximizing the net benefit of the power plant throughout its entire lifecycle as the core optimization objective, every link in cost control is deeply bound to the actual operating conditions of equipment operation and maintenance and the target power generation revenue. This fundamentally solves the two-way problems of unplanned equipment downtime, irreversible power generation loss, and shortened equipment service life caused by excessive cost reduction, as well as the indiscriminate over-maintenance driving up the entire lifecycle cost. The pain point is that this invention breaks down cross-regional, multi-scenario, and dynamically updated compliance requirements into quantifiable, calculable, and verifiable rigid constraint parameters. It accurately quantifies the cost impact coefficients of different compliance clauses on single-site operation and maintenance procedures, data collection, personnel scheduling, and storage solutions. Then, the compliance constraints are fully embedded into the entire process control link of target cost calculation, decomposition, execution, and optimization. From the pre-compliance verification in the data collection stage to the compliance constraint penalty mechanism in the cost calculation stage, and then to the rigid compliance verification in the execution stage, the regional differences and dynamic updates of compliance requirements are transmitted to the entire cost control link in real time. This transforms compliance control from an independent additional function into an endogenous constraint of cost control, which not only avoids compliance red line risks from the source, but also achieves precise control of compliance costs, and completely solves the problems of uncontrolled growth of compliance costs and serious deviations from target cost execution. Attached Figure Description

[0018] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts:

[0019] Figure 1 This is a schematic diagram of the new energy service target cost control system of the present invention; Figure 2 This is a schematic diagram of the new energy service target cost control method of the present invention. Detailed Implementation

[0020] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0021] Reference Figure 1-2As shown, the new energy service target cost control system of this invention, when running, first executes the step of generating a standardized, homogeneous compliance dataset. Through SCADA systems, IoT sensor platforms, operation and maintenance work order systems, financial ERP systems, SRM supply chain systems, and compliance management systems deployed at photovoltaic / wind power plant sites, it collects real-time raw data from all dimensions of the power plant. The collected raw data includes equipment technical parameters reflecting the operating status of the power plant equipment, business process data reflecting the operation and maintenance execution of the power plant, financial accounting data reflecting the cost occurrence of the power plant, and compliance rule data reflecting regional compliance requirements. A pre-verification gateway set up at the access port of each data source performs compliance pre-verification on the collected raw data. This verification gateway, as the core lower-level execution module of this step, adopts a conventional technical architecture combining embedded hardware deployment and a software rule engine. The rule engine uses Drools rule management technology to dynamically update the verification rules. The verification content includes verification of the legal authorization of data collection, verification of the minimum necessary principle of the collection scope, and verification of the matching degree of regional compliance requirements. Raw data that does not conform to the verification rules is directly intercepted and an immutable dataset is generated. The compliance risk log performs irreversible desensitization on the verified raw data. This desensitization process uses the industry-standard SHA-256 hash algorithm combined with masking technology. Personal sensitive information such as maintenance personnel's ID numbers and mobile phone numbers is masked, while owner's trade secrets and core operational data are desensitized using irreversible hashing. The verified raw data is then arranged according to the time sequence of data collection, and a conventional time series alignment technique using linear interpolation is employed to unify the timestamps of data from different sampling frequencies. This generates a standardized, homogeneous compliance dataset with time-series relationships. Each data unit in the standardized homogeneous compliance dataset contains corresponding timestamp information and a globally unified master data code. This master data code uses a segmented unique coding rule to ensure that data across systems and data sources can be accurately correlated through unified coding. This step achieves simultaneous completion of compliance control and data collection from the data source, solving the technical problems of data collection and compliance control being disconnected, business, financial, and equipment data not originating from the same source, and data silos in existing technologies. This ensures that all data input into the system is compliant, homogeneous, and traceable, providing reliable underlying data support for subsequent feature extraction and cost calculation.

[0022] After generating the standardized, homogeneous compliance dataset, the system executes the coupling feature set generation step. Feature extraction is performed on the generated standardized, homogeneous compliance dataset. First, the standardized, homogeneous compliance dataset is screened. Then, using conventional statistical techniques such as Pearson correlation coefficient analysis, the correlation between each parameter and the target cost control is calculated. The formula for calculating the Pearson correlation coefficient is: ,in, These are the timing values ​​of the parameters. This corresponds to the time series cost value. The parameters are taken as the average of the two values. Redundant parameters with a correlation of less than 0.1 with the target cost control are removed, and key parameters that directly reflect the operation and maintenance status of the power plant are retained. Then, the key parameters are standardized using the conventional data preprocessing technique of Min-Max normalization. The normalization formula is as follows: ,in, The original parameter values ​​are the maximum and minimum values ​​of the parameter within the statistical period. All parameter values ​​are mapped to the [0,1] interval to eliminate the dimensional differences between different parameters and avoid interference from numerical range differences in subsequent model training, resulting in standardized O&M full-dimensional parameters. These standardized O&M full-dimensional parameters are then processed. Based on the power plant's commissioning years, equipment design life, actual operating conditions, and performance degradation data, the entire O&M lifecycle of the power plant is divided into three core stages: commissioning period, stable period, and aging period. Conventional classification feature processing techniques using one-hot encoding are applied to these three lifecycle stages: commissioning period is encoded as [1,0,0], stable period as [0,1,0], and aging period as [0,0,1], generating basic features for each lifecycle stage. Finally, the standardized O&M full-dimensional parameters are divided into multiple parameter times according to the time series. For each parameter time period, the mean, variance, rate of change, extreme values, and volatility of the parameter are calculated as basic features of the operation and maintenance status. Simultaneously, the variation patterns of all dimensions of operation and maintenance parameters across different parameter time periods are analyzed. Conventional time-series signal processing techniques, such as Fast Fourier Transform, are used to extract periodic features reflecting the cyclical changes in the power plant's operation and maintenance status. The basic features of the lifecycle stage, the basic features of the operation and maintenance status, and the periodic features are combined to generate basic feature sets corresponding one-to-one with the cost, power generation, equipment lifespan, and compliance dimensions. A multi-head attention mechanism network is then constructed. This network is a conventional technical architecture in the fields of natural language processing and time-series data analysis. The multi-head attention mechanism is used to learn the intrinsic correlation weights of the four dimensions of cost, power generation, equipment lifespan, and compliance, generating coupled features that can be used for target cost calculation. This differs from general feature extraction; it serves the subsequent dual-rigid-constraint optimization model and is an organic whole.

[0023] This invention adapts the number of network heads and layers for new energy operation and maintenance scenarios, setting up a 4-attention head, 3-layer network structure. Its core attention calculation formula is as follows: Where Q is the query matrix, K is the key matrix, and V is the value matrix. As a feature dimension, the basic feature sets of the four dimensions are input into a multi-head attention mechanism network. The network automatically learns the intrinsic correlation weights between features of different dimensions and between different features, identifies strongly correlated feature pairs that significantly affect target cost control, and jointly encodes the identified strongly correlated feature pairs to generate coupled feature vectors. At the same time, the basic features of the life cycle stage are integrated into the coupled feature vectors, and the feature weights of different life cycle stages are adjusted to form a coupled feature matrix adapted to different stages of the power plant's entire life cycle. Finally, the conventional data dimensionality reduction technique of principal component analysis (PCA) is used to standardize and reduce the dimensionality of the coupled feature matrix. PCA transforms a set of potentially correlated variables into a set of linearly uncorrelated variables through orthogonal transformation, removes redundant features, and sets the cumulative variance contribution rate to 95%, generating a coupled feature set that can be directly input into the dual-constraint target cost calculation model. This step solves the technical problem of existing technologies that extract single-dimensional features and cannot capture the intrinsic coupling relationship of multiple variables. The generated coupled feature set can truly reflect the intrinsic linkage relationship of the four dimensions of cost, power generation, equipment life, and compliance in the operation and maintenance of the power plant, providing accurate feature input for subsequent dual-constraint target cost calculation, and significantly improving the accuracy and adaptability of the cost calculation model.

[0024] After the coupling feature set is generated, the system executes the calculation step of the dual-constraint target cost benchmark value. The generated coupling feature set is input into the preset dual-constraint target cost calculation model. First, a sub-module is constructed through the model's built-in coupling constraints. This sub-module is the core lower-level execution unit of the model, used to construct the coupled marginal benefit function of cost-power generation-equipment lifespan, clarifying the marginal gain of power generation and the marginal change of equipment remaining lifespan corresponding to different operation and maintenance cost inputs. Simultaneously, based on power plant design specifications, industry standards, and owner profit targets, three rigid constraint thresholds are set: a minimum equipment health threshold, a minimum power generation guarantee threshold, and a target cost red line threshold, forming internal operating condition coupling constraints. Finally, the model's built-in compliance constraints are applied... The dynamic adaptation submodule, another core lower-level execution unit of the model, receives compliance dimension features from the coupled feature set, analyzes the rigid compliance requirements of the corresponding region and scenario, decomposes compliance clauses into quantifiable constraint parameters, sets risk level coefficients and minimum execution requirement thresholds for each compliance clause, forming external rigid compliance constraints. When national or local compliance requirements are revised, the constraint parameters and risk level coefficients are automatically updated, achieving real-time linkage between compliance requirements and cost calculation. The model aims to maximize the net revenue of the power plant throughout its entire life cycle. Under the dual rigid constraints of coupled constraints and compliance constraints, it completes the target cost calculation through a core optimization objective function, which is: The physical meaning and value rules of each parameter in the formula are as follows: T is the total number of years in the entire life cycle of the power station, T=25 for photovoltaic / offshore wind power and T=20 for onshore wind power; t is the t-th year in the entire life cycle, with a value range of [1,T]; S is the total number of life cycle stages, with a fixed value of 3; s is the s-th life cycle stage, s=1 is the commissioning period, s=2 is the stable period, and s=3 is the aging period; K is the total number of operation and maintenance cost items, with a fixed value of 8, namely labor cost, spare parts cost, outsourced service cost, travel and scheduling cost, compliance cost, technical transformation cost, and management cost; k is the k-th cost item, with a value range of [1,8]. The core decision-making variable is the target cost value of the cost item in year t, stage s, and category k, expressed in ten thousand yuan. represents the benchmark on-grid tariff for photovoltaic / wind power in year t, in yuan / kWh, and is determined according to national and local electricity pricing policies; r is the industry benchmark discount rate, which is fixed at 6%. The penalty weight coefficient for coupling constraints is fixed at 100 to ensure the rigidity of the constraints; The base value for the compliance constraint penalty weighting coefficient is 200, and it can be further amplified for high-risk compliance clauses. Let be the power generation coupling function, representing the annual power generation of the power plant in year t and stage s, in ten thousand kilowatt-hours; and let be the function of operation and maintenance cost input, calculated as follows: ;in, The rated power generation of the power plant in year t and stage s. This is the baseline power generation loss rate under conditions of no operation and maintenance investment. Let be the power generation gain coefficient for the k-th type of cost input. The minimum baseline operation and maintenance investment is defined as the cost for category k, where M represents the total number of core equipment in the power plant. Let be the health gain coefficient of the m-th device, and be a function of operation and maintenance cost input: total penalty term for violation of coupling constraints. The calculation formula is: in, The minimum health threshold for equipment in year t and phase s. For the overall health of the power plant equipment, This represents the minimum power generation guarantee threshold for year t and phase s. Let be the total target cost for year t and stage s. The target cost threshold for year t and phase s; total penalty for violations of compliance constraints. The calculation formula is: Where N represents the total number of compliant clauses. This represents the minimum compliance requirement threshold for the nth compliance clause. To determine the degree of actual compliance satisfaction of the nth compliance clause, The risk level coefficient for the nth compliance clause is calculated by multiplying the penalty value quadratically when the calculation result violates the constraints, thus ensuring the rigid enforcement of the constraints.

[0025] During the forward propagation of the model, the prediction error is calculated using an improved loss function. The formula for the loss function is: ; among which the basic predicted loss A robust loss function, Huber loss, is used with a parameter set to 1.345. This effectively reduces the interference of outliers on model training and ensures model robustness. Then, the AdamW optimizer, a standard deep learning optimization algorithm, is used for backpropagation to update the model weights. The learning rate is set to 1e-4, and the batch size is... The size is set to 32, the maximum number of iterations is set to 200, and the gradient pruning threshold is set to 1.0 to avoid gradient explosion. At the same time, the conventional model overfitting suppression technique with early stopping mechanism is adopted. When the validation set loss does not decrease for 15 consecutive epochs, training is stopped, the optimal model weights are saved, and after the model training and convergence are completed, the output target cost calculation results are verified for dual constraint satisfaction. Calculation results that do not meet the constraints are returned to the model for re-optimization. For the calculation results that pass the verification, three versions of the calculation scheme are generated: a baseline scheme, a conservative scheme, and an aggressive scheme, for management to approve at different levels. The target cost baseline value is locked after hierarchical approval. This step solves the technical problems of existing technology cost calculation having no effective constraints, being detached from operation and maintenance conditions, and being detached from compliance requirements. For the first time, coupling constraints and compliance constraints are used as intrinsic constraints of the model, rather than external additional conditions. The output target cost baseline value achieves the optimal balance between operation and maintenance cost investment, power generation gain, and equipment life protection, and fully meets dynamic compliance requirements. It avoids the industry pain points of over-maintenance, under-maintenance, compliance risks, and uncontrolled compliance costs from the root.

[0026] After the target cost baseline value is locked, the system executes the steps of target cost hierarchical decomposition and ledger generation. Based on the approved and locked target cost baseline value, dual constraints are applied throughout the entire decomposition process. First, based on conventional project management techniques using the Work Breakdown Structure (WBS) for operation and maintenance, a four-level WBS decomposition structure is constructed. The first level is the overall operation and maintenance target of the power plant; the second level is the operation and maintenance professional classification; the third level is the major categories of operation and maintenance procedures; and the fourth level is the smallest operation and maintenance procedure unit. The total target cost is decomposed to the smallest operation and maintenance procedure unit. The cost amount corresponding to each procedure is strictly matched with the dual rigid constraints of coupling constraints and compliance constraints. All are associated with the corresponding equipment master data code and compliance clause master data code to ensure that the decomposed cost amount corresponds one-to-one with the equipment operation and maintenance needs and compliance requirements. Then, based on the conventional cost management technology of CBS cost decomposition structure, a four-level CBS decomposition structure is constructed. The first level is the total target cost of the power plant, the second level is the cost category, the third level is the cost sub-category, and the fourth level is the smallest cost accounting unit. The target cost of each operation and maintenance process is further decomposed to the corresponding cost item. The budget amount, scope of use, constraint rules, and prohibited use scenarios are clearly defined for each cost item to ensure that every budget has a clear usage boundary.

[0027] The decomposed process costs and cost items are then mapped to specific power plants, equipment units, and operation and maintenance networks, and simultaneously bound to a three-tier responsibility system of "Group Headquarters - Regional Company - Power Plant." Corresponding budget approval authority, control authority, and performance indicators are set for each responsible entity, clearly defining "who executes, who is responsible, and who controls." This architecture adopts a fully modular and flexibly configurable design, perfectly adapting to the company's current management granularity down to the power plant level, without requiring any organizational restructuring. Furthermore, this architecture directly assigns cost control responsibility to the power plant management entity, fundamentally addressing industry pain points such as unfulfilled responsibilities, budget misappropriation, and untraceable cost overruns. It also reserves complete architectural space for future refined management upgrades and compliance requirements, allowing for the direct activation of lower-level control functions without system reconstruction, significantly saving upgrade costs.

[0028] After the target cost control ledger is generated, the system executes a closed-loop control and dynamic adaptation process. Based on the generated target cost control ledger, it connects with the maintenance work order system, mobile operation APP, SRM supply chain system, financial ERP system, and OA approval system to build a fully online control link for pre-application, in-process execution, and post-verification. All maintenance work order initiation, spare parts requisition, purchase application, travel expense reimbursement, and outsourced service settlement actions must first be initiated within the system, associating with the budget amount, work unit, and responsible entity in the corresponding budget control ledger. For each online application, the system will... First, a rigid dual-constraint check is performed, including budget amount check, constraint rule check, and usage matching check. Applications without a budget, that do not meet constraint conditions, or that misappropriate budgets are directly blocked. Applications exceeding the budget automatically trigger a tiered approval process. For spare parts costs, a core component of maintenance costs, spare parts inventory data is supplemented directly through Excel ledger imports, integration with existing ERP system inventory modules, and manual entry. The system is fully compatible and does not affect the normal use of any functions. The system will collect real-time dynamic changes in power plant equipment operating conditions, compliance requirements, and weather conditions. When equipment health anomalies occur... In four scenarios—updated compliance requirements, sudden malfunctions, and sudden changes in extreme weather conditions—the system automatically triggers a dynamic budget adjustment assessment. It invokes a lightweight inference model of the dual-constraint target cost calculation model to recalculate the optimal budget allocation scheme for the corresponding scenario. Without exceeding the annual total target cost limit or violating the dual constraints, the system automatically performs dynamic adjustments to the budget across processes and accounts, updates the target cost control ledger, and sets three levels of early warning thresholds based on budget execution rate, constraint satisfaction, and risk level. Level one early warning is indicated by a budget execution rate ≥ 70%, and level two early warning is indicated by a budget execution rate ≥ 90%. The three-level early warning system is triggered when budget overruns or violations of the dual-constraint rigid red line. When the corresponding level of early warning is triggered, the warning information is automatically pushed to the corresponding responsible party. Applications that violate the dual-constraint rigid red line are permanently blocked, and an immutable risk log is generated. This step solves the technical problems of rigid in-process control and disconnection from actual operating conditions in existing technologies. It achieves a balance between rigid control and dynamic adaptation of target costs. It ensures that the total target cost does not exceed the red line, while adapting to changes in the operating conditions and updates to compliance requirements of power plant operation and maintenance. The response latency of overrun early warning can be controlled within 500ms, and unplanned downtime can be reduced by more than 50%.

[0029] During the control and management process, the system simultaneously executes deviation attribution and end-to-end self-iterative steps. Based on the locked target cost benchmark and actual execution data, it compares the target cost benchmark, constraint thresholds, and actual execution data. Using conventional statistical techniques such as analysis of variance, it calculates multi-dimensional deviation data, including total cost deviation, process cost deviation, item cost deviation, power generation deviation, equipment health deviation, and compliance execution deviation. It distinguishes between positive and negative deviations, calculates the impact and contribution of each deviation, and then, based on the coupled feature coding matrix, it identifies core feature variables strongly correlated with deviation data by tracing back the feature weight data generated by the coupled feature coding module throughout the power plant's entire lifecycle. Based on these core feature variables, it distinguishes five major categories of core causes of deviations: changes in operating conditions, updates to compliance requirements, model calculation errors, inadequate execution, and sudden changes in the external environment. For each category of deviation cause, it generates a corresponding deviation analysis report and actionable steps. The optimization suggestions clearly define the optimization direction, execution path, and responsible parties. Finally, the optimization suggestions and parameter adjustment results are fed back to all front-end processes. For model calculation errors, the model weights, loss function coefficients, and constraint parameters of the dual-constraint target cost calculation model are optimized. For feature weight deviations, the encoding logic of coupled feature extraction and the network parameters of the attention mechanism are optimized. For execution deficiencies, the target cost decomposition rules and control processes are optimized. For data collection deficiencies, the data collection frequency, range, and verification rules are optimized. Simultaneously, historical data on deviation attribution, optimization schemes, and execution effects are accumulated, constructing a knowledge base for new energy power plant cost control optimization. This step solves the technical problem that existing static control technologies cannot continuously optimize, achieving self-learning and self-iteration of the entire cost control system. The accuracy of deviation root cause location can reach over 92%, and the system control precision continuously improves with operating time, achieving continuous optimization and reduction of operation and maintenance costs annually.

[0030] In the final stage of system operation, a closed-loop process of visualized control and performance evaluation is implemented. Based on end-to-end control data, a comprehensive control indicator system centered on dual-constraint objectives is constructed. Using conventional information security technologies based on RBAC (Role-Based Access Control), corresponding visualized control interfaces are configured for responsible entities at different levels. From the panoramic control dashboard at the group headquarters level to the detailed control interfaces at the regional company and power plant levels, all support one-click drill-down from panoramic data to detailed data at the regional, power plant, equipment, and work order levels. Simultaneously, daily, weekly, monthly, and annual reports, as well as compliance audit reports, are automatically generated and pushed to the corresponding responsible entities. Finally, a performance evaluation indicator system centered on dual-constraint objectives is constructed, with core indicators including target cost completion rate, budget execution compliance rate, and other metrics. The system automatically captures real data for all indicators, including power consumption achievement rate, equipment health compliance rate, compliance requirement execution rate, and cost optimization results, eliminating the need for manual data entry. It sets corresponding indicator weights and assessment rules for different levels of responsible entities, automatically generating performance evaluation results and ratings for each entity. Finally, based on the performance evaluation results, it matches preset reward and punishment rules to automatically generate reward and punishment schemes, linking these results with budget approval authority, job promotion, and salary adjustments to form a complete management loop. This step solves the technical problem of the disconnect between existing technical control and assessment, achieving visualization, assessability, and implementability of control objectives. The fairness and accuracy of the assessment results reach 100%, and the participation of all employees in cost control can be increased by more than 80%.

[0031] In the centralized photovoltaic power station scenario, applicable to centralized photovoltaic power stations with an installed capacity of ≥50MW located in remote areas such as deserts / mountains, when the system performs the standardized homogeneous compliance dataset generation step, it will add the collection of component IV curve data, drone inspection image data, and vegetation cover data, with the collection frequency set to once per week. At the same time, the front-end verification gateway will add verification of environmental compliance requirements for desert / mountain photovoltaic power stations to ensure that the data collection and operation related to vegetation clearing and soil and water conservation meet the regional environmental protection requirements. When performing the coupled feature set generation step, it will add the extraction of irradiance features, terrain features, component shading features, and cleaning cycle features. In the multi-head attention mechanism network, the weight of features related to component cleaning cost and vegetation clearing cost will be increased to strengthen the coupled feature learning of cleaning cost, component cleanliness, and power generation.

[0032] When performing the dual-constraint target cost benchmark calculation step, the coupling constraint conditions are optimized by incorporating the minimum cleanliness of the components into the rigid constraint threshold and adjusting the power generation gain coefficient corresponding to the component cleaning cost in the power generation coupling function. The accuracy of the model's calculation of the relationship between cleaning costs and power generation has been improved to 0.15. At the same time, environmental compliance-related clauses have been added to the compliance constraints to ensure that the compliance cost investment for vegetation clearing and soil and water conservation meets the requirements. When performing the target cost hierarchical decomposition and ledger generation steps, special work units for component cleaning, vegetation clearing and soil and water conservation will be added to the WBS decomposition structure, special budget amounts will be set, and the corresponding operation and maintenance teams and responsible entities will be bound.

[0033] When implementing closed-loop management and dynamic adaptation steps throughout the entire process, dynamic adaptation trigger conditions for sandstorms and rainstorms will be added. When extreme weather conditions occur, the component cleaning budget and operation plan will be automatically adjusted to optimize the cleaning cycle without exceeding the total target cost. In this scenario, the system can achieve an average absolute percentage error of ≤2.6% in target cost calculation, reduce the overall cost of component cleaning by 18%, increase annual power generation by more than 2.5%, and achieve a 100% environmental compliance rate.

[0034] In the distributed photovoltaic power station scenario, applicable to distributed photovoltaic power stations with an installed capacity of ≤6MW, distributed on the rooftops of industrial and commercial plants / residential rooftops, and highly dispersed sites, when the system executes the standardized homogeneous compliance dataset generation step, it will optimize the data acquisition protocol in low-bandwidth environments, add rooftop safety status and owner electricity load data acquisition, and set the acquisition frequency to once / 30 minutes. At the same time, the front-end verification gateway will add owner data authorization verification and rooftop operation safety compliance requirement verification to ensure that data acquisition and operation meet compliance requirements.

[0035] When performing the coupled feature set generation step, new features such as site geographical location distribution, property owner characteristics, electricity load characteristics, and transportation cost characteristics are extracted. In the multi-head attention mechanism network, the weights of travel scheduling cost and labor cost-related features are increased, and the coupled feature learning of scheduling cost-site distribution-operation and maintenance response efficiency is strengthened.

[0036] When performing the dual-constraint target cost benchmark calculation step, the coupled constraint conditions are optimized by incorporating the operation and maintenance response time into the rigid constraint threshold, adjusting the weight of travel scheduling costs in the objective function, improving the model's accuracy in calculating cross-regional scheduling costs, and adding constraint parameters related to high-altitude operation safety and electricity safety in the compliance constraints. When performing the target cost hierarchical decomposition and ledger generation steps, a regional gridded decomposition structure is constructed, dividing the operation and maintenance grids according to geographical regions, decomposing the target cost to each grid, and binding it to the corresponding operation and maintenance team to achieve regional cost control.

[0037] When implementing closed-loop management and dynamic adaptation steps throughout the entire process, a new trigger condition for merging work orders from multiple sites in the same region will be added. This will automatically merge maintenance work orders of the same type and in the same region, optimize inspection routes and personnel scheduling, and reduce travel costs without exceeding the budget. In this scenario, the system can achieve an average absolute percentage error of ≤3.2% in target cost calculation, a 30% reduction in travel scheduling costs, a 40% reduction in single-site maintenance response time, and a 100% compliance rate in safety and compliance execution.

[0038] In the onshore wind power scenario, when the system performs the standardized homogeneous compliance dataset generation step, it will add the collection of wind turbine vibration data, oil monitoring data, and blade acoustic detection data. The key component status data collection frequency is set to 1 time / minute. At the same time, the front-end verification gateway will add verification of high-altitude operation safety and special equipment operation and maintenance compliance requirements. When performing the coupled feature set generation step, it will add the extraction of wind turbine component health status features, vibration features, wind speed features, and large component life features. In the multi-head attention mechanism network, the weight of features related to large component maintenance costs and fault repair costs will be increased, and the coupled feature learning of maintenance costs, component health, and fault downtime losses will be strengthened.

[0039] When performing the dual-constraint target cost benchmark value calculation step, the coupling constraint conditions are optimized by incorporating the minimum health level of key components and the upper limit of fault downtime into the rigid constraint thresholds, and adjusting the power generation gain coefficient corresponding to the maintenance cost of large components in the power generation coupling function. The value was increased to 0.18, which improved the model's accuracy in calculating the relationship between maintenance costs and downtime losses, while also amplifying the penalty weights related to downtime in the coupling constraint penalty terms.

[0040] When performing the target cost hierarchy decomposition and ledger generation steps, special work units for preventive maintenance and condition-based maintenance of large components will be added to the WBS decomposition structure, special budget amounts will be set, and corresponding professional operation and maintenance teams will be bound to them.

[0041] When implementing closed-loop management and dynamic adaptation steps throughout the entire process, new abnormal component health trigger conditions will be added. When the component health is lower than the warning threshold, the preventive maintenance budget and work plan will be automatically adjusted to carry out maintenance work in advance and reduce downtime losses. In this scenario, the system can achieve an average absolute percentage error of ≤2.9% in target cost calculation, a 60% reduction in downtime due to major component failures, a reduction of more than 15% in annual operation and maintenance costs, and a 100% compliance rate in special equipment operation and maintenance.

[0042] In the offshore wind power scenario, when the system performs the standardized same-source compliance dataset generation step, it will add marine meteorological data, tide data, equipment corrosion status data, and submarine cable status data collection. The collection frequency is set to 1 time / hour. At the same time, the front-end verification gateway will add verification of marine data local storage, offshore operation safety, and marine environmental protection compliance requirements. Data and operation applications that do not meet the requirements will be directly intercepted.

[0043] When performing the coupled feature set generation step, new features such as marine environment features, corrosion rate features, ship scheduling cost features, and maintenance window period features are extracted. In the multi-head attention mechanism network, the weights of features related to ship scheduling cost, offshore operation cost, and corrosion prevention and maintenance cost are increased, and the coupled feature learning of maintenance cost, maintenance window period, ship scheduling, and equipment health is strengthened.

[0044] When performing the dual-constraint target cost benchmark calculation step, the coupled constraint conditions are optimized by incorporating equipment corrosion rate and maintenance window utilization into rigid constraint thresholds, adjusting the weights of offshore operation costs and vessel scheduling costs in the objective function, improving the model's accuracy in calculating specific costs in marine scenarios, and adding constraint parameters for high-risk clauses related to marine environmental protection, offshore operation safety, and marine data management to the compliance constraints, and adjusting the compliance constraint penalty weight coefficient. Scale up to 1000 to ensure the rigid enforcement of high-risk compliance provisions.

[0045] When performing the target cost hierarchy decomposition and ledger generation steps, special work units such as ship scheduling, marine corrosion prevention and maintenance, and submarine cable inspection will be added to the WBS decomposition structure, special budget amounts will be set, and corresponding marine operation and maintenance professional teams will be bound to them.

[0046] When implementing closed-loop management and dynamic adaptation of the entire process, new marine meteorological and tidal change trigger conditions will be added, the optimal operation and maintenance window will be automatically matched, multiple equipment operation and maintenance work orders will be merged, and the ship scheduling plan will be optimized. Under the premise of not exceeding the total target cost, the ship leasing and operation costs will be reduced. In this scenario, the system can achieve an average absolute percentage error of ≤3.5% in target cost calculation, a 25% reduction in ship scheduling costs, a 55% reduction in unplanned downtime, a reduction of more than 20% in annual operation and maintenance costs, and a 100% compliance rate in marine environmental protection and safety.

[0047] In the above scheme, the new energy service target cost control system used to execute the above method includes at least one processor, a memory coupled to the processor, at least one network interface, at least one input / output interface, and an edge acquisition terminal deployed at the power plant site. The processor may include at least one single-core or multi-core processor, and the processor may be a combination of a general-purpose processor and a dedicated AI accelerator processor. The general-purpose processor is responsible for system logic operations and process control, while the AI ​​accelerator processor is responsible for training and inference of the multi-head attention mechanism network and the dual-constraint calculation model. The memory includes volatile memory and non-volatile memory, and the memory stores machine-executable instructions. When the processor executes the machine-executable instructions, it implements all the steps of the aforementioned new energy service target cost control method.

[0048] The network interfaces include wired network interfaces and 5G / 4G wireless network interfaces. The wired network interfaces are used to connect to the enterprise intranet, the power plant SCADA system, and various business systems, while the wireless network interfaces are used to connect to the IoT sensor platform, edge acquisition terminals, and mobile operation APP at the power plant site. The input / output interfaces include HDMI display interfaces and USB interfaces, used to connect display devices and operating terminals to achieve visualized control and operation of the system. The edge acquisition terminals are deployed at the power plant site, using industrial-grade embedded edge computing gateways with built-in pre-verification gateway programs. They are responsible for collecting on-site data, performing compliance pre-verification, and edge preprocessing. Standardized data is uploaded to the system's main server through an encrypted channel. The system supports multiple deployment modes, including private cloud deployment, public cloud deployment, hybrid cloud deployment, and edge + cloud collaborative deployment, which can be flexibly selected according to the enterprise's scale and security needs.

[0049] This invention also discloses a computer-readable storage medium storing a computer program for electronic data interchange, wherein the computer program causes a computer to perform the steps in the new energy service target cost control method of the foregoing embodiments. This invention further discloses a computer program product comprising a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform the steps in the new energy service target cost control method of the foregoing embodiments.

[0050] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules. That is, they may be located in one place or distributed across multiple modules.

[0051] Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort. Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented using software plus necessary general-purpose hardware platforms, or of course, using hardware.

[0052] Based on this understanding, the above-mentioned technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory, random access memory, programmable read-only memory, erasable programmable read-only memory, electronically erasable rewritable read-only memory, read-only optical disc or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to have or store data.

[0053] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A target cost control system for new energy services, characterized in that, The system includes a multi-source heterogeneous compliance data foundation module, a power plant full lifecycle coupling feature encoding module, a dual-constraint target cost dynamic calculation core engine module, a target cost hierarchical decomposition and responsibility implementation module, a full-process closed-loop management and dynamic adaptation execution module, a multi-dimensional deviation attribution and full-link iterative optimization module, and a hierarchical visualization and performance evaluation closed-loop module. The multi-source heterogeneous compliance data foundation module outputs standardized homogeneous compliance data to the power plant full lifecycle coupling feature encoding module. The power plant full lifecycle coupling feature encoding module generates a coupling feature set based on the standardized homogeneous compliance data and outputs it to the dual-constraint target cost dynamic calculation core engine module. The dual-constraint target cost dynamic calculation core engine module incorporates both coupling constraints and compliance constraints, outputting a target cost benchmark value that satisfies both constraints. The coupling constraint mechanism characterizes the dynamic correlation between cost input, power generation, and equipment lifespan. The compliance constraint mechanism quantifies dynamically changing compliance requirements into calculable constraint parameters. The dual-constraint target cost dynamic calculation core engine module aims to maximize the net revenue of the power plant throughout its lifecycle, outputting a target cost benchmark value that simultaneously satisfies both coupling constraints and compliance constraints. The multi-dimensional deviation attribution and full-link iterative optimization module, based on the deviation between the actual execution data and the target cost benchmark value, locates the root cause of the deviation by backtracking the feature weights generated by the coupled feature encoding module, and feeds back the optimization results to the multi-source heterogeneous compliance data base module, the power plant full life cycle coupled feature encoding module, and the dual-constraint target cost dynamic calculation core engine module, to achieve full-link self-iteration.

2. The new energy service target cost control system according to claim 1, characterized in that: The multi-source heterogeneous compliance data base module is used to complete the pre-compliance verification of multi-source data, unified encoding of homogeneous master data, and full lifecycle compliance storage. Specifically, it includes: real-time collection of full-dimensional raw data from the power station through SCADA systems, IoT sensor platforms, operation and maintenance work order systems, financial ERP systems, SRM supply chain systems, and compliance management systems deployed at photovoltaic / wind power station sites. This raw data includes equipment technical parameters reflecting the operating status of the power station equipment, business process data reflecting the operation and maintenance execution of the power station, financial accounting data reflecting the cost occurrence of the power station, and compliance rule data reflecting regional compliance requirements. Each data source access port is equipped with a pre-verification gateway to perform compliance pre-verification on the collected raw data. The verification includes verification of the legal authorization of data collection, verification of the minimum necessary principle of the collection scope, and verification of the matching degree of regional compliance requirements. Raw data that does not meet the verification rules is directly intercepted and a compliance risk log is generated. Raw data that passes the verification is subjected to irreversible de-identification processing. The raw data that passes the verification is arranged according to the time sequence of data collection to generate a standardized homogeneous compliance dataset with time sequence relationship. Each data unit in the standardized homogeneous compliance dataset contains corresponding timestamp information and globally unified master data code.

3. The new energy service target cost control system according to claim 1, characterized in that: The power plant full lifecycle coupled feature encoding module is used to extract features across four dimensions: cost, power generation, equipment lifespan, and compliance. It employs a multi-head attention mechanism network to learn the inherent correlation weights between features of different dimensions, generating a cross-dimensional coupled feature encoding matrix to characterize the strong coupling relationship between cost, power generation, and equipment lifespan. The output is a coupled feature set adapted to different stages of the power plant's full lifecycle. Specifically, this includes: filtering standardized, homogeneous compliance datasets, removing redundant parameters, retaining key parameters that directly reflect the power plant's operation and maintenance status, standardizing the key parameters to obtain standardized operation and maintenance full-dimensional parameters; and then applying these standardized operation and maintenance full-dimensional parameters... The parameters are processed to divide the entire life cycle of the power plant into stages and complete stage coding. Four basic features related to cost control are extracted to generate cost feature sets, power generation feature sets, equipment lifespan feature sets, and compliance feature sets. Each feature set has a one-to-one correspondence with key parameters. The four basic features are jointly coded to extract coupling features that reflect the inherent correlation between the multi-dimensional features, generating a coupling feature set containing multiple dimensions. The coupling feature set is used to reflect the linkage changes of cost, power generation, equipment lifespan, and compliance of the photovoltaic / wind power plant throughout its entire operation and maintenance life cycle.

4. The new energy service target cost control system according to claim 3, characterized in that: The standardized O&M parameters are processed in the following ways: First, the standardized O&M parameters are divided into three lifecycle stages—commissioning period, stable period, and aging period—based on the power plant's commissioning years, equipment design life, and actual operating conditions. One-hot encoding is performed on each lifecycle stage to generate basic features for that stage. Second, the standardized O&M parameters are divided into multiple parameter time periods according to a time series. Each parameter time period contains multiple continuously collected O&M parameters. For each parameter time period, the mean, variance, and rate of change of the parameters are calculated as basic features of the O&M status. Third, the variation patterns of the O&M parameters across different parameter time periods are analyzed, and periodic features reflecting the cyclical changes in the power plant's O&M status are extracted. Basic feature sets corresponding to the cost dimension, power generation dimension, equipment lifespan dimension, and compliance dimension are generated. Fourth, the basic features of the lifecycle stages, the basic features of the O&M status, and the periodic features are combined to generate a full set of basic features across the four dimensions.

5. The new energy service target cost control system according to claim 3, characterized in that: The four fundamental features are jointly encoded to extract coupling features that reflect the intrinsic relationships between the multi-dimensional features, generating a coupling feature set. Specifically, this includes: constructing a multi-head attention mechanism network; inputting the four fundamental feature sets into the multi-head attention mechanism network; automatically learning the intrinsic correlation weights between different dimensional features and between different features; identifying strongly correlated feature pairs that significantly affect target cost control; jointly encoding the identified strongly correlated feature pairs to generate a coupling feature vector; simultaneously integrating the fundamental features of the life cycle stage into the coupling feature vector; adjusting the feature weights of different life cycle stages to form a coupling feature matrix adapted to different stages of the power plant's entire life cycle; and standardizing and reducing the dimensionality of the coupling feature matrix to generate a coupling feature set that can be directly input into the core engine module for dynamic calculation of dual-constraint target costs. Each coupling feature in the coupling feature set corresponds to a clear dimensional relationship and physical meaning.

6. The new energy service target cost control system according to claim 1, characterized in that: The core optimization objective function of the dual-constraint target cost dynamic calculation core engine module is: Where T represents the total number of years in the power plant's entire lifecycle, t represents the t-th year in the entire lifecycle, S represents the total number of lifecycle stages, s represents the s-th lifecycle stage, K represents the total number of operation and maintenance cost items, and k represents the k-th cost item. The core decision variables are the target cost values ​​for the cost items in year t, stage s, and category k. Let be the power generation coupling function, representing the annual power generation of the power plant in year t and stage s, and be a function of operation and maintenance cost input. The benchmark on-grid tariff for photovoltaic / wind power in year t. Here, represents the compliance requirement coefficient, reflecting the dynamic impact of compliance requirements on costs; and r represents the industry benchmark discount rate. The weight coefficients are the penalty coefficients for coupling constraints. For compliance constraint penalty weighting coefficient, For violations of coupling constraints, the total penalty term is applied. To comply with the total penalty for violating the regulations.

7. The new energy service target cost control system according to claim 1, characterized in that: The specific calculation methods for the power generation coupling function, the total penalty term for violation of coupling constraints, and the total penalty term for violation of compliance constraints in the core optimization objective function include: the power generation coupling function The calculation formula is: ;in, The rated power generation of the power plant in year t and stage s. This is the baseline power generation loss rate under conditions of no operation and maintenance investment. Let be the power generation gain coefficient for the k-th type of cost input. The minimum baseline operation and maintenance investment is defined as the cost for category k, where M represents the total number of core equipment in the power plant. Let be the health gain coefficient of the m-th device, and be a function of the operation and maintenance cost; the total penalty term for violating the coupling constraint. The calculation formula is: in, The minimum health threshold for equipment in year t and phase s. For the overall health of the power plant equipment, This represents the minimum power generation guarantee threshold for year t and phase s. Let be the total target cost for year t and stage s. The target cost threshold for year t and phase s; the total penalty for violating the compliance constraints. The calculation formula is: Where N represents the total number of compliant clauses; This represents the minimum compliance requirement threshold for the nth compliance clause. To determine the degree of actual compliance satisfaction of the nth compliance clause, The risk level coefficient for the nth compliance clause.

8. The new energy service target cost control system according to claim 1, characterized in that: The target cost hierarchy decomposition and responsibility assignment module integrates the target cost benchmark value and dual constraints throughout the entire decomposition process, generating a hierarchical budget control ledger that binds responsible entities. Specifically, this includes: receiving the approved and locked target cost benchmark value; decomposing the total target cost into core maintenance processes such as daily inspections, preventative maintenance, fault repair, technical upgrades, and compliance testing based on the WBS (Work Breakdown Structure); and ensuring that the cost amount for each process strictly matches the dual rigid constraints of coupling and compliance. Based on the CBS (Cost Breakdown Structure), the target cost for each maintenance process is further decomposed into labor costs, spare parts costs, etc. The cost items corresponding to outsourcing service costs, travel scheduling costs, compliance costs, technological transformation costs, and management costs are clearly defined, specifying the budget amount, scope of use, and constraint rules for each cost item. The decomposed process costs and cost items are mapped to specific power plants, equipment units, and operation and maintenance grids, and simultaneously bound to a flexibly configurable hierarchical responsibility system of group headquarters-regional companies-power plants, clarifying the budget amount, control authority, and performance indicators for each responsible entity. According to the preset ledger format, the decomposed cost amounts, constraint rules, and responsible entities are integrated to generate a hierarchical target cost control ledger, which serves as a rigid basis for closed-loop control of the entire process.

9. The new energy service target cost control system according to claim 1, characterized in that: The aforementioned full-process closed-loop control and dynamic adaptation execution module, and multi-dimensional deviation attribution and full-link iterative optimization module are used to achieve rigid control, deviation root cause localization, and full-link self-iteration throughout the entire operation and maintenance process. Specifically, this includes: receiving the target cost control ledger; constructing a full-process online control link from pre-application to in-process execution to post-verification; verifying the budget amount and the matching degree of the dual constraints for all operation and maintenance work order initiation, spare parts requisition, procurement application, travel expense reimbursement, and outsourced service settlement actions; automatically intercepting applications without budget or that do not meet the constraints, and automatically triggering a tiered approval process for applications exceeding the budget; and collecting real-time dynamic change data on power plant equipment operating conditions, compliance requirements, and weather conditions. When abnormal equipment health, updated compliance requirements, sudden failures, or sudden changes in weather conditions occur, the system automatically completes budget adjustments across work orders without exceeding the total target cost limit or violating the dual constraints. The system dynamically adjusts cross-subject data and updates the target cost control ledger. It compares the target cost benchmark, constraint thresholds, and actual execution data to calculate multi-dimensional deviation data. Based on the coupled feature coding matrix, it traces back to locate the core causes of deviations through feature weights, distinguishing five types of deviation causes: changes in operating conditions, updates to compliance requirements, model calculation errors, inadequate execution, and sudden changes in the external environment. It generates deviation analysis reports and optimization suggestions. The optimization suggestions and parameter adjustment results are fed back to all front-end modules. The system optimizes the model parameters of the dual-constraint target cost dynamic calculation core engine module for model calculation errors, optimizes the coding logic of the power plant full life cycle coupled feature coding module for feature weight deviations, optimizes the target cost decomposition and control execution rules for execution rule defects, and optimizes the data collection logic of the multi-source heterogeneous compliance data base module for data collection defects, achieving full-link self-iteration.

10. A method for controlling target costs of new energy services, characterized in that, The system for controlling the target cost of new energy services, as described in any one of claims 1-9, comprises the following steps: S1: Acquire multi-source raw data throughout the entire lifecycle of photovoltaic / wind power plant operation and maintenance, complete compliance pre-verification and unified encoding of source master data, and generate a standardized source compliance dataset with time-series relationships; S2: Perform feature extraction processing on the standardized source compliance dataset, divide the entire lifecycle of the power plant into stages and complete stage encoding, extract basic features of four dimensions: cost, power generation, equipment life, and compliance, construct a cross-dimensional coupled feature encoding matrix, and generate a coupled feature set; S3: Input the coupled feature set into a preset dual-constraint target cost calculation model, with the maximization of net revenue throughout the entire lifecycle of the power plant as the optimization objective, and calculate and generate a target cost benchmark value that satisfies the dual constraints under the dual rigid constraints of coupling constraints and compliance constraints; S4: Based on the target cost benchmark value, the dual constraints are applied throughout the entire decomposition process to complete the hierarchical decomposition of the target cost and bind it to the responsible entities, generating a hierarchical target cost control ledger; S5: Based on the target cost control ledger, complete the rigid control of the entire operation and maintenance process and the dynamic adaptation and execution under changes in working conditions / compliance, and collect actual execution data simultaneously; S6: Based on the target cost benchmark and actual execution data, complete multi-dimensional deviation calculation and root cause attribution, and feed the optimization results back to all front-end links to achieve full-link self-iteration; S7: Based on full-link management and control data, construct a hierarchical and visualized management and control system and a multi-dimensional performance evaluation system with dual-constraint objectives as the core, and realize the closed loop of management and control objectives.

Images