A method and system for predicting the number of electric meter failures

Abstract

The present invention combines on-site fault data to predict the number of faults in future time periods. By introducing the advantage ratio, the accuracy of the prediction is improved, which is conducive to solving the problem of early prediction of electricity meter faults, thereby pre-discovering the possible trend of changes in the number of faults in the future intervals of smart electricity meters in the same batch, and improving the power grid company's ability to predict the life of smart electricity meters, risk assessment, and management decision-making throughout the life cycle. Pre-discovering the possible trend of changes in the number of faults in the future intervals of smart electricity meters in the same batch can provide technical support for the operation and maintenance of electricity meters in the future time period, batch rotation, and hidden danger investigation, and improve the power grid company's ability to predict the life of smart electricity meters, risk assessment, and management decision-making throughout the life cycle.

Description

Technical Field

[0001] The present invention relates to the field of equipment life prediction, and in particular to a method and system for predicting the number of failures of an electric meter. Background Art

[0002] Current research and analysis schemes for electric meters based on historical failure data focus on reliability evaluation and lifespan prediction, with little focus on predicting the number of meter failures in future time periods. Furthermore, in frequency prediction schemes, the focus is on using prediction methods such as neural networks, time series, and gray models. However, these methods are not targeted at predicting the number of meter failures in future time periods and cannot effectively solve the practical engineering problem of pre-judging the number of meter failures, thereby pre-discovering the possible trend of the number of failures in the future intervals of batches of smart electric energy meters. They cannot effectively provide technical support for improving the power grid company's ability to manage and control the operation and maintenance of smart meters, identify potential hazards, and make reserve decisions. Therefore, how to design a method and system that can predict the number of meter failures in future time periods has become an urgent problem in this field. Summary of the Invention

[0003] The purpose of the present invention is to provide a method and system for predicting the number of meter failures. The present invention combines on-site meter failure data to predict the number of failures in a future time period. By introducing the advantage ratio, the accuracy of the prediction is improved, which is conducive to solving the problem of early prediction of meter failures, thereby discovering in advance the possible trend of changes in the number of failures of the meter in the future interval, and improving the power grid company's management and control capabilities for the operation and maintenance of smart meters, hidden danger detection and risk decision-making.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] A method for predicting the number of electric meter failures, the method comprising:

[0006] Obtaining on-site timed truncated fault data for electric meters; the timed truncated fault data includes the installation time of each batch of electric meters, the number of electric meters installed in each batch, the observation time, the number of electric meter failures during the observation time, the predicted time, and the time corresponding to the failure of each electric meter, wherein the starting time of the predicted time is the ending time of the observation time;

[0007] Constructing a Weibull distribution life model of the electric meter based on the timed truncated fault data, and estimating parameters of the Weibull distribution life model of the electric meter using maximum likelihood;

[0008] Determine whether to predict the failure data of the uninstalled electricity meter based on the installed electricity meter;

[0009] If not, then select batches of electricity meters within the service life within the prediction time, perform single-batch dynamic single-sample prediction on a certain batch of electricity meters, realize multi-batch single-sample prediction, and obtain single-sample prediction results;

[0010] If so, the model parameters of each batch of electricity meters installed within the prediction time are calculated, and a single-batch double-sample prediction is performed for a certain batch of electricity meters to achieve multi-batch double-sample prediction and obtain the double-sample prediction results;

[0011] According to the single-sample prediction result and the dual-sample prediction result, the total predicted number of failures within the prediction time and the total prediction interval are obtained.

[0012] The present invention also provides a system for predicting the number of electric meter failures, the system comprising:

[0013] a timed truncated fault data acquisition unit, configured to acquire timed truncated fault data of electric meters on site; the timed truncated fault data includes the installation time of each batch of electric meters, the number of electric meters installed in each batch, the observation time, the number of electric meter failures during the observation time, the predicted time, and the time corresponding to the occurrence of a failure of each electric meter, wherein the starting time of the predicted time is the ending time of the observation time;

[0014] A life model construction unit is used to construct a Weibull distribution life model of the electricity meter based on the timed truncated fault data, and estimate the parameters of the Weibull distribution life model of the electricity meter using maximum likelihood;

[0015] A judgment unit, configured to judge whether to predict fault data of an uninstalled electric meter based on an installed electric meter;

[0016] If not, then select batches of electricity meters within the service life within the prediction time, perform single-batch dynamic single-sample prediction on a certain batch of electricity meters, realize multi-batch single-sample prediction, and obtain single-sample prediction results;

[0017] If so, the model parameters of each batch of electricity meters installed within the prediction time are calculated, and a single-batch double-sample prediction is performed for a certain batch of electricity meters to achieve multi-batch double-sample prediction and obtain the double-sample prediction results;

[0018] The total prediction unit is used to obtain the total predicted number of failures and the total prediction interval within the prediction time based on the single sample prediction results and the double sample prediction results.

[0019] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0020] The meter failure frequency prediction method and system provided by the present invention can discover in advance the possible changing trend of the number of failures of smart meters in future time periods, and can provide technical support for the operation and maintenance, batch rotation and hidden danger investigation of meters in future time periods, thereby improving the power grid company's full life cycle management and control capabilities for smart meters in terms of life prediction, risk assessment and management decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 This is a flow chart of a method for predicting the number of electric meter failures provided in Example 1 of the present invention;

[0023] Figure 2 A diagram showing the structure of the fault data of a batch of electric meters that has been periodically truncated;

[0024] Figure 3 A structural diagram for predicting electric meter failures;

[0025] Figure 4 This is a structural diagram of an electric meter failure frequency prediction system provided by Example 2 of the present invention.

[0026] Explanation of symbols:

[0027] 1. Timed truncation fault data acquisition unit; 2. Life model construction unit; 3. Judgment unit; 4. Overall prediction unit. DETAILED DESCRIPTION

[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0029] The purpose of the present invention is to provide a method and system for predicting the number of meter failures. By combining the present invention with on-site failure data, the number of failures in a future time period is predicted. By introducing the advantage ratio, the accuracy of the prediction is improved, which is conducive to solving the problem of early prediction of meter failures, thereby discovering in advance the possible trend of changes in the number of failures in the future intervals of the same batch of smart meters, and improving the power grid company's full life cycle management and control capabilities for smart meters in terms of life prediction, risk assessment, and management decision-making.

[0030] The following are the explanations of the professional terms related to the present invention:

[0031] Weibull distribution: The Weibull distribution, also known as the Weibull distribution or Weibull distribution, is the theoretical foundation of reliability analysis and life testing. It is widely used in reliability engineering, particularly for the distribution of wear-induced cumulative failures in electromechanical products. Because its distribution parameters can be easily inferred from probability values, it is widely used in data processing for various life testing applications.

[0032] Single sample prediction, also known as within-sample prediction, is to give future observations of the same sample.

[0033] Double sample prediction, also known as new sample prediction, is to give the observation value of future samples based on past samples.

[0034] If the past samples or future samples in the new sample prediction contain multiple samples, it is called diverse prediction.

[0035] Odds ratio: The odds ratio (OR) is another way to describe probability. It tells us how much greater the probability of a hypothesis is than the opposite probability. In other words, the odds ratio is the ratio of the probability of a hypothesis being true to the probability of a hypothesis being false.

[0036] Timed truncated data: After the meter is put into operation, the meter fault data is collected immediately. The observation ends at time t. The meter fault data after time t cannot be obtained. This type of data is called timed truncated data.

[0037] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0038] Example 1:

[0039] Based on the above explanation, please refer to Figure 1 The present invention provides a method for predicting the number of electric meter failures, the method comprising:

[0040] S1: Obtaining on-site timed truncated fault data of electric meters; the timed truncated fault data includes the installation time of each batch of electric meters, the number of electric meters installed in each batch, the observation time, the number of electric meter failures during the observation time, the predicted time, and the time corresponding to the occurrence of each electric meter failure, wherein the starting time of the predicted time is the ending time of the observation time;

[0041] In practical applications, such as Figure 2 As shown, assuming that a batch of electric meters at time t z Sometimes there is N zThe meter was put into use and at the observation time (t z ,t s ]H z The time when each meter fails is t H1 , t H2 ,…,t Hz , if the prediction time t is given s , t e , in (t s ,t e ]The number of faults in the interval is J z In order to maintain the normal operation of the equipment, it is necessary to reserve J z This component.

[0042] S2: constructing a Weibull distribution life model of the electricity meter based on the timed truncated fault data, and estimating parameters of the Weibull distribution life model of the electricity meter using maximum likelihood;

[0043] In step S2, constructing a Weibull distribution life model of an electric meter based on the timed truncated fault data specifically includes:

[0044] S21: Determine the cumulative failure probability of the electric meter during the observation time based on the timed truncated failure data;

[0045] S22: Constructing the Weibull distribution life model of the electricity meter according to the cumulative failure probability of the electricity meter within the observation time. The Weibull distribution life model of the electricity meter is:

[0046]

[0047] Among them, h z is the cumulative failure probability of the electric meter within a certain batch of observation time, t zs is the observation time, η z is the scale parameter, m z is the shape parameter.

[0048] At the same time, the parameters of the Weibull distribution life model of the electric meter obtained by maximum likelihood estimation are:

[0049]

[0050]

[0051] Among them, η z is the scale parameter, m z is the shape parameter, t i is the time when the ith meter of the same installation time fails, N z is the total number of electric meters put into use during the observation period, H zis the number of meter failures during the observation time, t zs The observation time.

[0052] S3: Determine whether to predict the fault data of the uninstalled electricity meter based on the installed electricity meter;

[0053] If not, then select batches of electricity meters within the service life within the prediction time, perform single-batch dynamic single-sample prediction on a certain batch of electricity meters, realize multi-batch single-sample prediction, and obtain single-sample prediction results;

[0054] Specifically, the following steps are included:

[0055] S31: Determine the service life of each batch of electricity meters according to the parameters of the Weibull distribution life model of the electricity meters. The service life of each batch of electricity meters is:

[0056]

[0057] Among them, R is reliability (the State Grid stipulates that when the reliability R reaches 0.9, all the meters installed at the same time in the same batch will be replaced with new ones), t Rz is the service life of the zth batch of electric meters;

[0058] S32: Based on the service life of each batch of electricity meters and the predicted time, screen out the electricity meters that are still within their service life within the predicted time, and remove the electricity meters that are not within their service life;

[0059] Specifically, when the service life is greater than or equal to the observation time, that is, t Rz ≥t zs When , the meter is selected as the forecast meter, and the time length t is selected zs The largest batch of electricity meters is taken as the first batch of this prediction. All batches that meet the conditions are selected in turn, and the total batch is set to x; at the same time, the electricity meters that have exceeded their service life are eliminated.

[0060] S33: Determine the cumulative failure probability of a batch of electric meters within the observation time based on the screened electric meters that are still within their service life in the prediction time. The cumulative failure probability of the batch of electric meters within the observation time is:

[0061]

[0062] Among them, h z is the cumulative failure probability of a batch of electric meters during the observation time;

[0063] S34: Determine a point estimate of the cumulative failure probability of the batch of electricity meters during the observation time based on the cumulative failure probability of the batch of electricity meters during the observation time. The point estimate of the cumulative failure probability of the batch of electricity meters during the observation time is:

[0064]

[0065] in, is the point estimate of the cumulative failure probability of the electric meter within a certain batch of observation time;

[0066] S35: Determine the cumulative failure probability of a batch of electricity meters at the end of the prediction time under the Weibull distribution, where the cumulative failure probability of the batch of electricity meters at the end of the prediction time is:

[0067]

[0068] Among them, t ze It is the time from the start time of observation time to the end time of prediction time for a batch of electricity meters;

[0069] S36: Calculate the cumulative failure probability of a batch of electricity meters after the end of the prediction time based on the cumulative failure probability of the batch of electricity meters at the end of the prediction time. The cumulative failure probability of the batch of electricity meters after the end of the prediction time is:

[0070]

[0071] Among them, k z The cumulative failure probability of a batch of electric meters after the prediction time ends;

[0072] S37: Obtain a point estimate of the cumulative failure probability of the batch of electricity meters after the prediction time is over, based on the cumulative failure probability of the batch of electricity meters after the prediction time is over. The point estimate of the cumulative failure probability of the batch of electricity meters after the prediction time is:

[0073]

[0074] in, It is a point estimate of the cumulative failure probability of a batch of electricity meters after the end of the prediction time;

[0075] S38: Calculate the cumulative failure probability of a batch of electricity meters within the prediction time based on the cumulative failure probability of the batch of electricity meters within the observation time and the cumulative failure probability of the batch of electricity meters after the end of the prediction time. The cumulative failure probability of the batch of electricity meters within the prediction time is:

[0076]

[0077] Among them, j z is the cumulative failure probability of a batch of electricity meters within the prediction time;

[0078] S39: Obtain a point estimate of the cumulative failure probability of the batch of electricity meters within the prediction time based on the cumulative failure probability of the batch of electricity meters within the prediction time. The point estimate of the cumulative failure probability of the batch of electricity meters within the prediction time is:

[0079]

[0080] in, is the point estimate of the cumulative failure probability of a batch of electricity meters within the forecast time;

[0081] S310: Calculate the number of failures of the batch of electricity meters within the predicted time based on the cumulative failure probability of the batch of electricity meters within the predicted time. The number of failures of the batch of electricity meters within the predicted time is:

[0082]

[0083] Among them, J z The number of failures that occur in a batch of electricity meters within the predicted time;

[0084] S311: Obtain a point estimate of the number of failures that occur in the batch of electricity meters within the predicted time based on the number of failures that occur in the batch of electricity meters within the predicted time. The point estimate of the number of failures that occur in the batch of electricity meters within the predicted time is:

[0085]

[0086] in, It is a point estimate of the number of failures that will occur in a batch of electricity meters within the forecast time;

[0087] S312: Obtain an odds ratio based on the ratio of the cumulative failure probability of the batch of electricity meters within the observation time to the cumulative failure probability of the batch of electricity meters within the prediction time. The odds ratio is:

[0088]

[0089] Among them, R z is the odds ratio;

[0090] S313: A point estimate of the odds ratio is obtained according to the odds ratio:

[0091]

[0092] in, is the point estimate of the odds ratio;

[0093] S314: Obtain a two-sided confidence interval with a confidence level of γ based on the point estimate of the odds ratio. The two-sided confidence interval with a confidence level of γ is:

[0094]

[0095] Among them, χ 2 (1-γ) / 2 (2H z ) and χ 2 (1-γ) / 2 (2J z +2) is the χ with a confidence level of (1-γ) / 2 2 Quantile of the distribution, χ 2 (1+γ) / 2 (2H z +2) and χ 2 (1+γ) / 2 (2J z ) is the chi-squared value with a confidence level of (1+γ) / 2 2 Quantile of the distribution, R zL is the lower limit of the odds ratio of a batch, R zU is the upper limit of the odds ratio of a batch;

[0096] S315: Obtain an interval of the number of failures that occur in a batch of electricity meters within the predicted time according to the two-sided confidence interval with a confidence level of γ. The interval of the number of failures that occur in a batch of electricity meters within the predicted time is:

[0097]

[0098] Among them, J zL is the minimum number of failures of a batch of electric meters within the predicted time, J zU The maximum number of failures that occur in a batch of electricity meters within the predicted time;

[0099] S316: Based on the cumulative failure probability of the batch of electric meters within the prediction time, the predicted number of single-sample failures occurring in all batches of electric meters put into use within the prediction time is obtained. The predicted number of single-sample failures occurring in all batches of electric meters put into use within the prediction time is:

[0100]

[0101] Where J is the number of single sample predicted failures that occur in all batches of electricity meters put into use within the prediction time;

[0102] S317: Based on the single sample predicted number of failures that occur in all batches of electricity meters within the prediction time, a point estimate of the single sample predicted number of failures that occur in all batches of electricity meters put into use within the prediction time is obtained. The point estimate of the single sample predicted number of failures that occur in all batches of electricity meters put into use within the prediction time is:

[0103]

[0104] in, A point estimate of the number of failures that will occur within the forecast period for all batches of meters put into service;

[0105] S318: Calculate the interval of the number of single-sample predicted failures of the electric meters within the prediction time based on the point estimate of the number of failures that occur in all batches of electric meters put into use within the prediction time. The interval of the number of single-sample predicted failures of the electric meters within the prediction time is:

[0106]

[0107] Among them, J L is the minimum number of occurrences within the forecast period caused by the installed electricity meter, J U is the maximum number of occurrences within the forecast time caused by the installed and used electricity meters, and x is the total number of batches installed and used;

[0108] S319: Obtain a single sample prediction result according to the minimum number of occurrences within the predicted time caused by the installed and used electricity meter and the maximum number of occurrences within the predicted time caused by the installed and used electricity meter.

[0109] If so, the model parameters of each batch of electricity meters installed within the prediction time are calculated, and a single-batch double-sample prediction is performed for a certain batch of electricity meters to achieve multi-batch double-sample prediction and obtain the double-sample prediction results;

[0110] Specifically, the following steps are included:

[0111] S31`: Determine the life parameters of each batch of electric meters based on the parameters of the Weibull distribution life model of the electric meters, and screen out the life parameters of electric meters from the same batch or from the same manufacturer within the forecast period in combination with the State Grid installation plan, and calculate the average of the screened life parameters as the life parameters of the electric meters at a certain installation time within the forecast period;

[0112] Assume that in the future time period t se The electricity meters are installed in batches w times, and the installation time of each time is: t x+1 , t x+2 ,…,t x+w , let N x+1 、N x+2 ,…,N x+wis the number of installations in each batch. The life distribution of the electricity meters installed at a certain time in the future time period to be predicted is similar to that of the electricity meters that have been put into use in batches and belong to the same batch or the same manufacturer. They are collectively referred to as similar electricity meters. All similar electricity meters that have been put into use are screened out, and the distribution parameters of the screened similar electricity meters are reordered according to the installation time. Then, the life distribution parameters of all similar electricity meters are averaged to obtain the life distribution parameters of the electricity meters installed at a certain time in the prediction time:

[0113] If the installation time is t x+1 The number of installations of the batch of electric meters and the electric meters already in use that are of the same type is v, then

[0114]

[0115] In the above formula, ηx+1 The installation time is t x+1 Scale parameters of batch electric meters; mx+1 The installation time is t x+1 Shape parameters of batch electric meters; The filtered t x+1 The corresponding scale parameters of the batch of electricity meters installed at the time after sorting the same type of electricity meters; The corresponding shape parameters of the sorted similar electric meters;

[0116] Using this method, the life parameters of all batches of electricity meters installed within the forecast period can be calculated.

[0117] S32`: Based on the life parameter of the electric meter at a certain installation time within the predicted time, the cumulative failure probability from the certain installation time to the end time of the predicted time is obtained. The cumulative failure probability from the certain installation time to the end time of the predicted time is:

[0118]

[0119] Among them, h x+1 is the cumulative failure probability from a certain installation time to the end of the prediction time, t (x+1)e The time from a certain installation time to the end of the forecast time;

[0120] S33`: According to the cumulative failure probability from the installation time to the end of the prediction time, the predicted failure times of a batch of electric meters are predicted. The predicted failure times of the batch of electric meters are:

[0121]

[0122] Among them, J x+1 is the predicted failure times of a batch of electric meters, N x+1the total number of meters installed for a particular installation time;

[0123] S34': Calculate the number of failures before the end of the prediction time caused by the installation of the electricity meters within the prediction time based on the predicted number of failures of the batch of electricity meters. The number of failures before the end of the prediction time caused by the installation of the electricity meters within the prediction time is:

[0124]

[0125] Where w is the number of batches installed during the forecast period;

[0126] S35`: obtaining a dual-sample prediction result according to the number of faults before the end of the prediction time caused by the installation of the electricity meter within the prediction time.

[0127] S4: deriving the total predicted number of failures and the total prediction interval within the prediction time according to the single-sample prediction result and the dual-sample prediction result.

[0128] Specifically, the total number of predicted failures within the prediction time is:

[0129]

[0130] The overall prediction interval is:

[0131]

[0132] like Figure 3 As shown in the figure, t s is the cutoff time of historical fault data, t e is the deadline for predicting the future time interval. Let t1, t2, ..., t x ,…,t x+1 , t x+2 ,…,t x+w is the installation time of each batch of electric meters, where t1, t2, ..., t x They are the asynchronous installation times of each batch of electric meters that have been put into use before the future time to be predicted, and the time from the installation time of each batch to the truncation time of the fault data t s The length of time is t 1s , t 2s ,…,t xs , that is, t 1s =t s -t1,t 2s =t s -t2,…,t xs =t s -t x . Let N1, N2, ..., N x is the number of units installed in each batch, H1, H2, ..., H xis the number of faults corresponding to each time period, that is, the time period (t1,t s ]The number of failures that occurred is H1, (t2,t s ]The number of failures that occurred is H2, and the corresponding (t x ,t s ]The number of failures that occurred is H x At the same time, it is assumed that w meters are installed in the future time interval to be predicted, and the installation time of each time is: t x+1 , t x+2 ,…,t x+w , each installation time to the predicted deadline t e The number of fault predictions in the time period are: x+1 、J x+2 ,…,J x+w The present invention uses the data of the electric meters that have been put into use and the electric meters that have not been put into use but will be put into use before the deadline of the prediction to realize the multi-batch asynchronous dynamic prediction of the future time interval (t s ,t e ] number of failures.

[0133] Therefore, the prediction method provided by the present invention can predict the number of meter failures in future time periods based on the existing time failure data at this stage, which is conducive to solving the problem of early prediction of meter failures and improving the power grid company's full life cycle control capabilities for meter life prediction, risk assessment, and management decision-making.

[0134] Example 2:

[0135] See also Figure 4 The present invention also provides a system for predicting the number of electric meter failures, the system comprising:

[0136] The timed truncated fault data acquisition unit 1 is used to acquire the timed truncated fault data of the electric meter on-site; the timed truncated fault data includes the installation time of each batch of electric meters, the number of electric meters installed in each batch, the observation time, the number of electric meter failures during the observation time, the predicted time, and the time corresponding to the failure of each electric meter, wherein the starting time of the predicted time is the ending time of the observation time;

[0137] Life model construction unit 2, used to construct a Weibull distribution life model of the electricity meter based on the timed truncated fault data, and estimate the parameters of the Weibull distribution life model of the electricity meter using maximum likelihood;

[0138] A judgment unit 3, configured to judge whether to predict the fault data of the non-installed electric meter based on the installed electric meter;

[0139] If not, then select batches of electricity meters within the service life within the prediction time, perform single-batch dynamic single-sample prediction on a certain batch of electricity meters, realize multi-batch single-sample prediction, and obtain single-sample prediction results;

[0140] If so, the model parameters of each batch of electricity meters installed within the prediction time are calculated, and a single-batch double-sample prediction is performed for a certain batch of electricity meters to achieve multi-batch double-sample prediction and obtain the double-sample prediction results;

[0141] The total prediction unit 4 is used to obtain the total predicted number of failures and the total prediction interval within the prediction time based on the single sample prediction results and the double sample prediction results.

[0142] Specifically, the life model construction unit 2 includes:

[0143] A cumulative failure probability calculation module is used to determine the cumulative failure probability of the electric meter within the observation time based on the timed truncated failure data;

[0144] A life model construction module, used to construct a Weibull distribution life model of the electric meter according to the cumulative failure probability of the electric meter within the observation time;

[0145] The parameter estimation module is used to estimate the parameters of the Weibull distribution life model of the electricity meter by using maximum likelihood.

[0146] Specifically, the judging unit 3 includes:

[0147] A life determination module is used to determine the service life of each batch of electricity meters based on the parameters of the Weibull distribution life model of the electricity meters;

[0148] A screening module is used to screen out the meters that are still within their service life according to the service life of each batch of meters and the predicted time, and to eliminate the meters that are not within their service life;

[0149] The first cumulative failure probability determination module is used to determine the cumulative failure probability of a batch of electric meters within the observation time based on the screened electric meters that are still within their service life in the predicted time;

[0150] A first point estimation module is used to determine a point estimate of the cumulative failure probability of a batch of electricity meters within the observation time based on the cumulative failure probability of a batch of electricity meters within the observation time;

[0151] A second cumulative failure probability determination module is used to determine the cumulative failure probability of a batch of electricity meters at the end of the prediction time under the Weibull distribution;

[0152] The third cumulative failure probability determination module is used to calculate the cumulative failure probability of a batch of electricity meters after the end of the prediction time based on the cumulative failure probability of a batch of electricity meters at the end of the prediction time,

[0153] A second point estimation module is used to obtain a point estimate of the cumulative failure probability of a batch of electricity meters after the end of the prediction time based on the cumulative failure probability of a batch of electricity meters after the end of the prediction time;

[0154] a fourth cumulative failure probability determination module, configured to calculate the cumulative failure probability of a batch of electricity meters within the prediction time based on the cumulative failure probability of a batch of electricity meters within the observation time and the cumulative failure probability of a batch of electricity meters after the end of the prediction time;

[0155] A third point estimation module is used to obtain a point estimate of the cumulative failure probability of a batch of electricity meters within the prediction time based on the cumulative failure probability of a batch of electricity meters within the prediction time;

[0156] The module for determining the number of failures of each batch of electric meters is used to calculate the number of failures of a batch of electric meters within the predicted time based on the cumulative failure probability of the electric meters in a batch within the predicted time;

[0157] A fourth point estimation module is used to obtain a point estimate of the number of failures that occur in a batch of electricity meters within the predicted time based on the number of failures that occur in a batch of electricity meters within the predicted time;

[0158] an odds ratio determination module, configured to obtain an odds ratio based on a ratio of a cumulative failure probability of a batch of electric meters within an observation period to a cumulative failure probability of the batch of electric meters within a prediction period;

[0159] a fifth point estimation module, configured to obtain a point estimate of the odds ratio based on the odds ratio;

[0160] A two-sided confidence interval acquisition module is used to obtain a two-sided confidence interval with a confidence level of γ based on the point estimate of the odds ratio;

[0161] The module for determining the interval of the number of failures of each batch of electric meters is used to obtain the interval of the number of failures of a batch of electric meters within the predicted time based on a two-sided confidence interval with a confidence level of γ;

[0162] A module for determining the number of single sample predicted failures is used to obtain the number of single sample predicted failures that will occur in all batches of electric meters put into use within the predicted time based on the cumulative failure probability of a batch of electric meters within the predicted time;

[0163] A sixth point estimation module, configured to obtain a point estimate of the number of failures that occur in all batches of electricity meters put into use within the predicted time period based on the number of failures that occur in all batches of electricity meters put into use within the predicted time period;

[0164] A module for determining the interval of the number of single sample predicted failures is used to calculate the interval of the number of single sample predicted failures that will occur in the electricity meters within the predicted time period based on the point estimate of the number of failures that will occur in all batches of electricity meters put into use within the predicted time period;

[0165] The single sample prediction result determination module is used to obtain the single sample prediction result based on the minimum number of occurrences within the prediction time caused by the installed and used electricity meter and the maximum number of occurrences within the prediction time caused by the installed and used electricity meter.

[0166] Specifically, the judging unit 3 further includes:

[0167] The life parameter determination module is used to determine the life parameters of each batch of electricity meters based on the parameters of the Weibull distribution life model of the electricity meters. In combination with the State Grid installation plan, the module selects the life parameters of the electricity meters from the same batch or manufacturer within the forecast period, and calculates the average of the selected life parameters as the life parameters of the electricity meters at a certain installation time within the forecast period.

[0168] a fifth cumulative failure probability determination module, configured to obtain a cumulative failure probability from a certain installation time to an end time of the prediction time based on the life parameters of the electric meter at a certain installation time within the prediction time;

[0169] A module for determining the predicted number of failures of each batch of electric meters is used to predict the predicted number of failures of a batch of electric meters based on the cumulative failure probability from a certain installation time to the end time of the prediction time;

[0170] A module for determining the number of failures before the predicted time limit is used to calculate the number of failures before the predicted time limit caused by the installation of the electric meters within the predicted time limit based on the predicted number of failures of the electric meters in the batch;

[0171] The dual-sample prediction result determination module is used to obtain the dual-sample prediction results according to the number of faults before the end of the prediction time caused by the installation of the electricity meter within the prediction time.

[0172] The prediction system provided by the present invention can establish an emergency reserve library of faulty electric meters in the future time period, so as to prepare for the rotation of electric meters that may fail in the future predicted time period.

[0173] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0174] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for predicting the number of electric meter failures, characterized in that: include: Obtaining on-site timed truncated fault data for electric meters; the timed truncated fault data includes the installation time of each batch of electric meters, the number of electric meters installed in each batch, the observation time, the number of electric meter failures during the observation time, the predicted time, and the time corresponding to the failure of each electric meter, wherein the starting time of the predicted time is the ending time of the observation time; Constructing a Weibull distribution life model of the electric meter based on the timed truncated fault data, and estimating parameters of the Weibull distribution life model of the electric meter using maximum likelihood; Determine whether to predict the failure data of the uninstalled electricity meter based on the installed electricity meter; If not, then select batches of electricity meters within the service life within the prediction time, perform single-batch dynamic single-sample prediction on a certain batch of electricity meters, realize multi-batch single-sample prediction, and obtain single-sample prediction results; If so, the model parameters of each batch of electricity meters installed within the prediction time are calculated, and a single-batch double-sample prediction is performed for a certain batch of electricity meters to achieve multi-batch double-sample prediction and obtain the double-sample prediction results; Determining the total predicted number of failures and the total prediction interval within the prediction time based on the single-sample prediction result and the dual-sample prediction result; If not, batches of electricity meters within the service life within the prediction time are selected, and a single batch dynamic single sample prediction is performed for a certain batch of electricity meters to achieve multi-batch single sample prediction. The single sample prediction results specifically include: The service life of each batch of electric meters is determined according to the parameters of the Weibull distribution life model of the electric meters. The service life of each batch of electric meters is: Among them, R is reliability, t Rz is the service life of the zth batch of electric meters, η z is the scale parameter, m z is the shape parameter; Based on the service life of each batch of electric meters and the predicted time, electric meters that are still within the service life within the predicted time are screened out, and electric meters that are not within the service life are eliminated; Based on the screened electric meters that are still within their service life in the predicted time, the cumulative failure probability of a batch of electric meters during the observation time is determined. The cumulative failure probability of the electric meters in the batch during the observation time is: Among them, h z is the cumulative failure probability of a batch of electric meters during the observation time; t zs is the observation time; According to the cumulative failure probability of the batch of electricity meters during the observation time, a point estimate of the cumulative failure probability of the batch of electricity meters during the observation time is determined. The point estimate of the cumulative failure probability of the batch of electricity meters during the observation time is: in, is the point estimate of the cumulative failure probability of the electric meter within a certain batch of observation time; Determine the cumulative failure probability of a batch of electricity meters at the end of the prediction time under the Weibull distribution. The cumulative failure probability of the batch of electricity meters at the end of the prediction time is: Among them, t ze It is the time from the start time of observation time to the end time of prediction time for a batch of electricity meters; Based on the cumulative failure probability of the batch of electric meters at the end of the prediction time, the cumulative failure probability of the batch of electric meters after the end of the prediction time is calculated. The cumulative failure probability of the batch of electric meters after the end of the prediction time is: Among them, k z The cumulative failure probability of a batch of electric meters after the prediction time ends; According to the cumulative failure probability of the batch of electricity meters after the end of the prediction time, a point estimate of the cumulative failure probability of the batch of electricity meters after the end of the prediction time is obtained. The point estimate of the cumulative failure probability of the batch of electricity meters after the end of the prediction time is: in, It is a point estimate of the cumulative failure probability of a batch of electricity meters after the end of the prediction time; The cumulative failure probability of a batch of electric meters within the prediction time is calculated based on the cumulative failure probability of the batch of electric meters within the observation time and the cumulative failure probability of the batch of electric meters after the end of the prediction time. The cumulative failure probability of the batch of electric meters within the prediction time is: Among them, j z is the cumulative failure probability of a batch of electricity meters within the prediction time; According to the cumulative failure probability of the batch of electric meters within the prediction time, a point estimate of the cumulative failure probability of the batch of electric meters within the prediction time is obtained. The point estimate of the cumulative failure probability of the batch of electric meters within the prediction time is: in, is the point estimate of the cumulative failure probability of a batch of electricity meters within the forecast time; According to the cumulative failure probability of the batch of electric meters within the predicted time, the number of failures of the batch of electric meters within the predicted time is calculated. The number of failures of the batch of electric meters within the predicted time is: Among them, J z N is the number of failures that occur in a batch of electric meters within the predicted time; z is the total number of electric meters put into use during the observation period; According to the number of failures of the batch of electric meters within the predicted time, a point estimate of the number of failures of the batch of electric meters within the predicted time is obtained. The point estimate of the number of failures of the batch of electric meters within the predicted time is: in, It is a point estimate of the number of failures that will occur in a batch of electricity meters within the forecast time; An odds ratio is obtained based on the ratio of the cumulative failure probability of the batch of electric meters within the observation time to the cumulative failure probability of the batch of electric meters within the prediction time. The odds ratio is: Among them, R z is the odds ratio; The point estimate of the odds ratio obtained from the odds ratio is: in, is the point estimate of the odds ratio; A two-sided confidence interval with a confidence level of γ is obtained based on the point estimate of the odds ratio. The two-sided confidence interval with a confidence level of γ is: Among them, χ 2 (1-γ) / 2 (2H z ) and χ 2 (1-γ) / 2 (2J z +2) is the χ with a confidence level of (1-γ) / 2 2 Quantile of the distribution, χ 2 (1+γ) / 2 (2H z +2) and χ 2 (1+γ) / 2 (2J z ) is the chi-squared value with a confidence level of (1+γ) / 2 2 Quantile of the distribution, R zL is the lower limit of the odds ratio of a batch, R zU is the upper limit of the odds ratio of a certain batch; H z is the number of meter failures during the observation period; According to the two-sided confidence interval with the confidence level γ, the interval of the number of failures of a batch of electric meters within the predicted time is obtained. The interval of the number of failures of a batch of electric meters within the predicted time is: Among them, J zL is the minimum number of failures of a batch of electric meters within the predicted time, J zU The maximum number of failures that occur in a batch of electricity meters within the predicted time; Based on the cumulative failure probability of a batch of electric meters within the prediction time, the single sample predicted number of failures occurring in all batches of electric meters put into use within the prediction time is obtained. The single sample predicted number of failures occurring in all batches of electric meters put into use within the prediction time is: Where J is the number of single sample predicted failures that occur in all batches of electricity meters put into use within the prediction time; Based on the single sample predicted number of failures that occur in all batches of electricity meters within the prediction time, a point estimate of the single sample predicted number of failures that occur in all batches of electricity meters put into use within the prediction time is obtained. The point estimate of the single sample predicted number of failures that occur in all batches of electricity meters put into use within the prediction time is: in, A point estimate of the number of failures that will occur within the forecast period for all batches of meters put into service; Based on the point estimate of the number of failures that occur in all batches of electricity meters put into use within the predicted time, the interval of the number of single-sample predicted failures that occur in the electricity meters within the predicted time is calculated. The interval of the number of single-sample predicted failures that occur in the electricity meters within the predicted time is: Among them, J L is the minimum number of occurrences within the forecast period caused by the installed electricity meter, J U is the maximum number of occurrences within the forecast time caused by the installed and used electricity meters, and x is the total number of batches installed and used; Obtain a single sample prediction result based on the minimum number of occurrences within the predicted time caused by the installed and used electricity meter and the maximum number of occurrences within the predicted time caused by the installed and used electricity meter; If so, the model parameters of each batch of electricity meters installed within the prediction time are calculated, and a single-batch double-sample prediction is performed for a certain batch of electricity meters to achieve multi-batch double-sample prediction. The double-sample prediction results specifically include: Determine the life parameters of each batch of electric meters based on the parameters of the Weibull distribution life model of the electric meters, and screen out the life parameters of electric meters from the same batch or the same manufacturer within the forecast period in combination with the State Grid installation plan. Average the screened life parameters and use them as the life parameters of the electric meters at a certain installation time within the forecast period. According to the life parameters of the electric meter at a certain installation time within the predicted time, the cumulative failure probability from the certain installation time to the end time of the predicted time is obtained. The cumulative failure probability from the certain installation time to the end time of the predicted time is: Among them, h x+1 is the cumulative failure probability from a certain installation time to the end of the prediction time, t (x+1)e The time from a certain installation time to the end of the forecast time; The predicted number of failures of a batch of electric meters is predicted based on the cumulative failure probability from the installation time to the end of the prediction time. The predicted number of failures of the batch of electric meters is: Among them, J x+1 is the predicted failure times of a batch of electric meters, N x+1 the total number of meters installed for a particular installation time; Based on the predicted number of failures of a batch of electric meters, the number of failures before the predicted time due to the installation of electric meters within the predicted time is calculated. The number of failures before the predicted time due to the installation of electric meters within the predicted time is: Where w is the number of batches installed during the forecast period; A dual-sample prediction result is obtained according to the number of faults before the end of the prediction time caused by the installation of the electricity meter within the prediction time.

2. The method for predicting the number of electric meter failures according to claim 1, wherein: Constructing the Weibull distribution life model of the electric meter based on the timed truncated fault data specifically includes: Determining the cumulative failure probability of the electric meter within the observation time based on the timed truncated failure data; The Weibull distribution life model of the electric meter is constructed according to the cumulative failure probability of the electric meter within the observation time. The Weibull distribution life model of the electric meter is: Among them, h z is the cumulative failure probability of the electric meter within a certain batch of observation time, t zs The observation time.

3. The method for predicting the number of electric meter failures according to claim 1, wherein: The parameters of the Weibull distribution life model of the electric meter obtained by maximum likelihood estimation are: Among them, t i is the time when the ith meter of the same installation time fails, N z is the total number of electric meters put into use during the observation period, H z is the number of meter failures during the observation time, t zs The observation time.

4. The method for predicting the number of electric meter failures according to claim 1, wherein: The total predicted number of failures within the prediction time is: The overall prediction interval is:

5. A system for predicting the number of electric meter failures, characterized in that: include: a timed truncated fault data acquisition unit, configured to acquire timed truncated fault data of electric meters on site; the timed truncated fault data includes the installation time of each batch of electric meters, the number of electric meters installed in each batch, the observation time, the number of electric meter failures during the observation time, the predicted time, and the time corresponding to the occurrence of a failure of each electric meter, wherein the starting time of the predicted time is the ending time of the observation time; A life model construction unit is used to construct a Weibull distribution life model of the electricity meter based on the timed truncated fault data, and estimate the parameters of the Weibull distribution life model of the electricity meter using maximum likelihood; A judgment unit, configured to judge whether to predict fault data of an uninstalled electric meter based on an installed electric meter; If not, then select batches of electricity meters within the service life within the prediction time, perform single-batch dynamic single-sample prediction on a certain batch of electricity meters, realize multi-batch single-sample prediction, and obtain single-sample prediction results; If so, the model parameters of each batch of electricity meters installed within the prediction time are calculated, and a single-batch double-sample prediction is performed for a certain batch of electricity meters to achieve multi-batch double-sample prediction and obtain the double-sample prediction results; The total prediction unit is used to obtain the total predicted number of failures and the total prediction interval within the prediction time based on the single sample prediction results and the double sample prediction results; The judging unit includes: A life determination module is used to determine the service life of each batch of electricity meters based on the parameters of the Weibull distribution life model of the electricity meters; A screening module is used to screen out the meters that are still within their service life according to the service life of each batch of meters and the predicted time, and to eliminate the meters that are not within their service life; The first cumulative failure probability determination module is used to determine the cumulative failure probability of a batch of electric meters within the observation time based on the screened electric meters that are still within their service life in the predicted time; A first point estimation module is used to determine a point estimate of the cumulative failure probability of a batch of electricity meters within the observation time based on the cumulative failure probability of a batch of electricity meters within the observation time; A second cumulative failure probability determination module is used to determine the cumulative failure probability of a batch of electricity meters at the end of the prediction time under the Weibull distribution; The third cumulative failure probability determination module is used to calculate the cumulative failure probability of a batch of electricity meters after the end of the prediction time based on the cumulative failure probability of a batch of electricity meters at the end of the prediction time, A second point estimation module is used to obtain a point estimate of the cumulative failure probability of a batch of electricity meters after the end of the prediction time based on the cumulative failure probability of a batch of electricity meters after the end of the prediction time; a fourth cumulative failure probability determination module, configured to calculate the cumulative failure probability of a batch of electricity meters within the prediction time based on the cumulative failure probability of a batch of electricity meters within the observation time and the cumulative failure probability of a batch of electricity meters after the end of the prediction time; A third point estimation module is used to obtain a point estimate of the cumulative failure probability of a batch of electricity meters within the prediction time based on the cumulative failure probability of a batch of electricity meters within the prediction time; The module for determining the number of failures of each batch of electric meters is used to calculate the number of failures of a batch of electric meters within the predicted time based on the cumulative failure probability of the electric meters in a batch within the predicted time; A fourth point estimation module is used to obtain a point estimate of the number of failures that occur in a batch of electricity meters within the predicted time based on the number of failures that occur in a batch of electricity meters within the predicted time; an odds ratio determination module, configured to obtain an odds ratio based on a ratio of a cumulative failure probability of a batch of electric meters within an observation period to a cumulative failure probability of the batch of electric meters within a prediction period; a fifth point estimation module, configured to obtain a point estimate of the odds ratio based on the odds ratio; A two-sided confidence interval acquisition module is used to obtain a two-sided confidence interval with a confidence level of γ based on the point estimate of the odds ratio; The module for determining the interval of the number of failures of each batch of electric meters is used to obtain the interval of the number of failures of a batch of electric meters within the predicted time based on a two-sided confidence interval with a confidence level of γ; A module for determining the number of single sample predicted failures is used to obtain the number of single sample predicted failures that will occur in all batches of electric meters put into use within the predicted time based on the cumulative failure probability of a batch of electric meters within the predicted time; A sixth point estimation module, configured to obtain a point estimate of the number of failures that occur in all batches of electricity meters put into use within the predicted time period based on the number of failures that occur in all batches of electricity meters put into use within the predicted time period; A module for determining the interval of the number of single sample predicted failures is used to calculate the interval of the number of single sample predicted failures that will occur in the electricity meters within the predicted time period based on the point estimate of the number of failures that will occur in all batches of electricity meters put into use within the predicted time period; A single sample prediction result determination module is used to obtain a single sample prediction result based on the minimum number of occurrences within the prediction time caused by the installed and used electricity meter and the maximum number of occurrences within the prediction time caused by the installed and used electricity meter; The judging unit further includes: The life parameter determination module is used to determine the life parameters of each batch of electricity meters based on the parameters of the Weibull distribution life model of the electricity meters. In combination with the State Grid installation plan, the module selects the life parameters of the electricity meters from the same batch or manufacturer within the forecast period, and calculates the average of the selected life parameters as the life parameters of the electricity meters at a certain installation time within the forecast period. a fifth cumulative failure probability determination module, configured to obtain a cumulative failure probability from a certain installation time to an end time of the prediction time based on the life parameters of the electric meter at a certain installation time within the prediction time; A module for determining the predicted number of failures of each batch of electric meters is used to predict the predicted number of failures of a batch of electric meters based on the cumulative failure probability from a certain installation time to the end time of the prediction time; A module for determining the number of failures before the predicted time limit is used to calculate the number of failures before the predicted time limit caused by the installation of the electric meters within the predicted time limit based on the predicted number of failures of the electric meters in the batch; The dual-sample prediction result determination module is used to obtain the dual-sample prediction results according to the number of faults before the end of the prediction time caused by the installation of the electricity meter within the prediction time.

6. The electric meter failure frequency prediction system according to claim 5, characterized in that: The life model building unit includes: A cumulative failure probability calculation module is used to determine the cumulative failure probability of the electric meter within the observation time based on the timed truncated failure data; A life model construction module, used to construct a Weibull distribution life model of the electric meter according to the cumulative failure probability of the electric meter within the observation time; The parameter estimation module is used to estimate the parameters of the Weibull distribution life model of the electricity meter by using maximum likelihood.

Images