Noise specification support device

The noise identification support device uses regression analysis to efficiently identify noise sources in electrical circuits by generating models based on coefficients, reducing man-hours in complex circuit verification.

JP2024164352AInactive Publication Date: 2024-11-27YAZAKI CORP
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
JP2023079765
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-15
Publication Date
2024-11-27
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Identifying noise sources in complex electrical circuits is labor-intensive due to the vast amount of data that needs verification, increasing man-hours.

Method used

A noise identification support device that utilizes regression analysis to generate a model based on objective and explanatory variables, identifying coefficients to pinpoint circuit blocks affecting noise generation.

Benefits of technology

Reduces the number of work steps required to identify noise sources by estimating circuit blocks with significant noise influence, particularly effective in centralized vehicle systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024164352000001_ABST
    Figure 2024164352000001_ABST
Patent Text Reader

Abstract

To provide a noise specification support device capable of supporting specification of a noise source.SOLUTION: A noise specification support device 1 comprises a model generation part 41 and a coefficient specification part 42. The model generation part 41, by performing a regression analysis with a value representing noise generated in an electric circuit including a plurality of electronic components defined as objective variables and a value representing an operational state of each of a plurality of circuit blocks configured by dividing the electric circuit into a plurality of blocks defined as explanatory variables, generates a model expressed by a multiple regression equation representing the objective variables by explanatory variables respectively corresponding to each circuit block and a coefficient calculated for each explanatory variable. The coefficient specification part 42 specifies a coefficient for the model generated by the model generation part 41.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a noise identification support device. [Background technology]

[0002] Conventionally, for example, Patent Document 1 describes a machine learning device that makes it possible to suppress the influence of abnormal values ​​due to noise, etc. This machine learning device includes a state observation unit that acquires input data including at least one of internal data and external data of a manufacturing machine, an input safety circuit that detects an abnormality in the input data and outputs safe input data, a machine learning unit that executes inference using a learning model based on the safe input data and outputs inference data as the inference result, an output safety circuit that detects an abnormality in the inference data and outputs safe inference data, and an output unit that outputs output data based on the safe inference data. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2019-086928 A Summary of the Invention [Problem to be solved by the invention]

[0004] The machine learning device described in the above-mentioned Patent Document 1 suppresses the influence of abnormal values ​​due to noise, etc. Conventionally, workers may perform the task of identifying noise sources occurring in electric circuits. In this case, since electric circuits include multiple functions and there are various data patterns that operate the electric circuits, the workers need to verify a huge amount of data, which causes an increase in the number of work steps.

[0005] Therefore, the present invention has been made in consideration of the above, and has an object to provide a noise identification support device that can support identification of a noise source. [Means for solving the problem]

[0006] In order to solve the above-mentioned problems and achieve the object, the noise identification support device of the present invention is characterized in that it comprises: a model generation unit that performs regression analysis using a value representing noise generated in an electric circuit including a plurality of electronic components as a response variable, and values ​​representing the operating states of each of a plurality of circuit blocks formed by dividing the electric circuit into a plurality of blocks as explanatory variables, thereby generating a model in which the response variable is represented by the explanatory variables corresponding to each of the circuit blocks and coefficients determined for each of the explanatory variables; and a coefficient identification unit that identifies the coefficients of the model generated by the model generation unit. Effect of the Invention

[0007] The noise identification assistance device of the present invention can identify circuit blocks that are estimated to be affecting noise generated in an electric circuit, based on the coefficients identified by the coefficient identification unit, and can assist in identifying the noise source of the electric circuit, based on the identified circuit blocks. [Brief description of the drawings]

[0008] [Figure 1] FIG. 1 is a block diagram illustrating an example of the configuration of a noise identification support device according to an embodiment. [Diagram 2] FIG. 2 is a diagram showing a plurality of circuit blocks as explanatory variables according to the embodiment. [Diagram 3] FIG. 3 is a flowchart illustrating an example of the operation of the noise identification support device according to the embodiment. [Figure 4] FIG. 4 is a diagram showing a signal pattern according to the embodiment. [Diagram 5] FIG. 5 is a block diagram showing an example of the configuration of a noise identification support device according to a modified example of the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] The form (embodiment) for carrying out the present invention will be described in detail with reference to the drawings. The present invention is not limited to the contents described in the following embodiment. Furthermore, the components described below include those that a person skilled in the art can easily imagine and those that are substantially the same. Furthermore, the configurations described below can be appropriately combined. Furthermore, various omissions, substitutions, or modifications of the configurations can be made without departing from the gist of the present invention.

[0010] A noise identification support device 1 according to an embodiment will be described with reference to the drawings. The noise identification support device 1 supports, for example, identifying a source of noise occurring in an electric circuit EC of an in-vehicle device. The noise identification support device 1 is used, for example, in a test environment before the in-vehicle device is shipped. As shown in FIG. 1, the noise identification support device 1 includes an input unit 10, an output unit 20, a storage unit 30, and a regression analysis unit 40.

[0011] The input unit 10 is a device that performs various inputs to the noise identification support device 1. The input unit 10 is realized, for example, by an operation input unit that accepts various operation inputs from a user, a data input device that accepts data (information) input from another device, etc. The operation input unit is realized, for example, by a mouse, a keyboard, etc. The data input device is realized, for example, by a recording medium interface that reads out various data from a recording medium such as a USB memory, etc.

[0012] The output unit 20 is a device that performs various outputs from the noise identification support device 1. The output unit 20 is realized by, for example, a display that outputs and displays various types of image information, a speaker that outputs sound information, or the like.

[0013] The storage unit 30 is a circuit that stores various data. The storage unit 30 is realized by, for example, a semiconductor memory element such as a RAM (Random Access Memory), a flash memory, a hard disk, an optical disk, or the like. The storage unit 30 stores, for example, programs for the noise identification support device 1 to realize various functions. The programs stored in the storage unit 30 include programs for causing the input unit 10, the output unit 20, the regression analysis unit 40, and the like to function. Specifically, the storage unit 30 stores a program for regression analysis by the regression analysis unit 40 and various data used in the regression analysis. Here, the program for regression analysis is realized by, for example, a spreadsheet software in which a regression analysis program is incorporated, or a program using the R programming language for statistical analysis, which are appropriately selected according to the amount of calculation for the regression analysis. The various data are read out of the storage unit 30 by the regression analysis unit 40, or the like, as necessary.

[0014] The regression analysis unit 40 performs regression analysis using a response variable and an explanatory variable. In this example, the response variable is a value representing noise occurring in an electric circuit EC of an in-vehicle device that includes a plurality of electronic components P and a signal transmission path. Here, the plurality of electronic components P are, for example, a central processing unit (CPU), a field effect transistor (FET) as a switching element, a capacitor, an inductor, and the like, and constitute, for example, a power supply circuit such as a power distribution unit or a DC / DC converter. The response variable is, for example, a bit error rate (BER) representing the ratio of data errors as a value representing noise occurring in the electric circuit EC by these electronic components P. The bit error rate is the ratio of erroneous data to all data received at the receiving side when data is transmitted from the transmitting side to the receiving side via a signal transmission path provided in the electric circuit EC.

[0015] The explanatory variables are values ​​representing the operation state of each of the circuit blocks BL (BL1 to BLn). Here, as shown in FIG. 2, each circuit block BL is configured by dividing an electric circuit EC including a plurality of electronic components P into a plurality of blocks. For example, each circuit block BL is a block configured by dividing an electric circuit EC including a plurality of electronic components P into predetermined regions. Also, each circuit block BL may be a block configured by dividing an electric circuit EC including a plurality of electronic components P into blocks by function, or may be divided by other division methods. The value representing the operation state of the circuit block BL, i.e., the explanatory variable, is, for example, the current value of the current flowing through the circuit block BL. The regression analysis unit 40 performs regression analysis using the above-mentioned objective variable (for example, bit error rate) and explanatory variable (for example, current value). The regression analysis unit 40 is configured to include a model generation unit 41 and a coefficient specification unit 42.

[0016] The model generation unit 41 generates a model by performing regression analysis using a response variable and explanatory variables. For example, the model generation unit 41 performs multiple regression analysis using a value (e.g., bit error rate) representing noise occurring in the electric circuit EC as the response variable and a value (e.g., current value) representing the operating state of each of the multiple circuit blocks BL as the explanatory variable, thereby generating a model in which the response variable is represented by explanatory variables corresponding to each circuit block BL and coefficients obtained for each explanatory variable. This model is expressed by the following multiple regression formula (1) (n is a natural number) where the response variable (bit error rate) is y, the explanatory variable (current value) is Xn, and the coefficient is an. y=a1X1+a2X2+a3X3+···+anXn···(1)

[0017] A data set consisting of a response variable (y: bit error rate) and each explanatory variable (Xn: current value) used in the model generation unit 41 is acquired by randomly operating the electric circuit EC. This data set is big data consisting of a plurality of data sets, such as a data set including a first current value (explanatory variable X1) of a current flowing through the circuit block BL1, a first current value (explanatory variable X2) of a current flowing through the circuit block BL2, a bit error rate (objective variable y) when a current of the first current value is flowed through the circuit block BL1 and the circuit block BL2, a data set including a second current value (explanatory variable X3) of a current flowing through the circuit block BL3, a second current value (explanatory variable X4) of a current flowing through the circuit block BL4, and a bit error rate (objective variable y) when a current of the second current value is flowed through the circuit block BL3 and the circuit block BL4. This big data is stored in the storage unit 30.

[0018] The coefficient identification unit 42 identifies the coefficients of the model represented by the multiple regression equation (1) generated by the model generation unit 41. Here, it is estimated that the explanatory variable corresponding to the coefficient (a1 to an), i.e., the circuit block BL, has a stronger influence on the noise generated in the electric circuit EC as the coefficient (a1 to an) is relatively larger. For example, when a current is applied to the electric circuit EC, if the coefficient a1 of the explanatory variable X1 is larger than the coefficient a2 of the explanatory variable X2, it is estimated that the explanatory variable X1 corresponding to the coefficient a1, i.e., the circuit block BL1 to which the current is applied, has a stronger influence on the noise generated in the electric circuit EC than the explanatory variable X2 corresponding to the coefficient a2, i.e., the circuit block BL2 to which the current is applied. This allows the noise identification support device 1 to support identification of the noise source of the electric circuit EC. The coefficient identification unit 42 determines the significance of the coefficient based on a P value, which is an index indicating the significance of the coefficient. For example, when the P value is equal to or less than "0.05", the coefficient specifying unit 42 determines that the coefficient is significant, and when the P value exceeds "0.05", the coefficient specifying unit 42 determines that the coefficient is not significant.

[0019] Next, an example of the operation of the noise identification support device 1 will be described. The model generation unit 41 refers to the storage unit 30 and acquires a data set consisting of a response variable (y: bit error rate) and each explanatory variable (Xn: current value) (step S1). Next, the model generation unit 41 generates a model by performing a regression analysis using the response variable and explanatory variables of the acquired data set (step S2). For example, the model generation unit 41 generates a model expressed by the above-mentioned multiple regression equation (1) in which the response variable (bit error rate) is y, the explanatory variable (current value) is Xn, and the coefficient is an. Next, the coefficient identification unit 42 identifies the coefficients (a1 to an) of the model expressed by the above-mentioned multiple regression equation (1) generated by the model generation unit 41. It is estimated that the larger the value of the coefficients (a1 to an) identified by the coefficient identification unit 42 is, the stronger the explanatory variable Xn corresponding to the coefficient (a1 to an), i.e., the circuit block BL, has on the noise generated in the electric circuit EC.

[0020] As described above, the noise identification support device 1 according to the embodiment includes a model generation unit 41 and a coefficient specification unit 42. The model generation unit 41 performs regression analysis using a value representing noise generated in an electric circuit EC including a plurality of electronic components P as a response variable and a value representing an operating state of each of a plurality of circuit blocks BL formed by dividing the electric circuit EC into a plurality of blocks as an explanatory variable, thereby generating a model represented by multiple regression equation (1) in which the response variable is expressed by explanatory variables corresponding to each circuit block BL and coefficients obtained for each explanatory variable. The coefficient specification unit 42 specifies the coefficients of the model generated by the model generation unit 41.

[0021] With this configuration, the noise identification support device 1 can identify the circuit block BL estimated to have an effect on the noise generated in the electric circuit EC based on the coefficients identified by the coefficient identification unit 42, and can support identification of the noise source of the electric circuit EC based on the identified circuit block BL. This allows the noise identification support device 1 to suppress an increase in the number of man-hours when identifying the noise source of the electric circuit EC. In recent years, there has been a trend to consolidate the functions of a vehicle and integrate the functions that were distributed on one board. In this case, for example, noise generated in a power supply circuit may affect communication. In order to comprehensively verify such noise, a huge amount of verification is required, which is a concern regarding an increase in the number of man-hours. However, the noise identification support device 1 according to the embodiment is particularly effective for such verification because it can identify the circuit block BL estimated to have an effect on the noise.

[0022] In the above description, the explanatory variable Xn is an example of a current value flowing through the circuit block BL as a value representing the operating state of the circuit block BL, but is not limited to this. For example, the explanatory variable Xn may include, in addition to the current value flowing through the circuit block BL, a frequency for driving a switching element as an electronic component P and a value representing a signal pattern communicated in the circuit block BL as a value representing the operating state of the circuit block BL. Here, there are a plurality of different signal patterns, as shown in FIG. 4, for example, and each signal pattern is associated with a value represented by 4 bits. The explanatory variable is set to a value represented by 4 bits according to the signal pattern transmitted through the signal transmission path of the electric circuit EC.

[0023] The model generation unit 41 performs regression analysis on a plurality of circuit blocks BL using the values ​​of each circuit block BL (values ​​representing a current value, a frequency, and a signal pattern) as explanatory variables to generate a model in which the objective variable is represented by explanatory variables corresponding to the values ​​of each circuit block BL and coefficients obtained for each explanatory variable. This model is represented by the following multiple regression equation (2) (n is a natural number) where the objective variable (bit error rate) is y, the explanatory variable (current value) is Xn1, the explanatory variable (frequency) is Xn2, the explanatory variable (signal pattern) is Xn3, the coefficient of the explanatory variable (current value) Xn1 is an1, the coefficient of the explanatory variable (frequency) Xn2 is an2, and the coefficient of the explanatory variable (signal pattern) Xn3 is an3.

[0024] y=(a11X11+a12X12+a13X13)+(a21X21+a22X22+a33X33)+...+(an1Xn1+an2Xn2+an3Xn3)...(2)

[0025] A data set consisting of a response variable (y: bit error rate) and explanatory variables (Xn1: current value, Xn2: frequency, Xn3: signal pattern) used in the model generation unit 41 is acquired by randomly operating the electric circuit EC. Big data consisting of a plurality of data sets is stored in the storage unit 30.

[0026] The model generating unit 41 refers to the storage unit 30 and acquires a data set consisting of a response variable (y: bit error rate) and each explanatory variable (Xn1: current value, Xn2: frequency, Xn3: signal pattern). After acquiring the data set, the model generating unit 41 generates a model expressed by the above-mentioned multiple regression equation (2). The coefficient identifying unit 42 identifies coefficients (current values: a11 to an1, frequencies: a12 to an2, signal pattern: a13 to an3) of the model expressed by the above-mentioned multiple regression equation (2) generated by the model generating unit 41. It is estimated that the explanatory variable corresponding to the coefficient, that is, the value (current value, frequency, signal pattern) of the circuit block BL (current value, frequency, signal pattern) has a stronger influence on the noise generated in the electric circuit EC as the coefficient identified by the coefficient identifying unit 42 is relatively larger. The noise identification support device 1 can estimate the circuit block BL having a stronger influence on the noise for each value. For example, when a current is applied to the electric circuit EC, if the coefficient a11 of the explanatory variable X11 is greater than the coefficient a21 of the explanatory variable X21, the noise identification support device 1 can estimate that the explanatory variable X11 corresponding to the coefficient a11, i.e., the circuit block BL1 to which the current is applied, has a stronger effect on the noise generated in the electric circuit EC by the application of the current than the explanatory variable X22 corresponding to the coefficient a21, i.e., the circuit block BL2 to which the current is applied. Also, when a switching element of the electric circuit EC is driven at a predetermined frequency, if the coefficient a12 of the explanatory variable X12 is greater than the coefficient a22 of the explanatory variable X22, the noise identification support device 1 can estimate that the explanatory variable X12 corresponding to the coefficient a12, i.e., the circuit block BL1 that drove the switching element, has a stronger effect on the noise generated in the electric circuit EC by the driving of the switching element than the explanatory variable X22 corresponding to the coefficient a22, i.e., the circuit block BL2 that drove the switching element.Furthermore, when different signal patterns are transmitted through the signal transmission paths of the electric circuit EC, if the coefficient a13 of the explanatory variable X13 is greater than the coefficient a23 of the explanatory variable X23, the noise identification support device 1 can estimate that the explanatory variable X13 corresponding to the coefficient a13, i.e., the circuit block BL1, has a stronger effect on the noise generated in the electric circuit EC due to changes in the signal pattern than the explanatory variable X23 corresponding to the coefficient a23, i.e., the circuit block BL2.

[0027] In the above explanation, the objective variable is described as an example of a bit error rate, but is not limited to this and may be, for example, an EMI (Emission) measurement value measuring the electromagnetic waves emitted from the electronic component P, or a value representing other noise such as an error in the output voltage output from the power supply circuit of the electrical circuit EC.

[0028] Each explanatory variable may include at least one of the values ​​of a current flowing through the circuit block BL, a frequency for driving a switching element included in the circuit block BL, and a signal pattern communicated in the circuit block BL. Each explanatory variable may also be a value other than the above-mentioned current value, frequency, and signal pattern.

[0029] Based on the coefficients identified by the coefficient identification unit 42, the circuit block BL having a strong influence on the noise may be automatically estimated. FIG. 5 is a block diagram showing an example of the configuration of a noise identification support device 1A according to a modified example of the embodiment. The noise identification support device 1A further includes a noise influence estimation unit 43, for example, as shown in FIG. 5. The noise influence estimation unit 43 automatically estimates the circuit block BL having a strong influence on the noise. For example, when the coefficients identified by the coefficient identification unit 42 are equal to or greater than a predetermined reference value, the noise influence estimation unit 43 estimates that the explanatory variable corresponding to the coefficient, i.e., the circuit block BL, has a strong influence on the noise generated in the electric circuit EC. On the other hand, when the coefficients identified by the coefficient identification unit 42 are less than the reference value, the noise influence estimation unit 43 estimates that the explanatory variable corresponding to the coefficient, i.e., the circuit block BL, does not have a strong influence on the noise generated in the electric circuit EC. In this way, the noise identification support device 1A according to the modified example includes the noise influence estimation unit 43 that automatically estimates the circuit block BL having a strong influence on the noise, thereby making it possible to further suppress an increase in the number of work steps. [Explanation of symbols]

[0030] 1. Noise identification support device 41 Model Generation Unit 42 Coefficient specification section BL Circuit block EC Electric Circuit P Electronic parts

Claims

1. a model generating unit that performs a regression analysis using a value representing noise generated in an electric circuit including a plurality of electronic components as a response variable and a value representing an operating state of each of a plurality of circuit blocks formed by dividing the electric circuit into a plurality of blocks as an explanatory variable to generate a model in which the response variable is represented by the explanatory variables corresponding to each of the circuit blocks and a coefficient obtained for each of the explanatory variables; a coefficient specifying unit that specifies the coefficients of the model generated by the model generating unit.

2. the explanatory variables include at least one value representing an operating state of the circuit block, which is a current value flowing through the circuit block, a frequency at which the switching element serving as the electronic component is driven, and a value representing a signal pattern communicated by the circuit block; 2. The noise identification support device according to claim 1, wherein the model generation unit performs a regression analysis using the at least one value of each of the plurality of circuit blocks as the explanatory variable, to generate the model in which the objective variable is represented by the explanatory variable corresponding to each of the at least one value of each of the circuit blocks and the coefficient obtained for each of the explanatory variables.

Citation Information

Patent Citations

  • Control device and machine learning device

    JP2019086928A