EMI calculation program, EMI calculation method and EMI calculation device

Abstract

To generate a machine learning model which reduces a burden of machine learning processing, and accurately predicts an Electro Magnetic Interference (EMI) value of each frequency.SOLUTION: An EMI calculation device 100 inputs circuit information on a first circuit into a machine learning model, and acquires an EMI value at a specific frequency of the first circuit. The EMI calculation device 100 selects first EMI information, among the plurality of pieces of EMI information in which an impedance characteristic of each of the plurality of circuits is associated with the EMI value at each of the plurality of frequencies of each of the plurality of circuits, on the basis of the impedance characteristic of the first circuit and the EMI value at the specific frequency. The EMI calculation device 100 calculates the EMI values in other frequencies other than the specific frequency of the first circuit, on the basis of the EMI value at the specific frequency and the first EMI information.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an EMI calculation technique for calculating electromagnetic radiation from an electronic circuit board. [Background technology]

[0002] When a current flows through an electronic circuit board (hereafter referred to as a circuit), electromagnetic waves are emitted. The adverse effect of these electromagnetic waves on electronic devices as noise is called EMI (Electro Magnetic Interference). In the following explanation, the strength of the electromagnetic waves emitted when a current flows through a circuit will be referred to as the "EMI value" where appropriate.

[0003] Regulatory values ​​are set for each frequency, and when designing a circuit, it is required to design it so that the EMI value for each frequency falls within the regulatory value. For this reason, EMI values ​​are predicted from circuit information using simulation and deep learning technology.

[0004] Although it is possible to predict the EMI value with high accuracy by using a simulation, the cost is high, so a deep learning technique is sometimes used instead of the simulation. In the following, conventional technique 1 will be described as an example of a conventional technique for predicting the EMI value.

[0005] Conventional technique 1 will be described. FIG. 9 is a diagram for explaining conventional technique 1. In conventional technique 1, a simulation is performed using some circuit information 10 to calculate an EMI value, and the calculated EMI value is used as supervised data 11. In conventional technique 1, a set of circuit information 10 and supervised data 11 is used as training data, and training of a machine learning model 12 is performed. The machine learning model 12 corresponds to AI (Artificial Intelligence) or the like. In conventional technique 1, practical circuit information 13 is input to the machine learning model 12, and an EMI value corresponding to the circuit information 13 is calculated. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2002-90403 A [Patent Document 2] JP 2020-60877 A Summary of the Invention [Problem to be solved by the invention]

[0007] However, in the above-mentioned conventional technology, there is a problem in that generating a machine learning model that predicts the EMI value for each frequency imposes a heavy burden on the machine learning process.

[0008] 10 and 11 are diagrams for explaining the problems of the conventional technology. Let us now consider FIG. 10. The horizontal axis of graph G1 shown in FIG. 10 corresponds to frequency, and the vertical axis corresponds to EMI value. In graph G1, EMI values ​​are output in increments of 9.7 MHz from 30 MHz to 1 GHz. When EMI values ​​are output in increments of 9.7 MHz like this, if the conventional technology is applied as is, 101-dimensional regression will be performed, which will increase the burden of machine learning processing.

[0009] FIG. 11 will be described. The horizontal axis of graph G2 shown in FIG. 11 corresponds to frequency, and the vertical axis corresponds to EMI value. Line l1 indicates the EMI value for each frequency used when training the machine learning model. Line l2 indicates the EMI value for each frequency output by the trained machine learning model. The peak of the EMI value is often steep, and when the average error of 101 dimensions is used as a loss function, the peak is shifted because the model is trained to reduce the error in other parts rather than increasing the error in an attempt to hit the peak. In graph G2, the peak of line l1 and the peak of line l2 are also shifted.

[0010] In one aspect, the present invention aims to provide an EMI calculation program, an EMI calculation method, and an EMI calculation device that can reduce the burden of machine learning processing and generate a machine learning model that accurately predicts EMI values ​​at each frequency. [Means for solving the problem]

[0011] In a first proposal, the EMI calculation program causes the computer to execute a process of inputting circuit information of a first circuit into a machine learning model and acquiring an EMI value of the first circuit at a specific frequency. The EMI calculation program causes the computer to execute a process of selecting first EMI information from a plurality of pieces of EMI information in which the impedance characteristics of each of a plurality of circuits are associated with the EMI values ​​of each of the plurality of circuits at a plurality of frequencies, based on the impedance characteristics of the first circuit and the EMI value at the specific frequency. The EMI calculation program causes the computer to execute a process of calculating an EMI value of the first circuit at a frequency different from the specific frequency, based on the EMI value at the specific frequency and the first EMI information. Effect of the Invention

[0012] It is possible to reduce the burden of machine learning processing and generate a machine learning model that accurately predicts EMI values ​​at each frequency. [Brief description of the drawings]

[0013] [Figure 1] FIG. 1 is a diagram for explaining the process of the EMI calculation device according to the present embodiment. [Diagram 2] FIG. 2 is a diagram showing an example in which the EMI values ​​at the resonant frequencies are close to each other, but the EMI values ​​at other frequencies are different. [Diagram 3] FIG. 3 is a functional block diagram showing a configuration of the EMI calculation device according to this embodiment. [Figure 4] FIG. 4 is a diagram illustrating an example of a data structure of the EMI information table. [Diagram 5] FIG. 5 is a diagram illustrating an example of impedance characteristics. [Figure 6] FIG. 6 is a diagram illustrating an example of the EMI information. [Figure 7] FIG. 7 is a flowchart showing a processing procedure of the EMI calculation device according to the present embodiment. [Figure 8] FIG. 8 is a diagram illustrating an example of a hardware configuration of a computer that realizes the same function as the EMI calculation device. [Figure 9] FIG. 9 is a diagram for explaining the prior art 1. In FIG. [Figure 10] FIG. 10 is a diagram (1) for explaining the problems of the conventional technology. [Figure 11] FIG. 11 is a diagram (2) for explaining the problem of the conventional technology. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] Hereinafter, an embodiment of an EMI calculation program, an EMI calculation method, and an EMI calculation device disclosed in the present application will be described in detail with reference to the drawings. Note that the present invention is not limited to the embodiment. EXAMPLES

[0015] An example of the processing of the EMI calculation device according to the present embodiment will be described. Fig. 1 is a diagram for explaining the processing of the EMI calculation device according to the present embodiment. The horizontal axis of the graph G3 shown in Fig. 1 corresponds to the frequency, and the vertical axis corresponds to the EMI value.

[0016] The EMI calculating device 100 uses a machine learning model 141 that is trained in advance and outputs an EMI value corresponding to the input resonant frequency when the resonant frequency is input. The EMI calculating device also holds an EMI information table 142 that defines the relationship between a circuit with a known EMI value for each frequency and the frequency characteristics of the impedance of this circuit.

[0017] In the following explanation, the frequency characteristics of impedance will be referred to as "impedance characteristics" where appropriate. The EMI values ​​at each frequency will be collectively referred to as "EMI information." A circuit whose EMI values ​​and impedance characteristics at each frequency are known will be referred to as "analyzed circuit." Additionally, a circuit whose EMI value is to be estimated will be referred to as "circuit to be estimated."

[0018] When the EMI calculating device 100 receives an input of the circuit 40 under estimation, the EMI calculating device 100 inputs the resonant frequency of the circuit 40 under estimation to the machine learning model 141 to identify EMI values ​​40a, 40b corresponding to the resonant frequency of the circuit 40 under estimation.

[0019] The EMI calculation apparatus 100 identifies an analyzed circuit having impedance characteristics similar to the impedance characteristics of the estimation target circuit 40, and acquires EMI information 40c of the identified analyzed circuit.

[0020] The EMI calculating device 100 generates EMI information 40d by correcting the overall shape of the EMI information 40c so that the peak value of the EMI information 40c matches the EMI values ​​40a, 40b corresponding to the resonant frequencies. The EMI calculating device 100 outputs the EMI information 40d as the EMI value of each frequency of the estimation target circuit 40.

[0021] As described above, the EMI calculating device 100 according to this embodiment identifies the EMI value of the resonant frequency of the estimation target circuit by using the machine learning model 141. Furthermore, the EMI calculating device 100 calculates the EMI values ​​of the remaining frequencies other than the resonant frequency by using the EMI values ​​of each frequency of an analyzed circuit having characteristics similar to the impedance identification of the estimation target circuit. Since the machine learning model 141 is trained based on the relationship between the resonant frequency and the EMI value, it is possible to reduce the load required for machine learning of the machine learning model compared to the case where machine learning is performed by using the EMI value of each frequency.

[0022] When interpolating the EMI values ​​of the remaining frequencies based on the EMI value of the resonant frequency, it is possible to simply extract EMI information with a close EMI value of the resonant frequency from the EMI values ​​of each frequency of the analyzed circuit. However, even if the EMI values ​​of the resonant frequencies are close, it does not necessarily mean that the other frequencies match.

[0023] Figure 2 shows an example where the EMI values ​​of the resonant frequencies are close but the EMI values ​​are different at other frequencies. The horizontal axis of graphs G4 and G5 in Figure 2 corresponds to frequency, and the vertical axis corresponds to EMI value. The EMI information of graph G4 and the EMI information of graph G5 are EMI values ​​at each frequency of different analyzed circuits. Comparing graphs G4 and G5, the EMI value of the resonant frequency and the EMI value of the resonant frequency of G5 are similar, but the EMI values ​​of other frequencies do not match.

[0024] In other words, when interpolating the EMI values ​​of the remaining frequencies based on the EMI value of the resonant frequency, the EMI value of each frequency of the circuit to be estimated cannot be calculated with high accuracy using a method of extracting EMI information with a close EMI value to the resonant frequency from the EMI information of the analyzed circuit.

[0025] On the other hand, the EMI calculating device 100 according to the present embodiment calculates the EMI value by correcting the EMI information of an analyzed circuit having characteristics similar to the impedance characteristics of the estimation target circuit, based on the EMI value of the remaining frequencies other than the resonant frequency, as described in Fig. 1. This makes it possible to calculate the EMI value at all frequencies without impairing the estimation accuracy of the EMI value.

[0026] Since impedance affects current strength, and current strength affects electromagnetic noise strength, minute fluctuations in EMI values ​​other than the resonant frequency can be approximated by using EMI information from an analyzed circuit that has similar impedance characteristics.

[0027] Next, an example of the configuration of the EMI calculating apparatus 100 according to this embodiment will be described. Fig. 3 is a functional block diagram showing the configuration of the EMI calculating apparatus according to this embodiment. As shown in Fig. 3, the EMI calculating apparatus 100 includes a communication unit 110, an input unit 120, a display unit 130, a storage unit 140, and a control unit 150.

[0028] The communication unit 110 executes data communication with an external device (not shown) via a network. For example, the communication unit 110 receives information on the estimation target circuit from the external device.

[0029] The input unit 120 receives various data from a user or the like, and inputs the data to the EMI calculation device 100 .

[0030] The display unit 130 displays information output from the control unit 150. For example, the display unit 130 displays EMI information of the estimation target circuit output from the control unit 150.

[0031] The storage unit 140 stores data, programs executed by the control unit 150, and the like. The storage unit 140 stores a machine learning model 141 and an EMI information table 142.

[0032] The machine learning model 141 is assumed to have been trained in advance using training data. The machine learning model 141 is an example of a neural network. For example, the machine learning model 141 is trained based on a backpropagation method or the like.

[0033] The training data used for training the machine learning model 141 associates an input "resonant frequency of the circuit" with an output (correct answer data) "EMI value corresponding to the resonant frequency of the circuit." By inputting a resonant frequency to the trained machine learning model 141, an EMI value corresponding to the resonant frequency is output.

[0034] The EMI information table 142 holds, for a plurality of analyzed circuits, the impedance characteristics of the analyzed circuits in association with the EMI information. Fig. 4 is a diagram showing an example of the data structure of the EMI information table. As shown in Fig. 4, the EMI information table 142 associates circuit identification information, impedance characteristics, and EMI information. The circuit identification information is information that uniquely identifies an analyzed circuit.

[0035] The impedance characteristic indicates the frequency characteristic of the impedance of the analyzed circuit. Fig. 5 is a diagram showing an example of the impedance characteristic. The horizontal axis of the graph G6 in Fig. 5 corresponds to the frequency, and the vertical axis corresponds to the absolute value of the impedance Zin. As shown in Fig. 5, in the impedance characteristic, the impedance value is associated with each frequency.

[0036] The EMI information indicates the EMI value of each frequency of the analyzed circuit. Fig. 6 is a diagram showing an example of the EMI information. The horizontal axis of the graph G7 in Fig. 6 corresponds to the frequency, and the vertical axis corresponds to the EMI value. As shown in Fig. 6, in the EMI information, the EMI value corresponding to each frequency is associated with each frequency.

[0037] Returning to the explanation of Fig. 3, the control unit 150 includes an acquisition unit 151 and a calculation unit 152.

[0038] The acquisition unit 151 acquires information on the circuit to be estimated via the communication unit 110 or the input unit. The acquisition unit 151 generates equivalent circuit information equivalent to the circuit to be estimated based on the information on the circuit to be estimated. The equivalent circuit information is information in which elements of the circuit to be estimated are simplified into a minimum number of elements that represent the characteristics of the circuit to be estimated.

[0039] The acquisition unit 151 inputs the equivalent circuit information to a circuit simulator to acquire the resonant frequency and impedance characteristics of the estimation target circuit (equivalent circuit information). For example, the acquisition unit 151 uses SPICE (Simulation Program with Integrated Circuit Emphasis) or the like as the circuit simulator.

[0040] Next, the acquisition unit 151 inputs the resonant frequency of the estimation target circuit to the trained machine learning model 141, thereby acquiring an EMI value corresponding to the resonant frequency of the estimation target circuit.

[0041] The acquisition unit 151 outputs to the calculation unit 152 the EMI value at the resonant frequency of the circuit under estimation acquired by the above processing and the impedance characteristic of the circuit under estimation.

[0042] The calculation unit 152 compares the impedance characteristics of the estimation target circuit with the impedance characteristics of each analyzed circuit registered in the EMI information table 142, and identifies the impedance characteristics most similar to the impedance characteristics of the estimation target circuit. In the following description, the analyzed circuit having the impedance characteristics most similar to the impedance characteristics of the estimation target circuit is referred to as a "similar circuit."

[0043] For example, the calculation section 152 calculates the absolute values ​​of the impedance differences at each frequency of the impedance characteristics to be compared, and determines that the compared impedance characteristics are more similar as the total of the absolute values ​​is smaller.

[0044] The calculation section 152 selects EMI information of a similar circuit from the multiple pieces of EMI information included in the EMI information table 142, and acquires the selected EMI information.

[0045] Next, the calculation unit 152 calculates an EMI value at a frequency different from the resonant frequency of the circuit to be estimated, based on the EMI value at the resonant frequency of the circuit to be estimated and the EMI information of the similar circuit.

[0046] The processing of the calculation unit 152 will be described with reference to Fig. 1 described above. The EMI values ​​of the resonant frequencies of the estimation target circuit are EMI values ​​40a and 40b, and the EMI information of the similar circuit is EMI information 40c. The calculation unit 152 generates EMI information 40d by correcting the overall shape of the EMI information 40c so that the peak value of the EMI information 40c matches the EMI values ​​40a and 40b corresponding to the resonant frequencies. The EMI calculation device 100 calculates the EMI information 40d as the EMI value of each frequency of the estimation target circuit.

[0047] When correcting the overall shape of the EMI information 40c, the calculation unit 152 deforms the EMI information 40c in the frequency direction and the EMI value direction so that the peak value of the EMI information 40c matches the EMI values ​​40a and 40b. The calculation unit 152 may output the EMI information 40c of the estimation target circuit to the display unit 130 for display, or may output it to an external device via the communication unit 110.

[0048] Next, an example of a processing procedure of the EMI calculating apparatus 100 according to this embodiment will be described. Fig. 7 is a flowchart showing the processing procedure of the EMI calculating apparatus according to this embodiment. As shown in Fig. 7, the acquiring unit 151 of the EMI calculating apparatus 100 receives input of information on the estimation target circuit (step S101).

[0049] The acquiring unit 151 generates equivalent circuit information of the estimation target circuit (step S102). The acquiring unit 151 executes a circuit simulator and acquires the resonant frequency and impedance characteristics of the estimation target circuit (equivalent circuit information) (step S103).

[0050] The acquiring unit 151 inputs the resonant frequency to the machine learning model 141 and acquires an EMI value corresponding to the resonant frequency (step S104). The EMI calculating device 100 compares the impedance characteristic of the estimation target circuit with the impedance characteristic in the EMI information table 142 to identify similar circuits (step S105).

[0051] The calculation unit 152 acquires EMI information corresponding to the similar circuit from the EMI information table 142 (step S106). The calculation unit 152 calculates the EMI information of the estimation target circuit based on the EMI value of the resonant frequency of the estimation target circuit and the acquired EMI information (step S107). The calculation unit 152 outputs the EMI information of the estimation target circuit (step S108).

[0052] Next, the effect of the EMI calculating device 100 according to the present embodiment will be described. The EMI calculating device 100 uses the machine learning model 141 to identify the EMI value of the resonant frequency of the estimation target circuit. The EMI calculating device 100 also calculates the EMI values ​​of the remaining frequencies other than the resonant frequency by using EMI information of a similar circuit having characteristics similar to the impedance identification of the estimation target circuit. Since the machine learning model 141 is trained based on the relationship between the resonant frequency and the EMI value, it is possible to reduce the load required for machine learning of the machine learning model compared to the case where machine learning is performed using the EMI values ​​of all frequencies.

[0053] Furthermore, the EMI calculation apparatus 100 corrects the shape of the entire EMI information so that the peak value of the EMI information of the similar circuit matches the EMI value corresponding to the resonance frequency of the analyzed circuit, thereby making it possible to calculate the EMI value at all frequencies without impairing the estimation accuracy of the EMI value.

[0054] The EMI calculating device 100 generates equivalent circuit information from information on the circuit to be estimated, obtains a resonant frequency by inputting the equivalent circuit information to a circuit simulator, and specifies an EMI value by inputting the obtained resonant frequency to the machine learning model 141. In this way, the EMI value corresponding to the resonant frequency of the circuit to be estimated can be efficiently obtained.

[0055] The above-described process of the EMI calculating apparatus 100 is just an example, and the EMI calculating apparatus 100 may execute other processes. For example, the EMI calculating apparatus 100 trains the machine learning model 141 using training data, and calculates an EMI value corresponding to a resonant frequency using the trained machine learning model 141.

[0056] Furthermore, when training the machine learning model 141, the EMI calculating device 100 may train the model using training data that associates the inputs "resonant frequency of the circuit" and "image of current distribution of the circuit" with the output "EMI value corresponding to the resonant frequency of the circuit." In this case, the acquiring unit 151 acquires images of the resonant frequency and current distribution of the estimation target circuit (equivalent circuit information) using a circuit simulator, and inputs the acquired images of the resonant frequency and current distribution to the machine learning model 141 to calculate the EMI value corresponding to the resonant frequency.

[0057] Next, a description will be given of an example of a hardware configuration of a computer that realizes the same functions as the EMI calculation apparatus 100 shown in the above embodiment. Fig. 8 is a diagram showing an example of a hardware configuration of a computer that realizes the same functions as the EMI calculation apparatus.

[0058] 8, the computer 200 has a CPU 201 that executes various arithmetic processes, an input device 202 that accepts data input from a user, and a display 203. The computer 200 also has a communication device 204 that receives data from an external device, and an interface device 205 that connects to various devices. The computer 200 has a RAM 206 that temporarily stores various information, and a hard disk device 207. The devices 201 to 207 are connected to a bus 208.

[0059] The hard disk device 207 stores an acquisition program 207a and a calculation program 207b. The CPU 201 reads out the acquisition program 207a and the calculation program 207b and loads them in the RAM 206.

[0060] The acquisition program 207a functions as an acquisition process 206a, and the calculation program 207b functions as a calculation process 206b.

[0061] The processing of the acquisition process 206a corresponds to the processing of the acquisition unit 151. The processing of the calculation process 206b corresponds to the processing of the calculation unit 152.

[0062] It is not necessary that each of the programs 207a and 207b is stored in the hard disk device 207 from the beginning. For example, each of the programs may be stored in a "portable physical medium" such as a flexible disk (FD), a CD-ROM, a DVD disk, a magneto-optical disk, or an IC card that is inserted into the computer 200. Then, the computer 200 may read and execute each of the programs 207a and 207b.

[0063] The following supplementary notes are further disclosed regarding the embodiments including the above examples.

[0064] (Appendix 1) Input circuit information of a first circuit into a machine learning model to obtain an EMI value of the first circuit at a specific frequency; selecting a first EMI information from a plurality of pieces of EMI information in which the impedance characteristics of each of a plurality of circuits are associated with the EMI values ​​of each of the plurality of circuits at a plurality of frequencies, based on the impedance characteristics of the first circuit and the EMI value at the specific frequency; calculating an EMI value of the first circuit at a frequency different from the specific frequency based on the EMI value at the specific frequency and the first EMI information; An EMI calculation program that causes a computer to execute a process.

[0065] (Supplementary Note 2) The EMI calculation program according to Supplementary Note 1, characterized in that the selection process selects, from the plurality of pieces of EMI information, EMI information of a circuit corresponding to impedance characteristics similar to the impedance characteristics of the first circuit as the first EMI information.

[0066] (Appendix 3) The EMI calculation program according to appendix 1 or 2, characterized in that the calculation process identifies a maximum EMI value among EMI values ​​at multiple frequencies included in the first EMI information, and calculates EMI values ​​at frequencies other than the specific frequency by correcting the relationship between the multiple frequencies and EMI values ​​of the first EMI information so that the identified maximum EMI value approaches the EMI value at the specific frequency.

[0067] (Appendix 4) The EMI calculation program according to appendix 1, 2 or 3, characterized in that the acquiring process inputs the resonant frequency of the first circuit as the circuit information into the machine learning model, and acquires the EMI value at the resonant frequency of the first circuit.

[0068] (Appendix 5) Input circuit information of a first circuit into a machine learning model to obtain an EMI value at a specific frequency of the first circuit; selecting a first EMI information from a plurality of pieces of EMI information in which the impedance characteristics of each of a plurality of circuits are associated with the EMI values ​​of each of the plurality of circuits at a plurality of frequencies, based on the impedance characteristics of the first circuit and the EMI value at the specific frequency; calculating an EMI value of the first circuit at a frequency different from the specific frequency based on the EMI value at the specific frequency and the first EMI information; A method for calculating EMI, characterized in that the processing is executed by a computer.

[0069] (Appendix 6) The EMI calculation method according to appendix 5, characterized in that the selection process selects, from the plurality of pieces of EMI information, EMI information of a circuit corresponding to impedance characteristics similar to the impedance characteristics of the first circuit as the first EMI information.

[0070] (Appendix 7) The EMI calculation method according to appendix 5 or 6, characterized in that the calculation process identifies a maximum EMI value among the EMI values ​​at multiple frequencies included in the first EMI information, and calculates EMI values ​​at frequencies other than the specific frequency by correcting the relationship between the multiple frequencies and the EMI values ​​of the first EMI information so that the identified maximum EMI value approaches the EMI value at the specific frequency.

[0071] (Appendix 8) The EMI calculation method according to appendix 5, 6 or 7, characterized in that the acquiring process inputs the resonant frequency of the first circuit as the circuit information into the machine learning model, and acquires the EMI value at the resonant frequency of the first circuit.

[0072] (Appendix 9) Input circuit information of a first circuit into a machine learning model to obtain an EMI value at a specific frequency of the first circuit; selecting a first EMI information from a plurality of pieces of EMI information in which the impedance characteristics of each of a plurality of circuits are associated with the EMI values ​​of each of the plurality of circuits at a plurality of frequencies, based on the impedance characteristics of the first circuit and the EMI value at the specific frequency; calculating an EMI value of the first circuit at a frequency different from the specific frequency based on the EMI value at the specific frequency and the first EMI information; An EMI calculation device having a control unit for executing processing.

[0073] (Supplementary Note 10) The EMI calculation device according to Supplementary Note 9, characterized in that the selection process selects, from the plurality of pieces of EMI information, EMI information of a circuit corresponding to impedance characteristics similar to the impedance characteristics of the first circuit as the first EMI information.

[0074] (Supplementary Note 11) The EMI calculation device according to Supplementary Note 9 or 10, characterized in that the calculation process identifies a maximum EMI value among EMI values ​​at multiple frequencies included in the first EMI information, and calculates EMI values ​​at frequencies other than the specific frequency by correcting the relationship between the multiple frequencies and EMI values ​​of the first EMI information so that the identified maximum EMI value approaches the EMI value at the specific frequency.

[0075] (Appendix 12) The EMI calculation device according to appendix 9, 10 or 11, characterized in that the acquiring process inputs the resonant frequency of the first circuit as the circuit information into the machine learning model, and acquires an EMI value at the resonant frequency of the first circuit. [Explanation of symbols]

[0076] 100 EMI calculation device 110 Communications Department 120 Input section 130 Display section 140 Storage section 141 Machine Learning Models 142 EMI Information Table 150 Control section 151 Acquisition Department 152 Calculation section

Claims

1. A resonant frequency of a first circuit is input to a machine learning model as circuit information of the first circuit, and an EMI value at the resonant frequency of the first circuit is obtained; selecting a first EMI information from a plurality of pieces of EMI information in which the impedance characteristics of each of a plurality of circuits are associated with the EMI values ​​of each of the plurality of circuits at a plurality of frequencies, based on the impedance characteristics of the first circuit and the EMI value at the resonant frequency; calculating an EMI value at another frequency different from the resonant frequency of the first circuit based on the EMI value at the resonant frequency and the first EMI information; The process is executed by a computer, The process of selecting includes: selecting, from the plurality of pieces of EMI information, EMI information of a circuit corresponding to an impedance characteristic similar to the impedance characteristic of the first circuit as the first EMI information; EMI calculation program comprising:

2. A resonant frequency of a first circuit is input to a machine learning model as circuit information of the first circuit, and an EMI value at the resonant frequency of the first circuit is obtained; selecting a first EMI information from a plurality of pieces of EMI information in which the impedance characteristics of each of a plurality of circuits are associated with the EMI values ​​of each of the plurality of circuits at a plurality of frequencies, based on the impedance characteristics of the first circuit and the EMI value at the resonant frequency; calculating an EMI value at another frequency different from the resonant frequency of the first circuit based on the EMI value at the resonant frequency and the first EMI information; The process is executed by a computer, The calculation process includes: Identifying a maximum EMI value among EMI values ​​at a plurality of frequencies included in the first EMI information; calculating an EMI value at a frequency different from the resonant frequency by correcting a relationship between a plurality of frequencies of the first EMI information and an EMI value so that the specified maximum EMI value approaches an EMI value at the resonant frequency; EMI calculation program comprising:

3. A resonant frequency of a first circuit is input to a machine learning model as circuit information of the first circuit, and an EMI value at the resonant frequency of the first circuit is obtained; selecting a first EMI information from a plurality of pieces of EMI information in which the impedance characteristics of each of a plurality of circuits are associated with the EMI values ​​of each of the plurality of circuits at a plurality of frequencies, based on the impedance characteristics of the first circuit and the EMI value at the resonant frequency; calculating an EMI value at another frequency different from the resonant frequency of the first circuit based on the EMI value at the resonant frequency and the first EMI information; The computer executes the process, The process of selecting includes: selecting, from the plurality of pieces of EMI information, EMI information of a circuit corresponding to an impedance characteristic similar to the impedance characteristic of the first circuit as the first EMI information; EMI calculation method comprising the steps of:

4. Inputting a resonant frequency of a first circuit into a machine learning model as circuit information of the first circuit, and obtaining an EMI value at the resonant frequency of the first circuit; selecting a first EMI information from a plurality of pieces of EMI information in which the impedance characteristics of each of a plurality of circuits are associated with the EMI values ​​of each of the plurality of circuits at a plurality of frequencies, based on the impedance characteristics of the first circuit and the EMI value at the resonant frequency; calculating an EMI value at another frequency different from the resonant frequency of the first circuit based on the EMI value at the resonant frequency and the first EMI information; The computer executes the process, The calculation process includes: Identifying a maximum EMI value among EMI values ​​at a plurality of frequencies included in the first EMI information; calculating an EMI value at a frequency different from the resonant frequency by correcting a relationship between a plurality of frequencies of the first EMI information and an EMI value so that the specified maximum EMI value approaches an EMI value at the resonant frequency; EMI calculation method comprising the steps of:

5. A resonant frequency of a first circuit is input to a machine learning model as circuit information of the first circuit, and an EMI value at the resonant frequency of the first circuit is obtained; selecting a first EMI information from a plurality of pieces of EMI information in which the impedance characteristics of each of a plurality of circuits are associated with the EMI values ​​of each of the plurality of circuits at a plurality of frequencies, based on the impedance characteristics of the first circuit and the EMI value at the resonant frequency; calculating an EMI value at another frequency different from the resonant frequency of the first circuit based on the EMI value at the resonant frequency and the first EMI information; A control unit that executes processing, The control unit is selecting, from the plurality of pieces of EMI information, EMI information of a circuit corresponding to an impedance characteristic similar to the impedance characteristic of the first circuit as the first EMI information; An EMI calculation device that performs the process.

6. Inputting a resonant frequency of a first circuit into a machine learning model as circuit information of the first circuit, and obtaining an EMI value at the resonant frequency of the first circuit; selecting a first EMI information from a plurality of pieces of EMI information in which the impedance characteristics of each of a plurality of circuits are associated with the EMI values ​​of each of the plurality of circuits at a plurality of frequencies, based on the impedance characteristics of the first circuit and the EMI value at the resonant frequency; calculating an EMI value at another frequency different from the resonant frequency of the first circuit based on the EMI value at the resonant frequency and the first EMI information; A control unit that executes processing, The control unit is Identifying a maximum EMI value among EMI values ​​at a plurality of frequencies included in the first EMI information; calculating an EMI value at a frequency different from the resonant frequency by correcting a relationship between a plurality of frequencies of the first EMI information and an EMI value so that the specified maximum EMI value approaches an EMI value at the resonant frequency; An EMI calculation device that performs the process.

Images