ICIM-CI model modeling method based on particle swarm optimization

Abstract

The invention relates to the field of ICIM-CI model modeling, in particular to an ICIM-CI model modeling method based on particle swarm optimization. The modeling method mainly comprises the following steps: segmenting input signals based on particle swarm optimization; recombining the output signals; and carrying out segmented Volterra kernel function identification. According to the method, the nonlinear behavior of the chip is effectively described through the piecewise Volterra series, and the piecewise threshold is optimized by using the particle swarm optimization algorithm, so that the precision and universality of the model are remarkably improved. According to the method, the problems that linear hypothesis exists in the modeling process of a traditional ICIM-CI model, the nonlinear response of MADIC under electromagnetic interference is difficult to accurately describe, and the defects exist in the aspects of sensitivity quantitative analysis, quantitative information output and parameterized cascade simulation under the integration of multiple criteria are solved. The ICIM-CI model modeling method is suitable for ICIM-CI model modeling.

Description

Technical Field

[0001] The present invention relates to the field of ICIM-CI (Integrated Circuit Immunity Model-Conducted Immunity, conducted immunity of integrated circuit radio frequency immunity behavior simulation) model modeling, and in particular to an ICIM-CI model modeling method based on particle swarm optimization. Background Art

[0002] Integrated circuits are the core of electronic devices and systems, and their electromagnetic immunity directly affects the electromagnetic compatibility of the entire device and the safety of system operation. Analog and hybrid chips are integrated circuits with both nonlinear and linear characteristics. They are very sensitive to electromagnetic interference. Minor changes in circuit topology may also cause significant changes in the electromagnetic immunity of the device or the entire system. Therefore, predicting the impact of analog and hybrid chips on operation under RF interference and establishing accurate chip conduction sensitivity models have become key electromagnetic compatibility issues.

[0003] However, the accurate conducted immunity characteristics of the chip cannot be obtained from the simulation functional model provided by the chip manufacturer, because the functional model only focuses on the function and logic of the chip, and does not include the anti-interference characteristics such as the response characteristics and stability of the chip under external electromagnetic interference.

[0004] If the conducted immunity of the chip is not accurately modeled, the sensitivity of the circuit cannot be correctly defined, which will affect the electromagnetic compatibility design and circuit layout of the circuit. To solve this problem, it is urgent to study the modeling method of chip conducted sensitivity, especially "black box modeling". Because "black box modeling" does not require knowledge of the circuit structure or design of the chip, it only needs to model according to the input and output characteristics of the modeled chip, so the chip user does not need to understand the internal structure of the circuit to build the model. That is, the chip user can obtain the model through the test results without understanding the internal circuit structure of the chip.

[0005] The black box models used in previous studies, such as statistical models, envelope domain models, and artificial neural networks, etc., but these models focus more on the failure of the chip at a certain frequency point, but have some problems and shortcomings in terms of wide bandwidth and model universality.

[0006] In recent years, the ICIM-CI model based on the PDN (Passive Distribution Network) module and the IB (Immunity Behavior) module has obvious advantages in circuit simulation, and it can provide a chip broadband immunity simulation model. However, since the model requires linear assumptions when building IB, this is inconsistent with the nonlinear effects of analog and analog-to-digital hybrid chips after being subjected to electromagnetic interference, which also leads to differences between the model and the actual measurement results. Summary of the invention

[0007] The purpose of the present invention is to overcome the shortcomings of the prior art and provide an ICIM-CI modeling method based on particle swarm optimization, which can provide accurate nonlinear output response.

[0008] The present invention adopts the following technical scheme to achieve the above-mentioned purpose. The present invention provides an ICIM-CI modeling method based on particle swarm optimization, and the modeling method comprises:

[0009] The IB module of the chip is established by introducing the segmented Volterra series, and the particle swarm optimization algorithm is used to reduce the segmented Volterra series, including:

[0010] S1. Input signal segmentation based on particle swarm optimization:

[0011] S2, output signal recombines;

[0012] S3. Segmented Volterra kernel function identification.

[0013] Furthermore, step S1 specifically includes:

[0014] The input signal is decomposed according to the power size to determine the input power threshold. The threshold is expressed as follows:

[0015] Γ={λ 1 ,λ 2 ,L,λ s},λ 1 <λ 2 <L<λ s ;

[0016] where λ i ,i=1,2,L,s is the threshold of the input signal power amplitude, s is the number of thresholds set;

[0017] The input signal x(t) passes through the set threshold and is decomposed into s+1 sub-signals according to the following formula:

[0018]

[0019] The decomposed input sub-signals are sampled in discrete time. The input sub-signals have the same time dimension. Assuming the number of sampling points is n, the decomposed input sub-signals are represented by a (s+1)×n matrix:

[0020]

[0021] In the decomposition process, the initial value of the particle in the optimization space is randomly set according to the objective function space, and the particle initial value is set according to the individual extreme value p. best and the group extreme value g best Update its own speed and position, and find the optimal value in the objective function space:

[0022] Let p num is the total number of particles, D is the particle dimension, when the maximum number of iterations max is reached iter Stop iteration when v a (t) is the iteration speed of the ath particle at time t, x a (t) is the position of the ath particle in the objective function space at time t, pbest a (t) is the individual extreme value of the ath particle at time t;

[0023] v a (t) = [v a1 (t),v a2 (t),…,v aD (t)] T , x a (t) = [x a1 (t),x a2 (t),…,x aD (t)] T ;

[0024] The update method of particles in the objective function space is as follows:

[0025] v a+1 (t+1)=ωv a (t)+c 1 a 1 (pbest a (t)-x a (t))+c 2 a 2 (gbest a (t)-x a (t));

[0026] x a+1 (t+1)=x a (t)+v a+1 (t+1);

[0027] Among them, ω is the inertia weight coefficient, which represents the influence of the previous iteration on the current iteration. 1 is the individual learning factor, c 2 is the group learning factor. 1 and a 2 Is a random number.

[0028] Furthermore, step S2 specifically includes:

[0029] After completing the power threshold segmentation, the output is weighted combined to obtain a complete output signal;

[0030] The output signal corresponding to each input sub-signal is shown as follows:

[0031]

[0032] In the formula, y i (·) is the output sub-signal corresponding to the i-th input sub-signal, K i is the order of the Volterra series corresponding to the i-th input sub-signal, M i Memory length of the input sub-signal for the i-th segment;

[0033] The output signal is obtained by synthesizing each output sub-signal as follows:

[0034]

[0035] Furthermore, step S3 specifically includes:

[0036] The piecewise kernel function uses the following formula to predict the output waveform:

[0037]

[0038] Where X pre_,i i=1,2L,s,+ is the input matrix obtained by multiplying the i-th input sub-signal and the memory delay, is a piecewise kernel function.

[0039] The beneficial effects of the present invention are:

[0040] The present invention effectively describes the nonlinear behavior of the chip through the segmented Volterra series, and optimizes the segmented threshold using the PSO algorithm, which significantly improves the accuracy and versatility of the model. The model overcomes the linear assumptions in the modeling process of the traditional ICIM-CI model, which makes it difficult to accurately describe the nonlinear response of MADIC (Mixed Analog-Digital Integrated Circuit) under electromagnetic interference, and has deficiencies in sensitivity quantitative analysis, output quantitative information and parameterized cascade simulation under the comprehensive analysis of multiple criteria. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flow chart of the ICIM-CI modeling method based on particle swarm optimization provided by an embodiment of the present invention;

[0042] Figure 2 It is a schematic diagram of the ICIM-CI model structure provided by an embodiment of the present invention;

[0043] Figure 3 is an optimization flow chart of a particle swarm optimization algorithm provided by an embodiment of the present invention;

[0044] Figure 4 is a normalized mean square error convergence curve diagram of the particle swarm optimization provided by an embodiment of the present invention;

[0045] Figure 5 It is a comparison curve diagram of sensitivity prediction and actual output result when the output waveform is offset by 5% or turned off provided by an embodiment of the present invention;

[0046] Figure 6 It is a comparison curve diagram of sensitivity prediction and actual output result when the output waveform is offset by 10% or turned off provided by an embodiment of the present invention;

[0047] Figure 7 It is a comparison curve diagram of the model output and the actual output result when the chip provided by the embodiment of the present invention works normally (interference is injected at 90MHz and -10dBm);

[0048] Figure 8 It is a comparison curve diagram of the model output and the actual output provided by the embodiment of the present invention when a slight disturbance occurs to the chip output (interference is injected at 90MHz with 10dBm);

[0049] Fig. 9 It is a comparison curve diagram of the model output and the actual output result when the chip output is severely disturbed (interference injected at 90MHz, 25dBm) provided by the embodiment of the present invention;

[0050] Fig.10 It is a comparison curve diagram of the model output and the actual output result provided by the embodiment of the present invention when the chip output is severely disturbed and the waveform is severely distorted (interference is injected at 900MHz with 25dBm);

[0051] Fig.11 It is a comparison curve diagram of the sensitivity prediction result and the actual output result of the ICIM-CI (PSVIB) modeling method provided by the embodiment of the present invention. DETAILED DESCRIPTION

[0052] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0053] The ICIM-CI model of the present invention is composed of a PDN module and an IB module. The model structure is as follows: Figure 2 shown.

[0054] PDN describes the linear behavior of the chip, IB describes the behavior of the chip's in-band and out-of-band frequency response, and the information output by the monitoring port describes the chip's response to the interference signal applied to the disturbance injection port. IB contains parameters such as frequency, transmission power, and changes in the monitoring port. IB data is obtained by using pass / fail or fail sensitivity test standards.

[0055] There are two modeling methods for IB. The first is to use the go-no-go sensitivity test standard to directly obtain IB binary data. The second is to use non-pass or fail tests to extract fault detection data and then use sensitivity test standards (criteria) to predict the pass or fail results of the device. Compared with the first method, the second method predicts faults through additional monitoring and analysis steps, and can provide output data (such as power, voltage, etc.), and then apply sensitive standards to evaluate whether these data meet the fault conditions. Due to the introduction of additional analysis steps, the number of tests is indirectly reduced. At the same time, it also provides a more flexible means for fault analysis, which is the same as the direction of improvement of the present invention.

[0056] When using the second method, a linear fitting equation needs to be established to obtain fault detection simulation data. That is, at each frequency point, parameters such as interference injection power / voltage, frequency, and sensitive performance indicators are recorded, and a linear fitting equation is established. The model is ultimately expressed as a linearized sensitivity prediction model expressed in the form of a netlist, without containing the quantized output data of the monitoring port. But in reality, when analog and analog-to-digital hybrid chips are sensitive, they usually show nonlinearity, which is different from the linear assumption of the model. This also leads to differences between the model prediction results and the actual measurement results.

[0057] In order to improve the nonlinearity of the ICIM-CI model and increase the quantized waveform of the model output for anti-interference analysis, the present invention introduces the segmented Volterra series to establish the IB module of the chip, and uses the particle swarm optimization algorithm to reduce the segmented Volterra series to further improve the model accuracy.

[0058] The Volterra series is a functional series used to describe nonlinear systems. It establishes nonlinear systems based on input signals and output signals. The Volterra time domain kernel and frequency domain kernel both have corresponding physical meanings. The time domain kernel corresponds to the high-order impulse response function of the system, and the frequency domain kernel corresponds to the generalized frequency response function of the system. When the Volterra series is used to describe a nonlinear system, it can be described as follows:

[0059]

[0060] Where x(t),y(t)∈R, x(t) is the input signal of the system, y(t) is the output signal of the system, and h n (τ 1 ,τ 2 ,L,τ n ) is the Volterra time domain kernel of the described nonlinear system corresponding to the nth order. The above can also be solved in the frequency domain by performing a multidimensional Fourier transform.

[0061] When the Volterra series is actually used to describe a nonlinear system, a discrete model is often used and the order and memory length are truncated, taking the order as N and the memory length as M. To describe a nonlinear system, the corresponding Volterra series model is shown as follows:

[0062]

[0063] Where g(m 1 )…is the kernel function obtained by sampling the time domain kernel at time T, and its specific representation is as follows:

[0064] g(m 1 )=h 1 (m 1 T)·T

[0065] g(m 1 ,m 2 )=h 2 (m 1 T,m 2 T)·T 2

[0066] g(m 1 ,m 2 ,L,m N )=h N (m 1 T,m 2 T,L,m N T)·T N

[0067] For strongly nonlinear systems, because higher orders and memory lengths need to be used for fitting, the number of Volterra kernel functions will grow exponentially. To address this problem, the present invention proposes a piecewise Volterra series method for application in the modeling process of memristors, and the piecewise Volterra series can be used to reduce the order of kernel functions. For the application of the conductive sensitivity modeling problem of chips, the input power can be segmented, each segmented signal corresponds to a low-order Volterra kernel function, and then the output signals are comprehensively weighted to obtain the total output signal. This processing method reduces the complexity of the model while maintaining a high modeling accuracy. The piecewise Volterra series modeling process mainly includes: input signal segmentation, output signal recombination, and piecewise Volterra kernel function identification.

[0068] However, the segmented Volterra series modeling method does not have a centralized definition for the selection of power thresholds. Usually, it is necessary to continuously try according to the input waveform, and blindly setting the power threshold will lead to an increase in the system error, which reduces the versatility of the model. Therefore, in order to solve this problem, the present invention uses a particle swarm optimization algorithm to intelligently segment the power threshold to optimize the segmented Volterra series, ensure the accuracy in the segmented Volterra reduction process, and further improve the versatility and repeatability of the method in the IB module modeling process, so as to be more suitable for quickly building an IB module.

[0069] like Figure 1 As shown, the specific technical solution is as follows.

[0070] Input signal segmentation based on particle swarm optimization:

[0071] The input signal is decomposed according to the power size to determine the input power threshold. The threshold is expressed as follows:

[0072] Γ={λ 1 ,λ 2 ,L,λ s},λ 1 <λ 2 <L<λ s ;

[0073] where λ i ,i=1,2,L,s is the threshold of the input signal power amplitude, s is the number of thresholds set;

[0074] The input signal x(t) passes through the set threshold and is decomposed into s+1 sub-signals according to the following formula:

[0075]

[0076] The decomposed input sub-signals are sampled in discrete time. The input sub-signals have the same time dimension. Assuming the number of sampling points is n, the decomposed input sub-signals are represented by a (s+1)×n matrix:

[0077]

[0078] In the decomposition process, the initial value of the particle in the optimization space is randomly set according to the objective function space, and the particle initial value is set according to the individual extreme value p. best and the group extreme value g best Update its own speed and position, and find the optimal value in the objective function space:

[0079] Let p num is the total number of particles, D is the particle dimension, when the maximum number of iterations max is reached iter Stop iteration when v a (t) is the iteration speed of the ath particle at time t, x a (t) is the position of the ath particle in the objective function space at time t, pbest a (t) is the individual extreme value of the ath particle at time t;

[0080] v a (t) = [v a1 (t),v a2 (t),…,v aD (t)] T , x a (t) = [x a1 (t),x a2 (t),…,x aD (t)] T ;

[0081] pbest a (t) = [p a1 (t),p a2 (t),…,p aD (t)] T ;

[0082] gbest=[g 1 ,g 2 …,g d ] T ;

[0083] The update method of particles in the objective function space is as follows:

[0084] v a+1 (t+1)=ωv a (t)+c 1 a 1 (pbest a (t)-xa (t))+c 2 a 2 (gbest a (t)-x a (t));

[0085] x a+1 (t+1)=x a (t)+v a+1 (t+1);

[0086] Among them, ω is the inertia weight coefficient, which represents the influence of the previous iteration on the current iteration. 1 is the individual learning factor, c 2 is the group learning factor. 1 and a 2 Is a random number.

[0087] According to the above analysis, the optimization process of particle swarm optimization algorithm is as follows: Figure 3 As shown:

[0088] S1, setting the objective function space for the particle swarm optimization algorithm;

[0089] S2, initialization parameters, such as population size and inertia weight;

[0090] S3, calculate the function value of each particle;

[0091] S4, update individual optimal value and global optimal value;

[0092] S5, update the velocity and position of the particle;

[0093] S6, recalculate the function value of each particle;

[0094] S7. Set the iteration termination condition. If the termination condition is not met, return to step S5. If the termination condition is met, output the optimal solution.

[0095] Output signal reassembly:

[0096] After completing the power threshold segmentation, the output is weighted combined to obtain a complete output signal;

[0097] The output signal corresponding to each input sub-signal is shown as follows:

[0098]

[0099] In the formula, y i (·) is the output sub-signal corresponding to the i-th input sub-signal, K i is the order of the Volterra series corresponding to the i-th input sub-signal, M i Memory length of the input sub-signal for the i-th segment;

[0100] The output signal is obtained by synthesizing each output sub-signal as follows:

[0101]

[0102] Piecewise Volterra kernel function identification:

[0103] The piecewise kernel function uses the following formula to predict the output waveform:

[0104]

[0105] in is the input matrix obtained by multiplying the i-th input sub-signal and the memory delay, is a piecewise kernel function.

[0106] The solution of the present invention is further described below in conjunction with specific embodiments.

[0107] The following modeling work was carried out on the SF702AMC analog switch chip, SJ96F174MC four-channel differential driver, etc. to verify the versatility of the modeling method.

[0108] SF702AMC analog switch modeling example:

[0109] The SF702AMC analog switch is selected as the test and modeling object. The SF702AMC analog switch is a semiconductor device for analog signal control and transmission. It is a single-pole single-throw (SPST) analog switch with wide bandwidth (-3dB bandwidth greater than or equal to 110MHz), low power consumption (less than or equal to 1uA) and low on-resistance (less than or equal to 6Ω). It uses a 5-lead ceramic leadless chip carrier (LCC05B) package, supports signal switching based on control voltage levels, and can provide reliable signal routing in various applications. Its basic function is to select the on and off mode between the analog input terminal S and the analog output terminal D under the control of the switch control terminal IN. When the IN port is low, the S terminal and the D terminal are turned on; when the IN port is high, the S terminal and the D terminal are turned off. The symbol table of its corresponding lead-out port is shown in Table 6 below. According to the corresponding port relationship and basic functions, the conduction sensitivity modeling test board is designed.

[0110] Table 6 SF702AMC analog switch lead port function symbol table

[0111] Pin number symbol Function Pin number symbol Function 1 D Analog Output 4 IN Switch Control 2 S Analog Input 5 <![CDATA[V DD ]]> power supply 3 GND land

[0112] Modeling process: Select power meter and power amplifier according to the specific power size and port characteristics. When extracting the PSVIB module, set the analog switch to work normally with analog input frequency f = 1kHz and peak-to-peak value V PP =0.5V sine wave, the switch control terminal IN port is set to low level V = 0.8V, at this time, the analog switch is turned on, the S port and the D port are connected, and the analog output port should be output frequency f = 1kHz, peak value V PP = 0.5 V sine wave. Radio frequency interference is injected into the port where the model needs to be built, PSVIB is built using the solution provided by the present invention, and the PDN model is built using the active integrated device PDN vector fitting modeling method, and finally the ICIM-CI (PSVIB) model is generated.

[0113] Verification results: In the process of applying electromagnetic interference to all ports of the analog switch, the IN port responsible for controlling the switch state is most sensitive to EMI. Therefore, a detailed analysis is performed using the IN port as an example. When sinusoidal interference f=90MHz is injected into the analog switch, when the interference power reaches P=10dBm, the output of the analog switch shows obvious waveform distortion. This shows that the chip becomes sensitive to interference signals at 10dBm. Since the traditional ICIM-CI modeling method cannot generate output waveforms in the time domain, at this point, the IB module is built and the corresponding ICIM-CI model is established through three methods (original Volterra series, segmented Volterra series, and PSVIB), and it is compared with the output waveform obtained from the actual DPI test to verify the accuracy of the PSVIB method compared with other methods.

[0114] According to the requirements of the IEC62433 standard, the DPI test of the chip was carried out, and the obtained time domain measurement results were compared with the time domain results generated by the three established models to verify the accuracy of the model. When the interference was injected at 90MHz and 10dBm, the output of the chip showed obvious nonlinearity. At this time, when the original Volterra series was used to build the IB model, when the order N=3 and the memory length M=3 were set, it captured some nonlinear behavior, but the error was still large, with a normalized mean square error (NMSE) of -9.7755dB. When a higher order (N=5) and a longer memory length (M=4) were used, the model was improved, and the normalized mean square error was reduced to -13.4332dB, but the error was still large and 1362 kernel functions were required for identification. When using the segmented Volterra series to build the IB model, the input signal is divided into intervals {-0.15, 0, 0.15}, and the order and storage length (N=3, M=3) of each segment are kept the same. At this time, the model significantly improves the ability to capture nonlinear behavior, and the NMSE reaches -17.7621dB. Based on the segmented Volterra model, the PSO algorithm is used to optimize and automatically obtain the optimal power threshold. The model accuracy is further improved, and the NMSE is -19.9152dB. At this time, the optimization results of the power segment threshold are {0.2195, -0.1287, 0.2693}, and the normalized mean square error convergence of the solution process is shown in the figure. Figure 4 The number of time domain kernel identifications and the normalized root mean square error corresponding to the IB modules generated by different solution methods are shown in Table 1.

[0115] Table 1 Number of identified kernel functions and NMSE in the time domain for different Volterra series models

[0116]

[0117]

[0118] It can be found that in the ICIM-CI model of the IB module constructed by different methods, the traditional Volterra series model still has a relatively high model error even with high order and long memory depth, and the number of time-domain kernels to be identified is also large. However, when the segmented Volterra series is introduced, the model can more accurately describe the strong nonlinear characteristics, thereby significantly improving the model accuracy. Furthermore, by applying particle swarm optimization to the segmented power threshold, it can be observed that the model accuracy has been significantly improved again.

[0119] In order to verify the accuracy of the ICIM-CI (PSVIB) model in broadband, the sensitivity curves of the measured data, traditional ICIM-CI and the ICIM-CI (PSVIB) model proposed in this paper are compared within the frequency band of 1MHz-1GHz. In the actual test, the interference is injected into the IN port of the chip, and the maximum injection power is 25dBm. In the sensitivity assessment of the circuit board or system, the user needs to set a reasonable sensitivity standard by systematically quantifying the design of distributed electromagnetic compatibility parameters. In the present invention, the sensitivity curve is first drawn when the output waveform is offset by 5% or shut down as the sensitivity test standard (Table 3). At the same time, the sensitivity prediction curve of the traditional ICIM-CI model is constructed according to Table 2. For the ICIM-CI (PSVIB) model that has been constructed, the sensitivity prediction curve under its test standard (criteria) is also drawn according to Table 2. Finally, the three curves are compared (see Figure 5 ). The results show that compared with the traditional ICIM-CI modeling method, the results obtained by ICIM-CI (PSVIB) are closer to the sensitivity curve obtained by the test in a wide frequency band. Among them, the overall NMSE of the model is -18.3521dB when the traditional ICIM-CI modeling method is used, and the overall NMSE of the model is -25.6765dB when the ICIM-CI (PSVIB) modeling method is used, which is an improvement of 7.3dB. This shows that the ICIM-CI (PSVIB) method has higher accuracy in simulating the sensitivity of the chip to electromagnetic interference. In addition, when the traditional ICIM-CI model is used for modeling according to the sensitivity test criteria in Table 3, 4 prediction netlists need to be built and comprehensively judged, which means that the traditional ICIM-CI method must perform four independent modeling tasks to obtain the final result. In contrast, ICIM-CI (PSVIB) only needs one modeling task to obtain the same result because it can make a judgment based on the output waveform. This means that the significant reduction in the number of modeling iterations directly leads to an efficiency improvement of at least 75%.

[0120] Table 2 SF702AMC analog switch sensitivity test criteria

[0121] Characteristic parameters Limits Maximum Greater than 0.275V or less than 0.1V Minimum Less than -0.275V or greater than -0.1V

[0122] Table 3 SJ96F174MC type four-way differential driver lead port function symbol table

[0123]

[0124]

[0125] In order to verify the accuracy of the model generated by ICIM-CI (PSVIB) after adjusting the test standard, the ICIM-CI (PSVIB) model of the chip is plotted with a sensitivity curve when the output waveform is offset by 10% or turned off as the sensitivity test standard (Table 4). At this time, due to the change in the criterion, the original ICIM-CI model needs to be rebuilt. Similarly, the sensitivity curves of the two models are plotted according to the new test standard (criteria) and compared with the test results, as shown in Figure 4. Figure 6 As shown, it can be found that when using the traditional ICIM-CI modeling method, the overall NMSE of the model is -11.0895dB, and when using the ICIM-CI (PSVIB) modeling method, the overall NMSE of the model is -24.429dB. ICIM-CI (PSVIB) can still maintain a high prediction accuracy and does not require re-modeling.

[0126] Table 4 SJ96F174MC sensitivity test criteria

[0127] Characteristic parameters Limits Maximum Greater than or equal to 2.5V or less than or equal to 1.5V Minimum Greater than or equal to -1.5V or less than or equal to -2.5V average value Greater than or equal to 0.2V or less than or equal to -0.2V

[0128] Table 5 shows the performance comparison of the two modeling methods in different scenarios, including model error, complexity of handling multiple criteria, and whether remodeling is required when changing criteria. It can be seen that the ICIM-CI (PSVIB) model is superior to the traditional ICIM-CI model in multiple dimensions.

[0129] First, from the perspective of model error, in single-point analysis, the traditional ICIM-CI model cannot output single-point waveforms and can only provide some monitoring parameter points, while the ICIM-CI (PSVIB) model can output accurate waveforms with an error of -25.6765dB; and under the wide-band conducted sensitivity prediction modeling, the error of the traditional ICIM-CI model is -18.3521dB, while the error of the ICIM-CI (PSVIB) model is even smaller, at -25.6765dB. This shows that the ICIM-CI (PSVIB) model can more accurately predict the impact of electromagnetic interference on chips, both in single-point analysis and in a wider frequency range, and is closer to the actual test results.

[0130] Secondly, the ICIM-CI (PSVIB) model is more convenient in dealing with the complexity of multiple criteria. The ICIM-CI (PSVIB) model can directly provide detailed actual waveform output and then set the criteria, which directly simplifies the comprehensive processing of multiple criteria. The traditional ICIM-CI model needs to build multiple prediction models when dealing with complex criteria, which greatly increases the difficulty of modeling and the amount of calculation.

[0131] Finally, the ICIM-CI (PSVIB) model has the significant advantage of not requiring remodeling when changing the criteria. Users can directly adjust the criteria according to their needs without rebuilding the model, which greatly improves flexibility and efficiency. The traditional ICIM-CI model requires remodeling when changing the criteria, which limits its adaptability in practical applications.

[0132] Table 5 Performance comparison of two modeling methods in different scenarios

[0133]

[0134] SJ96F174MC quad differential driver modeling example:

[0135] This section uses the SJ96F174MC four-channel differential driver to verify the universality of the ICIM-CI (PSVIB) modeling method. The SJ96F174MC is a monolithic integrated circuit manufactured using a 0.6μm CMOS process. It is a data transmission interface circuit for the EIA-422A / 485 communication standard. Its main function is to convert the input TTL logic signal into a high-speed, high-drive capability differential signal to achieve long-line transmission. The device can achieve three-state output through the enable control terminal. Each enable terminal controls the output of two drivers respectively, and has the advantages of fast speed and strong drive capability. The device uses a 20-lead ceramic LCC package. In addition, Table 3 shows the port functions and symbols of the chip.

[0136] Modeling process: Set the normal working condition as the first driver input 1A port input: frequency is 1kHz, peak-to-peak value is 4V pp , offset 0V DC The power supply voltage Vcc = 5V, and the EN1 / 2 enable port is 3V high level. At this time, the 1Y port is connected to the IN1 port, and the normal output frequency of the 1Y port is 1kHz, the peak-to-peak value is 4Vpp, and the offset is 0VDC. Inject interference into the EN1 / 2 enable port, and collect the input time domain waveform of the 1A port and the output time domain waveform of the 1Y port. Use the theory in Chapter 2 to build the ICIM-CI (PSVIB) model.

[0137] By comparing the sensitivity curves obtained from DPI measurements with the time domain output results of the established ICIM-CI (PSVIB) model ( Figure 7-Figure 10 ). It can be found that when interference with different frequencies and different injection powers is injected into the EN1 / 2 enable port, the time domain output results of the model generated by the ICIM-CI (PSVIB) modeling method are highly consistent with the actual measurement results. The model can accurately provide the quantized waveform output of the chip during normal operation and when disturbances occur. Even if the output waveform is severely disturbed (see Fig. 9 and Fig.10 ), the output results of the model can still be accurately consistent with the measurement results. This shows that the time domain output results of the model have good accuracy and versatility under different interference frequencies and different interference injection powers.

[0138] When the sensitivity test standard of the chip is set according to the circuit or system design requirements, the present invention takes the sensitivity test standard set in Table 4 as an example to draw the sensitivity prediction curve of the ICIM-CI (PSVIB) model in the frequency range of 10MHz-1GHz (see Fig.11 ) and compared with the sensitivity curves of actual tests. It can be seen that the ICIM-CI (PSVIB) method proposed in this paper and the sensitivity curves obtained by the test still perform well in a wide frequency range, which shows that the model has high accuracy and good versatility in the frequency domain. In some high-frequency bands, errors occur in the high-frequency bands due to the same memory length settings. In the future, the automatic adjustment of memory length will be carried out to improve this problem.

[0139] Combining the time domain output results and the frequency domain broadband prediction results, the ICIM-CI (PSVIB) modeling method shows good accuracy and versatility in both the time domain and the frequency domain. There are slight differences between the model and the measured results under high frequency bands and high power injection conditions. This is because the enhancement of high-frequency noise levels requires a longer memory length to accurately characterize, and the unified memory length setting used to maintain modeling efficiency introduces modeling errors. At present, we are committed to exploring advanced methods to further improve the overall broadband modeling accuracy through program and model optimization.

[0140] In summary, the present invention improves the IB module of the ICIM-CI model and introduces a piecewise Volterra series improved black box modeling method based on particle swarm optimization, namely ICIM-CI (PSVIB). The method uses the piecewise Volterra series optimized by particle swarm optimization to improve the accuracy of the output waveform and the nonlinearity of the model, and the modeling efficiency under the comprehensive analysis of multiple sensitive criteria is also significantly improved. The improved model effectively solves the limitations of the traditional ICIM-CI model in describing the nonlinear response of MADIC under electromagnetic interference (EMI), and improves the model's capabilities in sensitivity quantitative analysis, output quantitative information and parameterized cascade simulation under the comprehensive analysis of multiple sensitivity criteria. Through experimental verification of analog switches and four-way differential drivers, the results show that the ICIM-CI (PSVIB) model has significant improvements in modeling accuracy, versatility and modeling efficiency, the broadband model accuracy is improved by 7.3dB, and the modeling efficiency is improved by 75%.

[0141] The above is only a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art shall not deviate from the spirit and scope of the present invention, and shall be within the scope of protection of the claims attached to the present invention.

Claims

1. The ICIM-CI modeling method based on particle swarm optimization is characterized by: The modeling method comprises: The IB module of the chip is established by introducing the segmented Volterra series, and the particle swarm optimization algorithm is used to reduce the segmented Volterra series, including: S1, input signal segmentation based on particle swarm optimization; S2, output signal recombines; S3. Segmented Volterra kernel function identification.

2. The ICIM-CI modeling method based on particle swarm optimization according to claim 1, characterized in that: Step S1 specifically includes: The input signal is decomposed according to the power size to determine the input power threshold. The threshold is expressed as follows: Γ={λ1,λ2,L,λ s },λ1<λ2<L<λ s ; where λ i ,i=1,2,L,s is the threshold of the input signal power amplitude, s is the number of thresholds set; The input signal x(t) passes through the set threshold and is decomposed into s+1 sub-signals according to the following formula: The decomposed input sub-signals are sampled in discrete time. The input sub-signals have the same time dimension. Assuming the number of sampling points is n, the decomposed input sub-signals are represented by a (s+1)×n matrix: In the decomposition process, the initial value of the particle in the optimization space is randomly set according to the objective function space, and the particle initial value is set according to the individual extreme value p. best and the group extreme value g best Update its own speed and position, and find the optimal value in the objective function space: Let p num is the total number of particles, D is the particle dimension, when the maximum number of iterations max is reached iter Stop iteration when v a (t) is the iteration speed of the ath particle at time t, x a (t) is the position of the ath particle in the objective function space at time t, pbest a (t) is the individual extreme value of the ath particle at time t; v a (t)=[v a1 (t),v a2 (t),…,v aD (t)] T ,x a (t)=[x a1 (t),x a2 (t),…,x aD (t)] T ; The update method of particles in the objective function space is as follows: v a+1 (t+1)=ωv a (t)+c1a1(pbest a (t)-x a (t))+c2a2(gbest a (t)-x a (t)); x a+1 (t+1)=x a (t)+v a+1 (t+1); Among them, ω is the inertia weight coefficient, which represents the influence of the previous iteration on the current iteration, c1 is the individual learning factor, c2 is the group learning factor, and a1 and a2 are random numbers.

3. The ICIM-CI modeling method based on particle swarm optimization according to claim 2 is characterized in that: Step S2 specifically includes: After completing the power threshold segmentation, the output is weighted combined to obtain a complete output signal; The output signal corresponding to each input sub-signal is shown as follows: In the formula, y i (·) is the output sub-signal corresponding to the i-th input sub-signal, K i is the order of the Volterra series corresponding to the i-th input sub-signal, M i Memory length of the input sub-signal for the i-th segment; The output signal is obtained by synthesizing each output sub-signal as follows:

4. The ICIM-CI modeling method based on particle swarm optimization according to claim 3 is characterized in that: Step S3 specifically includes: The piecewise kernel function uses the following formula to predict the output waveform: Where X pre_,i i=1,2L,s,+ is the input matrix obtained by multiplying the i-th input sub-signal and the memory delay, is a piecewise kernel function.

Images