CN113378453A - Intelligent electric energy meter failure rate online prediction method based on deep belief network - Google Patents

Intelligent electric energy meter failure rate online prediction method based on deep belief network Download PDF

Info

Publication number
CN113378453A
CN113378453A CN202110509527.7A CN202110509527A CN113378453A CN 113378453 A CN113378453 A CN 113378453A CN 202110509527 A CN202110509527 A CN 202110509527A CN 113378453 A CN113378453 A CN 113378453A
Authority
CN
China
Prior art keywords
electric energy
energy meter
failure rate
intelligent electric
basic
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202110509527.7A
Other languages
Chinese (zh)
Inventor
杨挺
李艺可
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tianjin University
Original Assignee
Tianjin University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tianjin University filed Critical Tianjin University
Priority to CN202110509527.7A priority Critical patent/CN113378453A/en
Publication of CN113378453A publication Critical patent/CN113378453A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/048—Activation functions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/08—Probabilistic or stochastic CAD
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/08—Thermal analysis or thermal optimisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention relates to an intelligent electric energy meter failure rate online prediction method based on a deep belief network, which comprises the following steps of: step 1, calculating the basic failure rate of a key component; step 2, calculating the basic failure rate of the intelligent electric energy meter; step 3, calculating the actual failure rate of the intelligent electric energy meter; step 4, building a deep belief network model, determining the input of the network as the basic failure rate, the field environment temperature and the field working electrical stress level of the intelligent electric energy meter in the step 2, and the output as the actual failure rate of the intelligent electric energy meter in the step 3; and 5, training the deep belief network model set up in the step 4, inputting the basic failure rate, the real-time field temperature and the real-time electric stress level of the electric energy meter to be predicted into the network, wherein the output of the network is the predicted failure rate of the electric energy meter. The failure rate of the intelligent electric energy meter according with the working profile can be obtained, and important references are provided for the work of field detection and maintenance, early-stage reliability design and the like of the intelligent electric energy meter.

Description

Intelligent electric energy meter failure rate online prediction method based on deep belief network
Technical Field
The invention belongs to the technical field of application of a deep learning algorithm in an intelligent electric energy meter, and relates to an online prediction method for failure rate of the intelligent electric energy meter, in particular to an online prediction method for failure rate of the intelligent electric energy meter based on a deep belief network.
Background
Along with the realization of intelligent functions such as the bidirectional multi-rate metering function of the intelligent electric energy meter, the user side control function, the bidirectional data communication function of various data transmission modes, the electricity larceny prevention function and the like, the increase of the types and the number of the internal electronic components and the complexity of the circuit structure bring great influence on the reliability of the intelligent electric energy meter. Therefore, the reliability analysis capability of the intelligent electric energy meter is improved, the failure rate of the intelligent electric energy meter is accurately predicted, and the economic loss of the intelligent electric energy meter to power departments and users is reduced.
At present, failure rate prediction methods for electronic devices are mainly divided into two types: a physical model method based on failure mechanism analysis and a component stress method based on a reliability prediction manual. The method starts from the physical, chemical, mechanical and other root causes in the product failure process, analyzes the main failure mechanism of the product, obtains a failure physical model through simulation or mathematical derivation, and further predicts the failure rate of the product. According to the method, the reliability prediction can be performed from the failure model without counting the failure data of the components in a large scale and performing an accelerated stress test, but the failure physical models corresponding to some failure mechanisms are still studied, so that the method cannot be applied to the reliability prediction of all products at present. The latter component stress method is a method for calculating the failure rate of the whole equipment by combining data in a reliability prediction manual and various influence factors of an electronic equipment application field and through a system reliability block diagram. At present, most of the failure rate of the intelligent electric energy meter is predicted by adopting a component stress method based on a reliability prediction manual, but the calculated value of the method is usually deviated from the failure rate of the electric energy meter in actual field application. Therefore, the method has important significance for accurately predicting the failure rate of the intelligent electric energy meter by a deep learning method on the basis of data.
Through searching, no prior art publication which is the same as or similar to the present invention is found.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides an intelligent electric energy meter failure rate online prediction method based on a deep belief network, which is reasonable in design, strong in practicability and accurate in prediction result.
The invention solves the practical problem by adopting the following technical scheme:
an intelligent electric energy meter failure rate online prediction method based on a deep belief network comprises the following steps:
step 1, calculating basic failure rates of key components, and acquiring basic failure rate values of other components for the intelligent electric energy meter;
step 2, calculating the basic failure rate of the intelligent electric energy meter;
step 3, acquiring field fault data of intelligent electric energy meters of the same model and different batches, and calculating the actual failure rate of the intelligent electric energy meters;
step 4, building a deep belief network model, determining the input of the network as the basic failure rate, the field environment temperature and the field working electrical stress level of the intelligent electric energy meter in the step 2, and the output as the actual failure rate of the intelligent electric energy meter in the step 3;
and 5, training the deep belief network model set up in the step 4, inputting the basic failure rate, the real-time field temperature and the real-time electric stress level of the electric energy meter to be predicted into the network, wherein the output of the network is the predicted failure rate of the electric energy meter.
Moreover, the method for calculating the basic failure rate of the component in the step 1 comprises the following steps:
aiming at key components of which the metering chip, the communication chip, the clock chip and the battery have important influence on the reliability of the intelligent electric energy meter, the basic failure rate is obtained through high-temperature working life detection;
according to the JESD74A standard, the basic failure rate model formula of the key components is as follows:
Figure BDA0003059747110000031
in the formula, x2Representing the chi-square distribution function, c is the confidence level, d is the degree of freedom, the degree of freedom is equal to 2 times the number of failures plus 2, A is the acceleration factor, N is the number of samples participating in the test, t is the number of samples participating in the testATo speed up test time; the acceleration factor a can be found from the following model:
the effect of temperature stress can be represented by an Arrhenius model, and the temperature acceleration coefficient AtThe calculation formula is as follows:
Figure BDA0003059747110000032
where Ea is activation energy of the reaction, k is boltzmann constant, Tu is absolute temperature in a normal state of the device, and Ta is absolute temperature in an accelerated state of the device.
The action of the electric stress can be represented by an Eying model, and the acceleration coefficient of the electric stress AvThe calculation formula is as follows:
Figure BDA0003059747110000033
in the formula, Z is a voltage acceleration constant, Vu is a voltage in a normal state of the component, and Va is a voltage in an accelerated state of the component.
The final acceleration factor a is the product of the above acceleration factors, i.e.:
A＝AtAv (4)
the basic failure rate of the rest components in the intelligent electric energy meter is obtained by inquiring the GJB/Z299C manual.
Moreover, the specific method of the step 2 is as follows:
adding the basic failure rates of all components in the intelligent electric energy meter, defining the added value as the basic failure rate of the intelligent electric energy meter, and adopting the expression as follows:
Figure BDA0003059747110000041
in the formula (I), the compound is shown in the specification,λbmis the basic failure rate, lambda, of the intelligent electric energy meterb_iThe number n is the number of the components contained in the intelligent electric energy meter.
The specific method of step 3 is:
the failure rate of the intelligent electric energy meter is calculated after the field fault data of the intelligent electric energy meter is preprocessed, the failure rate is defined as the actual failure rate of the intelligent electric energy meter, and the calculation method comprises the following steps:
the failure probability distribution function F (t) of the intelligent electric energy meter is as follows:
Figure BDA0003059747110000042
wherein s is a shape parameter, t0Is a scale parameter.
Aiming at the failure data of the electric energy meters in the same batch, the failure probability distribution function of the intelligent electric energy meter is calculated by adopting the following formula:
Figure BDA0003059747110000043
in the formula, tiFor the i-th detection time, riIs at tiCumulative number of failures, N, of electric energy meter in time0The total number of the electric energy meters in the batch.
Transformed by equation (6) and equation (7):
Figure BDA0003059747110000044
s and t can be estimated by using least square method0The parameter value of (2).
Reliable service life t of intelligent electric energy meterRThe calculation expression is:
Figure BDA0003059747110000045
actual failure rate lambda of intelligent electric energy metercThe calculation expression is:
Figure BDA0003059747110000051
moreover, the basic structure of the deep belief network model in the step 4 is as follows:
the basic structure of the deep belief network DBN is formed by stacking limited Boltzmann machines RBMs, each RBM consists of a visible layer and a hidden layer, and a bidirectional full-connection structure is formed between the layers; the visible layer is composed of m visible units viE {0,1} (i ═ 1,2, …, m), the hidden layer is composed of n hidden units hjE {0,1} (j ═ 1,2, …, n), and the connection weight between layers is W ═ Wij}∈Rm×nThe bias of the visible layer is B ═ Bi}∈RmThe offset of the hidden layer is C ═ Cj}∈Rn. RBM is determined by these three parameters θ ═ W, B, C.
Moreover, the specific method of the step 5 is as follows:
for visible and hidden layers that obey bernoulli distribution, the energy function of the RBM is:
Figure BDA0003059747110000052
the probability of the hidden layer unit being activated is:
Figure BDA0003059747110000053
wherein, sigma is Sigmoid function;
the probability of the visible layer unit being activated is:
Figure BDA0003059747110000054
updating parameters through a contrast divergence algorithm, and taking input data as a visible layer sheetYuan viCalculating a hidden layer unit h according to the formula (12)jThen, an updated visible layer unit v 'is calculated from formula (13)'iAnd then, the updated hidden unit h 'is calculated from the formula (12)'j(ii) a The expression of the parameter update is:
Figure BDA0003059747110000061
in the formula, epsilonCDIn order to obtain a learning rate,<·>is a mathematical expectation;
in order to accelerate the model training speed, dividing a training data set into a plurality of groups; for the pth group, the expression for the specific parameter update is:
Figure BDA0003059747110000062
the invention has the advantages and beneficial effects that:
the invention provides an intelligent electric energy meter failure rate on-line prediction method based on a deep belief network, which improves the numerical precision of key basic failure rate of an intelligent electric energy meter through high-temperature working life detection, calculates the basic failure rate of the intelligent electric energy meter according to a circuit schematic diagram, calculates the actual failure rate of the intelligent electric energy meter of the same model in different batches by taking field fault data capable of reflecting the actual failure condition of the electric energy meter as the basis, finally takes the basic failure rate of the intelligent electric energy meter, the field environment temperature and the field working electric stress level as the input of the network, and takes the actual failure rate of the intelligent electric energy meter as the output of the network to train the deep belief network, thereby obtaining the nonlinear relation between a theoretical value and an actual value. When the application fault data of the intelligent electric energy meter is difficult to obtain or the reliability analysis of a novel intelligent electric energy meter in the future is needed, the failure rate of the intelligent electric energy meter according with the working profile can be obtained through a trained network, important reference is provided for the work of field detection and maintenance, early-stage reliability design and the like of the intelligent electric energy meter, and contribution is made to the improvement of the reliability of the intelligent electric energy meter.
Drawings
FIG. 1 is a process flow diagram of the present invention;
FIG. 2 is a comparison of the predicted results of different algorithms;
fig. 3 is a comparison graph of prediction errors for different algorithms.
Detailed Description
The embodiments of the invention will be described in further detail below with reference to the accompanying drawings:
an online prediction method for failure rate of an intelligent electric energy meter based on a deep belief network is shown in fig. 1, and comprises the following steps:
step 1, calculating a temperature acceleration coefficient A according to a formula (2-4)tElectrical stress acceleration coefficient AvAnd an acceleration coefficient A, substituting into the formula (1) to calculate the basic failure rate of the key component; and obtaining basic failure rate values of other components for the intelligent electric energy meter through the GJB/Z299C manual.
The method for calculating the basic failure rate of the components in the step 1 comprises the following steps:
aiming at key components of which the metering chip, the communication chip, the clock chip and the battery have important influence on the reliability of the intelligent electric energy meter, the basic failure rate is obtained through high-temperature working life detection;
according to the JESD74A standard, the basic failure rate model formula of the key components is as follows:
Figure BDA0003059747110000071
in the formula, x2Representing the chi-square distribution function, c is the confidence level, d is the degree of freedom, the degree of freedom is equal to 2 times the number of failures plus 2, A is the acceleration factor, N is the number of samples participating in the test, t is the number of samples participating in the testATo speed up the test time. The acceleration factor a can be found from the following model:
temperature can cause parameter drift, surface expansion and even deformation of the physical structure of the component. The effect of temperature stress can be represented by an Arrhenius model, and the temperature acceleration coefficient AtThe calculation formula is as follows:
Figure BDA0003059747110000072
where Ea is activation energy of the reaction, k is boltzmann constant, Tu is absolute temperature in a normal state of the device, and Ta is absolute temperature in an accelerated state of the device.
The change of the electrical stress can directly affect the stability of the components, and the components are damaged when the threshold value of the components is exceeded. The action of the electric stress can be represented by an Eying model, and the acceleration coefficient of the electric stress AvThe calculation formula is as follows:
Figure BDA0003059747110000081
in the formula, Z is a voltage acceleration constant, Vu is a voltage in a normal state of the component, and Va is a voltage in an accelerated state of the component.
The final acceleration factor a is the product of the above acceleration factors, i.e.:
A＝AtAv (4)
the basic failure rate of the rest components in the intelligent electric energy meter is obtained by inquiring the GJB/Z299C manual.
Step 2, calculating the basic failure rate of the intelligent electric energy meter by combining a formula (5) through a circuit diagram of the intelligent electric energy meter;
the specific method of the step 2 comprises the following steps:
according to IEC62059-41 issued by the international electrotechnical commission, all components have the same importance for system reliability, namely, the failure of any component in the intelligent electric energy meter causes the failure of the whole meter. Adding the basic failure rates of all components in the intelligent electric energy meter, defining the added value as the basic failure rate of the intelligent electric energy meter, and adopting the expression as follows:
Figure BDA0003059747110000082
in the formula, λbmIs the basic failure rate, lambda, of the intelligent electric energy meterb_iThe number n is the number of the components contained in the intelligent electric energy meter.
Step 3, acquiring field fault data of the intelligent electric energy meter, and calculating the actual failure rate lambda of the intelligent electric energy meter through a formula (6-10)c；
The specific method of the step 3 comprises the following steps:
the failure rate of the electric energy meter obtained by calculating after preprocessing field fault data of the intelligent electric energy meter is closest to the actual working condition and is the only real basis for predicting the failure rate of the intelligent electric energy meter. The failure rate is defined as the actual failure rate of the intelligent electric energy meter, and the calculation result can be used as the target reference of the failure rate prediction of the intelligent electric energy meter.
The failure probability distribution function F (t) of the intelligent electric energy meter is as follows:
Figure BDA0003059747110000091
wherein s is a shape parameter, t0Is a scale parameter.
Aiming at the failure data of the electric energy meters in the same batch, the failure probability distribution function of the intelligent electric energy meter is calculated by adopting the following formula:
Figure BDA0003059747110000092
in the formula, tiFor the i-th detection time, riIs at tiCumulative number of failures, N, of electric energy meter in time0The total number of the electric energy meters in the batch.
Transformed by equation (6) and equation (7):
Figure BDA0003059747110000093
s and t can be estimated by using least square method0The parameter value of (2).
Reliable service life t of intelligent electric energy meterRCalculation tableThe expression is as follows:
Figure BDA0003059747110000094
actual failure rate lambda of intelligent electric energy metercThe calculation expression is:
Figure BDA0003059747110000095
step 4, building a deep belief network model, and determining the input, output, layer number and unit number of each layer of the network;
the basic structure of the Deep Belief Network (DBN) is formed by stacking limited Boltzmann machines (RBMs), wherein the RBMs are composed of a visible layer and a hidden layer, and a bidirectional full-connection structure is formed between the layers. The visible layer is composed of m visible units viE {0,1} (i ═ 1,2, …, m), the hidden layer is composed of n hidden units hjE {0,1} (j ═ 1,2, …, n), and the connection weight between layers is W ═ Wij}∈Rm×nThe bias of the visible layer is B ═ Bi}∈RmThe offset of the hidden layer is C ═ Cj}∈Rn. RBM is determined by these three parameters θ ═ W, B, C.
And 5, training the deep belief network according to a formula (11-15), and calculating to obtain a prediction result of the failure rate of the intelligent electric energy meter.
The specific method of the step 5 comprises the following steps:
for visible and hidden layers that obey bernoulli distribution, the energy function of the RBM is:
Figure BDA0003059747110000101
the probability of the hidden layer unit being activated is:
Figure BDA0003059747110000102
where σ is a Sigmoid function.
The probability of the visible layer unit being activated is:
Figure BDA0003059747110000103
updating parameters through a contrast divergence algorithm, and taking input data as a visible layer unit viCalculating a hidden layer unit h according to the formula (12)jThen, an updated visible layer unit v 'is calculated from formula (13)'iAnd then, the updated hidden unit h 'is calculated from the formula (12)'j. The expression of the parameter update is:
Figure BDA0003059747110000111
in the formula, epsilonCDIn order to obtain a learning rate,<·>is a mathematical expectation.
To speed up the model training, the training data set is divided into multiple groups. For the pth group, the expression for the specific parameter update is:
Figure BDA0003059747110000112
when training the network, the basic failure rate lambda of the intelligent electric energy meterbmThe field environment temperature T and the field work electric stress level S are used as the input of the network, and the actual failure rate lambda of the electric energy metercAnd as the output of the network, the layer number and the unit number of the deep belief network are determined in a traversal search mode, and a conjugate gradient descent method is adopted in the fine adjustment stage. When the failure rate of the intelligent electric energy meter needs to be pre-measured, the basic failure rate, the real-time field temperature and the real-time electric stress level of the electric energy meter only need to be input into a trained network, and the output of the network is the predicted failure rate of the electric energy meter.
In this embodiment, the implementation effect of the present invention is verified by practical calculation:
the Mean Absolute Percentage Error (MAPE) is selected as the evaluation index of the prediction result of the deep belief network. The average absolute percentage error can reflect the average deviation level of the more real result of the prediction result and can avoid mutual offset between the errors, and the calculation formula of the average absolute percentage error is as follows:
Figure BDA0003059747110000113
the method is characterized in that the actual failure rate of 15 batches of intelligent electric energy meters of a certain manufacturer A is used as a test set, and a DNN algorithm, a GRNN algorithm and a component stress method are added for comparison, so that the effectiveness of the method is verified. FIG. 2 shows the failure rate prediction results of different algorithms, from which it can be seen that the method of the present invention is closest to the true value, and meanwhile, after the model A intelligent electric energy meter is calculated and passes through the method of the present invention, DNN algorithm, GRNN algorithm and stress method, MAPE is respectively 4.54%, 6.84%, 9.65% and 10.77%, the method of the present invention is obviously superior to other algorithms, and has a good prediction effect.
According to the failure rate prediction process of the intelligent electric energy meter with the model A, the same prediction steps are carried out on the rest 9 models of intelligent electric energy meters, the models are numbered from B to J, and the prediction errors of different algorithms are shown in figure 3.
The MAPE range of the method is between 4% and 7%, and the MAPE average of the method is 2.27%, 3.39% and 4.83% smaller than that of DNN algorithm, GRNN algorithm and stress method respectively.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

Claims (6)

1. An intelligent electric energy meter failure rate online prediction method based on a deep belief network is characterized by comprising the following steps: the method comprises the following steps:
step 1, calculating basic failure rates of key components, and acquiring basic failure rate values of other components for the intelligent electric energy meter;
step 2, calculating the basic failure rate of the intelligent electric energy meter;
step 3, acquiring field fault data of intelligent electric energy meters of the same model and different batches, and calculating the actual failure rate of the intelligent electric energy meters;
step 4, building a deep belief network model, determining the input of the network as the basic failure rate, the field environment temperature and the field working electrical stress level of the intelligent electric energy meter in the step 2, and the output as the actual failure rate of the intelligent electric energy meter in the step 3;
and 5, training the deep belief network model set up in the step 4, inputting the basic failure rate, the real-time field temperature and the real-time electric stress level of the electric energy meter to be predicted into the network, wherein the output of the network is the predicted failure rate of the electric energy meter.
2. The intelligent electric energy meter failure rate online prediction method based on the deep belief network as claimed in claim 1, characterized in that: the method for calculating the basic failure rate of the components in the step 1 comprises the following steps:
aiming at key components of which the metering chip, the communication chip, the clock chip and the battery have important influence on the reliability of the intelligent electric energy meter, the basic failure rate is obtained through high-temperature working life detection;
according to the JESD74A standard, the basic failure rate model formula of the key components is as follows:
Figure FDA0003059747100000011
in the formula, x2Representing the chi-square distribution function, c is the confidence level, d is the degree of freedom, the degree of freedom is equal to 2 times the number of failures plus 2, A is the acceleration factor, N is the number of samples participating in the test, t is the number of samples participating in the testATo speed up test time; the acceleration factor a can be found from the following model:
the effect of temperature stress can be represented by an Arrhenius model, and the temperature acceleration coefficient AtThe calculation formula is as follows:
Figure FDA0003059747100000021
where Ea is activation energy of the reaction, k is boltzmann constant, Tu is absolute temperature in a normal state of the device, and Ta is absolute temperature in an accelerated state of the device.
The action of the electric stress can be represented by an Eying model, and the acceleration coefficient of the electric stress AvThe calculation formula is as follows:
Av＝eZ|Vu-Va| (3)
in the formula, Z is a voltage acceleration constant, Vu is a voltage in a normal state of the component, and Va is a voltage in an accelerated state of the component.
The final acceleration factor a is the product of the above acceleration factors, i.e.:
A＝AtAv (4)
the basic failure rate of the rest components in the intelligent electric energy meter is obtained by inquiring the GJB/Z299C manual.
3. The intelligent electric energy meter failure rate online prediction method based on the deep belief network as claimed in claim 1, characterized in that: the specific method of the step 2 comprises the following steps:
adding the basic failure rates of all components in the intelligent electric energy meter, defining the added value as the basic failure rate of the intelligent electric energy meter, and adopting the expression as follows:
Figure FDA0003059747100000022
in the formula, λbmIs the basic failure rate, lambda, of the intelligent electric energy meterb_iThe number n is the number of the components contained in the intelligent electric energy meter.
4. The intelligent electric energy meter failure rate online prediction method based on the deep belief network as claimed in claim 1, characterized in that: the specific method of the step 3 comprises the following steps:
the failure rate of the intelligent electric energy meter is calculated after the field fault data of the intelligent electric energy meter is preprocessed, the failure rate is defined as the actual failure rate of the intelligent electric energy meter, and the calculation method comprises the following steps:
the failure probability distribution function F (t) of the intelligent electric energy meter is as follows:
Figure FDA0003059747100000031
wherein s is a shape parameter, t0Is a scale parameter.
Aiming at the failure data of the electric energy meters in the same batch, the failure probability distribution function of the intelligent electric energy meter is calculated by adopting the following formula:
Figure FDA0003059747100000032
in the formula, tiFor the i-th detection time, riIs at tiCumulative number of failures, N, of electric energy meter in time0The total number of the electric energy meters in the batch.
Transformed by equation (6) and equation (7):
Figure FDA0003059747100000033
s and t can be estimated by using least square method0The parameter value of (2).
Reliable service life t of intelligent electric energy meterRThe calculation expression is:
Figure FDA0003059747100000034
actual failure rate lambda of intelligent electric energy metercThe calculation expression is:
Figure FDA0003059747100000035
5. the intelligent electric energy meter failure rate online prediction method based on the deep belief network as claimed in claim 1, characterized in that: the basic structure of the deep belief network model in the step 4 is as follows:
the basic structure of the deep belief network DBN is formed by stacking limited Boltzmann machines RBMs, each RBM consists of a visible layer and a hidden layer, and a bidirectional full-connection structure is formed between the layers; the visible layer is composed of m visible units viE {0,1} (i ═ 1,2, …, m), the hidden layer is composed of n hidden units hjE {0,1} (j ═ 1,2, …, n), and the connection weight between layers is W ═ Wij}∈Rm ×nThe bias of the visible layer is B ═ Bi}∈RmThe offset of the hidden layer is C ═ Cj}∈Rn. RBM is determined by these three parameters θ ═ W, B, C.
6. The intelligent electric energy meter failure rate online prediction method based on the deep belief network as claimed in claim 1, characterized in that: the specific method of the step 5 comprises the following steps:
for visible and hidden layers that obey bernoulli distribution, the energy function of the RBM is:
Figure FDA0003059747100000041
the probability of the hidden layer unit being activated is:
Figure FDA0003059747100000042
wherein, sigma is Sigmoid function;
the probability of the visible layer unit being activated is:
Figure FDA0003059747100000043
updating the parameters by contrast divergence algorithm, and taking the input data as visibleLayer unit viCalculating a hidden layer unit h according to the formula (12)jThen, an updated visible layer unit v 'is calculated from formula (13)'iAnd then, the updated hidden unit h 'is calculated from the formula (12)'j(ii) a The expression of the parameter update is:
Figure FDA0003059747100000044
in the formula, epsilonCDIn order to obtain a learning rate,<·>is a mathematical expectation;
in order to accelerate the model training speed, dividing a training data set into a plurality of groups; for the pth group, the expression for the specific parameter update is:
Figure FDA0003059747100000051
CN202110509527.7A 2021-05-11 2021-05-11 Intelligent electric energy meter failure rate online prediction method based on deep belief network Pending CN113378453A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110509527.7A CN113378453A (en) 2021-05-11 2021-05-11 Intelligent electric energy meter failure rate online prediction method based on deep belief network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110509527.7A CN113378453A (en) 2021-05-11 2021-05-11 Intelligent electric energy meter failure rate online prediction method based on deep belief network

Publications (1)

Publication Number Publication Date
CN113378453A true CN113378453A (en) 2021-09-10

Family

ID=77572403

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110509527.7A Pending CN113378453A (en) 2021-05-11 2021-05-11 Intelligent electric energy meter failure rate online prediction method based on deep belief network

Country Status (1)

Country Link
CN (1) CN113378453A (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104182772A (en) * 2014-08-19 2014-12-03 大连理工大学 Gesture recognition method based on deep learning
CN109101750A (en) * 2018-08-30 2018-12-28 电子科技大学 A kind of accelerated aging prediction technique under electronic device hygrothermal environment
US20190285694A1 (en) * 2016-08-03 2019-09-19 Stmicroelectronics (Crolles 2) Sas Method for estimating an operating profile of an integrated circuit of a system-on-a-chip, and corresponding system-on-a-chip
CN112363099A (en) * 2020-10-30 2021-02-12 天津大学 TMR current sensor temperature drift and geomagnetic field correction device and method
CN112667957A (en) * 2020-12-03 2021-04-16 国网天津市电力公司营销服务中心 Intelligent electric energy meter failure rate prediction method based on deep neural network

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104182772A (en) * 2014-08-19 2014-12-03 大连理工大学 Gesture recognition method based on deep learning
US20190285694A1 (en) * 2016-08-03 2019-09-19 Stmicroelectronics (Crolles 2) Sas Method for estimating an operating profile of an integrated circuit of a system-on-a-chip, and corresponding system-on-a-chip
CN109101750A (en) * 2018-08-30 2018-12-28 电子科技大学 A kind of accelerated aging prediction technique under electronic device hygrothermal environment
CN112363099A (en) * 2020-10-30 2021-02-12 天津大学 TMR current sensor temperature drift and geomagnetic field correction device and method
CN112667957A (en) * 2020-12-03 2021-04-16 国网天津市电力公司营销服务中心 Intelligent electric energy meter failure rate prediction method based on deep neural network

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
NANO_022: "JEDEC STANDARD", 《百度文库》 *

Similar Documents

Publication Publication Date Title
CN109214708B (en) Electric power system risk assessment method based on cross entropy theory optimization support vector machine
CN110555230B (en) Rotary machine residual life prediction method based on integrated GMDH framework
Cai et al. Artificial intelligence enhanced reliability assessment methodology with small samples
CN111222549A (en) Unmanned aerial vehicle fault prediction method based on deep neural network
CN111259551A (en) Inertial measurement unit fault prediction method based on confidence rule base
CN111523778A (en) Power grid operation safety assessment method based on particle swarm algorithm and gradient lifting tree
CN113591379B (en) Auxiliary decision-making method for transient stability prevention and emergency coordination control of power system
CN111967183A (en) Method and system for calculating line loss of distribution network area
CN112307677A (en) Power grid oscillation mode evaluation and safety active early warning method based on deep learning
CN111814284A (en) On-line voltage stability evaluation method based on correlation detection and improved random forest
CN112651519A (en) Secondary equipment fault positioning method and system based on deep learning theory
CN114089255B (en) Stability evaluation method for capacitor voltage transformer
CN115238573A (en) Hydroelectric generating set performance degradation trend prediction method and system considering working condition parameters
CN110889646B (en) Method for checking correctness of operating parameters of generator set adapting to working condition change
CN113378453A (en) Intelligent electric energy meter failure rate online prediction method based on deep belief network
CN105447228A (en) Earth disturbing gravity assignment model applicability evaluating method
Chen et al. A performance evaluation method based on interval evidential reasoning approach with function monotonicity
CN109101759A (en) A kind of parameter identification method based on forward and reverse response phase method
CN115062533A (en) Construction and prediction method and medium of multiple regression prediction model
Li et al. Method for predicting failure rate of airborne equipment based on optimal combination model
CN113901649A (en) Load parameter identification method, system and storage medium in frequency simulation
CN112667957A (en) Intelligent electric energy meter failure rate prediction method based on deep neural network
CN116595883B (en) Real-time online system state correction method for numerical reactor
CN117113886B (en) Pressure prediction method and device
CN109376957A (en) A kind of prediction technique of thermal power plant&#39;s load

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20210910

RJ01 Rejection of invention patent application after publication