CN117074913A - Circuit board V-I curve uncertainty measurement method based on multi-objective optimization interval - Google Patents

Circuit board V-I curve uncertainty measurement method based on multi-objective optimization interval Download PDF

Info

Publication number
CN117074913A
CN117074913A CN202311045960.5A CN202311045960A CN117074913A CN 117074913 A CN117074913 A CN 117074913A CN 202311045960 A CN202311045960 A CN 202311045960A CN 117074913 A CN117074913 A CN 117074913A
Authority
CN
China
Prior art keywords
interval
data
curve
training
prediction
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
CN202311045960.5A
Other languages
Chinese (zh)
Inventor
林鹏
赵岩
赵铮
曹九稳
潘庆国
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.)
Hangzhou Dianzi University
Original Assignee
Hangzhou Dianzi 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 Hangzhou Dianzi University filed Critical Hangzhou Dianzi University
Priority to CN202311045960.5A priority Critical patent/CN117074913A/en
Publication of CN117074913A publication Critical patent/CN117074913A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/28—Testing of electronic circuits, e.g. by signal tracer
    • G01R31/2801—Testing of printed circuits, backplanes, motherboards, hybrid circuits or carriers for multichip packages [MCP]
    • G01R31/281—Specific types of tests or tests for a specific type of fault, e.g. thermal mapping, shorts testing
    • G01R31/2812—Checking for open circuits or shorts, e.g. solder bridges; Testing conductivity, resistivity or impedance
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/28—Testing of electronic circuits, e.g. by signal tracer
    • G01R31/2801—Testing of printed circuits, backplanes, motherboards, hybrid circuits or carriers for multichip packages [MCP]
    • G01R31/281—Specific types of tests or tests for a specific type of fault, e.g. thermal mapping, shorts testing
    • G01R31/2817—Environmental-, stress-, or burn-in tests
    • 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
    • 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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent

Landscapes

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

Abstract

The invention discloses a circuit board V-I curve uncertainty measurement method based on a multi-objective optimization interval. Aiming at the condition that the V-I curve data of the same measuring point, which is acquired by compensating zero crossing points of the acquisition equipment, have the same shape but are offset up and down, a maximum and minimum normalization method is adopted to eliminate the offset. Aiming at the situation that the deviation of the predicted interval is large due to different data distribution, a loss function containing deviation information is constructed to cover the interval on the low-density data, and fault detection research of the circuit board is further facilitated.

Description

Circuit board V-I curve uncertainty measurement method based on multi-objective optimization interval
Technical Field
The invention relates to the field of fault diagnosis of circuit boards, in particular to uncertainty in voltage-current (V-I) curve test of a circuit board, and in the case of data deviation and low density, a section prediction uncertainty measurement method is adopted to obtain a multi-target V-I section with high coverage rate, narrow width and small deviation.
Background
The V-I curve test is a failure diagnosis technology of a circuit board without power-up, and is suitable for comparison test of any device. Since device failures are typically accompanied by variations in impedance characteristics between pins, by directly observing or comparing V-I curves between the same nodes of a normal circuit board and a failed circuit board, the node of the impedance characteristic variation can be found, thereby determining the failed device.
Uncertainty in the V-I curve test can lead to a deviation between the measured value and the true value, affecting the accuracy of normal circuit board measurement data and accurate assessment of V-I curve performance. Factors such as changes in the measurement environment, precision and accuracy of the measurement equipment, inaccuracy of calibration or instrument drift, batch differences in the quality and characteristics of the components, etc., cause fluctuations or drift in the measured values. Interval prediction is a method for considering uncertainty in statistics, and can provide an upper bound and a lower bound of corresponding current under each voltage of a V-I curve, and the upper bound and the lower bound have a certain probability to contain a true value of the current. Common interval prediction methods are Delta method, bayesian method, mean variance estimation method, bootstrap method and LUBE method (see Hadjicharalambous M, polycarbou M, panayiotou C.G.neural network-based construction of online prediction intervals [ J ]. Neural Computing and Applications, 2020). Among them, the Delta method requires a large amount of computation and is difficult to be effectively applied in practice. The bayesian method has low accuracy in calculating the prediction interval because each parameter needs to be pre-assigned with a distribution, which requires a large amount of calculation. The mean variance estimation method may result in a smaller coverage of the prediction window, which may result in too narrow window to cover the true value. Bootstrap methods may require significant computing resources to re-sample the process, require high computing devices, and are time consuming. The above methods all assume that the data follows some a priori distribution and then calculate the upper and lower bounds of the interval into which the data may fall at a given probability level. The LUBE (Lower-upper bound estimation) method is a section prediction method based on a neural network, and under any data distribution condition, sections with high coverage rate and narrow average width are directly output. However, the LUBE method loss function is not trivial, and only non-gradient algorithms can be used in training, while gradient descent is the standard method of training neural networks. In addition, the LUBE method only considers the width and coverage index of the prediction interval, ignoring the deviation of the prediction interval, namely the average error between the prediction value and the true value. In practical applications, even if the interval has a high coverage and a narrow width, if the deviation thereof is large, the predicted value is likely to be far from the true value in a plurality of predictions. Such prediction intervals may not be reliable in actual decisions and predictions.
Disclosure of Invention
Aiming at uncertainty of circuit board V-I curve data, the invention provides a circuit board V-I curve uncertainty measurement method based on a multi-objective optimization interval. The method can measure uncertainty of the V-I curve aiming at the condition that offset and low density exist in the V-I curve data, and can realize the determination of the upper bound and the lower bound of the V-I curve section. By constructing a new micro-loss function, the determination of the upper and lower boundaries of the interval is realized by adopting a gradient descent training neural network mode in consideration of three aspects of high coverage, narrow width and small deviation.
The technical problems to be solved by the invention are realized by adopting the following technical scheme:
the circuit board V-I curve uncertainty measurement method based on a multi-objective optimization interval, wherein the multi-objective optimization interval is an interval which is optimized through three indexes of coverage rate, interval width and deviation, and comprises the following steps:
step (1), collecting V-I curves of circuit boards with the same function and different batches, and obtaining V-I curve data of a plurality of measuring nodes of a plurality of normal circuit boards;
judging the type of the V-I curve data, wherein the type comprises two types of approximate single-value functions and multiple-value functions, and cutting off the multiple-value function type into two approximate single-value function types according to a voltage rising section and a voltage falling section;
and (3) in order to solve the problem of up-down offset of data caused by the possible overcompensation problem of the data acquisition equipment, normalizing the approximate single-value function type data obtained in the step (2) to eliminate the data offset problem. Dividing the preprocessed approximate single-value function type data into a training set and a testing set according to a ratio of 7:3 to obtain a training set S= { (v) 1 ,i 1 ),...,(v p ,i p ) Sum test set t= { (v) p+1 ,i p+1 ),...,(v n ,i n ) V represents voltage, i represents current, n represents the total number of data points;
step (4), introducing an optimization target of prediction interval deviation information into a loss function, constructing a deep neural network model to determine the upper boundary and the lower boundary of a V-I curve interval, and automatically acquiring the V-I curve interval with high coverage, narrow width and small deviation characteristics; the deep neural network model consists of an input layer, a plurality of hidden layers and an output layer.
Step (5), inputting the training data in the step (3) into a deep neural network model for training and optimizing network parameters, and then inputting the test data into the trained deep neural network model to obtain a prediction interval of a V-I curve;
and (6) performing multiple simulation analysis on the data in order to acquire the reliability of the interval, and taking the average value of the results as a final interval.
In the method, in the step (2), the voltage rising section and the voltage falling section of the V-I curve are divided for the multi-valued function type, and two approximate single-valued function type data are generated for training and prediction independently.
In the method, in the step (3), the maximum and minimum normalization methods are used to normalize the voltage and current values in the V-I curve to be within the interval (0, 1).
In the method, the loss function of the neural network in the step (4) is a multi-objective loss function construction process including coverage rate, interval width and data deviation information, and the multi-objective loss function construction process is as follows:
the upper and lower boundaries of the prediction interval are respectivelyAnd->The interval should have the desired observation ratio (1-a), a common choice of a being 0.01 or 0.05,
a vector k of length n indicates whether each data point is covered by the estimated interval, each element k i E {0,1} is determined by the following formula:
defining the data volume captured by the interval as c:
the prediction interval coverage and average width are defined as follows:
the prediction interval should be reduced as much as possible by the average width MPIW, on the premise of satisfying the coverage PICP ∈ (1-a), so that only the MPIW of the covered data point area needs to be considered, defined as,
boolean variable k exists in original PICP calculation mode i The loss function formed by the method can not be micro in the gradient descending process, so that the sigmoid function is adopted for softening, the loss function can be micro, and the calculation formula is as follows:
wherein σ is a sigmoid function, s 1 Is a super parameter of softening.
On the basis of meeting the coverage rate, the method realizes narrow width and small deviation. The high-quality optimization principle fully considers the deviation relation between data points and the upper and lower boundaries of a prediction interval, a PISD optimization target is added into a loss function, and in order to quantify uncertainty and provide the optimal PI based on the high-quality optimization principle, the loss function is defined as follows:
wherein lambda is 1 And lambda (lambda) 2 For coverage and data to interval deviation optimization objective balance parameters, PISD represents the sum of deviation of normalized data points from predicted interval, PISD for each data point is defined as follows:
wherein s is 2 Is a super parameter. PISD is defined as follows:
in the method, the neural network training in the step (5) includes the following steps:
(5.1) aiming at V-I data of corresponding measuring nodes, constructing a nonlinear mapping relation between voltage and upper and lower bounds by using the deep neural network in the step (4), wherein each layer uses the output of the previous layer as the input of the device, and two output units of the output layer are an upper bound and a lower bound of a section;
(5.2) selecting V-I curve training data to perform neural network training, wherein the training data comprises two processes of forward propagation of signals and backward propagation of errors;
(5.3) in forward propagation, the input signal acts on the output node through the hidden layer, and the output signal is generated through nonlinear transformation, if the actual output does not accord with the expected output, the reverse propagation process of the error is shifted;
during error back transmission, the output errors are back transmitted layer by layer to the input layers through the hidden layers, the errors are distributed to all units of each layer, and error signals obtained from each layer are used as the basis for adjusting the weight of each unit;
and (5.4) reducing the error along the gradient direction by adjusting the connection strength of the input node and the hidden layer node and the connection strength of the hidden layer node and the output node as well as the threshold value, and determining the weight and the threshold value corresponding to the minimum error through repeated learning and training, and stopping training.
In the above method, the specific method for generating the upper bound and the lower bound of the final interval in the step (6) is as follows:
and (3) performing multiple simulation analysis, namely performing the experimental process in the step (5), and then respectively taking average values of the upper bound and the lower bound of the prediction interval to represent the final prediction interval. The calculation process is as follows,
wherein,and->The upper and lower bounds of the final prediction interval are represented, and m represents the number of repetitions.
The beneficial effects of the invention are as follows:
1. by introducing the optimization target of the prediction interval deviation information into the loss function, an improved prediction interval optimization framework is established, and a new thought is provided for prediction interval optimization. The deviation relation between the data points and the upper and lower boundaries of the interval is considered, and the generated prediction interval can more comprehensively capture the appointed part of data, so that the prediction risk is reduced.
2. The construction of the optimal prediction interval is linked to the uncertainty estimate. By training the deep neural network by using the loss function, better prediction precision and uncertainty estimation can be obtained.
3. The problem of vertical offset in the V-I curve is solved by using a maximum and minimum normalization method, the interval prediction technology is combined with the strong learning capacity of the deep learning model and is introduced into the V-I curve detection of the circuit board, and a high-quality prediction interval is constructed to quantify the uncertainty of the V-I curve. A new V-I curve prediction framework is provided, providing more information about the uncertainty of the V-I curve. The research content of the V-I curve prediction is enriched, and the prediction interval can provide more comprehensive information for a decision maker.
4. The uncertainty of the V-I curve collected at the measuring point in the prediction interval quantification is used for assisting in judging whether the measuring point on the test circuit board fails or not, so that the risk caused by comparison and detection of the original V-I curve is reduced.
Drawings
FIG. 1 is a schematic diagram of a V-I curve uncertainty quantization process in an embodiment of the present invention;
FIG. 2 is a data sampling circuit board in an embodiment of the invention;
FIG. 3 is a graph showing the up-and-down shift of the raw data of the V-I curve according to the embodiment of the present invention.
FIG. 4 is a different illustration of the distribution of data points of the V-I curve in an embodiment of the present invention;
FIG. 5 is a block diagram of a deep neural network in an embodiment of the present invention;
FIG. 6 is a sample of prediction intervals of V-I curve data in an embodiment of the invention.
Detailed Description
The invention will be further described with reference to the drawings and examples.
In order to make the technical solution of the present invention better understood by a person skilled in the art, the present invention will be more clearly and more fully described below with reference to the accompanying drawings in the embodiments, and of course, the described embodiments are only a part of, but not all of, the present invention, and other embodiments obtained by a person skilled in the art without making any inventive effort are within the scope of the present invention.
The embodiment is a circuit board V-I curve uncertainty measurement method based on a multi-objective optimization interval, which is characterized in that a V-I curve of a normal circuit board measuring point is collected, data preprocessing is carried out, V-I curve data is divided into a training set and a testing set, an optimization objective of a deep neural network is a multi-objective loss function formed by coverage rate, interval width, data point and interval boundary deviation, the training data is used for training the neural network, the neural network has the capacity of quantifying the V-I curve uncertainty through a prediction interval, and the testing data is used for testing the capacity of the neural network for interval prediction. And the uncertainty of the V-I curve caused by the characteristics of components or environmental changes is quantified, and the reliability of the fault detection method based on the V-I curve is improved. FIG. 2 is a data sampling circuit board in an embodiment of the invention;
as shown in fig. 1, the method for measuring uncertainty of the V-I curve of the circuit board based on the multi-objective optimization interval comprises the following steps:
(1) And inputting sine voltage excitation of one period to normal circuit boards of different batches according to the same principle for measurement, obtaining batch circuit board V-I curve data, wherein each V-I curve is composed of 100 data points consisting of voltage and current, and 500V-I curves are acquired for the same measuring point in a plurality of circuit boards.
(2) Judging the type of the V-I curve of the circuit board, wherein the measuring points are connected with a plurality of components in parallel, so that the shape of the V-I curve can be in various conditions, but from the practical application perspective, the types are classified into two types of approximate single-value functions and multiple-value functions, the approximate single-value functions show that a large blank area does not exist between Y values corresponding to X, X and Y correspond to voltage and current in the V-I curve respectively, and the V-I curve approximates the single-value functions. The multi-valued function type indicates that the Y value corresponding to X is divided into two areas, and a blank area exists between the two areas. In order to avoid the influence of the blank area on detection, the multi-valued function type is truncated according to the rising stage of the voltage and the falling stage of the voltage to generate two single-valued function type data;
(3) There may be overcompensation of the acquisition device resulting in data up-down offset problems, as shown in fig. 3. Performing maximum and minimum normalization processing on the up-down offset problem, normalizing the voltage and current values in the approximate single-value function type data obtained in the step (2) into (0, 1), solving the up-down offset problem and accelerating the neural network calculation, dividing the preprocessed approximate single-value function type data into a training set and a testing set according to a ratio of 7:3, and obtaining a training set S= { (v) 1 ,i 1 ),...,(v p ,i p ) Sum test set t= { (v) p+1 ,i p+1 ),...,(v n ,i n ) The method comprises the steps of (1) setting a value of p to 0.7 x n, wherein a training set is used for training a current network model, optimizing neural network model parameters towards a direction of decreasing a loss function, and a testing set is used for testing the performance of an optimal model obtained by training of the training set; FIG. 4 is a different illustration of the distribution of data points of the V-I curve in an embodiment of the present invention;
(4) Introducing an optimization target of prediction interval deviation information into the loss function, constructing an interval prediction optimization algorithm based on a deep neural network, optimizing a prediction interval based on the principles of high coverage, narrow width and small deviation, and optimizing the coverage condition of a low-density region of V-I curve data. The network structure is shown in fig. 5, and the deep neural network model is composed of an input layer, a plurality of hidden layers and an output layer, wherein the hidden layers comprise a plurality of middle layers, and each middle layer comprises a plurality of neurons for calculation. The configuration parameters of the neural network are shown in table 1, wherein the optimizer uses adam, the learning rate is 0.03, the learning rate decay rate is 0.95, the activation function is relu, the depth of the hidden layer is 6 layers, the training mode adopts batch training, the data size of each batch is 100, and the iteration number is 50.
(5) Inputting the training data in the step (3) into a neural network for training and optimizing neuron parameters in the neural network, and then inputting the test data into the neural network to automatically obtain a prediction interval of the V-I curve.
(6) In order to obtain the reliability of the interval, the data is subjected to repeated simulation analysis for more than 5 times, and the embodiment adopts repeated simulation analysis for 20 times, and takes the weighted average of the prediction interval obtained for 20 times as a final interval. The comparison example of the experimental results of the V-I curve is shown in the figure 6, wherein the black star-shaped line is the implementation effect of the method, the gray common line is the implementation effect of the prior method, the situation that the upper bound and the lower bound of the black star-shaped line cover the low-density data area is obviously superior to the upper bound and the lower bound of the gray common line can be obviously observed from the figure, the method can effectively solve the problem of low density of the data, and the generated prediction interval can realize high coverage, narrow width and small deviation and is superior to the prior method.
The optimization algorithm in step (4) of this embodiment is to construct a loss function based on the coverage, interval width, data point and predicted interval deviation relationship The construction steps are as follows:
the upper and lower boundaries of the prediction interval areAnd->The interval should have an ideal observation ratio (1Alpha), a common alpha is chosen to be 0.01 or 0.05,
a vector k of length n indicates whether each point is covered by the estimated interval, each element k i E {0,1} is determined by the following formula,
the Sigmoid function formula is as follows,
the amount of data captured for the interval is defined as c,
the prediction interval coverage and the average width are defined as follows,
the prediction interval should be reduced as much as possible by the MPIW if PICP ∈ (1- α) is satisfied, so that only the MPIW of the covered point region needs to be considered, defined as,
boolean variable k exists in original PICP calculation mode i Constituted byThe loss function can cause irreducibility in the gradient descent process, so that the sigmoid function is adopted for softening, so that the loss function is differentiable, the calculation formula is as follows,
wherein σ is a sigmoid function, s 1 Is a super parameter of softening.
The high-quality optimization principle is defined as realizing narrow width and small deviation on the basis of meeting coverage rate. The high quality optimization principle fully considers the deviation relation between the data points and the upper and lower boundaries of the prediction interval, adds a PISD optimization target into the loss function, redefines the loss function as follows in order to quantify the uncertainty and provide the optimal PI based on the high quality optimization principle,
wherein lambda is 1 And lambda (lambda) 2 As a balance parameter of the coverage and data-to-interval deviation optimization target, PISD represents the sum of deviations of normalized data points from a predicted interval, PISD of each point is defined as follows,
wherein s is 2 Is a super parameter. The PISD is defined as follows,
the neural network calculation in step (5) of this embodiment includes the following steps:
(5.1) aiming at V-I data of corresponding measuring points, constructing a nonlinear mapping relation between voltage and upper and lower bounds by using the deep neural network in the step (4), wherein each layer uses the output of the previous layer as its own input, the input of the input layer is voltage data, and two output units of the output layer are interval upper bounds and lower bounds;
(5.2) selecting V-I curve training data to perform neural network training, wherein the training data comprises two processes of forward propagation of signals and backward propagation of errors;
(5.3) in forward propagation, the input signal acts on the output node through the hidden layer, and the output signal is generated through nonlinear transformation, if the actual output does not accord with the expected output, the reverse propagation process of the error is shifted;
during error back transmission, the output errors are back transmitted layer by layer to the input layers through the hidden layers, the errors are distributed to all units of each layer, and error signals obtained from each layer are used as the basis for adjusting the weight of each unit;
and (5.4) reducing the error along the gradient direction by adjusting the connection strength of the input node and the hidden layer node and the connection strength of the hidden layer node and the output node as well as the threshold value, and determining the weight and the threshold value corresponding to the minimum error through repeated learning and training, and stopping training.
In this embodiment, the final section process is generated in step (6). In order to reduce the randomness of model prediction and the uncertainty of learning models, the simulation experiment is repeated 20 times, and then the average value of the prediction interval of 20 times is used for representing the final prediction interval, the calculation process is as follows,
wherein,and->Representing the upper and lower bounds of the final prediction interval, m=20 represents the number of repetitions.
The foregoing has shown and described the basic principles, principal features and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and that the above embodiments and descriptions are merely illustrative of the principles of the present invention, and various changes and modifications may be made without departing from the spirit and scope of the invention, which is defined in the appended claims. The scope of the invention is defined by the appended claims and equivalents thereof.

Claims (6)

1. The circuit board V-I curve uncertainty measurement method based on the multi-objective optimization interval is characterized by comprising the following steps of:
step (1), collecting V-I curves of circuit boards with the same function and different batches, and obtaining V-I curve data of a plurality of measuring nodes of a plurality of normal circuit boards;
judging the type of the V-I curve data, wherein the type comprises two types of approximate single-value functions and multiple-value functions, and cutting off the multiple-value function type into two approximate single-value function types according to a voltage rising section and a voltage falling section;
step (3), in order to solve the problem of up-down offset of data caused by possible overcompensation of the data acquisition equipment, normalizing the approximate single-valued function type data obtained in the step (2) to eliminate the problem of data offset; dividing the preprocessed approximate single-value function type data into a training set and a testing set according to a ratio of 7:3 to obtain a training set S= { (v) 1 ,i 1 ),...,(v p ,i p ) Sum test set t= { (v) p+1 ,i p+1 ),...,(v n ,i n ) V represents voltage, i represents current, n represents the total number of data points;
step (4), introducing an optimization target of prediction interval deviation information into a loss function, constructing a deep neural network model to determine the upper boundary and the lower boundary of a V-I curve interval, and automatically acquiring the V-I curve interval with high coverage, narrow width and small deviation characteristics; the deep neural network model consists of an input layer, a plurality of hidden layers and an output layer;
step (5), inputting the training data in the step (3) into a deep neural network model for training and optimizing network parameters, and then inputting the test data into the trained deep neural network model to obtain a prediction interval of a V-I curve;
and (6) performing multiple simulation analysis on the data in order to acquire the reliability of the interval, and taking the average value of the results as a final interval.
2. The method for measuring uncertainty of a circuit board V-I curve based on a multi-objective optimization interval according to claim 1, wherein in the step (2), a voltage rising section and a voltage falling section of the V-I curve are divided for a multi-valued function type, and two approximate single-valued function type data are generated for training and prediction independently.
3. The method of claim 1, wherein the voltage and current values in the V-I curve are normalized to within the interval (0, 1) using a maximum and minimum normalization method in step (3).
4. The method for uncertainty measurement of a circuit board V-I curve based on a multi-objective optimization interval according to claim 1, wherein the loss function of the neural network in the step (4) is a multi-objective loss function including coverage, interval width and data deviation information, and the construction process is as follows:
the upper and lower boundaries of the prediction interval are respectivelyAnd->The interval should have an ideal observation ratio (1- α):
length ofThe vector k of n indicates whether each data point is covered by the estimated interval, each element k i E {0,1} is determined by the following formula:
defining the data volume captured by the interval as c:
the prediction interval coverage and average width are defined as follows:
the prediction interval should be reduced as much as possible by the average width MPIW, on the premise of satisfying the coverage PICP ∈ (1-a), so that only the MPIW of the covered data point area needs to be considered, defined as,
boolean variable k exists in original PICP calculation mode i The loss function formed by the method can not be micro in the gradient descending process, so that the sigmoid function is adopted for softening, the loss function can be micro, and the calculation formula is as follows:
wherein σ is a sigmoid function, s 1 Is a super parameter of softening;
on the basis of meeting the coverage rate, realizing narrow width and small deviation; the high-quality optimization principle fully considers the deviation relation between data points and the upper and lower boundaries of a prediction interval, a PISD optimization target is added into a loss function, and in order to quantify uncertainty and provide the optimal PI based on the high-quality optimization principle, the loss function is defined as follows:
wherein lambda is 1 And lambda (lambda) 2 For coverage and data to interval deviation optimization objective balance parameters, PISD represents the sum of deviation of normalized data points from predicted interval, PISD for each data point is defined as follows:
wherein s is 2 Is a super parameter; PISD is defined as follows:
5. the method for uncertainty measurement of a V-I curve of a circuit board based on a multi-objective optimization interval according to claim 1 or 4, wherein the training of the neural network in the step (5) comprises the following steps:
(5.1) aiming at V-I data of corresponding measuring nodes, constructing a nonlinear mapping relation between voltage and upper and lower bounds by using the deep neural network in the step (4), wherein each layer uses the output of the previous layer as the input of the device, and two output units of the output layer are an upper bound and a lower bound of a section;
(5.2) selecting V-I curve training data to perform neural network training, wherein the training data comprises two processes of forward propagation of signals and backward propagation of errors;
(5.3) in forward propagation, the input signal acts on the output node through the hidden layer, and the output signal is generated through nonlinear transformation, if the actual output does not accord with the expected output, the reverse propagation process of the error is shifted;
during error back transmission, the output errors are back transmitted layer by layer to the input layers through the hidden layers, the errors are distributed to all units of each layer, and error signals obtained from each layer are used as the basis for adjusting the weight of each unit;
and (5.4) reducing the error along the gradient direction by adjusting the connection strength of the input node and the hidden layer node and the connection strength of the hidden layer node and the output node as well as the threshold value, and determining the weight and the threshold value corresponding to the minimum error through repeated learning and training, and stopping training.
6. The method for uncertainty measurement of a circuit board V-I curve based on a multi-objective optimization interval according to claim 5, wherein the specific method for generating the upper bound and the lower bound of the final interval in the step (6) is as follows:
performing multiple simulation analysis, namely, the experimental process in the step (5), and then respectively taking average values of an upper bound and a lower bound of a prediction interval to represent a final prediction interval; the calculation process is as follows,
wherein,and->The upper and lower bounds of the final prediction interval are represented, and m represents the number of repetitions.
CN202311045960.5A 2023-08-18 2023-08-18 Circuit board V-I curve uncertainty measurement method based on multi-objective optimization interval Pending CN117074913A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311045960.5A CN117074913A (en) 2023-08-18 2023-08-18 Circuit board V-I curve uncertainty measurement method based on multi-objective optimization interval

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311045960.5A CN117074913A (en) 2023-08-18 2023-08-18 Circuit board V-I curve uncertainty measurement method based on multi-objective optimization interval

Publications (1)

Publication Number Publication Date
CN117074913A true CN117074913A (en) 2023-11-17

Family

ID=88703735

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311045960.5A Pending CN117074913A (en) 2023-08-18 2023-08-18 Circuit board V-I curve uncertainty measurement method based on multi-objective optimization interval

Country Status (1)

Country Link
CN (1) CN117074913A (en)

Similar Documents

Publication Publication Date Title
CN109359385B (en) Training method and device for service quality assessment model
CN108520357B (en) Method and device for judging line loss abnormality reason and server
CN110852515B (en) Water quality index prediction method based on mixed long-time and short-time memory neural network
CN114297036B (en) Data processing method, device, electronic equipment and readable storage medium
CN116380445B (en) Equipment state diagnosis method and related device based on vibration waveform
CN109389145A (en) Electric energy meter production firm evaluation method based on metering big data Clustering Model
CN110879927B (en) Sea clutter amplitude statistical distribution on-site modeling method for sea target detection
CN112116198A (en) Data-driven process industrial state perception network key node screening method
CN111931983A (en) Precipitation prediction method and system
CN110321493A (en) A kind of abnormality detection of social networks and optimization method, system and computer equipment
CN115130546A (en) Abnormal parameter detection method and device, computer readable medium and electronic equipment
CN117708669A (en) Multi-source data-based adaptive probability neural network high-voltage cable state evaluation method
CN117709392A (en) Automatic multichannel ADC chip performance evaluation test method and system
CN117556198B (en) Apparent resistivity denoising calculation method based on full-waveform well ground method
CN113151842B (en) Method and device for determining conversion efficiency of wind-solar complementary water electrolysis hydrogen production
CN113158448A (en) Self-adaptive calculation method for equipment threshold of ship system
CN117233638A (en) Lithium battery SOH detection method and device, electronic equipment and storage medium
CN115031794B (en) Novel gas-solid two-phase flow measuring method based on multi-feature graph convolution
CN115618743B (en) State evaluation method and state evaluation system of sighting telescope system
CN117074913A (en) Circuit board V-I curve uncertainty measurement method based on multi-objective optimization interval
CN116431346A (en) Compensation method for main memory capacity of electronic equipment
CN116128361A (en) Multistage efficiency calculation method, device, equipment and medium for intelligent reconnaissance system
CN115907178A (en) Clean ecosystem CO 2 Method for predicting exchange amount
CN115359197A (en) Geological curved surface reconstruction method based on spatial autocorrelation neural network
CN112580853A (en) Bus short-term load prediction method based on radial basis function neural network

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