Automatic control method of gradient drive system for ultra-low field magnetic resonance

Abstract

The present invention provides an automatic control method for a gradient drive system for ultra-low field magnetic resonance imaging, relating to the technical field of ultra-low field magnetic resonance imaging systems. The automatic control method comprises determining a target switching frequency of the gradient drive system; determining a corresponding target sampling period and initial control parameters of the gradient drive system based on the target switching frequency; determining a target control parameter of the gradient drive system based on a target current and the initial control parameter; and determining a target duty cycle signal of the gradient drive system based on the target control parameter, thereby dynamically adjusting the switching frequency of the gradient drive system and reducing switching losses.

Description

Technical Field

[0001] The present invention relates to the technical field of ultra-low field magnetic resonance imaging systems, and in particular to an automatic control method for a gradient drive system for ultra-low field magnetic resonance imaging. Background Art

[0002] Ultra-low-field magnetic resonance (ULF) systems include RF transmitter and receiver systems and a gradient drive system. The gradient drive system generates significant heat during operation, primarily from switching losses in the switches, which are proportional to the switching frequency. To achieve better dynamic performance and more compact designs (reducing the size of inductors and capacitors), the switching frequency often needs to be continuously increased, which increases the energy density of the switches. This accumulated heat increases the junction temperature of the switches, which in turn affects their service life. Traditional gradient drive system controller designs often focus on dynamic performance under high currents, without specific optimizations for low currents or low frequencies. This results in low energy conversion efficiency under light loads. Traditional gradient drive system controllers typically use a fixed switching frequency design, maintaining a high switching frequency even at low currents or low frequencies. Each switching operation generates losses (such as switch on / off losses and driver losses), and these losses are proportional to the frequency. Therefore, it is difficult for traditional gradient drive system controllers to achieve both high dynamic performance and low power consumption. Designing a gradient drive system that combines high integration and high performance at low fields requires adaptive dynamic adjustment of the switching rate, making it particularly important. Summary of the Invention

[0003] In view of this, an object of the present invention is to provide an automatic control method for a gradient drive system for ultra-low field magnetic resonance imaging, so as to adaptively and dynamically adjust the switching frequency of the gradient drive system and automatically match the corresponding PID control signal, thereby reducing switching losses.

[0004] An embodiment of the present invention provides an automatic control method for a gradient drive system for ultra-low field magnetic resonance imaging.

[0005] Provided is an automatic control method for a gradient drive system for ultra-low field magnetic resonance imaging, comprising:

[0006] determining a target switching frequency of the gradient drive system;

[0007] Determining a corresponding target sampling period and initial control parameters of the gradient drive system according to the target switching frequency;

[0008] determining a target control parameter of the gradient drive system according to the target current and the initial control parameter;

[0009] A target duty cycle signal of the gradient drive system is determined according to the target control parameter.

[0010] In some embodiments, determining a target switching frequency of the gradient drive system includes:

[0011] Determining a target operating mode of the ultra-low field magnetic resonance imaging among a plurality of operating modes, wherein the operating modes include: a standby mode, a conventional imaging mode, and a rapid diffusion weighted imaging mode;

[0012] Each of the operating modes corresponds to a switching frequency;

[0013] A corresponding target switching frequency is determined according to the target operating mode.

[0014] In some embodiments, determining a target switching frequency of the gradient drive system includes:

[0015] Real-time acquisition of the actual output current of the gradient drive system;

[0016] Determine the current slope based on the output current at the current sampling moment and the output current at the historical sampling moment with a preset time interval from the current sampling moment;

[0017] Acquire a plurality of preset current slope thresholds, each of the preset current slope thresholds corresponding to a switching frequency;

[0018] A target switching frequency is determined according to the current slope and a plurality of the preset current slope thresholds.

[0019] In some embodiments, determining a corresponding target sampling period and initial control parameters of the gradient drive system according to the target switching frequency includes:

[0020] The target sampling period and the target switching frequency are reciprocals of each other;

[0021] The target switching frequency has a corresponding initial duty cycle signal of the gradient drive system, and the initial duty cycle signal is determined according to the initial control parameters, which include an initial integral control parameter, an initial proportional control parameter, and an initial differential control parameter.

[0022] In some embodiments, the initial duty cycle signal is determined according to the initial control parameters, including

[0023] The duty cycle signal expression of the gradient drive system is:

[0024]

[0025]

[0026] in, represents the duty cycle signal at k sampling time, is the integral control parameter, is the proportional control parameter, is the differential control parameter, represents the sampling period, represents the target current, Indicates the actual output current, represents the current error at sampling time k, Indicates the cumulative current error from time 0 to k;

[0027] Determining the corresponding initial duty cycle signal according to the target switching frequency;

[0028] The initial duty cycle signal is determined according to the initial integral control parameter, the initial proportional control parameter, and the initial differential control parameter.

[0029] In some embodiments, determining the target control parameter of the gradient drive system according to the target current and the initial control parameter includes:

[0030] Updating the initial control parameters to adjust the duty cycle signal and collecting the corresponding updated output current;

[0031] The target current is a preset value of the gradient drive system;

[0032] A target control parameter of the gradient drive system is determined according to a current error determined by the updated output current and the target current.

[0033] In some embodiments, determining a target control parameter of the gradient drive system according to a current error determined by the updated output current and the target current includes:

[0034] Construct mean square current error function based on LMS algorithm;

[0035] Updating the initial control parameters using a gradient descent method;

[0036] When the mean square current error function value is minimum, the corresponding control parameter is the target control parameter.

[0037] In some embodiments, the mean square current error function constructed according to the LMS algorithm is: .

[0038] In some embodiments, determining the target control parameter of the gradient drive system according to the target current and the initial control parameter includes:

[0039] The initial control parameters are input into the trained neural network, and the target control parameters are output.

[0040] In some embodiments, the neural network training process includes:

[0041] Using multiple historical target control parameters as labels, each of the initial control parameters corresponds to a historical target control parameter;

[0042] Inputting the plurality of initial control parameters into a neural network to be trained, and determining a historical current error corresponding to each of the initial control parameters as a residual, wherein the historical current error is determined according to the target current and the updated output current;

[0043] Accumulate all the residuals to obtain a loss function;

[0044] The neural network is optimized according to the loss function until a preset condition is met, thereby obtaining the optimized neural network.

[0045] The embodiments of the present invention bring the following beneficial effects: dynamic adjustment of the switching frequency of the gradient drive system, thereby reducing switching losses.

[0046] Other features and advantages of the present disclosure will be set forth in the following description, or some features and advantages may be inferred or unambiguously determined from the description, or may be learned by practicing the above-mentioned technology of the present disclosure.

[0047] In order to make the above-mentioned objectives, features and advantages of the present disclosure more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0049] Figure 1 A flow chart of an automatic control method for a gradient drive system for ultra-low field magnetic resonance imaging provided by an embodiment of the present invention;

[0050] Figure 2 A flow chart for determining a target switching frequency according to an embodiment of the present invention;

[0051] Figure 3 A flow chart of determining target control parameters of the gradient drive system based on the LMS algorithm provided in an embodiment of the present invention; DETAILED DESCRIPTION

[0052] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0053] To facilitate understanding of this embodiment, a magnetic resonance image analysis system disclosed in an embodiment of the present invention is first introduced in detail.

[0054] Figure 1 This is a flow chart of an automatic control method for a gradient drive system for ultra-low field magnetic resonance imaging provided in an embodiment of the present application.

[0055] First, determine the target switching frequency of the gradient drive system. The specific method can be based on Figure 2 The provided method determines the target switching frequency. In an ultra-low field magnetic resonance device, the current collected in the present invention is the core physical quantity that drives the gradient coil to generate a spatial linear magnetic field gradient. The current slope is determined based on the output current at the current sampling moment and the output current at the historical sampling moment with a preset time interval from the current sampling moment. Specifically, the historical sampling moment can be the moment before the current sampling moment, which can ensure the real-time calculation of the current slope and the real-time control; it can also be a preset time interval from the current sampling moment, such as an interval of half a cycle, an interval of N moments, N is a positive integer greater than 1, etc., which can improve the control efficiency and have a higher tolerance for the computing power of the hardware. In the present invention, multiple current slope thresholds are preset, and the relationship between the calculated current slope and the current slope threshold is compared to determine the target switching frequency. Since different current slope thresholds correspond to different working modes, the target switching frequency can also be determined by the target working mode of ultra-low field magnetic resonance. The present invention provides an embodiment of a working mode, a current slope threshold, and a target switching frequency, as shown in Table 1:

[0056] Table 1

[0057]

[0058] Once the operating mode is determined, the target switching frequency is determined. The operating mode is determined based on the imaging mode selected by the user or by calculating the current slope.

[0059] The method for determining the corresponding target sampling period and initial control parameters of the gradient drive system according to the target switching frequency includes: when the target switching frequency is determined, the target sampling period is correspondingly determined, and the target sampling period and the target switching frequency are reciprocals of each other.

[0060] Each target switching frequency (which can also be understood as each operating mode) corresponds to an initial duty cycle signal. The expression of the duty cycle signal is:

[0061]

[0062]

[0063] in, represents the duty cycle signal at k sampling time, is the integral control parameter, is the proportional control parameter, is the differential control parameter, represents the sampling period, represents the target current, Indicates the actual output current, represents the current error at sampling time k, Indicates the cumulative current error from time 0 to k.

[0064] Therefore, each target switching frequency corresponds to a set of initial control parameters ( , , ), when the working mode is switched, or when the current slope is detected to fall to another current slope threshold, it quickly switches to the corresponding initial control parameters. For example, in low power mode, a smaller and To reduce response speed and power consumption; in high performance mode, a larger and smaller This improves tracking accuracy and anti-interference capabilities. Initial control parameters can be preset in the system based on experience. The target current is the given value of the gradient current of the magnetic resonance system, that is, the current value that the system expects to drive the gradient coil.

[0065] like Figure 3 As shown, the mean square current error function is constructed according to the LMS algorithm. For example, the function can be , use the gradient descent method to update the initial control parameters. When the initial control parameters are updated, the corresponding duty cycle signal is also updated. Synchronously, the output current of the system is also updated synchronously. The output current is collected and updated to ensure that the updated output current of the system tracks the target current, that is, it is getting closer and closer to the target current. When the mean square current error function is minimized, the determined control parameters are the target control parameters. The LMS algorithm ensures the continuous stability of the closed-loop performance at different switching frequencies by minimizing the mean square error, so that the actual output current of the system is as close to the desired target current as possible. To this end, it will continuously adjust , , These three parameters. When switching the switching frequency, the initial control parameters are switched synchronously to ensure that the switching action does not introduce additional interference. After the switching is completed, the parameters are continuously updated through LMS, making the control performance become better and better through continuous learning.

[0066] The present invention also provides a method for determining target control parameters using a neural network. The neural network training process includes: using multiple historical target control parameters as labels, with each initial control parameter corresponding to a historical target control parameter; inputting the multiple initial control parameters into the neural network to be trained, determining the historical current error corresponding to each initial control parameter as a residual, where the historical current error is determined based on the target current and the updated output current; accumulating all the residuals to obtain a loss function; and optimizing the neural network according to the loss function until a preset condition is met (e.g., minimizing the loss function), thereby obtaining the optimized neural network. When using this neural network, the initial control parameters are input into the trained neural network, and the target control parameters are output.

[0067] Unless otherwise specifically stated, the relative steps, numerical expressions and values ​​of the components and steps set forth in these embodiments do not limit the scope of the present invention.

[0068] In all examples shown and described herein, any specific values ​​should be interpreted as merely exemplary and not limiting, and thus other examples of the exemplary embodiments may have different values.

[0069] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the above-described embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-described embodiments within the technical scope disclosed by the present invention, or replace some of the technical features therein with equivalents. Such modifications, changes, or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. An automatic control method for a gradient drive system for ultra-low field magnetic resonance imaging, characterized in that: include: determining a target switching frequency of the gradient drive system; Determining a corresponding target sampling period and initial control parameters of the gradient drive system according to the target switching frequency; determining a target control parameter of the gradient drive system according to the target current and the initial control parameter; A target duty cycle signal of the gradient drive system is determined according to the target control parameter.

2. The method according to claim 1, characterized in that Determining the target switching frequency of the gradient drive system includes: Determining a target operating mode of the ultra-low field magnetic resonance imaging among a plurality of operating modes, wherein the operating modes include: a standby mode, a conventional imaging mode, and a rapid diffusion weighted imaging mode; Each of the operating modes corresponds to a switching frequency; A corresponding target switching frequency is determined according to the target operating mode.

3. The method according to claim 1, characterized in that Determining the target switching frequency of the gradient drive system includes: Real-time acquisition of the actual output current of the gradient drive system; Determine the current slope based on the output current at the current sampling moment and the output current at the historical sampling moment with a preset time interval from the current sampling moment; Acquire a plurality of preset current slope thresholds, each of the preset current slope thresholds corresponding to a switching frequency; A target switching frequency is determined according to the current slope and a plurality of the preset current slope thresholds.

4. The method according to claim 1, wherein The determining of the corresponding target sampling period and the initial control parameters of the gradient drive system according to the target switching frequency includes: The target sampling period and the target switching frequency are reciprocals of each other; The target switching frequency has a corresponding initial duty cycle signal of the gradient drive system, and the initial duty cycle signal is determined according to the initial control parameters, which include an initial integral control parameter, an initial proportional control parameter, and an initial differential control parameter.

5. The method according to claim 4, characterized in that The initial duty cycle signal is determined according to the initial control parameters, including The duty cycle signal expression of the gradient drive system is: in, represents the duty cycle signal at k sampling time, is the integral control parameter, is the proportional control parameter, is the differential control parameter, represents the sampling period, represents the target current, Indicates the actual output current, represents the current error at sampling time k, Indicates the cumulative current error from time 0 to k; Determining the corresponding initial duty cycle signal according to the target switching frequency; The initial duty cycle signal is determined according to the initial integral control parameter, the initial proportional control parameter, and the initial differential control parameter.

6. The method according to claim 5, characterized in that The determining the target control parameter of the gradient drive system according to the target current and the initial control parameter includes: Updating the initial control parameters to adjust the duty cycle signal and collecting the corresponding updated output current; A target control parameter of the gradient drive system is determined according to a current error determined by the updated output current and the target current, where the target current is a preset value of the gradient drive system.

7. The method according to claim 6, characterized in that The determining of a target control parameter of the gradient drive system according to a current error determined by the updated output current and the target current comprises: Construct mean square current error function based on LMS algorithm; Updating the initial control parameters using a gradient descent method; When the mean square current error function value is minimum, the corresponding control parameter is the target control parameter.

8. The method according to claim 7, characterized in that The mean square current error function constructed according to the LMS algorithm is: .

9. The method according to claim 6, characterized in that The determining the target control parameter of the gradient drive system according to the target current and the initial control parameter includes: The initial control parameters are input into the trained neural network, and the target control parameters are output.

10. The method according to claim 9, characterized in that The training process of the neural network includes: Using multiple historical target control parameters as labels, each of the initial control parameters corresponds to a historical target control parameter; Inputting the plurality of initial control parameters into a neural network to be trained, and determining a historical current error corresponding to each of the initial control parameters as a residual, wherein the historical current error is determined according to the target current and the updated output current; Accumulate all the residuals to obtain a loss function; The neural network is optimized according to the loss function until a preset condition is met, thereby obtaining the optimized neural network.

Images