Control method for dual-target time-interventional electrical stimulation devices for children with cognitive developmental disorders

Abstract

This application relates to the field of neurostimulation device control technology, specifically disclosing a control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders. The method involves configuring carrier groups for each of the two targets and constraining the fundamental frequency difference between the two groups to be greater than the low-pass response cutoff frequency of the child's neurons. Simultaneously, it utilizes high-frequency echo signals to calculate the dynamic complex impedance matrix of the scalp electrodes in real time, extracting the disturbance caused by the child's micro-movements. The spatial drift error of the focus is quantified through forward electromagnetic mapping, and then the phase compensation angle and amplitude correction coefficient of each channel are calculated using an inverse mapping model. The carrier groups are then subjected to amplitude-phase joint modulation to lock the interference focus at the preset target coordinates. In the disturbance extraction stage, an electrothermal self-excited causal decomposition mechanism based on Joule thermal accumulation is introduced to separate the deterministic thermal drift caused by the device current from exogenous random micro-movements, ensuring that compensation only responds to real physical off-target factors, thus maintaining the time-invariant target locking accuracy throughout the entire treatment process.

Description

Technical Field

[0001] This application relates to the field of neuro-electrical stimulation device control technology, and more specifically, to a control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders. Background Technology

[0002] Rehabilitation treatment for cognitive development disorders in children requires precise neuromodulation of specific functional areas deep in the brain. Time-interference electrical stimulation technology selectively activates neurons by injecting multiple high-frequency carrier currents into the scalp and utilizing the low-frequency difference envelope generated deep in the brain tissue. It has the unique advantages of being non-invasive and reaching deep brain regions.

[0003] However, when this technology is expanded from single-target to dual-target synchronous stimulation to meet the needs of multi-node coordinated regulation of children's cognitive networks, at least two sets of high-frequency carrier waves need to be introduced. The linear superposition of multiple electric fields within the whole-brain volume conductor will inevitably cause cross-interference in non-target brain regions, forming ectopic low-frequency envelopes, i.e., pseudo-foci, posing a risk of accidental electrical stimulation to children's brain tissue during critical periods of neural development. At the same time, children's poor compliance during treatment, their tendency to make micro-movements of the head and sweat, and the resulting drastic and asymmetrical dynamic changes in the complex impedance of the scalp electrode interface not only change the actual amplitude of the injected current, but also introduce unpredictable phase shifts, directly causing centimeter-level spatial drift of the deep interference electric field focus, causing it to deviate from the preset target. More seriously, during a 20-30 minute cognitive rehabilitation treatment for children, the high-frequency current output by the device generates a local thermal effect according to Joule's law when it flows through the electrode skin interface. Children have a thin dermis and a high density of sweat glands, so the interface temperature rise will deterministically activate sweat gland secretion, thereby causing a self-excited monotonic drift of impedance. This deterministic drift is fundamentally different from exogenous micro-motion perturbation in terms of physical attribution. If the two are confused, it will lead to a continuous amplification of the systematic deviation of subsequent spatial compensation as the treatment time progresses, causing the target locking accuracy to degrade significantly in the second half of the treatment.

[0004] Therefore, an optimized control scheme for a dual-target time-interventional electrical stimulation device for children with cognitive developmental disorders is desired. Summary of the Invention

[0005] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide a control method for a dual-target time-interventional electrical stimulation device for children with cognitive developmental disorders, comprising: Step 1: Based on the target stimulus parameter set containing the three-dimensional coordinates of the first cognitive target and the three-dimensional coordinates of the second cognitive target, perform orthogonal decoupled carrier configuration and reference current calculation on the two targets to obtain the orthogonal decoupled carrier group and the current injection reference vector. Step 2: Excite the scalp electrode array by using an orthogonal decoupled carrier group and simultaneously acquire a multi-channel echo sampling sequence. Then, perform real-time dynamic complex impedance calculation based on high-frequency echo on the multi-channel echo sampling sequence to obtain the electrode dynamic complex impedance matrix. Step 3: Based on the current injection reference vector, perform spatial drift prediction and evaluation of the electrode dynamic complex impedance matrix under the dynamic interference of children to obtain the complex impedance disturbance amount and the focus drift error vector. Step 4: Using the inverse mapping model of the cranial interference field, perform inverse phasor compensation calculation based on the interference physical field to obtain the phase compensation angle sequence and amplitude correction coefficient matrix for the focus drift error vector and complex impedance perturbation. Step 5: Based on the phase compensation angle sequence and amplitude correction coefficient matrix, the orthogonal decoupled carrier group is subjected to amplitude-phase joint modulation and synthesis to obtain the spatially locked stimulus output stream.

[0006] Compared with existing technologies, this application proposes a control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders. First, carrier groups are configured for each of the two targets, and the fundamental frequency difference between the two carrier groups is strictly constrained to be greater than the low-pass response cutoff frequency of the child's neurons. This physiologically shields the cross-interference envelope generated by non-target brain regions from the neural response blind zone, thereby eliminating accidental electrical stimulation from ectopic pseudo-foci. Based on this, the dynamic complex impedance matrix of the scalp electrode interface is calculated in real time using high-frequency carrier echo signals. The capacitive reactance and impedance disturbances caused by the child's micro-movements are extracted, and the spatial drift error of the interference focus is quantified through forward electromagnetic mapping. Then, the required phase compensation angle and amplitude correction coefficient for each channel are calculated in reverse using a brain interference field inverse mapping model. The carrier groups are then subjected to amplitude-phase joint modulation to precisely lock the low-frequency interference envelope of the dual targets to a preset three-dimensional spatial coordinate. Furthermore, an electrothermal self-excited coupling causal decomposition mechanism based on Joule thermal accumulation is introduced in the disturbance extraction stage. This mechanism accurately separates the deterministic thermal drift caused by the device's own current from the external random micro-motion disturbance, ensuring that the reverse compensation only responds to real physical off-target factors, thus maintaining the time-invariant characteristic of target locking accuracy throughout the entire rehabilitation process. Attached Figure Description

[0007] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0008] Figure 1 This is a flowchart of a control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders, according to an embodiment of this application. Figure 2 This is a schematic diagram of data flow for a control method of a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders according to an embodiment of this application; Figure 3 This is a flowchart illustrating a control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders, based on a target stimulation parameter set including the three-dimensional coordinates of a first cognitive target and the three-dimensional coordinates of a second cognitive target. The flowchart describes the process of configuring orthogonal decoupled carriers and calculating a reference current for the dual targets to obtain an orthogonal decoupled carrier group and a current injection reference vector. Figure 4 This is a flowchart illustrating a control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders according to an embodiment of this application. The method involves predicting and evaluating the spatial drift of the electrode dynamic complex impedance matrix under dynamic interference in children, based on the current injection reference vector, to obtain the complex impedance disturbance amount and the focus drift error vector. Figure 5 This is a flowchart illustrating the process of extracting the dynamic impedance reference difference and perturbation amount from the dynamic complex impedance matrix of the electrode and the pre-calibrated ideal contact impedance constant matrix of the child's skin-electrode, according to an embodiment of this application, to obtain the complex impedance perturbation amount. Figure 6 This is a flowchart illustrating a control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders, according to an embodiment of this application. The method involves using a brain interference field inverse mapping model to perform inverse phasor compensation calculations based on the interference physical field to obtain a phase compensation angle sequence and an amplitude correction coefficient matrix for the focus drift error vector and complex impedance disturbance. Detailed Implementation

[0009] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0010] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0011] While this application makes various references to certain modules of the systems according to embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The modules described are merely illustrative, and different aspects of the systems and methods may use different modules.

[0012] Flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0013] When time-interference electrical stimulation (TIS) technology is expanded from single-target to dual-target synchronous stimulation, the cross-superposition of multiple high-frequency carrier waves within the whole-brain volume conductor can generate ectopic low-frequency pseudo-foci in non-target brain regions. Simultaneously, poor patient compliance in children leads to drastic changes in the complex impedance at the scalp electrode interface, causing spatial drift of the deep interference focus, causing it to deviate from the preset target. Furthermore, the thermal effect of the device's own current induces impedance self-excitation drift, which accumulates continuously as treatment progresses, further deteriorating control accuracy. The combined effects of crosstalk, drift, and drift compensation degradation severely restrict the safety and effectiveness of dual-target TIS technology in the rehabilitation of children with cognitive impairment. Therefore, this application proposes a control method for a dual-target TIS device used in the rehabilitation of children with cognitive developmental disorders. This method eliminates spurious focal points by constraining the difference between the fundamental frequencies of the two target carriers to be greater than the low-pass response cutoff frequency of the child's neurons, causing the cross-interference envelope to fall into the neural response blind zone and be physiologically shielded. It then uses high-frequency echo signals to calculate the dynamic complex impedance of the electrodes in real time and extracts the perturbation quantity. After quantifying the focal drift error through forward electromagnetic mapping, the inverse mapping model calculates the phase compensation angle and amplitude correction coefficient of each channel, and performs amplitude-phase joint modulation on the carrier group to lock the interference focal point. In the perturbation quantity extraction stage, an electrothermal self-excited causal decomposition mechanism based on Joule thermal accumulation is introduced to precisely separate deterministic thermal drift from exogenous random micro-motions, ensuring that inverse compensation only responds to real physical off-target factors, thus maintaining the time-invariant characteristic of target locking accuracy throughout the entire rehabilitation process.

[0014] Figure 1 This is a flowchart of a control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders, according to an embodiment of this application. Figure 2 This is a schematic diagram of data flow for a control method of a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders, according to an embodiment of this application. Figure 1 and Figure 2As shown, a control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders according to an embodiment of this application includes: S1, based on a target stimulation parameter set including the three-dimensional coordinates of a first cognitive target and the three-dimensional coordinates of a second cognitive target, performing orthogonal decoupling carrier configuration and reference current calculation on the dual targets to obtain an orthogonal decoupling carrier group and a current injection reference vector; S2, applying excitation to the scalp electrode array through the orthogonal decoupling carrier group and simultaneously acquiring a multi-channel echo sampling sequence, and performing real-time dynamic complex impedance calculation based on high-frequency echoes on the multi-channel echo sampling sequence to obtain... S3, Based on the current injection reference vector, the spatial drift prediction and evaluation of the electrode dynamic complex impedance matrix under dynamic interference in children is performed to obtain the complex impedance disturbance amount and the focus drift error vector; S4, Through the inverse mapping model of the cranial interference field, the focus drift error vector and the complex impedance disturbance amount are solved by inverse phasor compensation based on the interference physical field to obtain the phase compensation angle sequence and amplitude correction coefficient matrix; S5, Based on the phase compensation angle sequence and amplitude correction coefficient matrix, the orthogonal decoupled carrier group is subjected to amplitude and phase joint modulation and synthesis to obtain the spatially locked stimulus output stream.

[0015] Specifically, in step S1, based on the target stimulation parameter set including the three-dimensional coordinates of the first and second cognitive target points, orthogonal decoupling carrier configuration and reference current calculation are performed on the two target points to obtain the orthogonal decoupling carrier group and the current injection reference vector. The target stimulation parameter set also includes the cutoff frequency threshold for the child's neural response and the target frequency of the interference envelope. It should be noted that, given that dual-target time-interference electrical stimulation requires the simultaneous injection of multiple high-frequency carriers, the linear superposition of multiple electric fields within the whole-brain volume conductor inevitably produces a cross-interference low-frequency envelope in non-target brain regions, forming an ectopic pseudo-focus. Since the child's brain is in a critical developmental period, accidental low-frequency electrical stimulation of non-target brain regions can easily trigger epilepsy or hinder normal neural development. Therefore, the technical solution of this application first performs orthogonal decoupling carrier configuration and reference current calculation on the two target points based on the target stimulation parameter set, constraining the fundamental frequency difference between the two carrier groups to be greater than the low-pass filter response cutoff frequency of the child's neuronal cell membrane, making the cross-interference difference frequency far exceed the frequency range that the neuron can respond to, and simultaneously calculating the optimal current output weight of each electrode channel under ideal conditions. Through the above processing, non-target brain regions can achieve natural physiological immune shielding against the cross-interference envelope from the perspective of frequency domain physical mechanisms, completely eliminating the risk of accidental electrical stimulation from ectopic pseudofoci, and providing a zero-disturbance benchmark reference for subsequent dynamic compensation control.

[0016] Figure 3This is a flowchart illustrating a control method for a dual-target temporal interference electrical stimulation device for children with cognitive developmental disorders, according to an embodiment of this application. The method involves configuring orthogonal decoupled carriers and calculating a reference current for the dual targets based on a target stimulation parameter set containing the three-dimensional coordinates of a first cognitive target and a second cognitive target, to obtain an orthogonal decoupled carrier group and a current injection reference vector. (See flowchart for details.) Figure 3 As shown, step S1 includes: S11, encapsulating the three-dimensional spatial coordinates of the first cognitive target and the second cognitive target into a dual-target spatial coordinate matrix; S12, based on the child's neural response cutoff frequency threshold, assigning a first carrier fundamental frequency to the first cognitive target and a second carrier fundamental frequency to the second cognitive target, and combining the frequency with the interference envelope target frequency to obtain an interference frequency domain configuration matrix; S13, based on the frequency parameters in the interference frequency domain configuration matrix, performing orthogonal decoupling carrier synthesis on each channel to obtain an orthogonal decoupling carrier group; S14, inputting the dual-target spatial coordinate matrix into the child's cranial lead field forward mapping model to perform optimal current weight calculation to obtain the current injection reference vector.

[0017] In step S11, the three-dimensional spatial coordinates of the first and second cognitive target points are encapsulated into a dual-target spatial coordinate matrix. It should be noted that, since dual-target temporal interference electrical stimulation requires simultaneous targeted modulation of two spatially independent cognitive brain regions, the three-dimensional localization information of each target point will be used throughout the entire calculation process of subsequent carrier frequency allocation and current weighting. If these are transmitted separately as discrete parameters, data index misalignment is likely to occur. Therefore, the technical solution of this application first encapsulates the three-dimensional spatial coordinates of the first and second cognitive target points into a dual-target spatial coordinate matrix. Through the above processing, the dispersed target spatial information can be unified into a structured matrix object, providing a standardized input interface for the subsequent batch forward mapping calculation of the lead field model.

[0018] More specifically, in a specific example of this application, the spatial coordinates of the first cognitive target are first extracted from the three-dimensional reconstruction results of a child's cranial MRI image. These coordinates are represented in a three-dimensional Cartesian coordinate system as a coordinate vector containing three components: anteroposterior, left-right, and vertical axes. For example, this could be the target coordinates for the dorsolateral prefrontal cortex. The spatial coordinates of the second cognitive target are extracted in the same manner, for example, the target coordinates for the hippocampus region. Subsequently, the three-dimensional coordinate vectors of the two targets are arranged in rows and encapsulated into a two-row, three-column dual-target spatial coordinate matrix. The first row of the matrix corresponds to the three-axis coordinate components of the first cognitive target, the second row corresponds to the three-axis coordinate components of the second cognitive target, and each column corresponds to the spatial position values ​​in the anteroposterior, left-right, and vertical axes, respectively.

[0019] In step S12, based on the cutoff frequency threshold of the child's neural response, a first carrier fundamental frequency is assigned to the first cognitive target, and a second carrier fundamental frequency is assigned to the second cognitive target. These frequencies are then combined with the interference envelope target frequency to obtain an interference frequency domain configuration matrix. It should be noted that when multiple high-frequency carriers injected simultaneously into both targets overlap in the brain, a difference frequency envelope is generated. If this difference frequency falls into a low-frequency band that the child's neurons can respond to, it will form an ectopic pseudo-foci in a non-target brain region. Therefore, the technical solution of this application further assigns a first carrier fundamental frequency to the first cognitive target and a second carrier fundamental frequency to the second cognitive target based on the cutoff frequency threshold of the child's neural response, and combines this with the interference envelope target frequency to obtain an interference frequency domain configuration matrix. Through the above processing, the cross-interference difference frequency of the two carriers can far exceed the upper limit of the polarization response of the child's neuronal cell membrane, thereby achieving orthogonal isolation of the electric fields of the two targets at the frequency domain level.

[0020] More specifically, in a particular example of this application, a first carrier fundamental frequency is first assigned to the first cognitive target. The selection of this fundamental frequency must satisfy the basic constraint of being higher than the threshold of the child's neural response cutoff frequency to ensure that the high-frequency carrier itself does not directly activate neurons; for example, the first carrier fundamental frequency is set to 2000Hz. Subsequently, a second carrier fundamental frequency is assigned to the second cognitive target. The determination of this second carrier fundamental frequency follows the orthogonal decoupling constraint condition, that is, the second carrier fundamental frequency is equal to the sum of the first carrier fundamental frequency, the child's neural response cutoff frequency threshold, and the safety margin.

[0021] in, The second carrier base frequency, The first carrier base frequency, The cutoff frequency threshold for neural responses in children is the highest critical value at which the cell membrane of a child's neuron can generate an action potential response. The safety margin refers to the additional frequency isolation protection bandwidth introduced to prevent frequency drift in actual operating conditions. According to the above constraints, if the first carrier fundamental frequency is 2000Hz, the cutoff frequency threshold for the child's neural response is 1000Hz, and the safety margin is 200Hz, then the second carrier fundamental frequency is 3200Hz. At this point, the cross-interference difference frequency between the two carriers is 1200Hz, far exceeding the upper limit of the low-frequency response that the child's neurons can generate. Non-target brain regions cannot produce an electrophysiological response to this cross-envelope. After completing the fundamental frequency allocation for the dual targets, further frequency combination is performed by combining the target frequencies of the interference envelope. That is, a low-frequency modulation frequency for cognitive network neural entrainment is superimposed on the carrier fundamental frequencies of each target point. For example, if the target frequency of the interference envelope is set to 40Hz for gamma-band cognitive modulation, then the carrier pair frequencies corresponding to the first cognitive target are 2000Hz and 2040Hz, and the carrier pair frequencies corresponding to the second cognitive target are 3200Hz and 3240Hz. All the frequency parameters mentioned above are arranged according to the two-dimensional index relationship between the target point and the carrier pair, and a two-row, two-column interferometric frequency domain configuration matrix is ​​constructed. The first row of the matrix contains the two carrier frequencies of the first cognitive target point, the second row contains the two carrier frequencies of the second cognitive target point, and each column corresponds to the fundamental frequency carrier and the modulation carrier after superimposing the envelope frequency, respectively.

[0022] In step S13, orthogonal decoupled carrier synthesis is performed on each channel based on the frequency parameters in the interferometric frequency domain configuration matrix to obtain an orthogonal decoupled carrier group. It should be noted that since the interferometric frequency domain configuration matrix only contains discrete frequency numerical parameters and has not yet been converted into a continuous time-domain electrical signal that can be applied to the physical electrodes, subsequent impedance echo detection and electric field interference both require actual simulated waveforms as the carrier. Based on this, the technical solution of this application further performs orthogonal decoupled carrier synthesis on each channel based on the frequency parameters in the interferometric frequency domain configuration matrix to obtain an orthogonal decoupled carrier group. Through the above processing, the frequency domain configuration parameters can be converted into a multi-channel continuous time-domain excitation signal with orthogonal isolation characteristics, providing a directly applicable high-frequency carrier waveform for the physical electrode array.

[0023] More specifically, in a concrete example of this application, all frequency elements in the interferometric frequency domain configuration matrix are first read. For the first cognitive target, two corresponding carrier frequencies, 2000Hz and 2040Hz, are extracted; for the second cognitive target, two corresponding carrier frequencies, 3200Hz and 3240Hz, are extracted, resulting in four independent frequency parameters. Subsequently, using the initial current amplitude set by the hardware constant current source as a unified amplitude reference and zero phase as the initial phase reference, continuous time-domain signal synthesis is performed on the four frequency parameters according to the standard time-domain expression of a sine function. Each parameter generates an independent sinusoidal carrier signal with a defined frequency, amplitude, and initial phase. Since the frequencies of the four carriers are 2000Hz, 2040Hz, 3200Hz, and 3240Hz respectively, the signals do not overlap in the spectrum and satisfy the orthogonal decoupling constraint. That is, the frequency difference between the two carriers within the same target point is 40Hz, which is used to generate the low-frequency interference envelope required for cognitive modulation deep within the target point, while the frequency difference between the carriers across target points is greater than 1000Hz, which exceeds the response range of neurons. The above four independent sinusoidal carrier signals are encapsulated into a multi-channel array according to the physical channel mapping relationship of the scalp electrode array to obtain an orthogonal decoupling carrier group. Each channel in this carrier group carries a high-frequency sinusoidal signal with a specific frequency, and the channels maintain a strict orthogonal isolation relationship in the frequency domain.

[0024] In step S14, the dual-target spatial coordinate matrix is ​​input into the forward mapping model of the pediatric cranial lead field to calculate the optimal current weights and obtain the current injection reference vector. It should be noted that due to the conductivity differences between the scalp electrode array and the deep brain target points (such as the skull and cerebrospinal fluid), the contribution of the current injected into each electrode channel to the target point varies after attenuation along different paths. Therefore, it is necessary to determine the optimal current distribution ratio for each channel to focus the interference electric field at the target point. Based on this, the technical solution of this application further inputs the dual-target spatial coordinate matrix into the forward mapping model of the pediatric cranial lead field to calculate the optimal current weights and obtain the current injection reference vector. Through the above processing, the optimal current amplitude and initial phase of each electrode channel can be determined under ideal impedance conditions, providing a reference under zero-disturbance conditions for subsequent dynamic compensation.

[0025] More specifically, in a concrete example of this application, the spatial coordinate matrix of the two target points is first imported into a pre-constructed forward mapping model of the pediatric cranial lead field. This model establishes the electric field transmission relationship from the scalp electrode position to any spatial coordinate point in the brain based on the conductivity parameters of the layered tissues of the child's skull. Its output is a lead field matrix, where each column corresponds to the unit electric field contribution vector generated by an electrode channel at the target point. Subsequently, with the maximization of the three-dimensional spatial electric field intensity at the center of the two target points as the objective function, the optimal current weight solution of the lead field matrix is ​​performed using the Tikhonov regularized inverse solution algorithm. The calculation process is expressed as follows:

[0026] in, The obtained current injection reference vector contains the current amplitude and initial phase information that each electrode channel should output under ideal conditions. The intracranial lead field matrix for children represents the positive transmission relationship between the injected current through the scalp and the electric field at the target point in the brain. This is the transpose of the field matrix of that lead. The regularization parameter is used to constrain ill-conditioned matrix divergence during the solution process to ensure numerical uniqueness. It is the identity matrix. This represents the desired electric field intensity vector of the target brain region corresponding to the spatial coordinate matrix of the dual target points. Through the above inverse solution, each element in the current injection reference vector corresponds to the optimal complex value of the current for a physical electrode channel. Its magnitude represents the absolute current intensity that the channel should output, and its phase angle represents the initial phase offset of the carrier signal of the channel. When each channel outputs in coordination according to this reference vector, the interference electric field can simultaneously reach the preset low-frequency envelope intensity peak at both the prefrontal cortex and the hippocampus.

[0027] Specifically, in step S2, an excitation is applied to the scalp electrode array using an orthogonal decoupled carrier group, and a multi-channel echo sampling sequence is simultaneously acquired. The dynamic complex impedance matrix of the electrode is then obtained through real-time dynamic complex impedance calculation based on high-frequency echoes. It should be noted that children are prone to head micro-movements and sweating during treatment, causing the complex impedance of the scalp electrode interface to be in a state of continuous dynamic change. Without real-time sensing, the actual attenuation and phase shift of the current transmission path cannot be quantified, and subsequent focus drift prediction will lack a physical basis. Therefore, the technical solution of this application further applies an excitation to the scalp electrode array using an orthogonal decoupled carrier group, simultaneously acquires a multi-channel echo sampling sequence, and performs real-time dynamic complex impedance calculation based on high-frequency echoes to obtain the electrode dynamic complex impedance matrix. Through the above processing, the transient impedance and capacitive reactance of each electrode channel can be obtained in real-time using the carrier itself as a probe signal without interrupting the therapeutic stimulation output.

[0028] More specifically, in a concrete example of this application, the excitation and echo signals of the orthogonal decoupled carrier group and the multi-channel echo sampling sequence are first synchronized and windowed in the time domain to obtain a windowed synchronized time series pair. While the orthogonal decoupled carrier group is applied to the scalp electrode array via a constant current source, the analog-to-digital converters of each channel synchronously acquire the echo voltage signal flowing through the electrode interface to form a multi-channel echo sampling sequence. Due to differences in hardware trace lengths and the sampling delay of the analog-to-digital converters, there is an inherent time deviation between the excitation signal and the echo signal. A sliding cross-correlation operation is needed to find the correlation peak between the two to determine the system-level time delay, and based on this, the echo sequence is shifted along the time axis to achieve nanosecond-level absolute clock alignment. After time-domain synchronization, a Hamming window is applied to the aligned sequence to suppress spectral leakage effects in subsequent spectral analysis. The synchronized and windowed excitation reference sequence and echo voltage sequence are then paired and encapsulated into a windowed synchronized time series pair.

[0029] Subsequently, frequency domain phasor separation and feature extraction were performed on the windowed synchronization time series pair to obtain the amplitude and phase attenuation feature vector. Fast Fourier Transform (FFT) was performed on the excitation reference sequence and echo voltage sequence in the windowed synchronization time series pair to convert the time-domain waveforms to frequency-domain representations. At each carrier frequency point set by the interferometric frequency domain configuration matrix, the reference phasor of the excitation current signal and the response phasor of the echo voltage signal were extracted. The reference phasor contains the amplitude and initial phase information of the excitation end, while the response phasor contains the amplitude and phase information transmitted through the electrode interface. Complex division was performed on the response phasor and reference phasor at the same frequency point to separate the amplitude attenuation ratio and absolute phase offset angle of each channel caused by scalp dynamic impedance. The attenuation ratios and offset angles of all channels were concatenated in channel index order to form the amplitude and phase attenuation feature vector.

[0030] Finally, complex impedance scalar calculations are performed on the amplitude and phase attenuation eigenvector, and a diagonal matrix is ​​constructed according to the spatial topology of the scalp electrode array to obtain the electrode dynamic complex impedance matrix. For the amplitude attenuation ratio and phase shift angle of each channel in the amplitude and phase attenuation eigenvector, based on the polar to rectangular coordinate transformation relationship of Ohm's law for AC circuits, the polar coordinate form of the impedance magnitude and phase angle is mapped to a complex impedance scalar composed of a real resistance component and an imaginary capacitive reactance component. The real part reflects the pure resistance characteristics of the electrode-skin interface, and the imaginary part reflects the interface capacitance effect caused by sweat penetration and changes in skin moisture content. The complex impedance scalars calculated independently for each channel are filled into the main diagonal elements of the diagonal matrix according to the spatial topological arrangement of the physical electrodes on the scalp. The off-diagonal elements are set to zero to characterize the independence of the impedance of each channel. Finally, the electrode dynamic complex impedance matrix is ​​constructed. The dimension of this matrix is ​​equal to the total number of channels in the scalp electrode array, and each complex element on its main diagonal completely characterizes the transient impedance state of the corresponding electrode at the current moment.

[0031] For example, during a treatment, the excitation current amplitude of channel 3 (located in the left forehead region) is 1.5mA and the initial phase is 0°. The echo voltage response phasor amplitude extracted at 2000Hz frequency point by fast Fourier transform is 0.45V and the phase is -12.3°. Then the amplitude attenuation ratio of this channel is 300Ω (impedance modulus) and the phase offset angle is -12.3° (negative value indicates capacitive hysteresis). Converting this polar coordinate impedance to rectangular coordinate form, the real part resistance component is about 293Ω and the imaginary part capacitive reactance component is about -64Ω. These are filled into the main diagonal position of the 3rd row and 3rd column of the electrode dynamic complex impedance matrix.

[0032] Specifically, in step S3, based on the current injection reference vector, the spatial drift prediction and evaluation of the electrode dynamic complex impedance matrix under dynamic interference in children is performed to obtain the complex impedance perturbation amount and the focus drift error vector. It should be noted that, given that the electrode dynamic complex impedance matrix only reflects the transient impedance state of each channel at the current moment, and has not yet revealed the actual impact of this impedance change on the spatial distribution of the deep interference electric field, and that impedance mutations caused by head micro-movements and sweat secretion during treatment in children can alter the actual amplitude of the injected current in each channel and introduce unexpected phase shifts, directly causing the interference focus to deviate from the preset target coordinates, it is necessary to quantify the electrical changes at the impedance level into the focus drift distance and direction at the spatial level to provide a clear correction target for subsequent inverse compensation. Based on this, the technical solution of this application further performs spatial drift prediction and evaluation of the electrode dynamic complex impedance matrix under dynamic interference in children based on the current injection reference vector to obtain the complex impedance perturbation amount and the focus drift error vector. Through the above processing, the electrical perturbation information on the scalp surface can be mapped to the focus shift amount in the three-dimensional space within the brain, providing accurate spatial error input for inverse phasor compensation calculation.

[0033] Figure 4 This is a flowchart illustrating a control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders, based on a current injection reference vector. The method involves predicting and evaluating the spatial drift of the electrode dynamic complex impedance matrix under dynamic disturbances in children to obtain the complex impedance disturbance amount and the focus drift error vector. Figure 4As shown, step S3 includes: S31, performing dynamic impedance benchmark difference and perturbation extraction on the dynamic complex impedance matrix of the electrode and the pre-calibrated ideal contact impedance constant matrix of the child's skin-electrode to obtain the complex impedance perturbation; S32, based on the perturbation coupling coefficient matrix, performing boundary condition correction on the complex impedance perturbation and the current injection benchmark vector to obtain the corrected actual current vector, and inputting the corrected actual current vector into the three-dimensional cranial lead field model of the child for positive electromagnetic radiation mapping to obtain the actual interference electric field distribution matrix; S33, searching for the node coordinates of the maximum low-frequency interference envelope intensity in the three-dimensional spatial grid of the actual interference electric field distribution matrix, and solving for the spatial drift error by comparing them with the coordinates of the theoretical cognitive target center to obtain the focus drift error vector.

[0034] In step S31, the dynamic impedance benchmark difference and perturbation amount are extracted from the electrode dynamic complex impedance matrix and the pre-calibrated ideal contact impedance constant matrix of the child's skin-electrode to obtain the complex impedance perturbation amount. It should be noted that, given that the electrode dynamic complex impedance matrix contains a mixture of static background impedance and dynamic perturbation impedance information, and as mentioned earlier, subsequent spatial drift prediction only needs to respond to the purely exogenous perturbation component, the dynamic change part needs to be separated from the overall measured value. Based on this, the technical solution of this application further extracts the dynamic impedance benchmark difference and perturbation amount from the electrode dynamic complex impedance matrix and the pre-calibrated ideal contact impedance constant matrix of the child's skin-electrode to obtain the complex impedance perturbation amount. Through the above processing, the impedance abrupt changes of each channel deviating from the ideal baseline can be accurately separated, providing a quantitative input that purely reflects external physical interference for subsequent forward electromagnetic mapping.

[0035] Figure 5 This is a flowchart illustrating the process of extracting the complex impedance perturbation amount by performing dynamic impedance reference difference and perturbation quantity extraction between the dynamic complex impedance matrix of the electrodes and the pre-calibrated ideal contact impedance constant matrix of the child's skin-electrode, according to an embodiment of this application, in a control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders. Figure 5 As shown, step S31 includes: S311, performing channel Joule thermal energy accumulation quantization and self-excited drift prediction on the current injection reference vector to obtain the self-excited thermal drift impedance matrix; S312, performing three-source causal decomposition and external disturbance isolation on the electrode dynamic complex impedance matrix, the pre-calibrated ideal contact impedance constant matrix of the child's skin-electrode, and the self-excited thermal drift impedance matrix to obtain the external disturbance candidate matrix; S313, based on the upper limit of the maximum allowable impedance change rate under the biomechanical constraints of the child's head movement, performing time-domain causal clamping verification and non-physical pseudo-component suppression on the channel disturbance values ​​in the external disturbance candidate matrix to obtain the complex impedance disturbance amount.

[0036] In step S311, channel Joule thermal energy accumulation quantization and self-excited drift prediction are performed on the current injection reference vector to obtain the self-excited thermal drift impedance matrix. It should be noted that, given the localized thermal effect generated by the continuously output high-frequency carrier current of the device when flowing through the electrode-skin interface according to Joule's law, and considering that children's dermal layer is thinner than adults' and their eccrine sweat gland density is much higher, the interface temperature rise will deterministically activate sweat gland secretion and change the interface ionic conductivity, causing a self-excited monotonic drift in impedance. This deterministic drift is fundamentally different in physical attribution from exogenous micro-motion disturbances. Conflating the two will lead to a systematic overestimation of the complex impedance disturbance, causing overcompensation or directional deviation in subsequent spatial compensation. Therefore, the technical solution of this application further performs channel Joule thermal energy accumulation quantization and self-excited drift prediction on the current injection reference vector to obtain the self-excited thermal drift impedance matrix. Through the above processing, the deterministic thermal drift component caused by the device's own current can be estimated a priori using a causal physical model, providing a subtractable deterministic baseline for subsequent mixed signal decomposition.

[0037] More specifically, in a particular example of this application, firstly, based on the current amplitude parameters of each channel in the current injection reference vector, and combined with the stimulation accumulation time counter, the scalar amount of heat energy accumulated and deposited on the child's skin electrode interface at the current moment is calculated according to Joule's law. Since the high-frequency carrier current continuously flows through the interface contact resistance, electrical energy is inevitably converted into heat energy and deposited in the superficial dermal tissue of the child. This heat load is the direct physical driver of subsequent sweat gland self-excitation, leading to deterministic impedance drift; therefore, it must be quantitatively calculated first. The accumulated Joule heat energy of each channel is calculated as follows:

[0038] in, This represents the cumulative joule heat energy of the k-th electrode channel at the current moment, expressed in joules. Let be the complex component of the current amplitude of the k-th channel in the current injection reference vector. Modulo operations on complex numbers are used to extract the effective value of the current. The initial contact resistance of the skin interface of the k-th channel electrode is obtained from the calibration during initialization. This is the cumulative elapsed time from the start of this stimulation treatment to the current moment, in seconds.

[0039] Subsequently, the cumulative thermal energy scalar of each channel is input into a pre-established nonlinear transfer function model of the electrothermal impedance of children's skin for self-excited drift prediction. The thermoelectric coupling response of children's skin exhibits nonlinear characteristics. The resistive component decreases exponentially due to the increase in skin conductivity caused by temperature rise, while the capacitive component increases logarithmically due to the expansion of the interface equivalent capacitance caused by continuous sweat penetration. This nonlinear transfer function model, based on pre-calibrated offline parameters of children's skin thermophysiological characteristics including dermal thickness, sweat gland density, and thermal diffusivity, can deterministically map the thermal energy input into a complex impedance drift prediction value. This allows for prior estimation of the self-excited drift component, which cannot be directly measured, using a causal physical model, providing a subtractable deterministic baseline for subsequent mixed-signal decomposition. The calculation of this transfer function is expressed as follows:

[0040] in, This is the predicted value of the complex impedance drift caused by the self-excited heating effect in the k-th channel. The resistive drift saturation coefficient characterizes the maximum decrease in skin resistance due to temperature rise, and is measured in ohms. The capacitive drift growth factor is a rate factor characterizing the increase in interfacial capacitance due to sweat secretion, expressed in ohms. It is a natural constant. The skin thermal time constant for children is a characteristic energy scale representing the steady-state thermal diffusion in the dermis, measured in joules. The energy threshold for sweat gland activation in children is the critical value of cumulative heat energy required for eccrine sweat glands to start secreting from rest, measured in joules. The imaginary unit, The natural logarithm function is used to describe the logarithmic saturation characteristics of sweat secretion. The exponential saturation characteristic of the real part of the formula reflects the physical constraint that skin resistance cannot decrease indefinitely, while the logarithmic growth characteristic of the imaginary part reflects the mass transfer law that the sweat diffusion rate slows down as the concentration gradient decreases. The self-excited thermal drift impedance matrix is ​​obtained by arranging the predicted self-excited drift complex impedance values ​​of each channel in a diagonal matrix format according to spatial topology.

[0041] For example, in a pediatric cognitive rehabilitation therapy targeting both the prefrontal cortex and hippocampus, at the 15th minute of treatment, the electrode channel near the hairline on the forehead had accumulated Joule heat energy exceeding the sweat gland activation energy threshold due to the high density of sweat glands and large current amplitude in that area. The transfer function model predicted that the resistive component of this channel had decreased to nearly 60% of its saturation value, while the capacitive component showed a continuous logarithmic growth trend. In contrast, the electrode channel located in the temporal region had accumulated heat energy before reaching the sweat gland activation threshold due to the smaller current amplitude and better heat dissipation conditions of the skin in that area. The predicted self-excited drift was close to zero. The differentiated self-excited drift prediction values ​​for each channel were filled into the corresponding diagonal positions of the self-excited thermal drift impedance matrix.

[0042] In step S312, the electrode dynamic complex impedance matrix, the pre-calibrated ideal contact impedance constant matrix of the child's skin-electrode, and the self-excited thermal drift impedance matrix are subjected to three-source causal decomposition and external disturbance isolation to obtain an external disturbance candidate matrix. It should be noted that, given that the measured electrode dynamic complex impedance matrix contains a mixture of components from three different physical sources—static ideal background impedance, device self-excited thermal drift impedance, and external physical disturbance impedance—if only a simple binary difference operation is performed and the difference is treated as an external disturbance, the self-excited thermal drift will be misjudged as an external micro-motion disturbance, resulting in a systematic overestimation of the complex impedance disturbance. Subsequent reverse compensation will inevitably lead to overcompensation or directional deviation. Therefore, the technical solution of this application further performs three-source causal decomposition and external disturbance isolation on the electrode dynamic complex impedance matrix, the pre-calibrated ideal contact impedance constant matrix of the child's skin-electrode, and the self-excited thermal drift impedance matrix to obtain an external disturbance candidate matrix. Through the above processing, impedance abrupt changes caused purely by the child's external physical micro-movements can be accurately isolated from the mixed observation signal, so that subsequent spatial compensation responds only to real physical off-target factors.

[0043] More specifically, in a concrete example of this application, a three-source causal decomposition model is established, treating the measured electrode dynamic complex impedance matrix as a linear superposition of three independent physical sources: static ideal background impedance, device self-excited thermal drift impedance, and external physical disturbance impedance. The measured matrix is ​​then subtracted sequentially by a pre-calibrated ideal skin-electrode contact impedance constant matrix to remove the static background, and then by subtracting the self-excited thermal drift impedance matrix predicted in the previous sub-step to remove the deterministic self-excited component. This accurately isolates impedance abrupt changes purely caused by external physical micro-movements of the child from the mixed observation signal, eliminating the systematic bias of misjudging self-excited drift as external disturbance, ensuring that subsequent spatial compensation responds only to real physical off-target factors. The process of three-source causal decomposition is as follows:

[0044] in, The candidate matrix for exogenous perturbations includes only pure external impedance abrupt changes caused by children's physical micro-movements. This represents the dynamic complex impedance matrix of the electrodes, i.e., the measured value of the current transient impedance across the entire scalp channel, calculated in real time. For the pre-calibrated ideal contact impedance constant matrix of the child's skin-electrode, This is the self-excited thermal drift impedance matrix. The formula sequentially separates the static background impedance baseline calibrated before the start of treatment and the deterministic self-excited drift component caused by the thermal effect of the device's own current from the real-time acquired full-channel transient impedance measurement values. The final residual difference is the impedance mutation amount caused purely by external random physical events such as micro-movements of the child's head.

[0045] For example, at the 15th minute of treatment, the measured complex impedance of an electrode channel in the forehead hairline area decreased by 12 ohms compared to the initial calibration value. Of this, self-excited thermal drift, predicted by the transfer function model, contributed about 8 ohms to the decrease in resistance. After three-source causal decomposition, the candidate value of the exogenous disturbance of this channel was only 4 ohms, instead of the 12 ohms obtained by simple difference. This avoids the problem of misattributing the thermal drift component to the child's micro-movement, which could lead to three times overcompensation.

[0046] In step S313, based on the upper limit of the maximum permissible impedance change rate under the biomechanical constraints of children's head movements, time-domain causality clamping verification and non-physical pseudo-component suppression are performed on the perturbation values ​​of each channel in the exogenous perturbation candidate matrix to obtain the complex impedance perturbation amount. It should be noted that, given that the exogenous perturbation candidate matrix after three-source causal decomposition may still contain non-physical pseudo-components introduced by transfer function model prediction errors or sensor noise, and considering that children's head movements are physically constrained by the mechanical characteristics of neck muscles and the moment of inertia of the skull, the resulting electrode displacement rate cannot exceed the biomechanical limit, it is necessary to perform physical rationality verification on the candidate perturbation values ​​accordingly. Based on this, the technical solution of this application further performs time-domain causality clamping verification and non-physical pseudo-component suppression on the perturbation values ​​of each channel in the exogenous perturbation candidate matrix based on the upper limit of the maximum permissible impedance change rate under the biomechanical constraints of children's head movements to obtain the complex impedance perturbation amount. Through the above processing, non-physical pseudo-perturbation components caused by model prediction errors or residual sensor noise can be eliminated, ensuring that the final output perturbation amount has a true physical correspondence.

[0047] More specifically, in a concrete example of this application, a time-domain causality check is performed on the decomposed candidate matrix of exogenous disturbances. First, the rate of change of each channel's candidate disturbance value relative to the previous control cycle is calculated. The absolute value of this rate of change is then compared channel by channel with the pre-calibrated upper limit of the maximum permissible rate of impedance change under the biomechanical constraints of a child's head movement. For channels that pass the check, their disturbance values ​​are retained; for channels that fail, the disturbance values ​​are clamped to the biomechanically permissible maximum rate of change boundary value. This eliminates non-physical spurious disturbance components caused by model prediction errors or residual sensor noise, ensuring that the final output disturbance has a true physical correspondence. The time-domain causality clamping check is represented as follows:

[0048] in, This represents the final confirmed value for the k-th channel in the complex impedance disturbance. Let be the candidate perturbation value for the k-th channel in the candidate matrix of exogenous perturbations. Let be the time rate of change of the candidate value of the exogenous disturbance in the k-th channel. This represents the upper limit of the maximum permissible rate of impedance change under biomechanical constraints on children's head movements, measured in ohms per second. It is calibrated jointly by the maximum angular acceleration of the child's neck muscles and the spatial sensitivity of the electrodes. This is a sign function used to preserve the direction of change. This represents the time step of the system control cycle. The final matrix after the above clamping verification is the complex impedance perturbation quantity. This formula uses the head movement limit rate determined by the mechanical characteristics of children's neck muscles as the physical criterion to make a reasonable judgment on the time change rate of exogenous perturbation candidate values. If the change rate does not exceed the biomechanical limit, the perturbation is considered to be caused by real children's micro-movements and is retained. If the change rate exceeds the limit, it is judged as model residual error or noise pseudo-component and forcibly clamped to the maximum boundary value allowed by biomechanics, thereby ensuring that each component in the final output complex impedance perturbation quantity has a traceable real physical cause.

[0049] For example, during a certain control cycle in the treatment process, the candidate value of an exogenous perturbation in a certain electrode channel in the temporal region suddenly showed a rate of change of 50 ohms per second relative to the previous cycle. However, the upper limit of the maximum allowable impedance change rate, which is jointly calibrated based on the maximum angular acceleration of the child's neck muscles and the spatial sensitivity of the electrode position, is 20 ohms per second. The rate of change of this channel exceeds the biomechanical limit and is determined to be a non-physical pseudo-component introduced by model residual error or noise. It is clamped to the boundary value corresponding to 20 ohms per second multiplied by the control cycle time step and the original direction of change is retained. Meanwhile, the rate of change of another channel in the forehead region is 15 ohms per second, which does not exceed the upper limit constraint, so its candidate perturbation value is directly retained as the final confirmed value.

[0050] In step S32, based on the perturbation coupling coefficient matrix, the boundary conditions of the complex impedance perturbation and the current injection reference vector are corrected to obtain the corrected actual current vector. This corrected actual current vector is then input into a three-dimensional cranial lead field model of a child for forward electromagnetic radiation mapping to obtain the actual interference electric field distribution matrix. It should be noted that since the complex impedance perturbation reflects the electrical changes at the scalp electrode interface and has not yet been converted into actual current deviations flowing into the brain tissue, and the influence of impedance changes in each channel on the injected current depends on the coupling relationship of that channel in the overall circuit network, it is necessary to map the surface impedance perturbation into actual current deviations and deduce the spatial distribution shift of the electric field generated within the brain. Based on this, the technical solution of this application further corrects the boundary conditions of the complex impedance perturbation and the current injection reference vector based on the perturbation coupling coefficient matrix to obtain the corrected actual current vector, and then inputs the corrected actual current vector into a three-dimensional cranial lead field model of a child for forward electromagnetic radiation mapping to obtain the actual interference electric field distribution matrix. Through the above processing, the electrical disturbances at the scalp impedance level can be transmitted step by step to the electric field distribution level in the three-dimensional space of the brain, providing a spatial reference for the precise localization of subsequent focus drift errors.

[0051] More specifically, in a concrete example of this application, the boundary conditions of the complex impedance perturbation and the current injection reference vector are first corrected using a pre-calibrated perturbation coupling coefficient matrix. The perturbation coupling coefficient matrix characterizes the conversion ratio of the scalp impedance change in each channel to the attenuation of the injected current amplitude and the phase hysteresis. The complex impedance perturbation is weighted by this coupling coefficient matrix and then subtracted from the current injection reference vector to obtain the actual current intensity and phase information that each channel can inject into the brain tissue under the current inferior impedance conditions, i.e., the corrected actual current vector. The calculation of this correction process is expressed as follows:

[0052] in, The corrected actual current vector represents the complex value of the actual current injected into the brain tissue through each channel under the influence of impedance perturbation. The current injection reference vector is the optimal current output weight under ideal conditions. This is the perturbation coupling coefficient matrix, used to convert the complex impedance change of the scalp surface into the amplitude attenuation and phase hysteresis ratio of the injected current. This represents the complex impedance disturbance.

[0053] Subsequently, the corrected actual current vector is input into a pre-defined three-dimensional cranial conduction field model for forward electromagnetic radiation mapping. This conduction field model establishes the electric field transmission relationship from scalp-injected current to arbitrary spatial grid points within the brain based on the conductivity parameters of the layered tissues of the child's skull. The corrected actual current vector is input into this model as a boundary excitation condition. Through forward electromagnetic field calculation, the physical interference superposition state of multiple high-frequency carrier waves within the whole-brain volume conductor is simulated, outputting a three-dimensional spatially gridded actual interference electric field distribution matrix. The calculation of this forward mapping is expressed as follows:

[0054] in, The actual interference electric field distribution matrix represents the true three-dimensional electric field intensity distribution generated at each spatial grid point in the brain under impedance perturbation. The matrix of the three-dimensional cranial lead field model for children represents the positive transmission operator from the scalp-injected current to the brain's electric field. This is the corrected actual current vector. Through the above two-stage cascaded calculations, the impedance perturbation information of the scalp surface is gradually transmitted and unfolded into a complete electric field distribution map in the three-dimensional space of the brain, providing a quantitative basis for the next step of searching for the actual focal location in this distribution and calculating its spatial error from the target point.

[0055] In step S33, the coordinates of the node with the maximum low-frequency interference envelope intensity are searched in the three-dimensional spatial grid of the actual interference electric field distribution matrix. These coordinates are then compared with the coordinates of the theoretical cognitive target center to calculate the spatial drift error vector. It should be noted that since the actual interference electric field distribution matrix only describes the electric field intensity distribution at each spatial grid point in the brain, the actual physical location of the interference focus and its specific distance and direction from the preset target point are not yet clearly defined. Subsequent inverse compensation calculations require a precise three-dimensional spatial error vector as input to calculate the correct correction parameters. Therefore, the technical solution of this application further searches for the coordinates of the node with the maximum low-frequency interference envelope intensity in the three-dimensional spatial grid of the actual interference electric field distribution matrix, and compares them with the coordinates of the theoretical cognitive target center to calculate the spatial drift error vector. Through the above processing, the information at the electric field distribution level can be transformed into a three-dimensional spatial error quantity with a clear direction and distance, providing a precise correction target for inverse phasor compensation.

[0056] More specifically, in a concrete example of this application, a spatial search for the low-frequency interference envelope intensity is first performed on the actual interference electric field distribution matrix. Since the therapeutic effect of dual-target time-interference electrical stimulation is determined by the spatial focusing position of the low-frequency difference frequency envelope, it is necessary to extract the spatial intensity distribution of the low-frequency envelope component from the three-dimensional electric field distribution of the whole brain. Envelope demodulation is performed on the multi-frequency electric field signals at each spatial grid node in the actual interference electric field distribution matrix to extract the amplitude of the low-frequency component corresponding to the interference envelope target frequency. The global maximum value of this low-frequency interference envelope intensity is searched throughout all three-dimensional spatial grid nodes, and the three-dimensional coordinates of the spatial grid node corresponding to this maximum value are extracted as the actual interference focus coordinates. Subsequently, a three-dimensional spatial vector subtraction operation is performed between these actual interference focus coordinates and the preset theoretical cognitive target center coordinates in the target stimulation parameter set to solve for the focus drift error vector containing offset components in three dimensions: front-back axis, left-right axis, and top-bottom axis. The magnitude of this vector represents the absolute distance of the focus missing the target, and its direction represents the spatial orientation of the focus offset. Since this scheme involves simultaneous stimulation of two targets, the above search and error solving processes are performed independently for the first cognitive target and the second cognitive target, and the drift errors of the two targets are spliced ​​together to form a complete focus drift error vector.

[0057] Specifically, in step S4, the focus drift error vector and complex impedance disturbance are calculated using the inverse mapping model of the cranial interference field, based on the inverse phasor compensation of the interference physical field, to obtain the phase compensation angle sequence and amplitude correction coefficient matrix. It should be noted that, given that the focus drift error vector and complex impedance disturbance describe the deviation state of the current interference electric field from the spatial and electrical domains respectively, but have not yet been converted into current control parameters that can be directly applied to the underlying hardware, and that sudden changes in capacitive reactance of the scalp impedance introduce a capacitor charging and discharging time delay leading to carrier phase loss, while impedance changes cause energy heat loss leading to carrier amplitude attenuation, it is necessary to jointly couple and calculate the spatial error and electrical disturbance into precise phase correction angles and amplitude gain coefficients for each channel to achieve focus pull-back at the physical excitation end. Based on this, the technical solution of this application further uses the inverse mapping model of the cranial interference field to calculate the focus drift error vector and complex impedance disturbance using the inverse phasor compensation of the interference physical field, to obtain the phase compensation angle sequence and amplitude correction coefficient matrix. Through the above processing, the focal offset information in the spatial domain and the impedance perturbation information in the electrical domain can be fused and transformed into phase and amplitude dual-dimensional control parameters that can be directly executed by the underlying hardware, providing accurate correction instructions for the subsequent amplitude-phase joint modulation of the carrier signal.

[0058] Figure 6This document describes a control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders, based on an embodiment of this application. The method involves using a brain interference field inverse mapping model to perform inverse phasor compensation calculations based on the interference physical field to obtain the phase compensation angle sequence and amplitude correction coefficient matrix, thereby achieving the desired results. Figure 6 As shown, step S4 includes: S41, inputting the focus drift error vector into the brain interference field inverse mapping neural network model to derive the spatial electromagnetic field reconstruction to obtain the target electric field reconstruction weight vector; S42, based on the capacitive reactance mutation and impedance mutation parameters in the complex impedance disturbance, performing physical phasor pre-compensation calculation under impedance coupling on the target electric field reconstruction weight vector to obtain the multi-channel complex phasor compensation matrix; S43, performing phasor separation and control parameter formatting extraction on the multi-channel complex phasor compensation matrix to obtain the phase compensation angle sequence and amplitude correction coefficient matrix.

[0059] In step S41, the focus drift error vector is input into the brain interference field inverse mapping neural network model for spatial electromagnetic field reconstruction derivation to obtain the target electric field reconstruction weight vector. It should be noted that since the focus drift error vector describes the distance and direction information of the focus deviating from the target point in the three-dimensional space within the brain, it belongs to the spatial domain error representation. However, the actual control requires adjusting the current output of each scalp electrode channel to pull the focus back to the target point. Therefore, inverse mapping is needed to reverse-derive the spatial error into the current adjustment weights at the scalp end. Based on this, the technical solution of this application further inputs the focus drift error vector into the brain interference field inverse mapping neural network model for spatial electromagnetic field reconstruction derivation to obtain the target electric field reconstruction weight vector. Through the above processing, the focus offset information in the brain's spatial domain can be reverse-transformed into the current adjustment ratio of each channel at the scalp electrode, providing a theoretical reconstruction benchmark for subsequent physical phasor compensation combined with impedance perturbation.

[0060] More specifically, in a concrete example of this application, the focus drift error vector is fed as input into a pre-trained neural network model for inverse mapping of the cranial interference field. This neural network model is trained offline based on a large amount of finite element simulation data of children's craniums. Its training samples are the optimal current adjustment weight combinations corresponding to different spatial drift directions and distances. The model learns the nonlinear inverse mapping relationship from three-dimensional spatial error to multi-channel current weights. After inputting the focus drift error vector into the model, the model derives, based on the drift direction and distance, the proportion of electromagnetic field energy that needs to be redistributed for each scalp electrode channel under the current topology to generate an inverse electric field potential energy gradient opposite to the drift direction and at the same distance. The output is the target electric field reconstruction weight vector. Each element in this vector corresponds to the gain adjustment factor of a physical electrode channel. Its value represents the relative weight of the channel's contribution to focus pull-back. A positive value indicates that the channel's output needs to be enhanced to pull the focus towards the target point, while a negative value indicates that the channel's output needs to be suppressed to reduce its thrust contribution to the focus offset direction.

[0061] Specifically, this brain interference field inverse mapping neural network model employs a multi-layer fully connected feedforward network architecture. The network input layer contains 6 neurons, corresponding to the drift error components of the first and second cognitive targets in the anterior-posterior axis, left-right axis, and upper-low axis, respectively. The network contains 4 hidden layers, each containing 256, 512, 512, and 256 neurons, respectively. Each hidden layer uses a modified linear unit as the activation function to introduce non-linear mapping capability, and a Dropout regularization layer (dropout rate of 0.2) is set between each hidden layer to prevent overfitting. The output layer contains N neurons (N is the total number of channels in the scalp electrode array, N=32 in this embodiment), and uses a linear activation function to allow positive and negative weights in the output. During the training phase, mean squared error is used as the loss function, and the optimizer uses the Adam algorithm (initial learning rate 0.001, β1=0.9, β2=0.999), with a batch size of 64 and 20,000 training iterations. The training dataset was generated using a finite element model of a child's brain. Specifically, drift direction and distance combinations were uniformly sampled within a ±5mm range around the two target points with a step size of 0.5mm. For each drift condition, the optimal current was calculated using forward finite element simulation to adjust the weights as labels, generating approximately 50,000 training sample pairs. 80% of these were used for training, 10% for validation, and 10% for testing. Model training stopped when the loss on the validation set stopped decreasing after 500 consecutive epochs. The final mean absolute error on the test set was less than 0.01mA.

[0062] In step S42, based on the capacitive reactance and impedance abrupt change parameters in the complex impedance perturbation, the physical phasor pre-compensation solution under impedance coupling is performed on the target electric field reconstruction weight vector to obtain a multi-channel complex phasor compensation matrix. It should be noted that since the target electric field reconstruction weight vector is a theoretical adjustment ratio derived through inverse mapping under ideal impedance conditions, it does not consider the capacitance charging and discharging delay caused by the actual capacitive reactance abrupt change at the current scalp electrode interface, nor the energy loss and attenuation caused by impedance abrupt change. If the current output is directly adjusted according to the theoretical weight, the compensation signal will experience another amplitude drop and phase shift after transmission through the poor-state impedance interface, failing to generate the expected correction electric field at the deep target point. Therefore, the technical solution of this application further performs a physical phasor pre-compensation solution under impedance coupling on the target electric field reconstruction weight vector based on the capacitive reactance and impedance abrupt change parameters in the complex impedance perturbation to obtain a multi-channel complex phasor compensation matrix. Through the above processing, the theoretical space reconstruction weights can be transformed into complex electrical compensation quantities that have been pre-offset to offset the effects of the current impedance degradation, ensuring that the compensation signal can still produce an accurate correction effect at the target point after crossing the degraded interface.

[0063] More specifically, in a concrete example of this application, the capacitive reactance mutation parameter and impedance mutation parameter of each channel are first extracted from the complex impedance disturbance. The capacitive reactance mutation parameter corresponds to the change in the imaginary part of the complex impedance disturbance, and its physical effect is to introduce an additional capacitor charging and discharging time delay when the high-frequency carrier passes through the electrode interface, which manifests as a phase lag in the carrier signal. The impedance mutation parameter corresponds to the change in the real part of the complex impedance disturbance, and its physical effect is to generate additional electrical energy heat loss when the carrier current flows through the increased interface resistance, which manifests as an amplitude attenuation in the carrier signal. Subsequently, the above two types of physical effects are coupled with the target electric field reconstruction weight vector in the complex domain to perform algebraic operations. A pre-compensation increment proportional to the current impedance disturbance is superimposed on the theoretical reconstruction weight of each channel, so that the compensation signal contains enough additional energy and leading phase to offset the interface transmission loss when it is emitted. This coupled solution maps the theoretical reconstruction weight in the spatial dimension to the complex electrical compensation amount that needs to be pre-filled at the underlying physical excitation end. The compensation amounts of each channel are arranged in a diagonal structure according to the spatial topology, generating a multi-channel complex phasor compensation matrix. Each element on the main diagonal of the matrix is ​​a complex value. Its magnitude represents the additional gain that the channel needs to add to offset impedance heat loss, and its phase angle represents the lead angle that the channel needs to pre-padded to offset capacitive reactance charging and discharging delay.

[0064] In step S43, phasor separation and control parameter formatting extraction are performed on the multi-channel complex phasor compensation matrix to obtain the phase compensation angle sequence and amplitude correction coefficient matrix. It should be noted that since the multi-channel complex phasor compensation matrix uniformly encodes the amplitude gain and phase lead information of each channel in complex form, and the underlying hardware's digital multiplier and digital phase shifter independently accept real-valued amplitude weights and angle-type phase parameters as control commands, the complex encoding needs to be split into two independent control parameter formats that can be directly executed by the hardware. Based on this, the technical solution of this application further performs phasor separation and control parameter formatting extraction on the multi-channel complex phasor compensation matrix to obtain the phase compensation angle sequence and amplitude correction coefficient matrix. Through the above processing, the unified complex compensation information can be decoupled into phase control sequences and amplitude control matrices that can be independently loaded at the hardware level, completing the parameter format conversion from the algorithm domain to the hardware control domain.

[0065] More specifically, in a concrete example of this application, polar coordinate decoupling is first performed on each complex element on the main diagonal of the multi-channel complex phasor compensation matrix. The phase angle component of each diagonal complex element is extracted. This phase angle represents the lead angle that needs to be pre-padded to compensate for capacitive reactance charging and discharging delays or the hysteresis angle that needs to be applied to compensate for inductive effects. The phase angle parameters of all channels are concatenated into a one-dimensional sequence according to the channel index order to obtain the phase compensation angle sequence. Subsequently, the magnitude component of each diagonal complex element is extracted. This magnitude represents the current intensity gain ratio that needs to be applied to compensate for impedance thermal loss attenuation of the corresponding channel. Further, the extracted magnitude is subjected to safety limiting processing. The gain ratio of each channel is compared element-by-element with the preset maximum safe stimulation current amplitude gain limit for children, and the smaller value is taken as the final effective gain, preventing the injected current from exceeding the safe tolerance threshold of children's brain tissue due to overcompensation. After the gain ratio of each channel is limited, it is arranged into a diagonal matrix according to the spatial topology to obtain the amplitude correction coefficient matrix. Each element on the main diagonal of the matrix corresponds to the current amplitude multiplicative gain coefficient that a physical electrode channel should execute in the current control cycle.

[0066] Specifically, in step S5, based on the phase compensation angle sequence and amplitude correction coefficient matrix, the orthogonal decoupled carrier group is subjected to amplitude-phase joint modulation and synthesis to obtain a spatially locked stimulus output stream. It should be noted that since the phase compensation angle sequence and amplitude correction coefficient matrix are only numerical control parameters in the algorithm domain and have not yet been applied to the actual carrier physical signal, the final driving electrode array to generate a spatially locked interference electric field must be a continuous excitation waveform after amplitude-phase dual correction. Therefore, the compensation parameters need to be superimposed on the original carrier signal and converted into a hardware-executable output stream. Based on this, the technical solution of this application further performs amplitude-phase joint modulation and synthesis on the orthogonal decoupled carrier group based on the phase compensation angle sequence and amplitude correction coefficient matrix to obtain a spatially locked stimulus output stream. Through the above processing, the compensation parameters in the algorithm domain can be converted into actual excitation signals in the physical domain, enabling the low-frequency envelope focus to be precisely locked at the preset target coordinates after the multi-channel carriers pass through the inferior impedance interface and intersect in the deep brain region.

[0067] More specifically, in a particular example of this application, the amplitude correction coefficient matrix is ​​first converted into digital multiplier weights, and multiplicative gain modulation is performed on the instantaneous amplitude of each channel of the orthogonal decoupled carrier group to obtain an amplitude precalibration carrier group. The gain coefficients of each channel on the main diagonal of the amplitude correction coefficient matrix are loaded into the digital multiplier register of the corresponding channel. Multiplicative operation is performed on the instantaneous amplitude of the carrier signal of each channel in the orthogonal decoupled carrier group on a sampling point-by-sampling basis. The carrier amplitude of the channel with a gain coefficient greater than 1 is proportionally amplified to compensate for the current attenuation caused by the increase in impedance of the channel, and the carrier amplitude of the channel with a gain coefficient less than 1 is proportionally reduced to avoid the current overshoot caused by the decrease in impedance of the channel. After amplitude modulation, the current intensity of each channel carrier has pre-included an additional energy margin sufficient to offset the interface transmission loss, thus obtaining an amplitude precalibration carrier group.

[0068] Subsequently, the lead or lag angle parameters in the phase compensation angle sequence are extracted, and the amplitude precalibrated carrier group is subjected to microsecond-level time axis translation servo processing through a digital phase shifter network to obtain the amplitude-phase joint modulation analog signal. The angle parameters of each channel in the phase compensation angle sequence are converted into time shifts at the corresponding carrier frequencies. Positive angles correspond to time axis lead shifts to pre-compensate for phase lag caused by capacitive reactance charging and discharging, while negative angles correspond to time axis lag shifts to cancel phase lead caused by inductive effects. The calculated time shifts are loaded into the digital phase shifters of each channel, and microsecond-level time axis translation operations are performed on the signals of each channel in the amplitude precalibrated carrier group that have undergone amplitude correction. This ensures that the multi-channel carrier signals can achieve nanosecond-level phase synchronization again when they reach the deep target point after passing through their respective inferior impedance interfaces, ensuring that the low-frequency envelope peak generated by interference superposition accurately falls at the preset target point coordinates. The time-domain expression of the output signal of each channel after amplitude gain modulation and phase translation servo processing is as follows:

[0069] in, It is an amplitude-phase combined modulation analog signal. The initial constant current amplitude constant vector set for each channel hardware. This is an element-wise multiplication operation. This is the column vector of gain coefficients for each channel extracted from the main diagonal of the amplitude correction coefficient matrix. The high-frequency carrier fundamental frequency column vector assigned to each channel, For continuous time variables, The phase compensation angle sequence is the initial phase offset angle superimposed by each channel. This formula shows that the output signal of each channel simultaneously carries the gain correction in the amplitude dimension and the lead-lag compensation in the phase dimension. The two work together to ensure that the carrier can still participate in the interference superposition with the correct intensity and phase at the deep target point after crossing the inferior impedance interface, thus obtaining the amplitude-phase joint modulation analog signal.

[0070] Finally, the amplitude-phase co-modulated analog signal is time-discrete sampled and amplitude-quantized encoded to obtain a spatially locked stimulus output stream. The continuous amplitude-phase co-modulated analog signal is sampled at equal intervals according to the fixed clock cycle of the digital-to-analog converter hardware drive protocol. The continuous analog level values ​​at each sampling moment are mapped to discrete digital encoded values ​​with a fixed quantization bit width. The discrete data after multi-channel quantization is then encapsulated with frame headers and arranged with channel identifiers to form a spatially locked stimulus output stream that can be directly sent to the constant current source drive base for physical output. This output stream is then converted back to analog current by the digital-to-analog converter and drives the scalp electrode array to generate a spatially locked low-frequency interference envelope electric field at a deep target point in the child's brain.

[0071] For example, after reverse compensation calculation, the amplitude correction factor for channel 3 is 1.15 (meaning the current amplitude needs to be increased by 15% to compensate for the attenuation caused by the increased impedance of this channel), and the phase compensation angle is +8.7° (meaning an 8.7° lead is needed to pre-compensate for the phase lag caused by capacitive reactance). The original carrier frequency of this channel is 2000Hz, and the initial amplitude is 1.5mA. Therefore, the modulated output signal amplitude is 1.725mA, and the time shift corresponding to the phase lead is approximately 12.1μs. After 16-bit quantization encoding at a sampling rate of 250kHz by a digital-to-analog converter, the spatially locked stimulus output stream component of this channel is formed.

[0072] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A control method for a dual-target time-interventional electrical stimulation device for children with cognitive developmental disorders, characterized in that, include: Step 1: Based on the target stimulus parameter set containing the three-dimensional coordinates of the first cognitive target and the three-dimensional coordinates of the second cognitive target, perform orthogonal decoupled carrier configuration and reference current calculation on the two targets to obtain the orthogonal decoupled carrier group and the current injection reference vector. Step 2: Excite the scalp electrode array by using an orthogonal decoupled carrier group and simultaneously acquire a multi-channel echo sampling sequence. Then, perform real-time dynamic complex impedance calculation based on high-frequency echo on the multi-channel echo sampling sequence to obtain the electrode dynamic complex impedance matrix. Step 3: Based on the current injection reference vector, perform spatial drift prediction and evaluation of the electrode dynamic complex impedance matrix under the dynamic interference of children to obtain the complex impedance disturbance amount and the focus drift error vector. Step 4: Using the inverse mapping model of the cranial interference field, perform inverse phasor compensation calculation based on the interference physical field to obtain the phase compensation angle sequence and amplitude correction coefficient matrix for the focus drift error vector and complex impedance perturbation. Step 5: Based on the phase compensation angle sequence and amplitude correction coefficient matrix, the orthogonal decoupled carrier group is subjected to amplitude-phase joint modulation and synthesis to obtain the spatially locked stimulus output stream.

2. The control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders according to claim 1, characterized in that, The target stimulation parameter set also includes the threshold for the cutoff frequency of the child's neural response and the target frequency of the interference envelope.

3. The control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders according to claim 2, characterized in that, Step one includes: The three-dimensional spatial coordinates of the first cognitive target and the second cognitive target are encapsulated into a dual-target spatial coordinate matrix; Based on the cutoff frequency threshold of children's neural response, a first carrier fundamental frequency is assigned to the first cognitive target, a second carrier fundamental frequency is assigned to the second cognitive target, and the frequency is combined with the interference envelope target frequency to obtain the interference frequency domain configuration matrix. Based on the frequency parameters in the interferometric frequency domain configuration matrix, orthogonal decoupled carrier synthesis is performed on each channel to obtain an orthogonal decoupled carrier group; The dual-target spatial coordinate matrix is ​​input into the positive mapping model of the pediatric cranial lead field to calculate the optimal current weight and obtain the current injection reference vector.

4. The control method for the dual-target time-interference electrical stimulation device for children with cognitive developmental disorders according to claim 3, characterized in that, The second carrier fundamental frequency is equal to the sum of the first carrier fundamental frequency, the child's neural response cutoff frequency threshold, and the safety margin.

5. The control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders according to claim 1, characterized in that, Step two includes: The excitation and echo signals of the orthogonal decoupled carrier group and the multi-channel echo sampling sequence are synchronized in the time domain and windowed to obtain windowed synchronized time series pairs. Frequency domain phasor separation and feature extraction are performed on windowed synchronous time series pairs to obtain amplitude and phase attenuation feature vectors; Complex impedance scalar calculations are performed on the amplitude and phase attenuation eigenvectors, and a diagonal matrix is ​​constructed according to the spatial topology of the scalp electrode array to obtain the dynamic complex impedance matrix of the electrodes.

6. The control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders according to claim 1, characterized in that, Step three includes: The dynamic impedance reference difference and perturbation quantity were extracted from the dynamic complex impedance matrix of the electrode and the pre-calibrated ideal contact impedance constant matrix of the child's skin-electrode to obtain the complex impedance perturbation quantity; Based on the perturbation coupling coefficient matrix, the boundary conditions of the complex impedance perturbation and the current injection reference vector are corrected to obtain the corrected actual current vector. The corrected actual current vector is then input into the three-dimensional cranial lead field model of the child for positive electromagnetic radiation mapping to obtain the actual interference electric field distribution matrix. The coordinates of the node with the maximum low-frequency interference envelope intensity are searched in the three-dimensional spatial grid of the actual interference electric field distribution matrix. The spatial drift error is solved by comparing these coordinates with the coordinates of the theoretical target center to obtain the focus drift error vector.

7. The control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders according to claim 1, characterized in that, Step four includes: The focus drift error vector is input into the brain interference field inverse mapping neural network model to derive the spatial electromagnetic field reconstruction to obtain the target electric field reconstruction weight vector. Based on the capacitive reactance and impedance abrupt parameters in the complex impedance disturbance, the physical phasor pre-compensation solution under impedance coupling is performed on the target electric field reconstruction weight vector to obtain the multi-channel complex phasor compensation matrix. Phasor separation and control parameter formatting extraction are performed on the multi-channel complex phasor compensation matrix to obtain the phase compensation angle sequence and amplitude correction coefficient matrix.

8. The control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders according to claim 1, characterized in that, Step five includes: The amplitude correction coefficient matrix is ​​transformed into digital multiplier weights, and multiplicative gain modulation is performed on the instantaneous amplitude of each channel of the orthogonal decoupled carrier group to obtain the amplitude precalibrated carrier group. Extract the lead or lag angle parameters from the phase compensation angle sequence, and perform microsecond-level time axis translation servo processing on the amplitude precalibrated carrier group through a digital phase shifter network to obtain the amplitude-phase joint modulation analog signal. The amplitude-phase co-modulated analog signal is subjected to time discretization sampling and amplitude quantization encoding to obtain a spatially locked stimulus output stream.

9. The control method for a dual-target time-interference electrical stimulation device for children with cognitive developmental disorders according to claim 6, characterized in that, The dynamic impedance reference difference and perturbation quantity are extracted from the dynamic complex impedance matrix of the electrode and the pre-calibrated ideal contact impedance constant matrix of the child's skin-electrode to obtain the complex impedance perturbation quantity, including: The channel Joule thermal energy accumulation quantization and self-excited drift prediction are performed on the current injection reference vector to obtain the self-excited thermal drift impedance matrix. Three-source causal decomposition and external disturbance isolation were performed on the electrode dynamic complex impedance matrix, the pre-calibrated ideal contact impedance constant matrix of pediatric skin-electrode, and the self-excited thermal drift impedance matrix to obtain the candidate matrix of external disturbance. Based on the upper limit of the maximum allowable rate of impedance change under the biomechanical constraints of children's head movement, time-domain causal clamping verification and non-physical pseudo-component suppression are performed on the perturbation values ​​of each channel in the candidate matrix of exogenous perturbation to obtain the complex impedance perturbation amount.

Images