An ANN chip current fault prediction algorithm based on near-field scanning

Abstract

The application relates to the technical field of current fault prediction, in particular to an ANN chip current fault prediction algorithm based on near-field scanning, which is realized based on a near-field scanning system and characterized by the following steps: white-box testing and analysis model, current estimation principle, construction and training of an artificial neural network model, and fault detection application. The current estimation method of the application is applied to fault detection of a power supply network, the current distribution of the power supply network is estimated by solving a quadratic programming problem, the current flow direction, the conductor current amplitude and the via current amplitude are respectively obtained, the differences of the estimated current distribution with defects are detected, the positions of the coaxial via defects and the significant decrease of the current amplitude of the open circuit are detected, and the open circuit problems existing in the coaxial via or the circuit can be clearly detected through the application.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of current fault prediction, and particularly relates to an ANN chip current fault prediction algorithm based on near-field scanning. BACKGROUND

[0002] In recent years, super large scale integrated circuits generally adopt low-power supply voltages, so the influence of power supply noise is more and more significant.

[0003] In a digital circuit, a current surge synchronized with a clock signal often causes power supply noise of a system on chip, the current causes voltage drop due to parasitic elements of a power supply network, causes change of time delay, and even causes functional failure due to set / hold violation. A measurement method in the prior art mainly analyzes the mechanism of a damaged area of an integrated circuit after opening, the test method has irreversibility, and test time and economic cost are high. Therefore, at present, three problems exist in current fault detection of an integrated circuit, that is, a first problem is a destructive detection problem of detection being invasive, a second problem is a problem of detection cost being expensive, and a third problem is a problem of detection results not being universal. A method needs to be provided to measure and evaluate an actual power supply network after manufacturing, so as to evaluate a failure analysis method of chip current work in a printed circuit board.

[0004] In conclusion, in order to meet the above characteristics, the application provides an ANN current fault prediction algorithm based on near-field scanning technology, so that current fault problems caused by coaxial via holes or short circuits of a chip can be efficiently detected. SUMMARY

[0005] The application aims to provide an ANN chip current fault prediction algorithm based on near-field scanning, so as to solve the problems in the background.

[0006] To achieve the above object, the application provides the following technical scheme.

[0007] An ANN (artificial neural network) chip current fault prediction algorithm based on near-field scanning, the ANN chip current fault prediction algorithm is realized based on a near-field scanning system, and the ANN chip current fault prediction algorithm comprises the following steps.

[0008] S1. White box test and analysis model: a measured object with a known structure is prepared, including an opened and packaged sample, near-field scanning system is used to obtain electromagnetic interference cloud picture information of a surface of the measured object, and a field strength component is used to estimate a surface current of a device;

[0009] S2. Current estimation principle: based on Biot-Savart law, the magnetic field generated by the current segment at the measuring point is calculated, the induced electromotive force is calculated through Faraday's law of electromagnetic induction, and then the magnetic field is calculated, the relationship between the magnetic field matrix and the current segment matrix is established, and the current distribution is solved through the measured data;

[0010] S3. Construct and train artificial neural network model: construct input layer, hidden layer and output layer, adjacent layer neurons are fully connected, use training set to optimize parameters, avoid overfitting through validation set, test preliminary results through test set, predict field strength component through ANN, and then estimate current distribution;

[0011] S4. Fault detection application: inject test current into the power supply network of the measured object to induce magnetic field interference emission, apply current estimation method to detect the current distribution of the power supply network, and locate the coaxial hole defect or open circuit problem through the position of the significant decrease of the current amplitude.

[0012] Preferably, the near-field scanning system comprises a detection structure, a transmission structure and an output module;

[0013] The detection structure is used to induce the magnetic field perpendicular to the loop;

[0014] The transmission structure is connected with the detection structure through a beveled inner buckle to eliminate the resonant signal at the connection;

[0015] The output module is used to transmit the radio frequency signal to the SMA output end for measurement by a spectrum analyzer or a network analyzer.

[0016] Preferably, the implementation process of step S1 white box test and analysis model is as follows:

[0017] S1.1. First, prepare the open and packaged test analysis samples with the same fault under the condition that the physical wiring of the measured object is known (i.e. white box condition);

[0018] S1.2. Then, use the near-field scanning system to obtain the electromagnetic interference cloud map information of the surface of the measured object;

[0019] S1.3. Finally, obtain the field strength component of the surface of the measured object through the near-field scanning system, so as to estimate the current on the surface of the device;

[0020] S1.4. Use the near-field probe to measure the EMI interference source emission of the measured circuit;

[0021] S1.5. Characterize the EMI two-dimensional field strength pattern of the measured object through near-field scanning technology;

[0022] S1.6. Use the pattern of the field strength component captured by the above-mentioned near-field probe to estimate the distribution of the current on the measured object.

[0023] Preferably, the step S3 constructs and trains the artificial neural network model as follows:

[0024] S3.1. Constructing an artificial neural network comprising an input layer, one or more hidden layers, and an output layer, ensuring that the neurons of adjacent layers are usually fully connected;

[0025] S3.2. The dipole moment is implicitly represented by the weights, biases, and activation functions of the artificial neural network;

[0026] S3.3. The matrix-vector multiplication operation between the free-space Green's function and the dipole moment in the traditional dipole method is replaced by the mapping of the artificial neural network;

[0027] S3.4. After adding hidden layers to the neural network, the mapping ability of the neural network is used to process non-direct waves;

[0028] S3.5. Training the artificial neural network: the training data is randomly divided into three groups, 70% of the rows are selected as the training set to optimize the artificial neural network parameters, 15% of the rows are selected as the validation set to avoid overfitting, and the last 15% of the rows are used as the test set to test the preliminary results of the winner model;

[0029] S3.6. The corresponding field strength components are predicted by the artificial neural network algorithm by updating the weights and biases using the training set and the backpropagation learning algorithm.

[0030] Preferably, the step S4 performs the flow of the fault detection application as follows:

[0031] S4.1. Inject a test current between VDD (digital power pin) and GND (0V line) of the power supply network;

[0032] S4.2. Place the test current source between VDD and GND on the object being tested, inject appropriate test current to induce magnetic field interference emission, and estimate the surface current of the power supply network;

[0033] S4.3. Apply the current estimation method in steps S3.1 to S3.2 to fault detection of the power supply network;

[0034] S4.4. Estimate the current distribution of the power supply network by solving the quadratic programming problem;

[0035] S4.5. Obtain the current flow direction, wire current amplitude, and via current amplitude, respectively, and the difference in the estimated current distribution of the defect, detect the position of the coaxial via defect and the significant decrease in the current amplitude of the open circuit of the line.

[0036] Compared with the prior art, the beneficial effects of the present application are that the present application predicts the electromagnetic field data of the surface of the measured object on the basis of the near-field scanning technology and in combination with the ANN equivalent source reconstruction algorithm, and characterizes the equivalent current of each region through the Biot-Savart law, so as to evaluate the failure analysis method of the chip current work of the printed circuit board, the current estimation method of the present application is applied to the fault detection of the power supply network, the current distribution of the power supply network is estimated by solving the quadratic programming problem, the current flow direction, the wire current amplitude and the via current amplitude are obtained respectively, the difference of the estimated current distribution with defects, the position of the significant decrease of the current amplitude of the coaxial via defect and the open circuit of the line is detected, and the coaxial via or the open circuit problem existing in the circuit can be detected clearly through the present application. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 The main flow chart of the ANN chip current fault prediction algorithm model of the present application is shown in the figure.

[0038] Figure 2 The Biot-Savart law calculation schematic diagram of the present application is shown in the figure.

[0039] Figure 3 The surface current estimation schematic diagram of the measured object based on the near-field scanning technology of the present application is shown in the figure.

[0040] Figure 4 The artificial neural network structure of the equivalent source reconstruction problem of the present application is shown in the figure.

[0041] Figure 5 The power supply network estimation schematic diagram of the measured object of the present application is shown in the figure. DETAILED DESCRIPTION

[0042] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0043] Embodiment 1

[0044] Please refer to Figure 1 , Figure 2 , Figure 3 and Figure 4 , the present application proposes an ANN chip current fault prediction algorithm based on near-field scanning, the ANN (artificial neural network) chip current fault prediction algorithm is realized based on a near-field scanning system, and the ANN chip current fault prediction algorithm comprises the following steps:

[0045] S1. White box test and analysis model: prepare the known structure of the measured object, including the open cover and packaged samples, use the near-field scanning system to obtain the electromagnetic interference cloud map information of the surface of the measured object, and estimate the surface current of the device through the field strength component;

[0046] Specifically, the process is as follows:

[0047] S1.1. First, in the case of known physical wiring of the measured object (i.e. white box condition), prepare open cover and packaged test analysis samples with the same fault respectively;

[0048] S1.2. Then, use the near-field scanning system to obtain the electromagnetic interference cloud map information of the surface of the measured object;

[0049] S1.3. Finally, obtain the field strength component of the surface of the measured object through the near-field scanning system, so as to estimate the surface current of the device;

[0050] S1.4. Use the near-field probe to measure the EMI interference source emission of the measured circuit;

[0051] S1.5. Characterize the EMI two-dimensional field strength pattern of the measured object through near-field scanning technology;

[0052] S1.6. Use the field strength component pattern captured by the above-mentioned near-field probe to estimate the current distribution on the measured object;

[0053] S2. Current estimation principle: based on the Biot-Savart law, calculate the magnetic field generated by the current segment at the measurement point, calculate the induced electromotive force through Faraday's law of electromagnetic induction, and then calculate the magnetic field, establish the relationship between the magnetic field matrix and the current segment matrix, and solve the current distribution through the measurement data;

[0054] S3. Construct and train artificial neural network model: construct input layer, hidden layer and output layer, adjacent layer neurons are fully connected, use training set to optimize parameters, validation set to avoid overfitting, test set to test preliminary results, predict field strength component through ANN, and then estimate current distribution;

[0055] Specifically, the process is as follows:

[0056] S3.1. Construct an artificial neural network including an input layer, one or more hidden layers and an output layer, and ensure that the neurons of adjacent layers are usually fully connected;

[0057] S3.2. Dipole moment is implicitly represented by the weight, bias and activation function of artificial neural network;

[0058] S3.3. The matrix-vector multiplication operation between the free space Green function and the dipole moment in the traditional dipole method is replaced by artificial neural network mapping;

[0059] S3.4. After adding hidden layers in the neural network, the mapping ability of the neural network is used to process the non-direct wave;

[0060] S3.5. Training of the artificial neural network: the training data is randomly divided into three groups, 70% of the rows are selected as the training set to optimize the parameters of the artificial neural network, 15% of the rows are selected as the validation set to avoid overfitting, and the last 15% of the rows are used as the test set to test the preliminary results of the winner model;

[0061] S3.6. The corresponding field strength components are predicted by the artificial neural network algorithm by updating the weights and biases using the training set and the backpropagation learning algorithm

[0062] S4. Fault detection application: inject test current into the power supply network of the object under test to induce magnetic field interference emission, apply current estimation method to detect the current distribution of the power supply network, and locate the coaxial via defect or open circuit problem by detecting the position of significant decrease in current amplitude;

[0063] Specifically, the process is as follows:

[0064] S4.1. Inject test current between VDD (digital power pin) and GND (0V line) of the power supply network;

[0065] S4.2. Place the test current source between VDD and GND on the object under test, inject appropriate test current to induce magnetic field interference emission, and estimate the surface current of the power supply network;

[0066] S4.3. Apply the current estimation method in steps S3.1 to S3.2 to fault detection of the power supply network;

[0067] S4.4. Estimate the current distribution of the power supply network by solving the quadratic programming problem;

[0068] S4.5. Obtain the current flow direction, wire current amplitude and via current amplitude respectively, and the difference in estimated current distribution with defects, and detect the position of significant decrease in current amplitude of the coaxial via defect and open circuit of the line.

[0069] Example 2

[0070] In practical application, the high-performance magnetic field near-field probe design method and application steps of the ANN chip current fault prediction algorithm based on near-field scanning are as follows:

[0071] Please refer to Figure 1The application of the artificial intelligence algorithm in the white-box reconstruction of the EMI interference source is that the artificial intelligence algorithm is used to predict the field strength of a measured object, and the actual current of the measured object is estimated through field strength component inverse problem solving, which can be applied to fault analysis and positioning of the measured object, and a main flowchart is shown in Figure 1 First, the surface electromagnetic field components of the measured object of a known structure at a certain height are obtained through a high-performance near-field scanning system, then an equivalent dipole array is obtained through an artificial neural network model, and the EMI component information is reconstructed, and finally the current on the surface of the white-box EMI interference source is estimated through the reconstructed EMI component information, and is used for analyzing the open circuit position in the coaxial via and the wire of the measured object.

[0072] The transmission structure and the detection structure are connected through a beveled inner buckle structure, the detection loop part of the probe is inductively detected through the structure composed of the designed second layer coil, the third layer coil and the through hole, the magnetic field change perpendicular to the detection loop is derived according to the expression of the magnetic flux, and the corresponding frequency response is obtained, in the transmission process, six coaxial through hole arrays are used at the connection between the first layer and the second layer to eliminate the resonance signal at the connection, finally the radio frequency signal is transmitted to the SMA output end, so that the data is measured by the spectrum analyzer or the network analyzer, specifically, comprising the following steps.

[0073] 1) White-box test and analysis model of EMI interference source of the measured object based on near-field scanning technology:

[0074] First, under the condition that the physical wiring of the measured object is known (i.e. white-box condition), prepare the open cover and packaged test analysis samples with the same fault respectively;

[0075] Then, the electromagnetic interference cloud information of the surface of the measured object is obtained by using the near-field scanning system;

[0076] Finally, the field strength components of the surface of the measured object are obtained through the near-field scanning system, so as to estimate the current on the surface of the device, and the specific estimation principle is as follows:

[0077] As shown in Figure 2 The magnetic field measured at point P is described as the vector sum of the magnetic fields induced by all current paths on the measured object, in the current estimation, the power supply network is divided into a plurality of current segments, each current segment is modeled as a zero-width line with uniform current, and the magnetic field B l,p According to the Biot-Savart law, it is expressed as formula (1):

[0078]

[0079] In the formula, i l is the current on the line segment l, μ is the magnetic permeability of air, bl,p is a unit vector of the direction of the magnetic field generated by the current segment I at point p, r l,p is the perpendicular distance of the current segment I to point p, θ 1(l,p) and θ 2(l,p) is the angle formed by the current line segment I and the vector from the start (end) point of the current line segment to point p, the normal unit vector of the near-field probe at the measurement point p is defined as d, then the component of the magnetic field in the measurement direction is represented as formula (2):

[0080]

[0081] The entire magnetic field B d,p generated by the L current segments at point p is represented as formula (3):

[0082]

[0083] In the equation Y d,l,p is represented as formula (4):

[0084]

[0085] The magnetic flux φ p (t) obtained by the near-field scanning system at the measurement point p is represented as formula (5):

[0086] φ p (t) = N∫B d (t) dS (5);

[0087] In the above formula, N is the number of turns of the coil, Bd(t) is the component of the magnetic field passing through the coil in the measurement direction, S is the total area of the coil, and in this case, the measurement point means the center of the coil, the induced electromotive force v p (t) of the near-field probe at the measurement point p is represented by Faraday's law of electromagnetic induction as formula (6) and (7):

[0088]

[0089] Wherein:

[0090]

[0091] Because the amplitude of the magnetic field B p can be calculated from the measured induced electromotive force V d,p , the relationship between the magnetic field B matrix and the current segment I matrix can be represented as formula (8):

[0092]

[0093] In the formula, P is the total number of measurement points, and the current estimation flowchart of the physical wiring of the measured object using the above analysis model is as follows:Figure 3 As shown in FIG. 1, the EMI interference source emission of the measured circuit is measured by using the near-field probe, the EMI two-dimensional field strength pattern of the measured object is characterized by the near-field scanning technology, and the field strength component pattern captured by the near-field probe is used to estimate the current distribution on the measured object.

[0094] 2) EMI interference source white-box reconstruction model based on artificial intelligence algorithm:

[0095] Please refer to Figure 4 , an artificial neural network EMI interference source prediction model is established for the equivalent source reconstruction problem, the artificial neural network includes an input layer, one or more hidden layers and an output layer, the neurons of adjacent layers are usually fully connected, w, c and f are weight, bias and activation function respectively, the dipole moment is implicitly represented by the weight, bias and activation function of the artificial neural network, the artificial neural network mapping is used to replace the matrix vector multiplication operation between the free space Green function and the dipole moment in the traditional dipole method, after adding a hidden layer in the neural network, the mapping ability of the neural network can be used to process the non-direct wave, for the training of artificial neural network, the whole training data is randomly divided into three groups, 70% of the rows are selected as the training set to optimize the artificial neural network parameters, 15% of the rows are selected as the validation set to avoid overfitting, and the last 15% of the rows are used as the test set to test the preliminary results of the winner model, the weight w, the bias c and the activation function f are three undetermined parameters in the artificial neural network, the weight and the bias are updated by using the training set and the back propagation learning algorithm (gradient descent based on appropriate cost function), the corresponding field strength component can be predicted by the artificial neural network algorithm,

[0096] 3) Fault detection application based on current estimation, in order to control the magnetic field emission of the power supply network to achieve the test purpose, as shown in Figure 5 , a test current with a frequency of f is injected between VDD and GND of the power supply network, S3.2. The test current source is placed between VDD and GND on the measured object, and the appropriate test current is injected to induce magnetic field interference emission, and the surface current of the power supply network is estimated;

[0097] In order to accurately detect the through-hole defect or open circuit problem, the above current estimation method is applied to the fault detection of the power supply network, the current distribution of the power supply network is estimated by solving the quadratic programming problem, the current flow direction, the wire current amplitude and the via current amplitude are obtained respectively, the difference of the estimated current distribution with defects, by detecting the position of the coaxial via defect and the significant decrease of the current amplitude of the open circuit of the line, the coaxial via or the open circuit problem existing in the circuit can be clearly detected by the above analysis model estimation method.

[0098] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A near-field scanning-based ANN chip current fault prediction algorithm, wherein the ANN chip current fault prediction algorithm is implemented based on a near-field scanning system, characterized in that, The ANN chip current fault prediction algorithm includes the following steps: S1. White-box testing and analysis model: Prepare a test object with a known structure, including open and sealed samples. Use a near-field scanning system to obtain electromagnetic interference cloud map information on the surface of the test object, and estimate the surface current of the device through the field strength component. S2. Current estimation principle: Based on the Biot-Savart law, the magnetic field generated by the current segment at the measuring point is calculated, the induced electromotive force is calculated by Faraday's law of electromagnetic induction, and then the magnetic field is calculated. The relationship between the magnetic field matrix and the current segment matrix is ​​established, and the current distribution is solved by the measurement data. S3. Construct and train an artificial neural network model: Construct an input layer, hidden layer, and output layer, with neurons in adjacent layers fully connected. Optimize parameters using the training set, avoid overfitting using the validation set, test preliminary results using the test set, and predict the field strength components through the ANN to estimate the current distribution. S4. Fault detection application: Inject test current into the power supply network of the device under test to induce magnetic field interference emission. Apply current estimation method to detect the current distribution of the power supply network. By detecting the location of significant drop in current amplitude, locate coaxial through hole defects or open circuit problems.

2. The ANN chip current fault prediction algorithm based on near-field scanning according to claim 1, characterized in that, The near-field scanning system includes a detection structure, a transmission structure, and an output module; The detection structure is used to sense the magnetic field that passes vertically through the loop; The transmission structure and the detection structure are connected by an oblique inward connection to eliminate the resonant signal at the connection point; The output module is used to transmit radio frequency signals to the SMA output terminal for measurement by a spectrum analyzer or network analyzer.

3. The ANN chip current fault prediction algorithm based on near-field scanning according to claim 2, characterized in that, The implementation process of the white-box testing and analysis model in step S1 is as follows: S1.

1. First, under the condition that the physical traces of the object under test are known (i.e., white box condition), prepare test and analysis samples with the same fault, both with the cover open and the packaged. S1.

2. Then, a near-field scanning system is used to obtain electromagnetic interference cloud map information of the surface of the object under test; S1.

3. Finally, the field strength component of the surface of the object under test is obtained through a near-field scanning system, thereby estimating the current on the surface of the device; S1.

4. Measure the EMI interference source emission of the circuit under test using a near-field probe; S1.

5. The EMI two-dimensional field strength pattern of the test object is characterized by near-field scanning technology; S1.

6. Using the graph of the field strength components captured by the above near-field probe, the distribution of the current on the object under test can be estimated.

4. The ANN chip current fault prediction algorithm based on near-field scanning according to claim 3, characterized in that... The process of constructing and training the artificial neural network model in step S3 is as follows: S3.

1. Construct an artificial neural network consisting of an input layer, one or more hidden layers, and an output layer, ensuring that neurons in adjacent layers are usually fully connected; S3.

2. The dipole moment is implicitly represented by the weights, biases, and activation functions of the artificial neural network; S3.

3. Artificial neural network mapping is used to replace the matrix-vector multiplication operation between the free space Green's function and the dipole moment in the traditional dipole method; S3.

4. After adding hidden layers to the neural network, the mapping ability of the neural network is used to process indirect waves; S3.

5. Training the artificial neural network: The training data is randomly divided into three groups. 70% of the rows are selected as the training set to optimize the parameters of the artificial neural network, 15% of the rows are selected as the validation set to avoid overfitting, and the last 15% of the rows are used as the test set to test the preliminary results of the winning model. S3.

6. By using a training set and a backpropagation learning algorithm to update the weights and biases, the corresponding field strength components are predicted using an artificial neural network algorithm.

5. The ANN chip current fault prediction algorithm based on near-field scanning according to claim 4, characterized in that, The process for applying fault detection in step S4 is as follows: S4.

1. Inject test current between VDD (digital power pin) and GND (0V line) of the power supply network; S4.

2. Place the test current source between VDD and GND on the device under test, inject an appropriate test current to induce magnetic field interference emission, and estimate the surface current of the power network. S4.

3. Apply the current estimation method from steps S3.1 to S3.2 to fault detection in the power supply network; S4.

4. The current distribution of the power supply network can be estimated by solving a quadratic programming problem; S4.

5. Obtain the current flow direction, conductor current amplitude, and through-hole current amplitude respectively, estimate the difference in current distribution due to defects, and detect the location of significant drop in current amplitude at coaxial through-hole defects and open circuit locations in the circuit.

Images