Intelligent evaluation and dynamic training system for electric power engineering project management talents

Abstract

The invention discloses an electric power engineering project management talent intelligent evaluation and dynamic training system, and the system comprises a multi-dimensional quality capability model construction module which constructs a four-dimensional capability framework based on the characteristics of an electric power engineering industry, generates a differentiated post capability model through a dynamic weight distribution algorithm, and carries out the dynamic training of the differentiated post capability model; constructing an industry ability dictionary library by utilizing a knowledge graph technology; the intelligent evaluation and matching module integrates a multi-modal evaluation tool to collect dominant and implicit capability data, dynamically adjusts capability index weights through a pre-trained AI model based on project characteristic parameters, and calculates man-post adaptation indexes by adopting a cosine similarity algorithm; and the dynamic training and development module is used for generating a personalized training scheme by quantifying the capability difference, tracking the training effect in a closed-loop manner based on a Korotkoff four-level evaluation model, and dynamically optimizing a training path by adopting a Bayesian network. The technical defects of single evaluation dimension, low man-post matching degree and insufficient cultivation efficiency of traditional power engineering human resource management can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of human resource management, and more specifically to an intelligent assessment and dynamic training system for power engineering project management talents, which is suitable for talent selection, performance management, talent inventory and personalized training scenarios in power engineering enterprises. Background Art

[0002] In the current power engineering project management, traditional human resource management methods have the following problems: First, the evaluation dimension is single, mainly relying on explicit indicators such as academic qualifications and work experience, ignoring implicit characteristics such as motivation and values, and it is difficult to distinguish between high-performance and ordinary employees; second, the person-job matching is low, and recruitment and promotion lack the support of a scientific capability model, resulting in the frequent occurrence of "high scores and low abilities"; third, the training efficiency is insufficient, the training content is generalized, and the program is not customized for individual ability gaps, resulting in serious waste of resources.

[0003] Although there are assessment tools based on competency models in the existing technology, they do not take into account the characteristics of the power engineering industry and lack a dynamic adjustment mechanism. Summary of the Invention

[0004] In view of this, the present invention provides an intelligent assessment and dynamic training system for power engineering project management talents. By integrating a multidimensional quality and ability model, a dynamic weight algorithm and a multimodal assessment technology, it solves the technical defects of traditional methods in power engineering human resource management, such as a single assessment dimension, low person-job matching and insufficient training efficiency.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] An embodiment of the present invention provides an intelligent assessment and dynamic training system for power engineering project management personnel, comprising:

[0007] A multi-dimensional competency model construction module builds a four-dimensional competency framework based on the characteristics of the power engineering industry, generates differentiated job competency models through a dynamic weight allocation algorithm, and uses knowledge graph technology to build an industry competency dictionary.

[0008] The intelligent assessment and matching module integrates multimodal assessment tools to collect explicit and implicit ability data, dynamically adjusts the weight of ability indicators based on project characteristic parameters through a pre-trained AI model, and uses the cosine similarity algorithm to calculate the person-job fit index;

[0009] The dynamic training and development module generates personalized training plans by quantifying capability gaps, tracks training results in a closed loop based on Kirkpatrick's four-level evaluation model, and uses Bayesian networks to dynamically optimize training paths.

[0010] Furthermore, the multidimensional quality and ability model construction module includes:

[0011] The four-dimensional capability framework unit is used to construct work experience, knowledge and skills, general qualities, and management / professional competence dimensions. The professional competence dimension includes secondary indicators such as safety control and cost optimization.

[0012] Dynamic differentiation modeling unit, used to achieve dynamic weight allocation of job levels and professional sequences through gradient boosting decision tree algorithm;

[0013] The industry dictionary generation unit uses knowledge graph technology to construct a logical relationship network between capability indicators and generate an industry capability dictionary.

[0014] Furthermore, the intelligent evaluation and matching module includes:

[0015] Multimodal data collection unit, integrating CBI structured interview tools, voice emotion analysis devices and psychological assessment systems to collect multimodal data of talents;

[0016] AI dynamic weight adjustment unit, which maps project feature vectors to capability weight coefficients based on the random forest model;

[0017] The person-job matching prediction unit uses the cosine similarity algorithm to calculate the matching degree of the feature vectors between the candidate and the target job.

[0018] Furthermore, the multimodal assessment tool includes:

[0019] CBI interview tool: Analyze implicit ability indicators through power engineering scenario cases and NLP technology;

[0020] Leaderless group discussion tool: Evaluate emotional stability and team influence through voice sentiment analysis and LDA topic model;

[0021] OPT career personality test tool: uses a psychological scale to generate candidate personality type labels.

[0022] Furthermore, the person-job matching prediction unit is specifically used to:

[0023] Feature vector generation: The candidate assessment results and job competency model are mapped into n-dimensional feature vectors, where each dimension is the product of the competency index score and the weight.

[0024] Similarity calculation: by formula

[0025]

[0026] Among them, C i The score of the candidate's i-th ability, P i is the weight of the i-th ability of the position;

[0027] Result output: Generates a person-job adaptation index, divides the recommended level according to the threshold, and outputs a visual radar chart to show the ability gap.

[0028] Furthermore, the dynamic training and development module includes:

[0029] The personalized training program generation unit quantitatively analyzes the ability gap and compares the candidate output by the intelligent assessment and matching module with the target position ability model.

[0030]

[0031] Calculate the gap value of the i-th capability index, where S i Score the candidate, R i thresholds for job requirements;

[0032] The Koch four-level assessment tracking unit builds a multi-level assessment and feedback mechanism for training effectiveness and outputs assessment results;

[0033] The training path dynamic optimization unit uses the Bayesian network model to dynamically adjust the weight of training resources based on the evaluation results of the Kirkpatrick four-level evaluation tracking unit.

[0034] Furthermore, the Koch four-level assessment tracking unit includes:

[0035] The response layer integrates the BERT sentiment analysis model to parse negative keywords in open text reviews;

[0036] At the learning level, differentiated knowledge tests are conducted after training to verify the learning effect;

[0037] The behavioral layer integrates the T-test behavioral analysis module to compare and analyze the differences in behavioral scores before and after training;

[0038] At the result layer, the correlation coefficient between training input and business indicators is calculated through regression model.

[0039] Furthermore, the weight distribution of the dynamic differentiation modeling unit satisfies the following constraints:

[0040] The sum of the weights of each dimension is 100%;

[0041] The fluctuation range of individual weights shall not exceed ±30% of the benchmark value;

[0042] In the senior position model, the weight of strategic decision-making is ≥18%, and the weight of execution efficiency of grassroots positions is ≥20%;

[0043] The weight of safety risk identification indicators in the safety and environmental management sequence shall be ≥8%.

[0044] Furthermore, the industry dictionary library generating unit is specifically configured to perform the following operations:

[0045] Associating power industry-specific capability indicators with policy and regulatory semantics to form a standardized capability tag library;

[0046] The logical dependency between capability indicators is represented by the edge weights of the knowledge graph, among which the correlation strength between cost optimization capability and the compilation of the bill of quantities is ≥0.85.

[0047] Furthermore, the personalized training program generating unit includes predefined triggering rules:

[0048] When the capability gap value is ≥30%, based on the preset "capability-resource" mapping rule base;

[0049] 30%>When the ability gap value is ≥25%, a collaborative filtering algorithm is used to generate a personalized learning list and dynamically allocate resource priorities;

[0050] 25%>When the capability gap value is ≥20%, the safety VR simulation training and one-on-one tutoring tasks will be triggered.

[0051] It can be seen from the above technical solutions that, compared with the prior art, the present invention has the following technical effects:

[0052] (1) Realize the precise and dynamic construction of the talent capability model in the power engineering industry, solve the technical defects of the traditional model such as rigid dimension and low industry adaptability, and provide a standardized data foundation for job matching and talent training.

[0053] (2) Through multimodal data fusion and dynamic weight adjustment, the objectivity of power engineering talent assessment and the accuracy of person-job matching are significantly improved, the matching error rate is reduced, and the efficiency is improved compared with traditional methods.

[0054] (3) The utilization rate of training resources is improved, the period for employees to reach the required capabilities is shortened, and the improvement rate of business indicators is increased within a period of time after training, which is significantly better than the traditional general training model. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0056] Figure 1 This is a block diagram of the intelligent assessment and dynamic training system for power engineering project management talents provided by the present invention.

[0057] Figure 2This is a structural diagram of the multidimensional quality and ability model construction module provided by the present invention.

[0058] Figure 3 This is a structural block diagram of the intelligent evaluation and matching module provided by the present invention.

[0059] Figure 4 This is a structural diagram of the dynamic training and development module provided by the present invention. DETAILED DESCRIPTION

[0060] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0061] Reference Figure 1 As shown, the embodiment of the present invention discloses an intelligent assessment and dynamic training system for power engineering project management talents, including the following core modules: a multi-dimensional quality and ability model construction module, an intelligent assessment and matching module, and a dynamic training and development module.

[0062] A multi-dimensional competency model construction module builds a four-dimensional competency framework based on the characteristics of the power engineering industry, generates differentiated job competency models through a dynamic weight allocation algorithm, and uses knowledge graph technology to build an industry competency dictionary.

[0063] The intelligent assessment and matching module integrates multimodal assessment tools to collect explicit and implicit ability data, dynamically adjusts the weight of ability indicators based on project characteristic parameters through a pre-trained AI model, and uses the cosine similarity algorithm to calculate the person-job fit index;

[0064] The dynamic training and development module generates personalized training plans by quantifying capability gaps, tracks training results in a closed loop based on Kirkpatrick's four-level evaluation model, and uses Bayesian networks to dynamically optimize training paths.

[0065] Among them, the multidimensional quality and ability model construction module builds a multidimensional quality and ability model with hierarchical and professional characteristics based on the characteristics of the power engineering industry and the differentiated needs of positions. Figure 2 As shown, it specifically includes the following technical features:

[0066] Four-Dimensional Competency Framework Construction Unit: A four-dimensional competency framework is constructed, encompassing work experience, knowledge and skills, general competencies, and management / professional competencies. The work experience dimension quantifies candidates' historical project experience, tenure, and performance results to form standardized evaluation indicators. The knowledge and skills dimension covers power engineering professional and technical knowledge (e.g., project management, certification status), industry standards (e.g., GB / T 50932), and tool application capabilities (e.g., BIM modeling). The general competency dimension measures candidates' logical reasoning, stress tolerance, and professional values ​​using a psychological scale. The management / professional competency dimension is further divided into management-level competencies (e.g., team leadership, strategic decision-making) and professional-level competencies (e.g., safety and environmental management, project cost estimation). Based on the unique characteristics of the power industry, secondary specialized dimensions are added, including safety management capabilities and cost optimization capabilities, forming an industry-specific competency dictionary.

[0067] Dynamic differentiation modeling unit: Based on preset job level parameters (senior, middle, and grassroots) and professional sequence parameters (safety and environmental management, engineering economics, and technology research and development), a differentiated capability model is generated through a dynamic weight allocation algorithm. Specifically, the core capability weights are adjusted according to the job level parameters: in the senior job model, the weight of the strategic decision-making dimension is set to 20%, and the weight of the risk prevention dimension is 18%; in the middle-level job model, the weight of the team collaboration dimension is set to 15%, and the weight of the resource coordination dimension is 12%; in the grassroots job model, the weight of the execution efficiency dimension is set to 25%, and the weight of the technical specification compliance dimension is set to 20%;

[0068] Expand capability indicators based on professional sequence parameters: In the safety and environmental management sequence model, add "safety risk identification" and "emergency plan formulation" indicators, with weights of 10% and 8% respectively;

[0069] In the engineering economic sequence model, the indicators of “cost sensitivity analysis” and “compilation of bill of quantities” are added, with weights of 12% and 9% respectively.

[0070] Industry dictionary generation unit: semantically associates the capability indicators unique to the power engineering industry (such as "policy interpretation capability" and "organizational mobilization capability" in management) with general capability indicators to form a standardized capability label library, and constructs a logical relationship network between capability indicators through knowledge graph technology to support adaptive iteration and cross-scenario calling of the model.

[0071] Reference Figure 3 As shown, the intelligent assessment and matching module: Based on multimodal data fusion and dynamic weight optimization algorithm, it realizes intelligent assessment and job matching of power engineering project management talents, and specifically includes the following technical features:

[0072] Multimodal data acquisition unit: Integrates the following assessment tools and quantifies the output of explicit and implicit capability indicators:

[0073] 1. CBI (Context-Based Interview) interview tool: Through pre-set power engineering scenarios (such as "emergency response to sudden power outages"), the tool collects candidate behavioral event descriptions (STAR ​​model) and uses natural language processing (NLP) technology to analyze their implicit ability indicators, including risk prediction ability, decision-making logic, and compliance awareness, and generates a standardized score from 0 to 1.

[0074] 2. Leaderless Group Discussion Tool: Based on typical power engineering tasks (e.g., "Coordinating Schedule and Cost Conflicts"), this tool uses voice sentiment analysis techniques (e.g., MFCC feature extraction) to identify candidates' emotional stability. It also uses semantic analysis techniques (e.g., LDA topic modeling) to assess their communication and coordination skills and team influence, generating a multi-dimensional competency matrix.

[0075] 3. OPT professional personality test tool: Uses psychological scales (such as the Big Five personality model) to measure candidates' sense of responsibility, openness, and stress tolerance, and uses clustering algorithms (such as K-means) to divide personality type labels and associate them with job competency requirements.

[0076] AI dynamic weight adjustment unit: Based on the input of characteristic parameters of power engineering projects (including project scale, environmental complexity, and technical difficulty), the weights of capability indicators are dynamically adjusted through a pre-trained random forest model. The specific implementation method is as follows: Constructing a project feature vector: normalizing the project parameters into a feature vector. For example, the project scale is divided into three categories according to the contract amount: small (<5 million yuan), medium (5-20 million yuan), and large (>20 million yuan) and encoded as a one-hot vector; dynamic weight mapping: outputting weight adjustment coefficients through AI models (such as gradient boosting decision trees). For example, the weight of the "risk control" indicator in large projects is increased to 22% (the benchmark weight is 15%), and the weight of the "cross-departmental collaboration" indicator in a high-complexity environment is increased to 18%; weight constraint conditions: setting the total weight to 100%, and the fluctuation range of a single weight does not exceed ±30% of the benchmark value to ensure model stability.

[0077] Personnel-job matching prediction unit: uses the cosine similarity algorithm to calculate the ability match between the candidate and the target position. The specific steps include:

[0078] Feature vector generation: The candidate assessment results and job competency model are mapped into n-dimensional feature vectors, where each dimension is the product of the competency index score and the weight.

[0079] Similarity calculation: by formula

[0080]

[0081] Among them, C i The score of the candidate's i-th ability, P i is the weight of the i-th ability of the position;

[0082] Result output: Generates a person-job fit index (0-100 points), and divides the recommendation level according to the threshold (such as ≥85 points is "strong recommendation", 70-84 points is "recommended"), and outputs a visual radar chart to show the ability gap.

[0083] Reference Figure 4 As shown, the dynamic training and development module: Based on the closed loop of capability gap analysis and training effect feedback, it realizes the personalized training path planning and dynamic optimization of power engineering talents. Specifically, it includes the following technical features:

[0084] Personalized training program generation unit, capability gap quantitative analysis, compare the candidate output by the intelligent assessment module with the target position capability model, and use the formula

[0085]

[0086] Calculate the gap value of the i-th capability index, where S i Score the candidate, R i thresholds for job requirements;

[0087] Intelligent matching of learning resources: Based on a pre-set "capability-resource" mapping rule base (e.g., "Cost awareness gap ≥ 30% → push cost reduction and efficiency improvement case library"), a collaborative filtering algorithm is used to generate personalized learning checklists and dynamically prioritize resources. For example, when the "Safety Standards Application" gap is ≥ 20%, a "Safety Standards VR Simulation Training" and one-on-one tutoring are triggered.

[0088] The Koch four-level assessment tracking unit builds a multi-level assessment and feedback mechanism for training effectiveness, including:

[0089] 1) Response layer data collection: using online questionnaires to collect real-time student satisfaction ratings (1-5 points) on the training content, instructors, and format. Sentiment analysis models (such as BERT) are used to parse negative keywords (such as "abstract content" and "outdated cases") in open text reviews.

[0090] 2) Learning effectiveness verification: A knowledge test is conducted within 24 hours after the training. The difficulty of the test questions is matched to the level of ability gap (e.g., those with a high ability gap are given Paper A, which focuses on basic knowledge; those with a low ability gap are given Paper B, which focuses on comprehensive application). The degree of knowledge absorption is determined by the pass rate (≥80% is considered a pass);

[0091] 3) Behavioral tracking analysis: Within 3-6 months after training, based on the behavioral key points of the job competency model (e.g., "cost awareness → frequency of proactively proposing cost reduction plans"), multi-source evaluation data from superiors, colleagues, and subordinates was collected using a 360-degree assessment tool. T-tests were used to compare behavioral scores before and after training (significance level p < 0.05);

[0092] 4) Quantification of benefits at the result level: Linking to the business database, extracting key performance indicators (such as project cost deviation rate and safety accident rate) 6-12 months after training, and calculating the correlation coefficient (R) between training input and business results through regression model. 2 ≥0.7 is considered effective).

[0093] Dynamic optimization unit of training path: Based on the results of Kirkpatrick's evaluation, the Bayesian network model is used to update the training plan. Specifically, if there is no significant improvement in the behavior layer (p ≥ 0.05), "secondary reinforcement learning" is triggered to increase the weight of actual combat exercises (such as increasing the proportion of simulation projects to 60%); if the benefit of the result layer does not meet the standard (R 2 <0.7), the “capability-resource” mapping rules are recalibrated and an external expert intervention mechanism is introduced.

[0094] A preferred embodiment of the present invention takes "selection and training of project managers for the construction of a large-scale new energy base with a capacity of tens of millions of kilowatts" as an application scenario, and the system operation process is as follows:

[0095] (1) Construction of multidimensional quality and ability model:

[0096] During the job requirements analysis phase, based on the characteristics of the 10 million kilowatt-level new energy base construction project, the core competency requirements for the project manager position were extracted, including:

[0097] Project scale characteristics: Multi-energy complementary projects with a single unit capacity of ≥1GW (covering wind power / photovoltaic / energy storage systems);

[0098] Environmental constraint characteristics: involving ecological protection red line areas and desertification control areas;

[0099] Policy and regulatory characteristics: Must meet the national "dual carbon" strategic goals and renewable energy quota system requirements;

[0100] The capability model construction includes the following steps:

[0101] Step 1: Define the four-dimensional capability framework:

[0102] Quantitative indicators of work experience include: historical management project scale classification (GW-level project experience coefficient 0.8), proportion of new energy projects (≥70% is considered full score);

[0103] The knowledge and skills dimension adds power industry-specific indicators such as "multi-energy complementary dispatching technology" and "grid-type energy storage system specifications";

[0104] The management competency dimension has added secondary characteristic indicators of "wind, solar and storage coordination capability" and "ecological risk management capability", and is linked to technical specifications such as "GB / T 36547-2018 Technical Regulations for Electrochemical Energy Storage System Connection to the Grid".

[0105] Step 2: Dynamic weight configuration:

[0106] Based on the contract amount (RMB 5.2 billion) and the environmental complexity level (Level A) obtained by the project parameter input module, a dynamic weight adjustment plan is output through the pre-trained gradient boosting decision tree model:

[0107] - The weight of risk control indicators increased from the baseline value of 15% to 24%;

[0108] - The weight of the cross-departmental collaboration indicator was increased from the baseline value of 10% to 18%;

[0109] - Added the indicator "Wind, Solar and Storage Joint Dispatch Capability" with an initial weight of 12%;

[0110] Step 3: Generate industry dictionary library:

[0111] The knowledge graph technology is used to construct the capability indicator association network, where:

[0112] - The "Policy Interpretation Capability" node links to policy documents such as the "Implementation Plan for the Responsibility Weighting of Renewable Energy Power Consumption" and the "Detailed Rules for the Implementation of Green Power Trading";

[0113] - The “Ecological Risk Management” node is associated with practical knowledge bases such as the Environmental Impact Assessment Report Preparation Procedures and the Biodiversity Compensation Program Design Guidelines.

[0114] (2) Intelligent assessment and job matching:

[0115] Multimodal evaluation implementation phase:

[0116] a) Assessment data collection:

[0117] Using a contextual behavioral interview (CBI) tool, we set up a typical case scenario, "Wind, Solar, and Storage Output Fluctuations Leading to Exceeding Grid-Connected Power Limits." We used natural language processing technology to analyze the candidate's responses, extracting keywords such as "AGC Command Response Delay Control" and "Energy Storage SOC Balancing Strategy," and generated a technical decision-making ability scoring matrix.

[0118] Through leaderless group discussions, a scenario of "construction delays and ecological protection conflicts" was simulated. MFCC speech feature extraction technology was used to monitor emotional stability indicators, and semantic analysis models were used to identify innovative strategy keywords such as "segmented construction plan" and "vegetation transplantation technology."

[0119] The Big Five Personality Inventory was used to assess professional personality and generate quantitative scores for dimensions such as conscientiousness (94% percentile) and openness (88% percentile).

[0120] b) Dynamic matching calculation:

[0121] Construct the job capability model vector V_p = [(strategic decision-making, 20%, 85), (wind, solar and energy storage coordination, 15%, 90), (risk management, 24%, 80)...];

[0122] Generate candidate capability vector V_c = [(strategic decision-making, 20%, 82), (wind, solar, and storage coordination, 15%, 88), (risk management, 24%, 76)...];

[0123] Calculate the matching degree using the cosine similarity algorithm:

[0124] Similarity=Σ(V_c[i]×V_p[i]) / (||V_c||×||V_p||)=0.89;

[0125] V_c[i] is the candidate's i-th ability score, V_p[i] is the i-th ability weight of the position; V_c is the candidate's total ability, and V_p is the total ability weight of the position.

[0126] The output job adaptation index is 89 points (strong recommendation level), and the radar chart visualization shows that there is an 18% capability gap in the "ecological risk management" dimension.

[0127] (3) Dynamic training and effect verification:

[0128] Training program implementation phase:

[0129] a) Generation of personalized training paths:

[0130] For the ecological risk management capability gap value Δ=18%, the training resource matching rule is automatically triggered:

[0131] When Δ≥15%, the courses on “Ecological Restoration VR Simulation Training System” and “Technical Guidelines for Environmental Impact Assessment of Construction Projects” will be linked;

[0132] For the wind, solar and storage coordination capability gap value Δ=12%, the collaborative filtering algorithm is called to prioritize the push of the "Multi-energy Complementary Optimization Algorithm Practical Course" with a historical completion rate of 92%.

[0133] b) Keshi four-level evaluation mechanism:

[0134] Response layer: Collect students' satisfaction rating of VR training equipment (4.7 / 5.0), use the BERT model to analyze open text reviews, and identify the frequency of negative keywords such as "device delay" (less than 3%);

[0135] Learning level: A test paper A was administered within 24 hours of the training. The accuracy rate of the knowledge point "Eco-compensation Mechanism" increased from 55% to 91%, and the overall pass rate reached 88%;

[0136] Behavioral level: 360-degree assessment tools were used to quantify behavioral improvement data. Within three months after training, the frequency of proposing ecological solutions increased from 0.2 times / month to 1.5 times / month (p=0.032<0.05).

[0137] Result layer: Statistics from the associated business database show that the incidence of ecological violations in the project decreased by 67% after the training, and it was certified as a national green construction demonstration project.

[0138] c) Dynamic optimization mechanism:

[0139] When the Bayesian network detects that the improvement rate of “wind-solar-storage coordination capability” has not reached the target threshold (actual 12% vs. target 25%), the implementation plan is adjusted:

[0140] Increase the proportion of class hours for "virtual power plant scheduling sandbox simulation" from 40% to 60%;

[0141] Trigger the external expert intervention mechanism and introduce the special workshop "Technical Specifications for Collaborative Control of Grid-Based Energy Storage Systems".

[0142] The intelligent assessment and dynamic training system for power engineering project management talents provided by the present invention achieves improved assessment accuracy and reduced matching error rate; moreover, training efficiency is optimized, training resource utilization is improved, and the cycle for reaching capability standards is shortened; in addition, business indicators are also improved, and the project cost deviation rate within a period of months after training is reduced, further reducing the safety accident rate.

[0143] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0144] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An intelligent assessment and dynamic training system for power engineering project management talents, characterized by: include: A multi-dimensional competency model construction module builds a four-dimensional competency framework based on the characteristics of the power engineering industry, generates differentiated job competency models through a dynamic weight allocation algorithm, and uses knowledge graph technology to build an industry competency dictionary. The intelligent assessment and matching module integrates multimodal assessment tools to collect explicit and implicit ability data, dynamically adjusts the weight of ability indicators based on project characteristic parameters through a pre-trained AI model, and uses the cosine similarity algorithm to calculate the person-job fit index; The dynamic training and development module generates personalized training plans by quantifying capability gaps, tracks training results in a closed loop based on Kirkpatrick's four-level evaluation model, and uses Bayesian networks to dynamically optimize training paths.

2. The intelligent assessment and dynamic training system for power engineering project management talents according to claim 1 is characterized in that: The multidimensional quality and ability model construction module includes: The four-dimensional capability framework unit is used to construct work experience, knowledge and skills, general qualities, and management / professional competence dimensions. The professional competence dimension includes secondary indicators such as safety control and cost optimization. Dynamic differentiation modeling unit, used to achieve dynamic weight allocation of job levels and professional sequences through gradient boosting decision tree algorithm; The industry dictionary generation unit uses knowledge graph technology to construct a logical relationship network between capability indicators and generate an industry capability dictionary.

3. The intelligent assessment and dynamic training system for power engineering project management personnel according to claim 1 is characterized in that: The intelligent evaluation and matching module includes: Multimodal data collection unit, integrating CBI structured interview tools, voice emotion analysis devices and psychological assessment systems to collect multimodal data of talents; AI dynamic weight adjustment unit, which maps project feature vectors to capability weight coefficients based on the random forest model; The person-job matching prediction unit uses the cosine similarity algorithm to calculate the matching degree of the feature vectors between the candidate and the target job.

4. The intelligent assessment and dynamic training system for power engineering project management personnel according to claim 3 is characterized in that: The multimodal assessment tool includes: CBI interview tool: Analyze implicit ability indicators through power engineering scenario cases and NLP technology; Leaderless group discussion tool: Evaluate emotional stability and team influence through voice sentiment analysis and LDA topic model; OPT career personality test tool: uses a psychological scale to generate candidate personality type labels.

5. The intelligent assessment and dynamic training system for power engineering project management personnel according to claim 3 is characterized in that: The person-job matching prediction unit is specifically used to: Feature vector generation: The candidate assessment results and job competency model are mapped into n-dimensional feature vectors, where each dimension is the product of the competency index score and the weight. Similarity calculation: by formula Among them, C i The score of the candidate's i-th ability, P i is the weight of the i-th ability of the position; Result output: Generates a person-job adaptation index, divides the recommended level according to the threshold, and outputs a visual radar chart to show the ability gap.

6. The intelligent assessment and dynamic training system for power engineering project management personnel according to claim 1 is characterized in that: The dynamic training and development modules include: The personalized training program generation unit quantitatively analyzes the ability gap and compares the candidate output by the intelligent assessment and matching module with the target position ability model. Calculate the gap value of the i-th capability index, where S i Score the candidate, R i thresholds for job requirements; The Koch four-level assessment tracking unit builds a multi-level assessment and feedback mechanism for training effectiveness and outputs assessment results; The training path dynamic optimization unit uses the Bayesian network model to dynamically adjust the weight of training resources based on the evaluation results of the Kirkpatrick four-level evaluation tracking unit.

7. The intelligent assessment and dynamic training system for power engineering project management personnel according to claim 6 is characterized in that: The Koch four-level assessment tracking unit includes: The response layer integrates the BERT sentiment analysis model to parse negative keywords in open text reviews; At the learning level, differentiated knowledge tests are conducted after training to verify the learning effect; The behavioral layer integrates the T-test behavioral analysis module to compare and analyze the differences in behavioral scores before and after training; At the result layer, the correlation coefficient between training input and business indicators is calculated through regression model.

8. The intelligent assessment and dynamic training system for power engineering project management personnel according to claim 2 is characterized in that: The weight distribution of the dynamic differentiation modeling unit meets the following constraints: The sum of the weights of each dimension is 100%; The fluctuation range of individual weights shall not exceed ±30% of the benchmark value; In the senior position model, the weight of strategic decision-making is ≥18%, and the weight of execution efficiency of grassroots positions is ≥20%; The weight of safety risk identification indicators in the safety and environmental management sequence shall be ≥8%.

9. The intelligent assessment and dynamic training system for power engineering project management personnel according to claim 2 is characterized in that: The industry dictionary library generating unit is specifically used to perform the following operations: Associating power industry-specific capability indicators with policy and regulatory semantics to form a standardized capability tag library; The logical dependency between capability indicators is represented by the edge weights of the knowledge graph, among which the correlation strength between cost optimization capability and the compilation of the bill of quantities is ≥0.

85.

10. The intelligent assessment and dynamic training system for power engineering project management personnel according to claim 2, characterized in that: The personalized training program generation unit includes predefined triggering rules: When the capability gap value is ≥30%, based on the preset "capability-resource" mapping rule base; 30%>When the ability gap value is ≥25%, a collaborative filtering algorithm is used to generate a personalized learning list and dynamically allocate resource priorities; 25%>When the capability gap value is ≥20%, the safety VR simulation training and one-on-one tutoring tasks will be triggered.

Images