Software quality evaluation method and device for power grid dispatching automation system

Abstract

The invention provides a software quality evaluation method and device for a power grid dispatching automation system. Belongs to the technical field of power dispatching automation. The method comprises the following steps: acquiring running state data of power grid dispatching automation system software, and extracting a third-level index from the running state data; determining subjective and objective fusion weights of the third-level indexes; aggregating the third-level indexes related to the same quality feature according to the subjective and objective fusion weights of the third-level indexes to obtain corresponding second-level indexes; an index vector composed of the second-level indexes is input into the dynamic fuzzy neural network model, and the first-level indexes of the corresponding subsystems are output to serve as comprehensive quality evaluation scores; and determining the software quality grade of the business subsystem according to the comprehensive quality evaluation score and a preset grade threshold value. The method is suitable for power grid dispatching automation system software quality evaluation oriented to complex operation scenes, and comprehensive evaluation and self-adaptive optimization of dispatching software under multi-dimensional and dynamic conditions can be achieved.

Description

Technical Field

[0001] This invention belongs to the field of power dispatch automation technology, and in particular relates to a method and apparatus for evaluating the software quality of a power grid dispatch automation system. Background Technology

[0002] Against the backdrop of dual-carbon goals and the construction of a new power system, the power grid is evolving from centralized dispatch to a new operational paradigm of multi-source coordination and interaction between power generation, grid, load, and storage. Nationwide, new energy installed capacity has surpassed thermal power to become the main power source. The integration of new control entities such as large-scale distributed photovoltaics, centralized wind power, and virtual power plants has exacerbated system power fluctuations, significantly increased the number of control objects and constraints, and shortened action cycles. Dispatch automation systems face multi-objective, multi-path, and multi-frequency nonlinear control requirements, significantly increasing pressure on software quality and control reliability.

[0003] As a core information system supporting the safe and stable operation of the power grid, the dispatch automation system encompasses functions such as a unified platform, real-time monitoring, state estimation, automatic generation control, load forecasting, power flow calculation, and safety verification. It runs through measurement and acquisition, communication transmission, strategy determination, and terminal execution, exhibiting typical cyber-physical integration characteristics. With increased complexity in the operating environment, system operation highly depends on precise coordination between components. Under the influence of disturbances such as equipment state drift, communication anomalies, strategy conflicts, or response delays, the overall control behavior exhibits strong coupling, high dynamism, and high uncertainty. Traditional static testing and single-point evaluation methods are insufficient to accurately characterize software quality features and control robustness.

[0004] Currently, software quality evaluation mainly includes three types of methods: First, quality characteristic evaluation based on international standard models such as ISO / IEC 25010, which divides quality into dimensions such as functionality, reliability, and efficiency, but lacks specific definitions for scheduling operations and is difficult to cover indicators such as control chain real-time performance, policy coordination, and cross-system compatibility; Second, subjective evaluation based on expert experience and scoring, which relies on human judgment and is highly subjective, making it difficult to reflect the dynamic performance of the system under actual disturbance conditions; Third, multi-indicator fusion methods based on traditional fuzzy comprehensive evaluation, which can handle certain fuzziness and uncertainty, but are prone to information loss and failure of maximum membership determination, and are difficult to characterize the coupling relationship between indicators.

[0005] Existing research and engineering practices still have three shortcomings: First, there is a lack of hierarchical quality models oriented towards the characteristics of scheduling operations, which fail to fully reflect industry characteristics such as real-time performance, robustness, and control coordination, and are difficult to reflect the complexity of multi-source data interaction and dynamic decision-making. Second, there is a lack of dynamic evaluation mechanisms that combine fault injection with operational scenario simulation. Most methods remain at the level of static analysis or functional testing, and cannot characterize the response patterns and risk resistance capabilities under complex disturbances, policy linkages, and extreme scenarios. Third, the evaluation results are disconnected from the actual operating performance of the system. A closed-loop quality management process covering indicator collection, weight calculation, and result interpretation has not yet been formed, which cannot provide strong support for multi-dimensional quality verification and control before the scheduling software goes live. Summary of the Invention

[0006] The purpose of this invention is to provide a method and apparatus for evaluating the software quality of a power grid dispatch automation system. This method and apparatus are applicable to evaluating the software quality of power grid dispatch automation systems facing complex operating scenarios and can achieve comprehensive evaluation and adaptive optimization of dispatch software under multi-dimensional and dynamic conditions.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, this invention provides a method for evaluating the software quality of a power grid dispatch automation system. For each business subsystem of the software, a first-level indicator is preset to quantify the overall quality of the subsystem. The first-level indicator is determined by several second-level indicators, each used to quantify multiple quality characteristics of the subsystem. The second-level indicators are determined by several third-level indicators used to quantify the operational status of the business subsystem. The method includes: Obtain the operating status data of the power grid dispatch automation system software, and extract the third-level indicators from the operating status data; The third-level indicators are standardized, and the subjective and objective weights of each third-level indicator are determined. Based on the subjective and objective fusion weights of each third-level indicator, the third-level indicators related to the same quality characteristic are aggregated to obtain the corresponding second-level indicators; The index vector composed of each second-level index is input into a pre-trained dynamic fuzzy neural network model to output the first-level index of the corresponding subsystem as the comprehensive quality evaluation score. The software quality level of the business subsystem is determined based on the comprehensive quality evaluation score and the preset level threshold, and the software quality evaluation result, including the comprehensive quality evaluation score and the quality level, is output.

[0008] Furthermore, the software quality evaluation method for power grid dispatch automation systems provided by this invention also includes: An anomaly pattern set is constructed based on historical fault records, operation logs, and simulation results:

[0009] Each abnormal pattern Represented by the eigenvectors:

[0010] in, Fault type For the affected business subsystems, For the duration of the abnormality, To affect the strength coefficient; At least one abnormal pattern is selected from the set of abnormal patterns to construct a disturbance scenario. Under this scenario, a disturbance is applied to the operation of the power grid dispatch automation system software, causing the disturbed software inputs or critical operational variables to... Satisfy the following formula:

[0011] in, This refers to the software input or critical runtime variables before the disturbance. These are the perturbation amplitude coefficients associated with the anomalous mode. The disturbance signal function; Obtain the comprehensive quality evaluation score sequence corresponding to multiple disturbance scenarios and normal operation scenarios, calculate the quality stability index representing the quality fluctuation degree of the business subsystem under different operation scenarios based on the score sequence, and output the quality stability index as part of the software quality evaluation result.

[0012] Furthermore, the standardization process for the third-level indicators includes: Based on the business implications of the third-level indicators, the third-level indicators are divided into positive indicators and negative indicators; The standardized value of each third-level indicator is calculated using the following formula. :

[0013] The original values ​​for each third-level indicator are as follows: , and These are the maximum and minimum values ​​of the third-level indicator in the sample set, respectively. For positive indicators, the values ​​of each business subsystem under the indicator are mapped to the range of 0 to 1 according to a preset rule. For negative indicators, the values ​​of each business subsystem under the indicator are mapped to the range of 0 to 1 according to a preset rule, so that all third-level indicators after standardization have a uniform value range and maintain monotonicity. The standardized values ​​of each business subsystem on all third-level indicators are arranged according to the business subsystem dimension and the indicator dimension to construct a standardized indicator matrix. :

[0014] in, For the number of business subsystems, This refers to the number of third-level indicators.

[0015] Furthermore, the weights of each third-level indicator are determined based on a combination of subjective and objective factors, including: Calculate the entropy value of the third-level index using the following formula. :

[0016] No. The third-level indicator in the The proportion of each business subsystem, k=1 / ln n , Number of business subsystems; The objective weight is determined from the entropy value using the following formula. :

[0017] The number of third-level indicators; The Analytic Hierarchy Process (AHP) is used to construct pairwise comparison matrices for the third-level indicators, and the subjective weight vectors are obtained using the eigenvalue method. Make it satisfy the following formula:

[0018] in, This is a pairwise comparison matrix. for The largest eigenvalue, For corresponding The eigenvectors of , their components For the first Subjective weights of each third-level indicator; The subjective and objective fusion weights of the third-level indicators are calculated using the following formula:

[0019] in, For the first The subjective and objective weighting of the third-level indicators ∈[0,1] represents the preset fusion coefficient.

[0020] Furthermore, the pre-trained dynamic fuzzy neural network models include: The input layer is used to receive an indicator vector consisting of multiple second-level indicators; A membership function layer is used to map each second-level indicator to a membership degree of several linguistic variables, wherein each second-level indicator corresponds to at least one adjustable membership function. The fuzzy rule layer is used to calculate the activation degree of fuzzy rules based on the output combination of each membership function. The fuzzy rule defines the correlation between the second-level index and the first-level index. A normalization layer is used to normalize the activation level of each fuzzy rule in order to obtain the normalized weight of each fuzzy rule. And a weighted summation of the outputs of each fuzzy rule based on the normalized weights to obtain the first-level index as the output layer for the comprehensive quality evaluation score.

[0021] Furthermore, the training steps for the dynamic fuzzy neural network model include: A training sample set is constructed using historical operation data, simulation operation data, or manual test data. The second-level index vector corresponding to each training sample is used as the model input, and the target quality evaluation result corresponding to the training sample is used as the expected output of the model. The parameters of the membership function layer and the rule parameters of the fuzzy rule layer are iteratively updated based on the training sample set, so that the error between the first-level index output by the model and the expected output is gradually reduced. The model parameters are updated according to the following formula:

[0022] in, For a moment The model parameter vector, For learning rate, The difference between the actual output and the expected output. This refers to the prediction error or correction amount. Once the error meets the preset convergence condition or reaches the preset training rounds, the membership function and fuzzy rules obtained from the training are fixed as model parameters for use in the running phase.

[0023] Furthermore, the software quality level of the business subsystem is determined based on the comprehensive quality evaluation score and the preset level threshold, including: The overall quality evaluation score is pre-divided into multiple consecutive grade intervals, with each grade interval corresponding to a software quality grade; After obtaining the comprehensive quality evaluation score of the business subsystem, determine the level range in which the score falls, and use the corresponding quality level as the software quality level of the business subsystem. The number and boundaries of the grade intervals are configured according to the software quality management requirements of the power grid dispatch automation system to support the hierarchical management and comparative analysis of the software quality of different business subsystems.

[0024] In a second aspect, the present invention provides a software quality evaluation device for a power grid dispatch automation system. For each business subsystem of the software, a first-level indicator for quantifying the overall quality of the subsystem is preset. The first-level indicator is determined by several second-level indicators for quantifying multiple quality characteristics of the subsystem. The second-level indicators are determined by several third-level indicators for quantifying the operating status of the business subsystem. The device includes: The data acquisition module is used to acquire the operating status data of the power grid dispatch automation system software and extract third-level indicators from the operating status data. The third-level indicator processing module is used to standardize the third-level indicators and determine the subjective and objective fusion weights of each third-level indicator. The second-level indicator processing module is used to aggregate third-level indicators related to the same quality characteristic according to the subjective and objective fusion weight of each third-level indicator to obtain the corresponding second-level indicator. The first-level indicator processing module is used to input the indicator vector composed of each second-level indicator into the pre-trained dynamic fuzzy neural network model, so as to output the first-level indicator of the corresponding subsystem as the comprehensive quality evaluation score. The evaluation module is used to determine the software quality level of the business subsystem based on the comprehensive quality evaluation score and the preset level threshold, and output the software quality evaluation results including the comprehensive quality evaluation score and the quality level.

[0025] In a third aspect, the present invention provides an electronic device including a processor and a memory, wherein the processor is used to execute a computer program stored in the memory to implement a method for evaluating the software quality of a power grid dispatch automation system.

[0026] In a fourth aspect, the present invention provides a computer-readable storage medium storing at least one instruction, wherein the at least one instruction, when executed by a processor, implements a method for evaluating the software quality of a power grid dispatch automation system.

[0027] Compared with the prior art, the beneficial effects of the present invention are as follows: The software quality evaluation device, electronic device, and computer-readable storage medium for power grid dispatch automation system provided by this invention also solve the problems raised in the background section.

[0028] 1. This invention constructs a hierarchical quality structure consisting of third-level indicators, second-level quality characteristic indicators, and first-level comprehensive quality indicators. This structure enables the software quality evaluation process of the scheduling automation system to form a quantifiable and traceable logical chain, from underlying operational behavior and quality characteristic dimensions to the overall quality of subsystems. The third-level indicators are directly derived from real-world operational data, ensuring the objectivity of the evaluation basis. The second-level indicators are weighted and aggregated based on a fusion of subjective and objective factors, ensuring that the evaluation results consider both data characteristics and business experience. The first-level indicators are generated by a dynamic fuzzy neural network, which can handle non-linear relationships between quality characteristics, making the evaluation more closely reflect the software's performance in real-world scenarios. This invention not only unifies the evaluation criteria across different business subsystems, solving the problems of fragmentation and inconsistent dimensions in traditional indicator systems, but also improves the sensitivity and discriminative power of comprehensive evaluation through fuzzy inference mechanisms. This allows the system to achieve more stable and reliable quality judgments under version upgrades, changes in operating environments, or differences in functional configurations. The final comprehensive quality score and quality level can serve as the basis for software acceptance, version comparison, and operational health analysis.

[0029] 2. This invention, by constructing an anomaly pattern set and a disturbance injection mechanism, enables the quantifiable recording of the behavior of dispatch automation software under both typical and extreme operating conditions, thus overcoming the limitation of traditional testing that can only evaluate normal operation quality. Anomaly patterns are described in the form of feature vectors, representing fault type, affected modules, duration, and intensity, facilitating systematic management and expansion. By applying disturbances to software inputs or critical operational quantities in a simulation environment, real-world operational risks such as link delays, data fluctuations, and module anomalies can be simulated and repeatedly verified without affecting the actual power grid. The comprehensive quality score sequence obtained under disturbance scenarios is used to calculate stability indicators, quantifying the degree of quality fluctuation in multiple scenarios. This allows the evaluation to not only answer whether the quality is good but also whether the quality is stable and reliable when encountering disturbances. This mechanism enhances the robustness verification capability of dispatch automation software under complex power grid operation backgrounds, providing crucial support for software launch acceptance, critical link risk assessment, and operational strategy optimization.

[0030] 3. This invention employs a dynamic fuzzy neural network model consisting of an input layer, a membership function layer, a fuzzy rule layer, a normalization layer, and an output layer. This enables the quality evaluation process to simultaneously possess the interpretability of fuzzy logic and the learning ability of a neural network. The membership function layer maps quality feature indicators to linguistic variables, suitable for handling uncertainty; the fuzzy rule layer solidifies expert experience into computable rules, making the evaluation logic transparent and interpretable; the output layer generates a comprehensive quality score through weighted calculation, enabling the model to have end-to-end reasoning capabilities. During the training phase, the model constructs samples using historical operating data, simulation data, or manual test data, and iteratively adjusts the membership function and rule parameters through parameter update formulas, allowing the model to learn the relationship between real operating behavior and quality features, gradually converging to a reasonable evaluation mode. After training convergence, the parameters are frozen for rapid reasoning during the actual operation phase. This model structure and training method enable the evaluation system to adapt to different software versions and operating conditions while maintaining long-term optimizability, improving the accuracy and robustness of the comprehensive quality evaluation. Attached Figure Description

[0031] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart of the software quality evaluation method for a power grid dispatch automation system according to an embodiment of the present invention; Figure 2 This is a technical framework diagram of the software quality evaluation method for power grid dispatch automation system according to an embodiment of the present invention; Figure 3 This is a schematic diagram of a software quality evaluation device for a power grid dispatch automation system according to an embodiment of the present invention; Figure 4 This is a structural block diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0032] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0033] The following detailed description is exemplary and intended to provide further detailed explanation of the invention. Unless otherwise specified, all technical terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terminology used in this invention is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention.

[0034] Example 1 The following embodiments are described based on the actual scenario of the operation and simulation environment of a power grid dispatch automation system, illustrating the specific implementation process of a software quality evaluation method for a power grid dispatch automation system in engineering. During deployment, the system completes the training and parameter calibration of the quality evaluation model. During operation, it executes the quality evaluation task according to the physical structure of data input layer—analysis and calculation layer—quality evaluation layer. Through multi-source data acquisition, index calculation, fault injection and scenario fusion, and the inference output of the intelligent evaluation model, it achieves comprehensive quality evaluation and risk identification of the dispatch automation software under normal operation and abnormal disturbance scenarios.

[0035] like Figure 1 and Figure 2 As shown, the software quality evaluation method for power grid dispatch automation system provided in this embodiment of the invention includes steps S1 to S5.

[0036] S1. Obtain the operating status data of the power grid dispatch automation system software, and extract the third-level indicators from the operating status data.

[0037] In this embodiment, each business subsystem of the software is pre-defined with a first-level indicator for quantifying the overall quality of the business subsystem. The first-level indicator is determined by several second-level indicators, which are used to quantify multiple quality characteristics of the business subsystem. The second-level indicators are determined by several third-level indicators, which are used to quantify the operating status of the business subsystem. The third-level indicators can also be called sub-characteristic indicators, which are used to reflect the quantifiable attributes of the scheduling automation software's operating status, performance, and stability.

[0038] During the operational phase, the system first accesses multi-source operational data from the dispatch automation system's main station platform and related control software at the data input layer. This data includes model data, operational logs, interface interaction information, alarm records, control command processing records, and system resource usage information from the dispatch master station and its subordinate business systems. Simultaneously, operational data from the simulation verification environment is also accessed. This operational data encompasses real-time measurements, dispatch control commands, command response timing, power grid model parameters, communication link status, process running status, CPU and memory usage, network latency, and task execution records. The data input layer performs preprocessing on the raw data, including outlier removal, missing value imputation, timestamp alignment, and format standardization. Based on a sliding time window mechanism, operational samples are constructed to form an input dataset suitable for quality evaluation. This allows for a structured representation of the functional status, performance, and operational behavior of each business subsystem at different time periods, providing a unified data foundation for subsequent indicator calculations and quality evaluation.

[0039] After the data preprocessing is completed, the analysis and calculation layer extracts third-level indicators from the collected data according to a pre-established quality indicator system. Third-level indicators are quantifiable attributes that reflect the operational status, performance, and stability of the scheduling automation software, including but not limited to task response time, instruction execution accuracy, link availability, data synchronization latency, alarm handling success rate, and process stability. Each business subsystem corresponds to a set of third-level indicators, which belong to five quality characteristic dimensions: functional quality characteristics, reliability quality characteristics, efficiency quality characteristics, security quality characteristics, and standardization quality characteristics. Through the hierarchical definition of the indicator system, a software quality evaluation structure is formed: target layer—characteristic layer—third-level indicator layer. The target layer represents the overall quality of the scheduling automation business subsystems divided by business; the characteristic layer comprises the five categories of quality characteristic indicators mentioned above; and the third-level indicator layer includes quantifiable software quality elements such as functional correctness, fault tolerance, response latency, data consistency, and log completeness.

[0040] In a further embodiment, to evaluate the software's performance in typical and extreme scenarios, a disturbance scenario and quality stability evaluation can be introduced based on S1. Specifically, an abnormal mode set is constructed based on historical fault records, operation logs, and simulation results:

[0041] Each abnormal pattern Represented by the eigenvectors:

[0042] in, Fault type For the affected business subsystems, For the duration of the abnormality, To influence the intensity coefficient, the system abstracts the scheduling service link into a complex network model composed of nodes and edges. In this system, nodes represent key functional modules or operating units in the scheduling system, and edges represent data dependencies or control links between modules. Node importance is calculated using methods such as the Laplace matrix, and nodes are sorted according to their importance, with nodes of higher importance being prioritized for injection.

[0043] Based on the set of abnormal patterns, at least one abnormal pattern is selected to construct a disturbance scenario. Under this disturbance scenario, disturbances are applied to the input quantities of the power grid dispatch automation system software or the key operational quantities characterizing the operating status of the business subsystem, so that the corresponding operational state quantities satisfy:

[0044] in, The software inputs or critical runtime variables before the disturbance are at time [time]. The value of , The value obtained after applying the perturbation. The disturbance amplitude coefficients associated with the aforementioned anomalous mode. The disturbance signal function is defined as follows. The disturbance injection process is executed in an independent simulation verification environment. This simulation environment replicates the model, links, module interaction logic, and data structure of the dispatch master station, simulating actual input and output behavior through interfaces to ensure that it does not affect the actual power grid operation. Under each disturbance scenario, the system runs a preset dispatch business process, collects the software's response behavior and output results, extracts the corresponding third-level indicators, and calculates the corresponding quality feature vectors, providing a data foundation for subsequent quality stability analysis. Through the above design, high-value operational status data can be obtained under various typical and extreme operating conditions while ensuring the safety of the actual power grid, improving the comprehensiveness and robustness of the quality evaluation results.

[0045] S2. Standardize the third-level indicators and determine the subjective and objective fusion weights for each third-level indicator.

[0046] In this embodiment, to ensure that different types of third-level indicators are comparable under the same dimension, the system performs uniform standardization processing on all third-level indicators at the analysis and calculation layer. The business subsystem in the first The original values ​​under each third-level indicator are denoted as ,in Used to identify business subsystems or modules Third-level indicators are used to identify those belonging to a specific quality characteristic dimension. Based on the business meaning of the indicators, third-level indicators are divided into positive and negative indicators. A higher value for a positive indicator indicates better quality, while a higher value for a negative indicator indicates worse quality. Specifically, for each third-level indicator... Its standardized value is defined as:

[0047] in, and These are the maximum and minimum values ​​of the third-level indicator in the sample set, respectively. Through the above standardization process, all third-level indicators are unified to... The interval is defined, and the original monotonicity is maintained, avoiding the deviation caused by direct comparison between different dimensions and magnitudes.

[0048] After standardization is completed, the standardized values ​​of each business subsystem on all third-level indicators are arranged according to the business subsystem dimension and the indicator dimension to construct a standardized indicator matrix. :

[0049] in, For the number of business subsystems, For the number of third-level indicators, each element in the matrix Indicates the first The business subsystem in the first Standardized scores under each third-level indicator. Matrix As the core input dataset for subsequent weight calculation and intelligent evaluation modules, it enables weight learning and quality evaluation to be carried out in a unified index space.

[0050] To quantify the importance of each third-level indicator, this embodiment employs a combined subjective and objective weight determination method in the indicator weight calculation and analysis phase. The objective weighting part uses the information entropy method to analyze the distribution of the third-level indicators in the training samples. For the third... For each third-level indicator, first calculate its weight in each business subsystem:

[0051] Next, calculate the information entropy of this indicator:

[0052] in , This represents the number of business subsystems. The lower the information entropy, the stronger the indicator's ability to distinguish different sample quality levels. Based on this, the objective weight is calculated:

[0053] in For the number of third-level indicators, For the first The objective weight of each third-level indicator.

[0054] The subjective weighting part employs the analytic hierarchy process (AHP), where experts construct pairwise comparison matrices for the third-level indicators. The subjective weight vector is obtained by the eigenvalue method. To satisfy:

[0055] in, For the pairwise comparison matrix, for The largest eigenvalue, For corresponding The eigenvectors of , their components Constituting the first The subjective weights of each third-level indicator. Finally, based on the preset fusion coefficient. By linearly fusing objective and subjective weights, the first result is obtained. The subjective and objective combined weights of the third-level indicators:

[0056] The integrated weights utilize the discriminative information contained in the objective data while retaining the expert's understanding of the business characteristics, providing a reasonable weighting basis for subsequent indicator aggregation and quality evaluation.

[0057] S3. Aggregate the third-level indicators related to the same quality characteristic according to the subjective and objective fusion weights of each third-level indicator to obtain the corresponding second-level indicators.

[0058] In this embodiment, for the first For each business subsystem, the analysis and calculation layer utilizes a weighted approach that combines subjective and objective metrics to aggregate third-level indicators belonging to the same quality characteristic dimension, resulting in five second-level quality characteristic indicators: functionality, reliability, efficiency, security, and standardization. Taking the functional quality characteristic as an example, its score can be calculated using the following formula:

[0059] in, This is the set of third-level indicators belonging to the functional quality dimension. For the set of The subjective and objective weighting of the third-level indicators For the first The standardized scores of each business subsystem under this indicator can be calculated separately. Similarly, the reliability quality characteristic scores can be calculated separately. Performance quality characteristic score Safety and quality characteristic score and normative quality characteristic score This allows for the construction of a second-level quality feature vector for each business subsystem:

[0060] This aggregation process enables an effective mapping from a large number of fine-grained third-level indicators to a small number of quality feature indicators. While retaining key information, it reduces the input dimension of the evaluation model, improves computational efficiency, and facilitates the interpretation of evaluation results from the perspective of quality features.

[0061] S4. Input the index vector composed of each second-level index into the pre-trained dynamic fuzzy neural network model, and output the first-level index of the corresponding subsystem as the comprehensive quality evaluation score.

[0062] The quality evaluation layer calls upon the pre-trained dynamic fuzzy neural network model, using the quality feature vectors of each business subsystem. As model input, this enables a nonlinear comprehensive evaluation of software quality. The dynamic fuzzy neural network model employs a hierarchical structure, including an input layer, membership function layer, fuzzy rule layer, normalization layer, and output layer. The input layer receives... The second-level indicators are as follows: the membership function layer fuzzifies each input dimension using a triangular or Gaussian membership function to obtain the membership degree of the corresponding linguistic variable; the fuzzy rule layer combines the membership degrees of each quality feature according to preset "if-then" fuzzy rules to obtain the activation degree of each rule; the normalization layer normalizes the activation degree of the rules to obtain normalized weights; and the output layer calculates the suggested output value of each rule according to the rule output function. The data is then weighted according to normalized weights to obtain the first-level indicator of the business subsystem, namely the comprehensive quality evaluation score. .

[0063] For scenarios considering disturbances, in each operating scenario or disturbance condition... Below, the system uses the corresponding quality feature vector Using the input as the scene, the calculation is obtained. The following is the comprehensive quality evaluation score:

[0064] in, For the first The activation level of a fuzzy rule. This is the output value of the rule. Evaluation Score Values ​​located at Within the range, the closer to 1, the higher the software quality. By introducing a dynamic fuzzy neural network, the nonlinear relationships and coupling characteristics between various quality features can be fully utilized to improve the sensitivity and discriminative power of the comprehensive quality evaluation results to the actual operating state.

[0065] S5. Determine the software quality level of the business subsystem based on the comprehensive quality evaluation score and the preset level threshold, and output the software quality evaluation results including the comprehensive quality evaluation score and quality level.

[0066] The quality evaluation module determines the quality level based on the comprehensive quality evaluation score of each business subsystem and the preset quality level classification rules. For example, it can classify the intervals... Divided into four levels: "Excellent," "High," "Medium," and "Low." When judged as "excellent", When it is judged as "high", When it is determined to be "medium", The time frame is judged as "low". Considering perturbation scenarios, the system generates a comprehensive quality evaluation score sequence across multiple scenarios and time points. Statistical analysis is performed to construct a quality stability coefficient by calculating statistical measures such as variance, range, or standard deviation. This coefficient reflects the quality fluctuations of the business subsystem under fault injection, load change, or operational disturbance scenarios. A smaller stability coefficient indicates more stable quality performance and stronger resistance to disturbances. Finally, the quality evaluation module outputs multi-dimensional evaluation results, including a comprehensive quality evaluation score, quality level, and quality stability coefficient, and generates a visualized quality evaluation report. This provides a basis for the online acceptance testing of the scheduling software, version comparison, operational risk monitoring, and regulatory business decisions.

[0067] Steps S1 to S5 above describe the process by which the system performs quality evaluation using the trained model during the operational phase. To obtain a dynamic fuzzy neural network model and corresponding index weights suitable for engineering applications, the system also includes a model training and weight learning phase before deployment. The training phase uses the same index system and standardization rules as the operational phase, but the training phase focuses on learning the parameters and subjective / objective fusion weights of the quality evaluation model from historical and simulation data. The specific process of the training phase includes: collecting historical operational logs, simulation data, and manual test data; performing the same preprocessing operations on the data as in the operational phase; extracting third-level indicators to construct training samples; calculating objective and subjective weights using the information entropy method and the analytic hierarchy process (AHP); and using the fusion coefficient... Obtain the overall weight Quality feature vectors based on training samples Based on the quality assessment results, a loss function is constructed. Gradient descent or minimum mean square error methods are used to iteratively update the membership function parameters, rule antecedent parameters, and rule consequent parameters, thereby improving the model output. The error between the error and the reference value gradually decreases. When the error meets the preset convergence condition or reaches the preset training rounds, the membership function and fuzzy rules obtained from the training are fixed as model parameters used in the operation phase. Only periodic offline fine-tuning is performed in the simulation or pre-production environment based on new data to adapt to the long-term evolution trend of the system. Through the cooperation of the above-mentioned operation phase and training phase, this embodiment can realize dynamic quality evaluation of power grid dispatch automation system software under multi-dimensional operating conditions, and has adaptive learning capability, composite scenario verification capability, and closed-loop optimization capability.

[0068] As shown in Figure 2, the technical framework of the power grid dispatch automation system software quality evaluation method serves as an optional implementation environment for this method. The data input layer encompasses multi-source data, including power grid models, power grid parameters, power grid operation data, fault cases, and primary and secondary events; that is, the collected operational status data of the power grid dispatch automation system, providing raw data support for subsequent indicator extraction and analysis. The analysis and calculation layer serves as a key intermediate layer. The quality model construction module transforms multi-source data into a structured indicator matrix through a process of data acquisition → feature extraction → indicator classification → indicator quantification. The indicator weight calculation and analysis module integrates subjective and objective weights to obtain a weight vector. The processing logic for determining indicator weights through a subjective-objective fusion method is used. The indicator processing module completes indicator calculation, normalization, and quality classification, and outputs the indicators. As a result, the system simultaneously feeds back model parameter errors, enabling the second-level index aggregation and model inference stages. The fault injection and scenario fusion module simulates extreme scenarios, generating scenario features and robustness indicators, which are then passed upwards to the quality evaluation layer. It also receives weight vectors and indicator results downwards and feeds back response results, fulfilling the requirements for constructing abnormal pattern sets, constructing disturbed scenarios, and performing quality stability analysis. This addresses the pain point of traditional evaluation methods being unable to cover extreme scenarios. The quality evaluation layer integrates heterogeneous data streams to achieve comprehensive evaluation, risk identification and early warning, and quality report generation, along with the level determination step, outputting a final evaluation report containing scores, levels, and risk points. This technical framework realizes the entire process logic of multi-source data → index processing → scenario fusion → comprehensive evaluation.

[0069] Through the above implementation methods, the system achieves dynamic quality evaluation of power grid dispatch automation system software under multi-dimensional operating conditions. It possesses adaptive learning capabilities, composite scenario verification capabilities, and closed-loop optimization capabilities, providing reliable data support, quality basis, and risk assessment capabilities for the research, development, deployment, operation, and upgrade of dispatch business software. This realizes the upgrade of power grid dispatch software quality evaluation from static evaluation to dynamic, multi-scenario, and verifiable evaluation. Example 2 like Figure 3 As shown, based on the same inventive concept as the above embodiments, the present invention also provides a software quality evaluation device for a power grid dispatch automation system. For each business subsystem of the software, a first-level indicator for quantifying the overall quality of the subsystem is preset. The first-level indicator is determined by several second-level indicators for quantifying multiple quality characteristics of the subsystem. The second-level indicators are determined by several third-level indicators for quantifying the operating status of the business subsystem. The device includes: The data acquisition module is used to acquire the operating status data of the power grid dispatch automation system software and extract third-level indicators from the operating status data. The third-level indicator processing module is used to standardize the third-level indicators and determine the subjective and objective fusion weights of each third-level indicator. The second-level indicator processing module is used to aggregate third-level indicators related to the same quality characteristic according to the subjective and objective fusion weight of each third-level indicator to obtain the corresponding second-level indicator. The first-level indicator processing module is used to input the indicator vector composed of each second-level indicator into the pre-trained dynamic fuzzy neural network model, so as to output the first-level indicator of the corresponding subsystem as the comprehensive quality evaluation score. The evaluation module is used to determine the software quality level of the business subsystem based on the comprehensive quality evaluation score and the preset level threshold, and output the software quality evaluation results including the comprehensive quality evaluation score and the quality level.

[0070] Example 3 like Figure 4 As shown, the present invention also provides an electronic device 100 for implementing a software quality evaluation method for a power grid dispatch automation system; The electronic device 100 includes a memory 101, at least one processor 102, a computer program 103 stored in the memory 101 and executable on at least one processor 102, and at least one communication bus 104.

[0071] The memory 101 can be used to store computer program 103. The processor 102 implements the software quality evaluation method of power grid dispatch automation system in Embodiment 1 by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101.

[0072] The memory 101 may primarily include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created based on the use of the electronic device 100 (such as audio data), etc. In addition, the memory 101 may include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other non-volatile solid-state storage device.

[0073] At least one processor 102 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 102 may be a microprocessor or any conventional processor. Processor 102 is the control center of electronic device 100, connecting various parts of electronic device 100 via various interfaces and lines.

[0074] The memory 101 in the electronic device 100 stores multiple instructions to implement a software quality evaluation method for a power grid dispatch automation system, and the processor 102 can execute multiple instructions to achieve the following: Obtain the operating status data of the power grid dispatch automation system software, and extract the third-level indicators from the operating status data; The third-level indicators are standardized, and the subjective and objective weights of each third-level indicator are determined. Based on the subjective and objective fusion weights of each third-level indicator, the third-level indicators related to the same quality characteristic are aggregated to obtain the corresponding second-level indicators; The index vector composed of each second-level index is input into a pre-trained dynamic fuzzy neural network model to output the first-level index of the corresponding subsystem as the comprehensive quality evaluation score. The software quality level of the business subsystem is determined based on the comprehensive quality evaluation score and the preset level threshold, and the software quality evaluation result, including the comprehensive quality evaluation score and the quality level, is output.

[0075] Example 4 If the modules / units integrated in the electronic device 100 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, and read-only memory (ROM).

[0076] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0077] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0078] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0079] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0080] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0081] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A power grid dispatch automation system software quality evaluation method, characterized in that, For each business subsystem of software, a first-level indicator for quantifying the overall quality of the subsystem is preset, the first-level indicator is determined by a plurality of second-level indicators each for quantifying a plurality of quality characteristics of the subsystem, the second-level indicators are determined by a plurality of third-level indicators each for quantifying an operating state of the business subsystem, and the method comprises: acquiring operating state data of the power grid dispatching automation system software and extracting third-level indicators from the operating state data; standardizing the third-level indicators and determining subjective and objective fusion weights of the third-level indicators; aggregating third-level indicators related to the same quality characteristic according to the subjective and objective fusion weights of the third-level indicators to obtain corresponding second-level indicators; inputting an indicator vector composed of the second-level indicators into a pre-trained dynamic fuzzy neural network model to output a first-level indicator corresponding to the subsystem as a comprehensive quality evaluation score; determining a software quality level of the business subsystem according to the comprehensive quality evaluation score and a preset level threshold, and outputting a software quality evaluation result including the comprehensive quality evaluation score and the quality level.

2. The grid dispatch automation system software quality evaluation method of claim 1, wherein, Further comprising: constructing an abnormal pattern set based on historical fault records, operating logs and simulation results: wherein each anomaly pattern is represented by a feature vector as wherein, is a fault type, is an affected business subsystem, is an anomaly duration, is an impact intensity coefficient; select at least one abnormal mode based on the abnormal mode set, construct a disturbance scenario, and apply a disturbance to the operation of the power grid dispatch automation system software under the disturbance scenario, so that the software input quantity or key operation quantity after the disturbance satisfies the following formula: wherein, is a software input quantity or a key operating quantity before the disturbance, is a disturbance amplitude coefficient related to the abnormal pattern, is a disturbance signal function; obtaining a corresponding sequence of comprehensive quality evaluation scores under a plurality of disturbance scenarios and a normal operating scenario, and calculating a quality stability indicator representing the quality fluctuation degree of the business subsystem under different operating scenarios based on the score sequence, and outputting the quality stability indicator as part of the software quality evaluation result.

3. The grid dispatch automation system software quality evaluation method of claim 1, wherein, The standardization of the third-level indicators comprises: dividing the third-level indicators into positive indicators and negative indicators according to the business implications of the third-level indicators; The standardized value of each third-level indicator is calculated by the following formula : Wherein, the original value of each third-level index is , and are the maximum and minimum values of the third-level index in the sample set, respectively. for positive indicators, mapping the values of each business subsystem under the indicator to the interval of 0 to 1 according to a preset rule, and for negative indicators, mapping the values of each business subsystem under the indicator to the interval of 0 to 1 according to a preset rule, so that all third-level indicators after standardization have a unified value range and maintain monotonicity; The standardized values of each business subsystem on all third-level indexes are arranged according to the business subsystem dimension and the index dimension to construct a standardized index matrix : wherein, is the number of business subsystems, is the number of third level indicators.

4. The grid dispatch automation system software quality evaluation method of claim 1, wherein, determining the weights of the third-level indicators based on the subjective and objective fusion weights comprises: The entropy value of the third level index is calculated according to the following formula : The first third level index is the proportion of the first business subsystem; k = 1 / ln n , is the number of business subsystems; The objective weight is determined from the entropy value according to the following formula : is the third level indicator quantity; The analytic hierarchy process is used to construct a pairwise comparison matrix of the third level indexes, and a subjective weight vector is obtained by using an eigenvalue method so as to satisfy the following formula: wherein, is the pairwise comparison matrix, is the largest eigenvalue of is the eigenvector corresponding to whose components is the subjective weight of the first third-level indicator; calculating the subjective and objective fusion weights of the third-level indicators according to the following formula: wherein, is the subjective weight of the first secondary index, is the subjective weight of the second secondary index, is the subjective weight of the third secondary index, and ∈[0,1] is a preset fusion coefficient.

5. The grid dispatch automation system software quality evaluation method of claim 1, wherein, The pre-trained dynamic fuzzy neural network model comprises: an input layer for receiving an indicator vector composed of a plurality of second-level indicators; a membership function layer for mapping each second-level indicator to a plurality of linguistic variable membership degrees, wherein each second-level indicator corresponds to at least one adjustable membership function; a fuzzy rule layer for calculating the activation degree of a fuzzy rule according to the output of each membership function, the fuzzy rule defining the association between the second-level indicators and the first-level indicator; a normalization layer for normalizing each fuzzy rule activation degree to obtain a normalized weight of each fuzzy rule; and an output layer for weighted sum of the outputs of each fuzzy rule according to the normalized weights to obtain the first-level indicator as the comprehensive quality evaluation score.

6. The grid dispatch automation system software quality evaluation method of claim 5, wherein, The training steps of the dynamic fuzzy neural network model comprise: The training sample set is constructed by using historical operation data, simulation operation data or artificial test data, the second-level index vector corresponding to each training sample is taken as the model input, and the target quality evaluation result corresponding to the training sample is taken as the expected output of the model; The parameters of the membership function layer and the rule parameters of the fuzzy rule layer are iteratively updated based on the training sample set, so that the error between the first-level index output by the model and the expected output gradually decreases; The model parameters are updated according to the following formula: wherein, is a model parameter vector at time is a learning rate, is a difference between an actual output and an expected output, is a prediction error or a correction amount;​ After the error meets the preset convergence condition or reaches the preset training round, the training obtained membership function and fuzzy rule are fixed as the model parameters used in the running phase.

7. The grid dispatch automation system software quality evaluation method of claim 1, wherein, The software quality level of the business subsystem is determined according to the comprehensive quality evaluation score and a preset level threshold, including: A plurality of continuous level intervals are pre-divided for the comprehensive quality evaluation score, and each level interval corresponds to a software quality level; After obtaining the comprehensive quality evaluation score of the business subsystem, the level interval where the score is located is determined, and the corresponding quality level is taken as the software quality level of the business subsystem; The number and boundary of the level intervals are configured according to the software quality management requirements of the power grid dispatching automation system to support the hierarchical management and comparative analysis of the software quality of different business subsystems.

8. A power grid dispatch automation system software quality evaluation device, characterized by, For each business subsystem of the software, a first-level index for quantifying the overall quality of the subsystem is preset, the first-level index is determined by a plurality of second-level indexes for quantifying a plurality of quality characteristics of the subsystem, and the second-level indexes are determined by a plurality of third-level indexes for quantifying the running state of the business subsystem, and the device includes: A data acquisition module is configured to acquire running state data of the software of the power grid dispatching automation system and extract third-level indexes from the running state data; A third-level index processing module is configured to perform standardization processing on the third-level indexes and determine the subjective and objective fusion weights of the third-level indexes; A second-level index processing module is configured to aggregate third-level indexes related to the same quality characteristic according to the subjective and objective fusion weights of the third-level indexes to obtain corresponding second-level indexes; A first-level index processing module is configured to input an index vector composed of the second-level indexes into a pre-trained dynamic fuzzy neural network model to output the first-level index of the corresponding subsystem as a comprehensive quality evaluation score; An evaluation module is configured to determine the software quality level of the business subsystem according to the comprehensive quality evaluation score and a preset level threshold, and output a software quality evaluation result including the comprehensive quality evaluation score and the quality level.

9. An electronic device, comprising: The computer readable storage medium stores at least one instruction, and the at least one instruction is executed by the processor to implement the power grid dispatching automation system software quality evaluation method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores at least one instruction, and the at least one instruction is executed by the processor to implement the power grid dispatching automation system software quality evaluation method according to any one of claims 1 to 7.

Images