A terminal meter reliability dynamic evaluation method, system, device and medium based on multi-source data features
By integrating features from multiple data sources and applying Bayesian algorithms, the reliability of metering terminals is dynamically evaluated, overcoming the shortcomings of existing evaluation methods and enabling comprehensive reliability assessment and fault risk identification of terminal meters.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- YUNNAN POWER GRID CO LTD
- Filing Date
- 2026-01-21
- Publication Date
- 2026-05-12
AI Technical Summary
In the context of high penetration of new energy sources and complex environments, existing technologies fail to effectively integrate multi-source data characteristics in the reliability assessment methods of metering terminals, leading to frequent equipment failures. The assessment indicators cannot dynamically reflect changes in risk factors, resulting in wasted operation and maintenance resources and safety hazards.
By acquiring multi-source data from metering terminals and integrating it into a structured feature matrix, a hybrid feature screening mechanism and Bayesian algorithm are used for feature screening and weight allocation to construct a comprehensive reliability scoring model. Risk thresholds are dynamically set according to environmental parameters to achieve dynamic reliability assessment of terminal meters.
It enables dynamic reflection of the reliability assessment of terminal meters in different scenarios, identifies potential failure risks, improves the pertinence and security of operation and maintenance, and avoids waste of resources.
Smart Images
Figure CN121561539B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of terminal meter failure prediction technology, and in particular to a method, system, device and medium for dynamic evaluation of terminal meter reliability based on multi-source data characteristics. Background Technology
[0002] As core equipment of the smart grid, metering terminals undertake key functions such as electricity metering, data acquisition, and load control. Their operational reliability directly affects the operating efficiency of the power system, the quality of user service, and the economic benefits of power companies. However, the operational reliability of metering terminals is becoming increasingly prominent, especially in scenarios with high penetration of new energy sources and complex environments (such as coastal areas with high salt spray and industrial pollution). Frequent equipment failures lead to increased metering errors, data loss, and even safety accidents, seriously affecting the stable operation of the power system and the user's electricity experience. Existing research relies on partial electricity data to assess the operating status of electricity metering devices, without integrating multi-source data (equipment status, environmental factors, etc.), and the assessment indicators need further updates. Another type of assessment method establishes a more comprehensive indicator system, but the method only evaluates the operating status of electricity meters and the indicator weights are fixed, failing to dynamically reflect changes in risk factors under different scenarios. It lacks a comprehensive reliability assessment method for metering terminals, leading to wasted maintenance resources and unnecessary safety hazards. Considering the above problems, this invention proposes a dynamic reliability assessment method for terminal meters based on multi-source data characteristics. Summary of the Invention
[0003] In view of the aforementioned existing problems, the present invention is proposed.
[0004] Therefore, the technical problem solved by this invention is: how to achieve dynamic reliability assessment of metering terminal meters by integrating multi-source data features in scenarios with high penetration of new energy and complex environments, and at the same time, conduct risk factor change analysis in different scenarios based on the dynamic adjustment mechanism of weights to identify potential failure risks and weak links in operation and maintenance.
[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a dynamic reliability assessment method for terminal meters based on multi-source data characteristics, comprising,
[0006] Acquire multi-source data from metering terminals, integrate the multi-source data to form a structured feature matrix, and preprocess the structured feature matrix;
[0007] A hybrid feature filtering mechanism is introduced to perform hybrid feature filtering on the preprocessed structured feature matrix to obtain the key feature set of the terminal meter;
[0008] A Bayesian algorithm is used to assign weights to the key feature set of terminal meters to obtain a comprehensive weight.
[0009] A comprehensive reliability scoring model is constructed based on the comprehensive weights, and the comprehensive reliability score of the terminal meter is output.
[0010] The risk threshold is dynamically set based on environmental parameters, and the overall risk level of the terminal meter is output.
[0011] As a preferred embodiment of the dynamic reliability assessment method for terminal meters based on multi-source data features described in this invention, the step of acquiring multi-source data from the metering terminal and integrating the multi-source data to form a structured feature matrix includes:
[0012] Sensors are installed in the metering terminal to detect environmental features, static features are obtained by searching historical files, and dynamic features are obtained based on the feedback from the equipment.
[0013] All data is aligned by device ID and timestamp to generate a structured feature matrix. ,in Indicates the time dimension. For the number of devices, For the number of features, It is the field of real numbers.
[0014] As a preferred embodiment of the dynamic reliability assessment method for terminal meters based on multi-source data features described in this invention, the preprocessing of the structured feature matrix includes:
[0015] For outliers in the structured feature matrix, the sliding window mean method is used to suppress burst noise, as expressed in the following expression:
[0016] ,
[0017] in, for The smoothing value at time, for The original data value at that moment, This represents a temporary index variable used for summation during iteration within the sliding window.
[0018] As a preferred embodiment of the dynamic reliability assessment method for terminal meters based on multi-source data features described in this invention, the method involves introducing a hybrid feature filtering mechanism to perform hybrid feature filtering on the preprocessed structured feature matrix to obtain a key feature set for the terminal meters, including:
[0019] Hybrid feature filtering includes key static feature filtering and key dynamic feature filtering;
[0020] Key static features are filtered by constructing a random forest model, expressed as follows:
[0021] ,
[0022] in, Static features that indicate the importance to be evaluated Representing static features The consequences of splitting a single decision tree node The decrease in coefficient, Representing static features In all decision trees The sum of the decreases in coefficients This represents all static features in all decision trees. The total decrease in coefficients Representing static features The global importance score;
[0023] The correlation between dynamic characteristics and fault states is quantified using the mutual information method, and key dynamic characteristics are selected. The expression is:
[0024] ,
[0025] in, For the dynamic characteristics to be evaluated, This is a fault condition. For dynamic features The marginal probability distribution, i.e., the dynamic characteristics without considering fault states. Values The probability, Fault status The marginal probability distribution, i.e., the fault state without considering dynamic characteristic values. Values The prior probability, For dynamic features Fault status The probability of them happening simultaneously For dynamic features Fault status The probability of occurring independently, For dynamic features Fault status Mutual information value;
[0026] The key feature set of the terminal meter is obtained by integrating key static features and key dynamic features.
[0027] As a preferred embodiment of the dynamic reliability assessment method for terminal meters based on multi-source data features described in this invention, the step of using a Bayesian algorithm to assign weights to the key feature set of the terminal meters to obtain a comprehensive weight includes:
[0028] Input the key feature set of the terminal meter and construct a Bayesian linear regression model, the expression of which is:
[0029] ,
[0030] in, To score the reliability of the terminal, For the first key feature set The weights of key features, including key static features and key dynamic features. For the first key feature set The values of the key features, For noise terms, For noise variance, This represents the total number of key features in the key feature set. Indicates a normal distribution. Index of key features in the key feature set;
[0031] The Markov chain Monte Carlo method was used to sample 1000 times to obtain the mean of the posterior distribution of the key feature weights as the baseline weights.
[0032] Based on the baseline weights, a dynamic adjustment factor under the influence of photovoltaic scenarios is introduced to generate a comprehensive weight, expressed as follows:
[0033] ,
[0034] in, As a dynamic adjustment factor, For the first key feature set Key features The overall weight of time, for Photovoltaic volatility at any given time For the first key feature set The baseline weights of the key features, For the first key feature set Key features Adjustment weights at different times.
[0035] As a preferred embodiment of the dynamic reliability assessment method for terminal meters based on multi-source data features described in this invention, the step of constructing a comprehensive reliability scoring model according to comprehensive weights and outputting the comprehensive reliability score of the terminal meters includes:
[0036] The comprehensive reliability scoring model consists of the sum of weighted characteristic functions, expressed as:
[0037] ,
[0038] in, The overall reliability score for terminal meters, This represents the feature standardization function.
[0039] As a preferred embodiment of the dynamic reliability assessment method for terminal meters based on multi-source data features described in this invention, the step of dynamically setting a risk threshold based on environmental parameters and outputting the comprehensive risk level of the terminal meter includes:
[0040] The risk threshold expression is:
[0041] ,
[0042] in, express Risk threshold at any time An index for environmental risk conditions. This represents the total number of environmental risk condition types. For conditional triggering indicator functions, This is an adjustment factor for environmental risk conditions.
[0043] This invention provides a dynamic reliability evaluation system for terminal meters based on multi-source data characteristics.
[0044] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a dynamic evaluation system for the reliability of terminal meters based on multi-source data features, comprising: a data acquisition and preprocessing module, a feature filtering module, a weight allocation module, an output module, and a risk level determination module;
[0045] The acquisition and preprocessing module acquires multi-source data from the metering terminal, integrates the multi-source data to form a structured feature matrix, and preprocesses the structured feature matrix.
[0046] The feature filtering module introduces a hybrid feature filtering mechanism to perform hybrid feature filtering on the preprocessed structured feature matrix to obtain the key feature set of the terminal meter.
[0047] The weight allocation module uses a Bayesian algorithm to allocate weights to the key feature set of the terminal meter to obtain a comprehensive weight.
[0048] The output module constructs a comprehensive reliability scoring model based on comprehensive weights and outputs the comprehensive reliability score of the terminal meter.
[0049] The risk level determination module dynamically sets risk thresholds based on environmental parameters and outputs the comprehensive risk level of the terminal meter.
[0050] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the method for dynamic evaluation of terminal meter reliability based on multi-source data features.
[0051] The present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of the method for dynamic evaluation of terminal meter reliability based on multi-source data characteristics.
[0052] The beneficial effects of this invention are as follows: This invention acquires multi-source data from metering terminals, integrates the multi-source data to form a structured feature matrix, and integrates comprehensive data dimensions, thus solving the problem of single data dimensions. By introducing a dynamic and static feature selection mechanism on the preprocessed structured feature matrix, the accuracy of feature selection is improved. By introducing a Bayesian algorithm to assign weights to the selected structured feature matrix and obtain comprehensive weights, the invention dynamically reflects the changes in the importance of risk factors under different scenarios, achieving the technical effect of adaptive adjustment of weights according to environmental fluctuations. A reliability scoring model is constructed through comprehensive weights, and risk thresholds are dynamically set in combination with environmental parameters, ultimately outputting the comprehensive risk level of the terminal meter. Attached Figure Description
[0053] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0054] Figure 1 This is a schematic diagram of the overall process of a dynamic evaluation method for the reliability of terminal meters based on multi-source data features, as described in one embodiment of the present invention. Detailed Implementation
[0055] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.
[0056] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for dynamic evaluation of terminal meter reliability based on multi-source data features is provided, comprising:
[0057] S1: Acquire multi-source data from metering terminals, integrate the multi-source data to form a structured feature matrix, and preprocess the structured feature matrix.
[0058] It should be noted that obtaining multi-source data from the metering terminal includes: installing sensors in the metering terminal to detect environmental features, searching historical files to obtain static features, obtaining dynamic features based on device feedback, aligning all data by device ID and timestamp, and generating a structured feature matrix. ,in Indicates the time dimension. For the number of devices, For the number of features, It is the field of real numbers.
[0059] It should be noted that static characteristics include, but are not limited to: equipment batch and enclosure corrosion level; dynamic characteristics include, but are not limited to: voltage over-limit duration and power flow reversal frequency; and environmental characteristics include, but are not limited to: temperature, humidity, and salt spray concentration.
[0060] Furthermore, the preprocessing of the structured feature matrix includes using a sliding window mean method to suppress burst noise for outliers in the structured feature matrix. The window length is set to 1 hour (6 sampling points, step size 10 minutes), and the smoothing formula is as follows:
[0061] ,
[0062] in, for The smoothing value at time, for The original data value at that moment, This represents a temporary index variable used for summation within the sliding window, with a value range from... arrive There are a total of 6 index values, corresponding to The time and the five consecutive sampling points prior to that time.
[0063] It should be noted that constructing a structured feature matrix integrates multi-source data, combines static features, dynamic features, and environmental parameters, and stores them in a structured manner to achieve comprehensive data coverage and avoid the impact of a single data dimension on the detection results.
[0064] S2: Introduce a hybrid feature filtering mechanism to perform hybrid feature filtering on the preprocessed structured feature matrix to obtain the key feature set of the terminal meter.
[0065] Hybrid feature screening includes key static feature screening and key dynamic feature screening.
[0066] It should be noted that key static features are selected by constructing a random forest model, the expression of which is:
[0067] ,
[0068] in, Static features that indicate the importance to be evaluated Representing static features The consequences of splitting a single decision tree node The decrease in coefficient, Representing static features In all decision trees The sum of the decreases in coefficients This represents all static features in all decision trees. The total decrease in coefficients Representing static features The global importance score.
[0069] Furthermore, a random forest model is used to select key static features. A random forest model containing 100 decision trees is trained, and the ranking of each static feature across all trees is statistically analyzed. The total reduction in coefficients is used to calculate the global importance score of each static feature. A global importance threshold of 0.1 is set. Static features with a global importance score below 0.1 typically contribute little and are unstable to the model's predictions. Static features with a global importance score greater than or equal to 0.1 are selected as key static features. The global importance threshold is set empirically.
[0070] Furthermore, the correlation between dynamic characteristics and fault states is quantified using the mutual information method to identify key dynamic characteristics, expressed as:
[0071] ,
[0072] in, For the dynamic characteristics to be evaluated, This is a fault condition. For dynamic features The marginal probability distribution, i.e., the dynamic characteristics without considering fault states. Values The probability, Fault status The marginal probability distribution, i.e., the fault state without considering dynamic characteristic values. Values The prior probability, For dynamic features Fault status The probability of them happening simultaneously For dynamic features Fault status The probability of occurring independently, For dynamic features Fault status The mutual information value.
[0073] Furthermore, the mutual information method is used to screen key dynamic features, and the dynamic features and fault states are discretized. The joint probability distribution is statistically analyzed, and the mutual information value of each dynamic feature is calculated. A mutual information threshold is set, which is determined empirically. The mutual information values of most weakly correlated or non-contributing dynamic features are concentrated below 0.05, while the mutual information values of key dynamic features are mostly distributed between 0.1 and 0.5. Dynamic features with mutual information values greater than or equal to 0.1 and less than or equal to 0.5 are selected as key dynamic features.
[0074] Furthermore, by integrating key static features and key dynamic features, a key feature set for the terminal meter is obtained.
[0075] It should be noted that this invention uses dynamic features and fault states as variables, calculates their joint probability distribution, substitutes it into the mutual information formula, sets a mutual information threshold to screen out key dynamic features, and improves the reliability assessment model's ability to capture dynamic operational risks of terminal meters.
[0076] S3: The Bayesian algorithm is used to assign weights to the key feature set of the terminal meter to obtain the comprehensive weight.
[0077] It should be noted that, in order to establish the mapping relationship between the key feature set of the terminal meter and the reliability score of the terminal meter, a Bayesian linear regression model is constructed, with the expression as follows:
[0078] ,
[0079] in, To score the reliability of the terminal, For the first key feature set The weights of key features, including key static features and key dynamic features. For the first key feature set The values of the key features, For noise terms, For noise variance, This represents the total number of key features in the key feature set. Indicates a normal distribution. This is the index of key features in the key feature set.
[0080] It should be noted that the weight of each key feature in the key feature set includes a baseline weight and an adjustment weight. The baseline weight represents a fixed weight component learned from historical data, reflecting the importance of that feature to the reliability baseline, and it does not change over time. The adjustment weight refers to the weight affected by photovoltaic volatility. The adjusted weight components change over time.
[0081] Furthermore, the key features are concentrated in the first... Weights of key features Follows prior distribution The mean of the posterior distribution of the key feature weights is obtained by sampling 1000 times using the Markov chain Monte Carlo method, and used as the baseline weight of the key feature.
[0082] Based on the baseline weights, an adjustment factor under the influence of photovoltaic scenarios is introduced to generate a comprehensive weight, expressed as follows:
[0083] ,
[0084] in, As a dynamic adjustment factor, For the first key feature set Key features The overall weight of time, for Photovoltaic volatility at time, mixing coefficient and These represent the contribution ratios of the baseline weight and the adjustment weight to the overall weight, respectively, and are set empirically. This is an empirical adjustment coefficient used to normalize the physical quantity value of photovoltaic volatility to a reasonable weighting adjustment range. For the first key feature set The baseline weights of the key features, For the first key feature set Key features Adjustment weights at different times.
[0085] It should be noted that the adjustment weight is obtained through the average rate of change of the key feature. For a key feature in the set of key features, the past hour is set as a fixed time window. The mean rate of change of this key feature within the fixed time window is calculated, and the mean rate of change is normalized to a value range of 0 to 1, which is used as the adjustment weight of this key feature at the current time.
[0086] It should be noted that by constructing a Bayesian linear regression model and combining the baseline weights and adjustment weights to generate comprehensive weights, the adaptive dynamic adjustment of the key feature weights is achieved, which solves the problems of fixed indicator weights and inability to reflect changes in the importance of risk factors under different scenarios in traditional assessment methods.
[0087] S4: Construct a comprehensive reliability scoring model based on the comprehensive weights and output the comprehensive reliability score of the terminal meter.
[0088] It should be noted that the expression for the comprehensive reliability scoring model is as follows:
[0089] ,
[0090] in, The overall reliability score for terminal meters, This represents the feature standardization function.
[0091] Furthermore, the feature standardization function in the reliability scoring model is defined. The specific form of the standardization function is as follows: different standardization functions are selected for different feature types, aiming to transform the values of key features in the key feature set into a unified, unitless standardized score to accurately reflect the nonlinear relationship with fault risk.
[0092] Different feature types include cumulative features and event-based features.
[0093] Cumulative characteristics include, but are not limited to: voltage over-limit duration and corrosion level; event-type characteristics include, but are not limited to: number of power flow reversals.
[0094] The cumulative features are standardized using an exponential function, expressed as follows:
[0095] ,
[0096] Exponential functions map eigenvalues to Interval normalization is achieved.
[0097] By standardizing event-type features using a linear function, the expression is:
[0098] ,
[0099] Event-driven characteristics are typically approximately directly proportional to risk. Each event (such as a current reversal) is considered an independent unit of risk. A linear function can reflect the cumulative contribution of event frequency to total risk. (Coefficient) Based on experience, a baseline contribution weight for event-type features in the total score is set, which also serves to scale the original values to an appropriate range. Overall, this provides real-time, adaptive prediction results for terminal meter reliability assessment.
[0100] It should be noted that the characteristic standardization function maps characteristic values of different dimensions and ranges (such as corrosion level, voltage over-limit duration, and salt spray concentration) to a unified numerical range so as to perform weighted summation.
[0101] S5: Dynamically set risk thresholds based on environmental parameters and output the overall risk level of terminal meters.
[0102] It should be noted that the risk threshold is dynamically set based on environmental parameters, and the expression is:
[0103] ,
[0104] in, express Risk threshold at any time An index for environmental risk conditions. Each index in the database corresponds to a specific environmental risk condition (such as high temperature or high salt spray). This represents the total number of environmental risk condition types. This is a conditional trigger indicator function (1 if the condition is met, 0 otherwise). This is an adjustment factor for environmental risk conditions, set based on experience. This is an empirical constant, representing the baseline risk line under no special environmental pressures.
[0105] Furthermore, based on the obtained comprehensive reliability score of the faulty electricity meter, combined with the risk threshold, the comprehensive risk level of the terminal meter is output, as shown in Table 1:
[0106] Table 1. Comprehensive Risk Level Assessment Table for Terminal Meters
[0107]
[0108] It should be noted that the comprehensive reliability scoring model and risk threshold setting are the core of the entire evaluation process. The comprehensive reliability scoring model is used to calculate and combine the risk threshold to output the comprehensive risk level of the terminal meter. This comprehensive risk level integrates static factors, dynamic factors and environmental factors, and can reflect the reliability and lifespan of the terminal meter.
[0109] Example 2 is an embodiment of the present invention. This embodiment provides a dynamic evaluation system for the reliability of terminal meters based on multi-source data features, including: a data acquisition and preprocessing module, a feature filtering module, a weight allocation module, an output module, and a risk level determination module.
[0110] The acquisition and preprocessing module acquires multi-source data from the metering terminal, integrates the multi-source data to form a structured feature matrix, and preprocesses the structured feature matrix.
[0111] The feature filtering module introduces a hybrid feature filtering mechanism to perform hybrid feature filtering on the preprocessed structured feature matrix to obtain the key feature set of the terminal meter.
[0112] The weight allocation module uses a Bayesian algorithm to allocate weights to the key feature set of the terminal meter to obtain a comprehensive weight.
[0113] The output module constructs a comprehensive reliability scoring model based on comprehensive weights and outputs the comprehensive reliability score of the terminal meter.
[0114] The risk level determination module dynamically sets risk thresholds based on environmental parameters and outputs the comprehensive risk level of the terminal meter.
[0115] This embodiment also provides an electronic device applicable to a dynamic reliability evaluation method for terminal meters based on multi-source data features, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the dynamic reliability evaluation method for terminal meters based on multi-source data features proposed in the above embodiment.
[0116] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a dynamic evaluation method for the reliability of terminal meters based on multi-source data characteristics as proposed in the above embodiments.
[0117] The storage medium proposed in this embodiment belongs to the same inventive concept as the method for dynamic evaluation of terminal meter reliability based on multi-source data characteristics proposed in the above embodiments. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.
[0118] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.
[0119] 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A method for dynamic evaluation of terminal meter reliability based on multi-source data features, characterized in that: include, Acquire multi-source data from metering terminals, integrate the multi-source data to form a structured feature matrix, and preprocess the structured feature matrix; A hybrid feature filtering mechanism is introduced to perform hybrid feature filtering on the preprocessed structured feature matrix to obtain the key feature set of the terminal meter; A Bayesian algorithm is used to assign weights to the key feature set of terminal meters to obtain a comprehensive weight. A comprehensive reliability scoring model is constructed based on the comprehensive weights, and the comprehensive reliability score of the terminal meter is output. The risk threshold is dynamically set based on environmental parameters, and the overall risk level of the terminal meter is output. The introduced hybrid feature filtering mechanism performs hybrid feature filtering on the preprocessed structured feature matrix to obtain the key feature set of the terminal meter, including: Hybrid feature filtering includes key static feature filtering and key dynamic feature filtering; Key static features are filtered by constructing a random forest model, expressed as follows: ， in, Static features that indicate the importance to be evaluated Representing static features The reduction in Gini coefficient when a single decision tree node splits. Representing static features The sum of the reductions in Gini coefficients across all decision trees. This represents the sum of the reductions in Gini coefficients for all static features across all decision trees. Representing static features The global importance score; The correlation between dynamic characteristics and fault states is quantified using the mutual information method, and key dynamic characteristics are selected. The expression is: ， in, For the dynamic characteristics to be evaluated, Let p(x) represent the fault state, p(x) be the marginal probability distribution of the dynamic feature X (i.e., the probability that the dynamic feature X takes the value x without considering the fault state), and p(y) be the marginal probability distribution of the fault Y (i.e., the prior probability that the fault state Y takes the value y without considering the dynamic feature value). Let X be the probability that dynamic feature X and fault state Y occur simultaneously. Let X be the probability that the dynamic feature X and the fault state Y occur independently. The mutual information value between dynamic feature X and fault state Y; By integrating key static features and key dynamic features, a key feature set for terminal meters is obtained. The method employs a Bayesian algorithm to assign weights to the key feature set of the terminal meter, resulting in a comprehensive weight, including: Input the key feature set of the terminal meter and construct a Bayesian linear regression model, the expression of which is: ， Where y is the terminal reliability score. This represents the weight of the i-th key feature in the key feature set. Key features include key static features and key dynamic features. Let be the value of the i-th key feature in the key feature set. For noise terms, For noise variance, This represents the total number of key features in the key feature set. represents a normal distribution, and i is the index of the key feature in the key feature set; The Markov chain Monte Carlo method was used to sample 1000 times to obtain the mean of the posterior distribution of the key feature weights as the baseline weights. Based on the baseline weights, a dynamic adjustment factor under the influence of photovoltaic scenarios is introduced to generate a comprehensive weight, expressed as follows: ， ， Where β(t) is the dynamic adjustment factor, The comprehensive weight of the i-th key feature in the key feature set at time t. Let be the photovoltaic volatility at time t. Let i be the baseline weight of the i-th key feature in the key feature set. Let be the adjustment weight of the i-th key feature in the key feature set at time t.
2. The method for dynamic evaluation of terminal meter reliability based on multi-source data features as described in claim 1, characterized in that: The process of acquiring multi-source data from metering terminals and integrating the multi-source data to form a structured feature matrix includes: Sensors are installed in the metering terminal to detect environmental features, static features are obtained by searching historical files, and dynamic features are obtained based on the feedback from the equipment. All data is aligned by device ID and timestamp to generate a structured feature matrix. ,in Indicates the time dimension. For the number of devices, For the number of features, It is the field of real numbers.
3. The method for dynamic evaluation of terminal meter reliability based on multi-source data characteristics as described in claim 2, characterized in that: The preprocessing of the structured feature matrix includes: For outliers in the structured feature matrix, the sliding window mean method is used to suppress burst noise, as expressed in the following expression: ， in, for The smoothing value at time, For time points The original data values, This represents a temporary index variable used for summation during iteration within the sliding window.
4. The method for dynamic evaluation of terminal meter reliability based on multi-source data features as described in claim 3, characterized in that: The process of constructing a comprehensive reliability scoring model based on comprehensive weights and outputting a comprehensive reliability score for the terminal meter includes: The comprehensive reliability scoring model consists of the sum of weighted characteristic functions, expressed as: ， Where R is the overall reliability score of the terminal meter. This represents the feature standardization function.
5. The method for dynamic evaluation of terminal meter reliability based on multi-source data features as described in claim 4, characterized in that: The process of dynamically setting risk thresholds based on environmental parameters and outputting the comprehensive risk level of terminal meters includes: The risk threshold expression is: ， in, express Risk threshold at any time An index for environmental risk conditions. This represents the total number of environmental risk condition types. For conditional triggering indicator functions, This is an adjustment factor for environmental risk conditions.
6. A dynamic reliability assessment system for terminal meters based on multi-source data features, employing the dynamic reliability assessment method for terminal meters based on multi-source data features as described in any one of claims 1-5, characterized in that, include: The system includes a data acquisition and preprocessing module, a feature filtering module, a weight allocation module, an output module, and a risk level determination module. The acquisition and preprocessing module acquires multi-source data from the metering terminal, integrates the multi-source data to form a structured feature matrix, and preprocesses the structured feature matrix. The feature filtering module introduces a hybrid feature filtering mechanism to perform hybrid feature filtering on the preprocessed structured feature matrix to obtain the key feature set of the terminal meter. The weight allocation module uses a Bayesian algorithm to allocate weights to the key feature set of the terminal meter to obtain a comprehensive weight. The output module constructs a comprehensive reliability scoring model based on comprehensive weights and outputs the comprehensive reliability score of the terminal meter. The risk level determination module dynamically sets risk thresholds based on environmental parameters and outputs the comprehensive risk level of the terminal meter.
7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method for dynamic evaluation of terminal meter reliability based on multi-source data characteristics as described in any one of claims 1 to 5.
8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for dynamic evaluation of terminal meter reliability based on multi-source data characteristics as described in any one of claims 1 to 5.