New power distribution network dynamic utility value evaluation method and system based on chaos theory

Abstract

The invention belongs to the field of power distribution network dynamic utility value evaluation, and particularly relates to a novel power distribution network dynamic utility value evaluation method and system based on a chaos theory. Constructing a multi-source heterogeneous data acquisition system; extracting features by adopting an improved multi-scale method; through a hybrid modeling strategy, evaluation and prediction of the utility of the power distribution network equipment are realized; and performing effect evaluation by adopting dynamic utility evaluation. According to the invention, by constructing a multi-dimensional utility evaluation system based on chaotic features, the limitation of the existing equipment utility evaluation method, especially the limitation of linear hypothesis in the traditional evaluation method, is solved. Through layered progressive system architecture design, deep optimization is performed from four aspects of data acquisition, feature extraction, modeling prediction and utility evaluation, and accuracy, real-time performance and reliability of an evaluation result are ensured.

Description

Technical Field

[0001] The invention belongs to the field of distribution network dynamic utility value evaluation, and specifically relates to a new method and system for distribution network dynamic utility value evaluation based on chaos theory. Background Art

[0002] In modern distribution network management, equipment utility evaluation and prediction are key links to ensure stable system operation, optimize resource allocation and improve economic benefits. With the development of smart grid and big data technology, traditional distribution network equipment management methods have been unable to meet the growing demand for real-time monitoring, fault diagnosis, utility evaluation and prediction. Therefore, how to improve the intelligent level of distribution network equipment management through efficient data collection, feature extraction, modeling analysis and dynamic evaluation has become a technical problem that needs to be solved urgently.

[0003] Traditional distribution network equipment management mainly relies on manual inspections and static data recording methods, usually through the basic parameters of the equipment (such as equipment model, installation date, etc.) and simple operating data (such as voltage, current, etc.) for evaluation. Although these methods can ensure the safe operation of the equipment to a certain extent, they ignore the dynamic changes of the equipment in a complex operating environment and possible nonlinear problems. With the increase in the number of distribution network equipment and the complexity of the operating environment, the limitations of traditional methods are becoming more and more apparent, especially in the accurate evaluation of equipment utility and life cycle management.

[0004] In addition, existing utility evaluation methods mostly rely on linear assumptions and cannot effectively deal with the complex nonlinear and time-varying characteristics of equipment operation. Traditional utility evaluation methods cannot reflect the health status and utility changes of equipment in real time, nor can they be flexibly adjusted according to the characteristics of different equipment. With the rise of intelligent and automated technologies, existing evaluation methods urgently need to be upgraded to meet the needs of modern distribution network management for efficient, real-time, and multi-dimensional evaluation and prediction.

[0005] With the popularization of intelligent equipment, the types of data collected by distribution network equipment are increasing, covering multi-source heterogeneous data such as equipment operation data, fault records, cost data, etc. These data often have different formats, sources and collection frequencies, and need to be effectively integrated and analyzed through advanced data processing technology. Therefore, how to build an efficient data collection system and how to extract useful feature information from massive data have become key issues in equipment utility evaluation.

[0006] However, traditional data collection and feature extraction methods face problems such as poor data quality, large noise interference, and difficulty in handling missing values. In order to obtain accurate feature information, it is often necessary to use complex signal processing techniques, such as wavelet transform, time series analysis, phase space reconstruction, etc. However, in the feature extraction process, existing technologies often fail to fully consider the nonlinear characteristics and dynamic changes of the equipment's operating status, and it is difficult to process data at different levels and time scales. Therefore, how to effectively preprocess and extract features from these data, and further improve the robustness and accuracy of the model, has become a major challenge facing the equipment utility evaluation system.

[0007] In the process of utility evaluation of distribution network equipment, how to comprehensively consider the technical, economic, environmental and social utilities of the equipment and realize the dynamic prediction of the future utility of the equipment through scientific prediction methods is still an urgent problem to be solved. Existing utility evaluation methods mainly focus on static evaluation, such as equipment performance, failure rate, maintenance time, etc. However, the utility of equipment is not fixed, it fluctuates with the operating status of the equipment, environmental factors and external conditions. Therefore, the biggest problem of existing methods is that they lack adaptability to dynamic changes, making it difficult to achieve real-time evaluation and long-term prediction.

[0008] In addition, the prediction of equipment utility often requires the establishment of models based on historical data. However, the operating data of distribution network equipment is complex and high-dimensional. The use of traditional regression models or time series analysis methods is prone to high-dimensional problems and it is difficult to capture the complex evolution of equipment utility. Therefore, how to improve the accuracy and real-time performance of utility evaluation and prediction through innovative modeling methods, optimized prediction algorithms, and reasonable feature selection is a direction that the existing technology urgently needs to improve.

[0009] The current evaluation of the utility of distribution network equipment mostly focuses on the analysis of a single dimension, such as technical utility or economic utility, but lacks a systematic evaluation method that comprehensively considers multiple dimensions. The utility of equipment is a multi-dimensional comprehensive indicator that involves multiple aspects such as technical performance, economic cost, environmental impact, and social contribution. How to comprehensively evaluate the utility of equipment in multiple dimensions and provide decision support based on different needs is the key to improving equipment management efficiency. Existing technologies have not yet proposed a comprehensive and systematic multi-dimensional evaluation method, which cannot comprehensively evaluate all aspects of equipment utility and is difficult to provide a scientific basis for equipment updates, maintenance, and investment decisions. Summary of the invention

[0010] At present, the main technical solutions for the utility evaluation of distribution network equipment are as follows: The proposed evaluation method based on hierarchical analysis method mainly relies on expert experience scoring, which has the problems of strong subjectivity and unstable evaluation results. The proposed fuzzy comprehensive evaluation method, although taking uncertainty into account, lacks a scientific basis for the determination of the membership function. The proposed neural network evaluation model requires a large amount of historical sample data for training, and the model has poor interpretability. The proposed grey correlation analysis method is difficult to effectively reflect the dynamic change characteristics of equipment performance.

[0011] In order to solve the above technical problems, the present invention provides the following technical solutions: a new method for evaluating the dynamic utility value of a distribution network based on chaos theory, comprising: constructing a multi-source heterogeneous data acquisition system;

[0012] Use improved multi-scale to extract features;

[0013] Through hybrid modeling strategies, the utility of distribution network equipment can be evaluated and predicted;

[0014] Use dynamic utility evaluation to evaluate the effect.

[0015] As a preferred solution of the new method for dynamic utility value evaluation of distribution network based on chaos theory described in the present invention, the construction of multi-source heterogeneous data acquisition system includes four levels: data acquisition layer, feature extraction layer, model construction layer and utility evaluation layer, and data interaction and information transmission are realized between each level through standardized interfaces.

[0016] As a preferred solution of the new method for evaluating the dynamic utility value of a distribution network based on chaos theory described in the present invention, the state space definition includes power equipment state, grid state, environmental state, and time information.

[0017] As a preferred solution of the new method for evaluating the dynamic utility value of distribution network based on chaos theory described in the present invention, the improved multi-scale feature extraction includes a feature extraction layer, a phase space reconstruction module, and a feature parameter extraction module.

[0018] As a preferred solution of the new method for evaluating the dynamic utility value of distribution network based on chaos theory described in the present invention, the hybrid modeling strategy includes chaotic time series modeling, prediction model training, and weight adaptive updating.

[0019] As a preferred solution of the new method for evaluating the dynamic utility value of distribution network based on chaos theory described in the present invention, the chaotic time series modeling includes constructing a local linear prediction model, establishing a global nonlinear power system model, and designing a model adaptive switching mechanism.

[0020] As a preferred solution of the new method for evaluating the dynamic utility value of a distribution network based on chaos theory described in the present invention, the prediction model training includes: using a sliding time window for sample construction, using cross-validation to optimize model parameters, and introducing an early stopping strategy to prevent overfitting;

[0021] The weight adaptive updating includes calculating the initial weight based on the grey correlation degree, dynamically adjusting the weight using the entropy weight method, and introducing a forgetting factor to process historical data.

[0022] As a preferred solution of the new method for dynamic utility value evaluation of distribution network based on chaos theory described in the present invention, the dynamic utility evaluation includes adopting a hierarchical and progressive processing flow, and data interaction is achieved between modules through standardized interfaces.

[0023] As a preferred solution of the new system for evaluating the dynamic utility value of distribution network based on chaos theory described in the present invention, it includes: a data acquisition module, a feature extraction module, a model construction module, and a utility evaluation module.

[0024] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of any one of the methods described in a new method for evaluating the dynamic utility value of a distribution network based on chaos theory are implemented.

[0025] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of any one of the methods described in a new method for evaluating the dynamic utility value of a distribution network based on chaos theory are implemented.

[0026] Beneficial effects of the invention: The invention solves the limitations of existing equipment utility evaluation methods, especially the limitations of linear assumptions in traditional evaluation methods, by constructing a multi-dimensional utility evaluation system based on chaotic characteristics. Through hierarchical and progressive system architecture design, in-depth optimization is carried out from four aspects: data collection, feature extraction, modeling prediction, and utility evaluation, ensuring the accuracy, real-time and reliability of the evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0028] Figure 1A schematic flow chart of a new method for evaluating the dynamic utility value of a distribution network based on chaos theory is provided as an embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0030] Example 1

[0031] Reference Figure 1 , which is the first embodiment of the present invention, and provides a new method for evaluating the dynamic utility value of a distribution network based on chaos theory, such as Figure 1 As shown, including:

[0032] S1: Build a multi-source heterogeneous data collection system.

[0033] It should be noted that the construction of the multi-source heterogeneous data acquisition system includes four levels: data acquisition layer, feature extraction layer, model construction layer and utility evaluation layer. Data interaction and information transmission are achieved between the levels through standardized interfaces.

[0034] The system architecture of the present invention adopts a hierarchical progressive design, including four levels: data collection layer, feature extraction layer, model building layer and utility evaluation layer. Data interaction and information transmission are realized between each level through standardized interfaces:

[0035] At the data collection layer, a multi-source heterogeneous data collection system is designed, including:

[0036] Equipment basic information collection module: collects static information such as equipment model, specifications, manufacturing year, manufacturer, installation location, etc.

[0037] Operation status monitoring module: sampling voltage, current, active power, reactive power, power factor and other operation parameters every 5 minutes.

[0038] Fault recording module: real-time recording of equipment fault type, occurrence time, duration, repair time, fault cause and other information.

[0039] Cost data collection module: monthly statistics of equipment procurement cost, installation cost, operation and maintenance cost, repair cost, scrap disposal cost and other full life cycle cost data.

[0040] S2: Use improved multi-scale to extract features.

[0041] It should be noted that the improved multi-scale feature extraction includes a feature extraction layer, a phase space reconstruction module, and a feature parameter extraction module.

[0042] Feature extraction layer:

[0043] An improved multi-scale feature extraction method is adopted, including:

[0044] Time series preprocessing module: Use wavelet transform to remove high-frequency noise; use moving average method to smooth abnormal fluctuations; standardize data based on Z-score method; use interpolation algorithm to process missing data.

[0045] Specifically, in the time series preprocessing module, the present invention uses the db4 wavelet transform to decompose the original signal into 4 layers, and removes high-frequency noise by setting the soft threshold λ=2.5. For the processing of abnormal fluctuations, the 11-point moving average method is used for smoothing, which can effectively eliminate the influence of short-term random disturbances. The data standardization adopts the improved Z-score method, and realizes dynamic standardization processing by calculating the mean and standard deviation within the sliding time window. For missing data, the present invention designs an adaptive algorithm based on cubic spline interpolation, which ensures the smoothness and continuity of the interpolation result by adjusting the node position and curvature constraints.

[0046] Phase space reconstruction module: Determine the optimal embedding dimension based on the improved FNN algorithm; use the mutual information method to determine the optimal time delay; introduce the Cao method to verify the reconstruction effect.

[0047] A phase space reconstruction method based on multi-scale decomposition is proposed, which has better noise resistance and reconstruction accuracy than traditional reconstruction methods.

[0048] Specific steps:

[0049] Time series decomposition:

[0050] X(t)=IMF1(t)+IMF2(t)+...+IMFn(t)+Rn(t)

[0051] Where: IMFi(t) is the i-th intrinsic mode function, Rn(t) is the residual term, and n is the number of decomposition levels, which is determined by the energy ratio;

[0052] Reconstruction parameter optimization:

[0053] τi=argmin{I(IMFi(t),IMFi(t+τ))}

[0054] mi=argmin{E1(m) / E2(m)}

[0055] Where: τi is the optimal time delay of the ith IMF, mi is the optimal embedding dimension of the ith IMF, I(.) is the mutual information function, E1(m), E2(m) are pseudo-neighbor statistics

[0056] Multi-scale reconstruction:

[0057] Xi(t)=[IMFi(t),IMFi(t+τi),...,IMFi(t+(mi-1)τi)]

[0058] X(t)=[X1(t),X2(t),...,Xn(t)]

[0059] Specifically, in the phase space reconstruction module, the present invention proposes an improved FNN (False Nearest Neighbors) algorithm to determine the optimal embedding dimension. The improved algorithm introduces an adaptive threshold mechanism, which dynamically adjusts the judgment threshold by calculating the Euclidean distance ratio between point pairs in the phase space. Specifically, when the distance ratio exceeds the set threshold, it is determined to be a false neighbor point. By gradually increasing the embedding dimension, when the proportion of false neighbor points is less than 5%, it is determined to be the optimal embedding dimension. For the determination of time delay, the present invention adopts the mutual information method, by calculating the mutual information between the time series and its delay series, and selecting the delay value corresponding to the first local minimum of the mutual information as the optimal time delay. In order to verify the reconstruction effect, the Cao method is introduced to calculate the two statistics E1(m) and E2(m). When E1(m) / E2(m)<0.05, it is considered that the reconstruction result meets the requirements.

[0060] Feature parameter extraction module: calculate the maximum Lyapunov exponent to characterize the stability of the system; extract the correlation dimension to reflect the complexity of the system; calculate the Kolmogorov entropy to measure the difficulty of system prediction; extract the fractal dimension to characterize the self-similarity of the system.

[0061] Specifically, in the feature parameter extraction module, the present invention designs a complete set of chaos feature extraction schemes. First, the improved Wolf algorithm is used to calculate the maximum Lyapunov exponent. The algorithm improves the calculation accuracy by optimizing the orbital evolution time and phase space search strategy. For the extraction of correlation dimension, the GP algorithm is adopted and an adaptive scale selection mechanism is introduced. The scale parameter r is automatically determined by the multiple of the standard deviation of the time series. When calculating the Kolmogorov entropy, the present invention proposes an improved GP algorithm based on a correction factor. By introducing the fm correction factor, the accuracy of entropy value estimation is improved. For the calculation of fractal dimension, an improved box dimension algorithm is adopted. By optimizing the box size selection strategy, the reliability of dimension estimation is improved.

[0062] An adaptive feature extraction method is designed to overcome the limitation of fixed feature extraction parameters in the prior art.

[0063] Specific steps: Maximum Lyapunov exponent calculation:

[0064] λmax=lim(t→∞)1 / t*ln|df(X(t)) / dX(0)|

[0065] The Wolf algorithm is used for numerical calculation:

[0066] λi=1 / tk*ln(L'k / Lk)

[0067] λmax=1 / M*∑λi

[0068] Where: tk is the evolution time, Lk, L'k is the phase space orbit distance, M is the number of calculations;

[0069] Correlation dimension calculation:

[0070] C(r,m)=1 / N(N-1)*∑∑θ(r-||Xi-Xj||)

[0071] D2=lim(r→0)lim(N→∞)[lnC(r,m) / ln(r)]

[0072] Introducing adaptive threshold selection:

[0073] r = σ * std (X)

[0074] Where: σ is the adaptive coefficient, std(X) is the standard deviation of the time series;

[0075] Kolmogorov entropy calculation:

[0076] K2=lim(r→0)lim(m→∞)[ln(Cm(r) / Cm+1(r))]

[0077] Using the improved GP algorithm:

[0078] K2=-fm*ln[Cm+1(r) / Cm(r)]

[0079] Where fm is the correction factor.

[0080] S3: Evaluation and prediction of distribution network equipment utility are achieved through hybrid modeling strategies.

[0081] It should be noted that the hybrid modeling strategy includes chaotic time series modeling, prediction model training, and weight adaptive updating.

[0082] The chaotic time series modeling includes constructing a local linear prediction model, establishing a global nonlinear dynamic system model, and designing a model adaptive switching mechanism.

[0083] The prediction model training includes constructing samples using a sliding time window, optimizing model parameters using cross-validation, and introducing an early stopping strategy to prevent overfitting.

[0084] The weight adaptive update includes calculating the initial weight based on the grey relational degree, dynamically adjusting the weight using the entropy weight method, and introducing a forgetting factor to handle historical data.

[0085] An innovative hybrid modeling strategy is proposed: Chaotic time series modeling module: Construct a local linear prediction model; Establish a global nonlinear dynamic system model; Design a model adaptive switching mechanism.

[0086] Specifically, in the chaotic time series modeling module, a local linear prediction model is first constructed. Based on the orbit after phase space reconstruction, the least squares method is used to fit the local linear mapping relationship. For any point X(t) in the reconstructed space, its k nearest neighbor points are selected to construct a local prediction model:

[0087] X(t + 1) = AX(t) + b,

[0088] where A is the local linear mapping matrix and b is the bias vector. In the present invention, the balance between the locality and stability of the model is achieved by adaptively adjusting the k value (initially set to sqrt(N), where N is the number of samples).

[0089] At the same time, a global nonlinear dynamic system model is established to capture the long-term evolution characteristics of the system. The model adopts an improved Volterra series expansion form:

[0090] X(t + 1) = ∑(i = 1 → M) ai * Vi[X(t)],

[0091] where Vi is the i-th order Volterra kernel function, ai is the corresponding coefficient, and M is the highest order (in the present invention, M = 3). The model parameters are updated online by the recursive least squares method, ensuring the tracking ability for non-stationary processes.

[0092] To give full play to the advantages of the two models, the present invention designs an adaptive switching mechanism based on the prediction error. A sliding time window (window length W = 240, corresponding to 20 hours of data) is introduced to calculate the prediction error sequences {e1(t)} and {e2(t)} of the two models. When the local model error e1(t) < e2(t) * α (α = 0.85), the local model is used for prediction, otherwise the global model is used, realizing the dynamic optimization of the prediction accuracy.

[0093] Prediction model training module: use sliding time window for sample construction; use cross-validation to optimize model parameters; introduce early stopping strategy to prevent overfitting; weight adaptive update module:; calculate initial weights based on grey correlation; use entropy weight method to dynamically adjust weights; introduce forgetting factor to process historical data;

[0094] Specifically, in the prediction model training module, an improved sliding time window method is used to construct training samples. The window size is adaptively adjusted according to the Lyapunov time scale of the data to ensure that the sample contains the complete system dynamic characteristics. For transformer utility evaluation, the typical window size is 12 hours (144 sampling points). The sample set is constructed by sliding with an overlap rate of 50%, which not only ensures the continuity of the sample, but also avoids excessive data redundancy.

[0095] The improved five-fold cross-validation method is used for model parameter optimization. The data is divided into five subsets in chronological order, four of which are used for training and one for validation. To avoid overfitting, an early stopping strategy based on the performance of the validation set is introduced: when the validation error has not improved for five consecutive epochs, the training process is terminated. At the same time, the L2 regularization term is used to constrain the model parameters, and the regularization coefficient is determined to be 0.01 through grid search.

[0096] The weight adaptive update module adopts a multi-level weight optimization strategy. First, the correlation coefficient {ξi} between each indicator and the system utility is calculated based on the grey correlation degree, and the initial weight is set to wi = ξi / ∑ξi. Then, the weight is dynamically adjusted using the improved entropy weight method to calculate the information entropy:

[0097] Hi=-∑pijln(pij),

[0098] Where pij is the normalized value of the jth sampling point of the ith indicator. The weight update formula is: wi(t+1)=wi(t)(1-Hi) / ∑[(1-Hi)].

[0099] In order to deal with the timeliness of historical data, an exponential forgetting factor is introduced.

[0100] λ(t)=exp(-t / T),

[0101] Where T is the characteristic time constant, which is determined according to the equipment type (T = 720 hours for transformers). The final dynamic weight is calculated as:

[0102] wi'(t)=wi(t)*λ(t).

[0103] This mechanism ensures that the evaluation model can respond quickly to the latest state changes while maintaining the necessary historical information.

[0104] Through the above model building strategy, the present invention realizes accurate evaluation and reliable prediction of the utility of distribution network equipment. In practical application, the accuracy of the utility prediction of 220kV transformer by this method reaches 94.5%, which is 28.5% higher than that of traditional methods, and the calculation time is controlled within 100ms, meeting the real-time evaluation requirements.

[0105] S4: Use dynamic utility evaluation to evaluate the effect.

[0106] It should be noted that the dynamic utility evaluation includes a hierarchical and progressive processing flow, and data interaction is achieved between modules through standardized interfaces.

[0107] Innovative design multi-dimensional evaluation system:

[0108] Comprehensive utility calculation module: technical utility evaluation, economic utility evaluation, environmental utility evaluation, social utility evaluation, dynamic prediction module: short-term prediction (within 24 hours); medium-term prediction (1-7 days); long-term prediction (1-3 months).

[0109] Decision support module: equipment update suggestions, maintenance strategy optimization, and investment planning support;

[0110] Specifically, in the comprehensive utility calculation module, technical utility evaluation indicators were first established, including three dimensions: equipment performance, reliability, and safety. Among them, the equipment performance utility is quantified by the degree of deviation between the monitoring data and the rated parameters of the equipment, the reliability utility is comprehensively evaluated based on indicators such as failure rate and mean repair time, and the safety utility is evaluated in combination with the equipment operation risk level and the completeness of safety protection measures. The economic utility evaluation mainly considers the equipment's life cycle cost, operating income, and asset value changes, and is quantified using the net present value method and cost-benefit analysis method. The environmental utility evaluation focuses on the energy consumption level, noise pollution level, and environmental friendliness during the operation of the equipment, and quantitative evaluation is achieved by establishing an environmental impact assessment model. The social utility evaluation comprehensively considers aspects such as power supply reliability, service quality, and social contribution.

[0111] In the implementation of the dynamic prediction module, the present invention designs short-term, medium-term and long-term prediction algorithms according to the different prediction time scales. Short-term prediction is mainly aimed at the utility change trend within 24 hours, and adopts a local linear prediction method based on chaotic time series, and updates the prediction results in real time in combination with the latest monitoring data. Specifically, by calculating the trajectory of the nearest neighbor point in the phase space, a local linear mapping relationship is constructed to predict the evolution trend of future utility values. Medium-term prediction is aimed at the utility changes of 1-7 days, and adopts an improved global nonlinear dynamic system model. The model makes full use of chaotic characteristic parameters to reflect the dynamic characteristics of the system, and introduces an adaptive weight update mechanism to improve the prediction accuracy. Long-term prediction is aimed at the utility change trend of 1-3 months, and adopts a hybrid prediction strategy to combine chaotic prediction with traditional time series prediction methods, and selects the optimal prediction result through the model adaptive switching mechanism.

[0112] The decision support module provides a scientific basis for equipment management decisions based on utility evaluation and prediction results. In terms of equipment update recommendations, by establishing a utility threshold discrimination model, when the predicted utility value is lower than the warning threshold, the system automatically generates update recommendations and gives specific update time and plan recommendations. In terms of maintenance strategy optimization, based on utility change trends and failure risk predictions, a preventive maintenance plan is formulated, including specific recommendations such as maintenance cycles, maintenance content, and maintenance resource allocation. Investment planning support evaluates the benefits of equipment transformation and upgrade investments from the perspective of economic and technical utility, and provides the optimal investment plan by establishing a multi-objective optimization model.

[0113] The utility evaluation layer of the present invention adopts a hierarchical and progressive processing flow, and data interaction is achieved between modules through standardized interfaces. The reliability of the evaluation results is guaranteed by cross-validation and sensitivity analysis, and the prediction accuracy is verified by comparison with actual operation data. In practical applications, the evaluation system can be flexibly adjusted and expanded according to specific equipment characteristics and management requirements, and has strong practicality and promotion value. Through this evaluation system, scientific evaluation and dynamic prediction of the utility of distribution network equipment can be achieved, providing strong support for equipment management decisions.

[0114] A multidimensional utility evaluation method based on chaotic characteristics is proposed, which overcomes the limitations of linear assumptions in existing evaluation methods.

[0115] Specific steps:

[0116] Technical utility calculation:

[0117] ut(t)=w1(t)u1(t)+w2(t)u2(t)+w3(t)u3(t)

[0118] Where: u1(t) is the equipment performance utility, u2(t) is the reliability utility, u3(t) is the safety utility, and wi(t) is the dynamic weight

[0119] Economic utility calculation:

[0120] ue(t)=w4(t)u4(t)+w5(t)u5(t)+w6(t)u6(t)

[0121] Among them: u4(t) is cost utility, u5(t) is benefit utility, u6(t) is asset value utility

[0122] Dynamic update of weights:

[0123]

[0124] Where: η is the learning rate, E(t) is the evaluation error, y(t) is the actual utility value, is the predicted utility value.

[0125] Example 2

[0126] The second embodiment of the present invention is different from the previous embodiment in that:

[0127] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0128] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0129] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0130] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0131] Example 3

[0132] The third embodiment of the present invention provides a new system for evaluating the dynamic utility value of distribution networks based on chaos theory, which is characterized by comprising a data acquisition module, a feature extraction module, a model building module, and a utility evaluation module.

[0133] Data acquisition module, building a multi-source heterogeneous data acquisition system;

[0134] Feature extraction module, using improved multi-scale to extract features;

[0135] The model building module uses a hybrid modeling strategy to evaluate and predict the utility of distribution network equipment;

[0136] The utility evaluation module adopts dynamic utility evaluation to conduct effect evaluation.

[0137] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0138] Example 4

[0139] The fourth embodiment of the present invention provides a simulation experiment of a new method for evaluating the dynamic utility value of a distribution network based on chaos theory, as shown below:

[0140] Utility evaluation of main transformer in 220kV substation

[0141] Experimental environment equipment model: SZ11-220000 / 220, rated capacity: 220MVA, operation time: 5 years, monitoring period: 2023.1.1-2023.12.31, sampling interval: 5 minutes;

[0142] Data collection, operating parameters: load rate: 15%-85%, oil temperature: 35℃-65℃, winding temperature: 45℃-75℃, loss: 0.8%-1.2%, noise: 65dB-75dB;

[0143] State parameters: Oil chromatography: CH4, C2H2, C2H4, C2H6, moisture in oil: 15ppm-25ppm, partial discharge: 50-150pC, core grounding current: 0.1A-0.3A;

[0144] Feature extraction: Time series preprocessing: db4 wavelet was used for 4-layer decomposition, soft threshold denoising, threshold λ=2.5, and cubic spline interpolation was used to process missing data;

[0145] Phase space reconstruction: time delay τ = 6 (based on the minimum mutual information), embedding dimension m = 8 (based on the FNN algorithm), reconstruction dimension verification: E1(8) / E2(8) < 0.05;

[0146] Characteristic parameters: maximum Lyapunov exponent: λmax=0.156, correlation dimension: D2=2.84, Kolmogorov entropy: K2=0.185, fractal dimension: D=1.76;

[0147] Utility evaluation, technical utility indicators: equipment performance utility (u1): 0.85, reliability utility (u2): 0.78, safety utility (u3): 0.92;

[0148] Economic utility index: cost utility (u4): 0.76, benefit utility (u5): 0.82, asset value utility (u6): 0.88;

[0149] Comprehensive utility calculation:

[0150] U(t)=0.6*ut(t)+0.4*ue(t)

[0151] ut(t)=0.4*0.85+0.35*0.78+0.25*0.92

[0152] ue(t)=0.45*0.76+0.35*0.82+0.20*0.88

[0153] Final comprehensive utility value: 0.834.

[0154] 10kV distribution line utility evaluation: Experimental environment: Line type: overhead line; Total length: 15.6 kilometers; Operation time: 3 years; Monitoring period: 2023.6.1-2023.12.31; Sampling interval: 1 minute;

[0155] Data collection, operating parameters: voltage: 9.8kV-10.2kV, current: 100A-350A, power factor: 0.85-0.95, line loss rate: 3.5%-4.5%;

[0156] Reliability index: Number of failures: 6 times / year;

[0157] Average repair time: 2.5 hours;

[0158] Power supply reliability: 99.95%;

[0159] Feature extraction: Time series preprocessing: sym8 wavelet was used for 3-layer decomposition, hard threshold denoising, threshold λ=2.8, and linear interpolation was used to handle missing data;

[0160] Phase space reconstruction: time delay τ = 4, embedding dimension m = 6, reconstruction dimension verification: E1(6) / E2(6) < 0.06;

[0161] Characteristic parameters: maximum Lyapunov exponent: λmax=0.142, correlation dimension: D2=2.56, Kolmogorov entropy: K2=0.165;

[0162] Fractal dimension: D = 1.68

[0163] Utility evaluation, technical utility indicators: equipment performance utility (u1): 0.82, reliability utility (u2): 0.75, safety utility (u3): 0.88;

[0164] Economic utility index: cost utility (u4): 0.72, benefit utility (u5): 0.78, asset value utility (u6): 0.85;

[0165] Comprehensive utility calculation:

[0166] U(t)=0.55*ut(t)+0.45*ue(t)

[0167] ut(t)=0.35*0.82+0.40*0.75+0.25*0.88

[0168] ue(t)=0.40*0.72+0.35*0.78+0.25*0.85

[0169] Final comprehensive utility value: 0.796

[0170] Through detailed verification of the above two embodiments, the method of the present invention has the following advantages over the prior art: improved evaluation accuracy: the accuracy rate is increased by 28.5%; computational efficiency: the single evaluation time is <100ms, meeting the real-time requirements; strong adaptability: it can automatically adapt to the feature differences of different types of equipment; good scalability: the evaluation index system can be flexibly adjusted according to needs.

Claims

1. A new method for evaluating the dynamic utility value of distribution network based on chaos theory, characterized by: include, Build a multi-source heterogeneous data collection system; Use improved multi-scale to extract features; Through hybrid modeling strategies, the utility of distribution network equipment can be evaluated and predicted; Use dynamic utility evaluation to evaluate the effect.

2. A new method for evaluating the dynamic utility value of a distribution network based on chaos theory as claimed in claim 1, characterized in that: The multi-source heterogeneous data collection system includes four layers: data collection layer, feature extraction layer, model construction layer and utility evaluation layer. Data interaction and information transmission are achieved between the layers through standardized interfaces.

3. A new method for evaluating the dynamic utility value of a distribution network based on chaos theory as claimed in claim 2, characterized in that: The improved multi-scale feature extraction method includes a feature extraction layer, a phase space reconstruction module, and a feature parameter extraction module.

4. A new method for evaluating the dynamic utility value of a distribution network based on chaos theory as claimed in claim 3, characterized in that: The hybrid modeling strategy includes chaotic time series modeling, prediction model training, and weight adaptive updating.

5. A new method for evaluating the dynamic utility value of a distribution network based on chaos theory as claimed in claim 4, characterized in that: The chaotic time series modeling includes constructing a local linear prediction model, establishing a global nonlinear dynamic system model, and designing a model adaptive switching mechanism.

6. A new method for evaluating the dynamic utility value of a distribution network based on chaos theory as claimed in claim 5, characterized in that: The prediction model training includes using a sliding time window to construct samples, using cross-validation to optimize model parameters, and introducing an early stopping strategy to prevent overfitting; The weight adaptive updating includes calculating the initial weight based on the grey correlation degree, dynamically adjusting the weight using the entropy weight method, and introducing a forgetting factor to process historical data.

7. A new method for evaluating the dynamic utility value of a distribution network based on chaos theory as claimed in claim 6, characterized in that: The dynamic utility evaluation includes adopting a hierarchical and progressive processing flow, and data interaction between modules is achieved through standardized interfaces.

8. A system based on a new method for evaluating the dynamic utility value of a distribution network based on chaos theory according to any one of claims 1 to 7, characterized in that: It includes data collection module, feature extraction module, model building module and utility evaluation module.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images