A temperature monitoring based field stabilization method and system

Abstract

This invention provides a temperature-monitored field stabilization method and system, relating to the field of permanent magnet magnetic resonance (PMMR). The method acquires temperature data from a magnet and simultaneously measures the proton resonance signal frequency using a magnetic resonance probe to obtain temperature-frequency data. The data is preprocessed and feature extracted, and a specially tuned feature extraction neural network and a time-series analysis neural network structure are introduced to construct a nonlinear mapping model between the spatiotemporal distribution of the temperature field and frequency drift, enabling frequency prediction. A Z0 compensation coil then generates a compensation magnetic field to fine-tune the magnetic field. This invention significantly reduces frequency drift caused by ambient temperature fluctuations without requiring a complex hardware platform, compressing the drift from hundreds of Hz to the order of several Hz. This effectively improves the testing accuracy, long-term stability, and repeatability of permanent magnet PMMR systems, and is applicable to scenarios such as permanent magnet PMMR analysis instruments, permanent magnet magnetic field control systems, and long-term online monitoring equipment.

Description

Technical Field

[0001] This invention relates to the field of permanent magnet magnetic resonance technology, and in particular to a field stabilization method and system based on temperature monitoring. Background Technology

[0002] Permanent magnets are widely used in low-field magnetic resonance systems. However, the remanence of their core materials (such as rare-earth permanent magnet materials like NdFeB and SmCo) is significantly affected by temperature: for example, the remanence temperature coefficient of NdFeB is approximately −0.12% / ℃, and for every 1℃ change in ambient temperature, the magnetic field strength can change by up to 0.12%, directly causing a shift in the magnetic resonance center frequency of approximately 1200 ppm.

[0003] Traditional temperature compensation methods mainly include: static magnetic material compensation, such as the research by Danieli et al. (Smallmagnets for portable NMR spectrometers. Angew Chem Int Ed Engl 49, 4133-4135 (2010)) and the US patent US20100013473, which proposed a design for constructing a reverse magnet to solve the phenomenon of magnet temperature drift and reduce magnetic field drift by an order of magnitude. However, this scheme requires multiple magnets with different magnetization directions and ingenious tooling, making it difficult to apply. Chinese patent application CN202310009238.X proposes a temperature control system for magnets: using a temperature controller to control the magnet temperature. However, this method is limited by the temperature control accuracy, and the final stability of this scheme still cannot meet the long-term stability requirements of magnetic resonance analysis instruments.

[0004] None of the above methods take into account the nonlinear coupling relationship between magnet temperature distribution and center frequency change, and therefore cannot accurately predict the impact of magnet temperature fluctuations on the center frequency.

[0005] Therefore, there is an urgent need for a technical solution that can sense the temperature field distribution of a magnet in real time, intelligently predict frequency drift, and actively compensate for it. Summary of the Invention

[0006] To achieve the above-mentioned objectives and other advantages of the present invention, a first objective of the present invention is to provide a field stabilization method based on temperature monitoring, comprising the following steps: Temperature data at multiple spatial locations on the permanent magnet are collected in real time, and the center frequency of the proton resonance signal is measured using a magnetic resonance probe. The temperature data and center frequency data are preprocessed, and feature vectors characterizing the spatiotemporal distribution of the temperature field and the frequency variation trend are extracted based on the preprocessed data. The feature vector is input into a pre-trained machine learning model, which outputs a predicted value of the center frequency drift in the future or the current compensation period; wherein, the machine learning model is trained based on historical temperature-frequency data and is used to establish a nonlinear mapping relationship between the spatiotemporal characteristics of the temperature field and the center frequency drift. The target compensation magnetic field strength is calculated based on the predicted center frequency drift value, and the Z0 compensation coil, which is coaxial with the main magnetic field, is driven to generate a compensation magnetic field to counteract the center frequency drift caused by temperature changes.

[0007] Furthermore, temperature data is collected by a temperature sensor array arranged at both ends of the permanent magnet in the axial direction and on the opposite side in the radial direction, wherein the temperature sensor array includes at least four temperature sensors.

[0008] Furthermore, the preprocessing includes at least time alignment, denoising, and interpolation; the extraction of the feature vector includes constructing a sample sequence of temperature field time-series data and frequency change data based on a preset time window, and calculating the statistical characteristics and / or spatial distribution characteristics of the temperature data.

[0009] Furthermore, the step of preprocessing the temperature data and center frequency data, and extracting feature vectors characterizing the spatiotemporal distribution features and frequency variation trends of the temperature field based on the preprocessed data, further includes: The features of the temperature data are subjected to dimensionality reduction processing, and the principal components after dimensionality reduction are used as input features of the machine learning model.

[0010] Furthermore, the machine learning model is specifically a neural network model, which is a hybrid structure comprising a temperature feature extraction neural network and a time series analysis neural network; wherein, the temperature feature extraction neural network is used to extract long-range dependency features of the temperature field sequence, and the time series analysis neural network is used to fuse and model multi-scale time series information.

[0011] Furthermore, the neural network model supports online updates: during system operation, the model is periodically or triggered for fine-tuning based on newly collected temperature-frequency data.

[0012] Furthermore, it also includes a closed-loop calibration step: After applying compensation, the updated center frequency is measured, and the driving current of the Z0 compensation coil is finely adjusted based on the deviation between the measured center frequency and the target center frequency.

[0013] Further, in the step of calculating the target compensation magnetic field strength based on the predicted center frequency drift value, and driving the Z0 compensation coil coaxial with the main magnetic field to generate a compensation magnetic field to counteract the center frequency drift caused by temperature changes, the formula is used... The center frequency drift prediction value Converted to target magnetic field compensation quantity Where γ is the gyromagnetic ratio of the proton; then, based on the magnetic field-current coefficient of the Z0 compensation coil, Converted into the target drive current.

[0014] A second objective of this invention is to provide a temperature-monitored field stabilization system, which implements the above-mentioned method, comprising: A temperature sensor array is used to collect temperature data at multiple spatial locations on the permanent magnet in real time. A magnetic resonance probe is used to measure the center frequency of a proton resonance signal. The data processing and prediction module is used to preprocess the temperature data and center frequency data, and extract feature vectors representing the spatiotemporal distribution characteristics of the temperature field and the frequency change trend based on the preprocessed data; the feature vectors are input into a pre-trained machine learning model to output the predicted center frequency drift value for the future or the current compensation period; wherein, the machine learning model is trained based on historical temperature-frequency data to establish a nonlinear mapping relationship between the spatiotemporal characteristics of the temperature field and the center frequency drift; the compensation control module is used to generate compensation control commands based on the predicted center frequency drift value; A programmable current source is used to respond to the compensation control command and output a corresponding compensation current; The Z0 compensation coil is arranged in the uniform field region of the permanent magnet and is used to generate a compensation magnetic field coaxial with the main magnetic field under the drive of the compensation current.

[0015] Furthermore, the data processing and prediction module integrates a temperature acquisition circuit, a frequency measurement circuit, a data preprocessing unit, a feature extraction unit, and the machine learning model.

[0016] Furthermore, the Z0 compensation coil is fabricated using a printed circuit board or wire winding, and its coil shape, number of turns, and spatial position are optimized to generate a uniform axial compensation magnetic field in the target area.

[0017] Furthermore, it also includes a human-machine interface for setting system parameters and displaying temperature distribution, center frequency drift, and compensation status information.

[0018] A third object of the present invention is to provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0019] Compared with the prior art, the beneficial effects of the present invention are: This invention provides a field stabilization method and system based on temperature monitoring. By deploying an array of temperature sensors at multiple key locations on the magnet, the spatial-temporal distribution characteristics of the magnet's temperature field can be sensed in real time. Compared to solutions that only monitor the temperature at a single point, this method more accurately reflects the overall working state of the magnet. For the first time, a specially tuned feature extraction neural network and a time-series analysis neural network are applied to model the magnet's temperature-frequency relationship, fully exploring the nonlinear coupling characteristics between the temperature field and center frequency changes, resulting in high prediction accuracy and a model with good generalization ability. A compensating magnetic field with a spatially controllable distribution and aligned with the main magnetic field is generated using a Z0 coil, and automatic adjustment is achieved through a programmable current source. This allows for active compensation of center frequency drift without altering the magnet's structure, and the installation and modification are simple. The invention supports online model fine-tuning and long-term operation, suppressing performance degradation caused by slow-changing factors such as magnet aging and environmental changes, thus improving the system's long-term frequency stability and reliability. Experimental results show that under fluctuating ambient temperature conditions, this invention can significantly reduce the center frequency drift from hundreds of Hz to the order of several Hz, greatly improving the testing accuracy and long-term repeatability of low-field magnetic resonance systems.

[0020] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0021] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 The flowchart shows the field stabilization method based on temperature monitoring. Figure 2 This is a schematic diagram of the field stabilization principle based on temperature monitoring; Figure 3 A schematic diagram showing the arrangement of the permanent magnet and the temperature sensor; Figure 4 This is a diagram of the neural network model structure. Figure 5 The image shows a comparison of the center frequency changes of the permanent magnet before and after compensation. Figure 6 This is a block diagram of a field stabilization system based on temperature monitoring. Figure 7 A schematic diagram of computer equipment; Figure 8 This is a schematic diagram of a computer-readable storage medium. Detailed Implementation

[0022] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0023] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0024] The drawing numbers in this application are only used to distinguish the steps in the scheme and are not used to limit the execution order of the steps. The specific execution order is as described in the specification.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0026] This invention provides a field stabilization method and system based on temperature monitoring. By constructing a temperature sensor array covering the magnet, real-time data on the magnet's temperature field distribution is acquired. A specially tuned feature extraction neural network and a time-series analysis neural network are introduced to explore the nonlinear mapping relationship between the spatiotemporal characteristics of the temperature field and changes in the center frequency. A programmable current source drives the Z0 compensation coil to achieve high-precision active compensation of the magnet's center frequency, ensuring the stability of the center frequency during long-term continuous operation of the low-field magnetic resonance system. The specific scheme is as follows: Example 1 A field stabilization method based on temperature monitoring, such as Figure 1 , Figure 2 As shown, it includes the following steps: S100: Real-time acquisition of temperature data at multiple spatial locations on the permanent magnet, while simultaneously measuring the center frequency of the proton resonance signal using a magnetic resonance probe; like Figure 3 As shown, temperature data at different locations on the permanent magnet are collected in real time by an array of temperature sensors arranged at both ends of the permanent magnet's axis and the opposite side of its radial direction. The temperature sensor array includes at least four temperature sensors. A magnetic resonance probe is arranged within the uniform field region of the magnet, and the center frequency of the proton resonance signal is measured via a radio frequency link, achieving synchronous monitoring and recording of the magnet's temperature and center frequency.

[0027] Preferably, the temperature sensor array uses 4, 6, or 8 NTC temperature sensors, which are evenly arranged at both ends of the magnet's axis and on the opposite radial side to improve the spatial resolution of the temperature field.

[0028] The S100 step involves arranging an array of temperature sensors, such as four NTC / pt100 / , on the magnet. The PT1000 temperature sensor collects temperature data at different locations on the magnet in real time; at the same time, it uses a nuclear magnetic resonance probe to measure the center frequency of the proton (¹H) resonance signal, realizing synchronous monitoring and recording of magnet temperature and center frequency.

[0029] S200. The temperature data and center frequency data are preprocessed, and feature vectors characterizing the spatiotemporal distribution of the temperature field and the frequency change trend are extracted based on the preprocessed data. The preprocessing includes at least time alignment, denoising, and interpolation (i.e., missing data completion). Feature vector extraction involves constructing a sample sequence of temperature field time-series data and frequency variation data based on a preset time window, and calculating the statistical characteristics and / or spatial distribution characteristics of the temperature data. For example, the collected temperature data and center frequency data are denoised, filtered, interpolated, normalized, and time-aligned. Based on the preset time window, the temperature field time-series data and frequency variation data are constructed into a sample sequence, and feature vectors characterizing the spatial-temporal distribution characteristics of the temperature field and the frequency variation trend are extracted.

[0030] The preprocessing of the temperature data and center frequency data in step S200 also includes: selecting the time window length and sliding step size for neural network modeling based on the thermal inertia and frequency response characteristics of the magnet, and dividing the temperature-frequency sequence into time windows to obtain the input and output samples required for multi-step prediction.

[0031] In some embodiments, step S200 further includes: The temperature data features are subjected to dimensionality reduction processing, and the dimensionality-reduced principal components are used as input features of the neural network model. For example, principal component analysis (PCA) is used to reduce the dimensionality of high-dimensional temperature features, and several principal components (usually 3 to 5 dimensions) with a cumulative contribution rate greater than 95% are used as input features of the neural network, thereby reducing model complexity and improving prediction efficiency and robustness while ensuring information integrity.

[0032] Optionally, time synchronization: The temperature sensor data and the center frequency data are aligned according to the acquisition timestamp to ensure that they correspond to the same physical moment, and the time alignment error is no more than 100ms; Noise suppression: The temperature and frequency sequences are smoothed using moving average filtering or other low-pass filtering methods to remove transient disturbances and measurement noise; Missing data imputation: If there is missing data at a certain time, it is imputed by linear interpolation or adjacent data interpolation to ensure the continuity of the time series.

[0033] After completing the basic preprocessing, the temperature array is further enhanced in terms of features. Specifically, this includes: Statistical characteristic calculation: Calculate the mean, variance, range, and other statistical measures of the temperature sensor readings at the current moment to characterize the overall temperature level and the degree of temperature distribution unevenness; Spatial feature extraction: Based on the spatial position of each sensor on the magnet, indicators such as axial temperature difference and radial temperature difference are constructed to characterize the spatial gradient of the temperature field; Feature dimensionality reduction: Input the above multidimensional temperature features into the principal component analysis (PCA) module, perform linear dimensionality reduction on them, and retain several principal components with a cumulative contribution rate greater than 95%. Generally, the dimensionality can be reduced to 3 to 5 dimensions, which are then used as the input feature vectors of the neural network model.

[0034] Through the above steps S100 and S200, the temperature feature sequence and corresponding center frequency change data that change over time are obtained, providing a foundation for subsequent neural network modeling and compensation control.

[0035] S300. Input the feature vector into a pre-trained machine learning model and output the predicted value of the center frequency drift in the future or current compensation period; wherein, the machine learning model is trained based on historical temperature-frequency data and is used to establish a nonlinear mapping relationship between the spatiotemporal characteristics of the temperature field and the center frequency drift. In this embodiment, the machine learning model is specifically a neural network model. For example... Figure 4 As shown, the neural network model is a hybrid structure that includes a temperature feature extraction neural network and a time series analysis neural network. The feature vector is used as input, and the corresponding center frequency drift is used as a supervision signal to perform offline training and parameter optimization on the neural network. During system operation, the trained neural network is inferred online using the temperature field features collected in real time, and the predicted value of the center frequency drift in the future compensation period is output.

[0036] The temperature feature extraction neural network is used to extract long-range dependency features of the temperature field sequence, i.e., to model the long- and short-term dependencies of the time series. The time series analysis neural network is used to fuse and model multi-scale time series information, i.e., to fuse temperature evolution information at multiple time scales. The two are connected in series or in parallel to form a hybrid neural network structure, which is used to output the predicted change in center frequency. (unit: Hz) to improve the fitting accuracy and model generalization ability of the temperature-frequency nonlinear mapping relationship.

[0037] This embodiment constructs a neural network structure optimized for the temperature-frequency relationship of magnets. This structure includes a temperature feature input layer, a specially tuned temperature feature extraction neural network module, a time-series analysis neural network module, and an output module. The number of nodes in the input layer is equal to the temperature feature dimension after PCA dimensionality reduction, for example, 4 nodes. The input layer feeds the current time and necessary historical temperature feature sequences into the neural network module to mine the time-series characteristics and nonlinear relationships of temperature changes, and outputs the predicted change in center frequency. The training dataset consists of approximately 10,000 temperature-frequency samples collected under normal magnet operating conditions, within a temperature range of 28℃ to 30℃, covering various typical operating conditions such as uniform temperature change, rapid heating, and natural cooling. This dataset was used to train the neural network offline until it reached the preset error requirement on the validation set. During system operation, an online fine-tuning update was performed every 24 hours using the latest 5,000 data points to suppress long-term drift and maintain prediction accuracy.

[0038] The neural network model supports online updates (fine-tuning): during long-term system operation, the model can be periodically or triggered for fine-tuning based on newly collected temperature-frequency data.

[0039] To ensure long-term prediction accuracy, the system also supports online updates during actual operation. Newly acquired temperature-frequency data is used as incremental samples to periodically or trigger-based fine-tuning of the neural network. This compensates for the impact of magnet aging, environmental changes, and slow drift in system parameters on model accuracy. For example, approximately every 24 hours, the latest 10,000 sets of temperature-frequency samples are used to fine-tune the existing model, thereby compensating for model mismatch caused by slowly changing factors such as magnet aging and environmental variations, and preventing prediction accuracy from decreasing over time.

[0040] Through the modeling and training described above, this embodiment can predict the change in center frequency in the future or at the present moment in real time, given the current temperature characteristics. This provides input for the Z0 coil compensation control.

[0041] S400. Calculate the target compensation magnetic field strength based on the predicted center frequency drift value, and drive the Z0 compensation coil, which is coaxial with the main magnetic field, to generate a compensation magnetic field to counteract the center frequency drift caused by temperature changes.

[0042] In this embodiment, the Z0 compensation coil is driven to perform magnetic field and center frequency compensation control based on the prediction results. Specifically, the target compensation magnetic field strength is calculated based on the predicted center frequency drift value, and a target driving current command for the Z0 compensation coil is generated. A programmable current source outputs a corresponding compensation current to the Z0 compensation coil, generating a compensation magnetic field in the uniform field region of the magnet that is consistent with the direction of the main magnetic field. This counteracts the center frequency drift caused by temperature changes, thereby achieving dynamic and stable control of the center frequency.

[0043] In step S400, according to the formula The center frequency drift prediction value Converted to target magnetic field compensation quantity Where γ is the gyromagnetic ratio of the proton; then, based on the magnetic field-current coefficient of the Z0 compensation coil, Converted into the target drive current.

[0044] This embodiment is based on the relationship between nuclear magnetic resonance frequency and magnetic field strength. , where the gyromagnetic ratio of ¹H The change in center frequency predicted by the neural network Converted into the required magnetic field compensation amount Based on the magnetic field-current coefficient of the compensation coil, The amount of current change that should be adjusted in coil Z0 , k is the magnetic field strength coefficient k (T / A) generated by a unit current in coil Z0, and is given by the initial bias current. Generate new driving current based on A programmable current source is electrically connected to the Z0 coil, outputting a corresponding compensation current to the Z0 coil according to a preset compensation cycle (e.g., updating the current setpoint every 1 second). This causes the Z0 coil to generate a compensation magnetic field in the uniform field region of the magnet, with the direction consistent with the main magnetic field. The compensation magnetic field generated by the Z0 coil is superimposed on the original magnetic field of the magnet, thereby offsetting the center magnetic field drift caused by temperature changes and achieving real-time correction of the center frequency drift. Since the system continuously measures the resonant frequency, the compensation parameters can be further corrected based on the actual frequency feedback when needed, forming a complete link of temperature-frequency-compensation current.

[0045] In some embodiments, a closed-loop calibration step is also included: After applying compensation, the updated center frequency is measured, and the driving current of the Z0 compensation coil is finely adjusted based on the deviation between the measured center frequency and the target center frequency.

[0046] In this embodiment, after applying compensation current to the Z0 compensation coil, the updated center frequency is measured in real time, the measured center frequency is compared with the target center frequency, and the target drive current command is finely adjusted according to the frequency deviation, so as to realize closed-loop control of temperature prediction-magnetic field compensation-frequency feedback.

[0047] The compensation period is set to several seconds to tens of seconds based on the thermal response time constant of the magnet and the frequency stability requirements of the low-field magnetic resonance system, and can be adaptively adjusted according to the center frequency drift rate.

[0048] The method provided in this embodiment is applied to a low-field magnetic resonance analyzer to achieve long-term stable control of the center frequency during sample detection under fluctuating ambient temperature conditions, reducing the center frequency drift from hundreds of Hz in the uncompensated state to within ±5 Hz.

[0049] To verify the compensation effect of the method of the present invention, a typical magnet system was selected: the magnet has a diameter of approximately 230 mm and a height of approximately 180 mm, and the central magnetic field strength is approximately 0.5 T at room temperature of 25 °C, corresponding to a ¹H nuclear magnetic resonance center frequency of approximately 22.475 MHz. During the experiment, the ambient temperature fluctuated naturally within the range of approximately 23.5 °C to 24.5 °C.

[0050] Without the compensation method of this invention, the experimentally recorded center frequency change slowly drifted from 0Hz to -155Hz, with a total frequency drift of about 155Hz and a relative drift of about 7.04ppm, which has significantly affected the long-term measurement accuracy and repeatability of the low-field magnetic resonance system.

[0051] After activating the temperature-neural network-Z0 coil compensation method of this invention, the center frequency was measured again under the same environmental conditions. The results are as follows: Figure 5 As shown, the fluctuation range of the center frequency during the entire experiment was compressed to within ±5Hz (within about 0.5ppm), and the total drift was significantly reduced by about 92.9%, effectively suppressing the influence of temperature fluctuations on the center frequency.

[0052] from Figure 5 It can be seen that, under the premise of ensuring a simple system structure and without requiring significant modifications to the original magnet, this invention achieves efficient compensation for center frequency drift, proving that the technical route based on temperature field monitoring and neural network prediction combined with Z0 coil control has good engineering feasibility and application value.

[0053] It should be noted that the number and arrangement of temperature sensors, the specific number of layers and parameters of the neural network, the number of training samples, the compensation period, and the Z0 coil structure in the above embodiments can all be adjusted and optimized within the scope of the present invention. For example, the number of temperature sensors can be increased to dozens to improve the spatial resolution of the temperature field, the number of hidden layers in the neural network can be increased or the network structure can be changed to adapt to more complex working conditions, and the compensation period can also be shortened or extended according to the dynamic response characteristics of the system. These changes should not be construed as limiting the scope of protection of the present invention.

[0054] The method provided by this invention can be applied to scenarios requiring long-term stable operation, such as low-field magnetic resonance analysis instruments, permanent magnet magnetic field control systems, and long-term online monitoring equipment. Experimental results under typical ambient temperature fluctuation conditions demonstrate that, without requiring additional complex field-frequency interlocking system hardware, this invention can significantly reduce frequency drift with changes in ambient temperature, achieving high frequency stability control under long-term operating conditions.

[0055] Example 2 A temperature-monitoring-based field stabilization system implements the above-described method. For a detailed description of the method, please refer to the corresponding description in the above method embodiments; it will not be repeated here. Figure 6 As shown, the system 500 includes: Temperature sensor array 510 is used to collect temperature data at multiple spatial locations on the permanent magnet in real time; The 520 magnetic resonance probe is used to measure the center frequency of proton resonance signals. The data processing and prediction module 530 is used to preprocess the temperature data and center frequency data, and extract feature vectors representing the spatiotemporal distribution characteristics of the temperature field and the frequency change trend based on the preprocessed data; input the feature vectors into a pre-trained machine learning model, and output the predicted value of the center frequency drift in the future or the current compensation period; wherein, the machine learning model is trained based on historical temperature-frequency data and is used to establish a nonlinear mapping relationship between the spatiotemporal characteristics of the temperature field and the center frequency drift; the compensation control module is used to generate compensation control commands according to the predicted value of the center frequency drift; The programmable current source 540 is communicatively connected to the data processing and prediction module and is used to respond to the compensation control command and output the corresponding compensation current to the Z0 compensation coil. The Z0 compensation coil 550 is arranged at the center of the uniform field region of the permanent magnet and is coaxial with the direction of the main magnetic field. It is used to generate a compensation magnetic field in the same direction as the main magnetic field without destroying the original uniformity of the magnet, so as to achieve active compensation for the center frequency drift.

[0056] The system achieves dynamic compensation of the magnet's center frequency by comprising a compensation process involving temperature-frequency monitoring, data feature extraction, neural network prediction, Z0 coil compensation, and frequency feedback. For a detailed description of the method steps, please refer to the corresponding descriptions in the above method embodiments; they will not be repeated here.

[0057] like Figure 3 As shown, the permanent magnet adopts a cylindrical structure. Multiple temperature sensors are arranged on the magnet, forming a temperature sensor array, to measure the temperature of the magnet at different spatial locations. The sensor sampling frequency is set to 1~10kHz. Specifically, the temperature sensor array is arranged at both ends of the permanent magnet's axis and on opposite radial sides to collect temperature data at key locations in real time. A magnetic resonance probe is arranged within the uniform field region of the permanent magnet to collect proton resonance signals and measure the center frequency.

[0058] The temperature sensor array is connected to the data processing and prediction module through a multi-channel temperature acquisition circuit, and outputs the temperature values ​​of each measuring point in real time. , … (n is the number of sensors). Magnetic resonance probes are arranged inside the magnet to measure the center frequency in real time. The frequency measurement resolution is preferably no less than 0.1Hz, and the measurement frequency is no less than 1Hz. The data processing and prediction module synchronously records the temperature array and the corresponding center frequency measurement values.

[0059] To minimize the impact on the magnet structure, the size and installation method of the Z0 compensation coil are designed with compatibility with the existing magnet structure in mind, avoiding the introduction of additional magnetic materials or mechanical stress, thereby ensuring that the field uniformity of the magnet itself is not compromised.

[0060] Specifically, a Z0 compensation coil is placed at the axial center of the uniform field region of the magnet, and its arrangement is as follows: Figure 3 As shown. The Z0 compensation coil is preferably fabricated using flexible circuit board printing or wire winding. The shape, number of turns, and distribution (spatial position) of the coil can be optimized based on the magnetic field simulation results to generate a uniform axial compensation magnetic field in the target area. This ensures that the generated compensation magnetic field is as uniform as possible within the sample area, improving the spatial uniformity and efficiency of the compensation magnetic field. It provides a highly efficient compensation magnetic field without compromising the original uniformity of the magnet, and its direction is coaxial with the main magnetic field of the magnet.

[0061] The data processing and prediction module integrates a temperature acquisition circuit, a frequency measurement circuit, a data preprocessing unit, a feature extraction unit, and the machine learning model. It is connected to the programmable current source via wired or wireless communication to form an integrated intelligent compensation control device system, making the entire device compact, easy to install, and highly automated.

[0062] In some embodiments, the system further includes a human-machine interface for setting system parameters and displaying operational information such as temperature distribution, center frequency drift, and compensation status. The system parameters include the target center frequency, compensation period, model update strategy, and alarm threshold.

[0063] The system is used in low-field magnetic resonance analyzers, permanent magnet magnetic field control systems, or long-term online monitoring equipment to maintain the long-term stability of the magnet's center frequency under fluctuating ambient temperature conditions.

[0064] Example 3 A computer device 600, such as Figure 7 As shown, the system includes a memory 610, a processor 620, and a computer program 630 stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a field stabilization method based on temperature monitoring. For a detailed description of the method, please refer to the corresponding description in the above method embodiments; it will not be repeated here.

[0065] Example 4 A computer-readable storage medium, such as Figure 8 As shown, a computer program is stored thereon, which, when executed by a processor, implements the steps of a field stabilization method based on temperature monitoring. For a detailed description of the method, please refer to the corresponding description in the above method embodiments, and will not be repeated here.

[0066] The number of devices and processing scale described herein are for the purpose of simplifying the description of the invention. Applications, modifications, and variations of the invention will be readily apparent to those skilled in the art.

[0067] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

[0068] The apparatus, computer device, and non-volatile computer storage medium and method provided in the embodiments of this specification are corresponding. Therefore, the apparatus, computer device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, computer device, and non-volatile computer storage medium will not be repeated here.

[0069] Those skilled in the art will also know that, besides implementing the controller in the form of purely computer-readable program code, the same functions can be achieved by logically programming the method steps, making the controller take the form of logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers (PLCs), and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the devices included within it for implementing various functions can also be considered structures within that hardware component. Alternatively, the devices for implementing various functions can be considered as both software units implementing the method and structures within a hardware component.

[0070] The systems, apparatuses, or units described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. For ease of description, the above apparatuses are described separately as various units based on their functions. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0071] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0072] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0073] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0074] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0075] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0076] This specification may be described in the general context of computer-executable instructions, such as program units, that are executed by a computer. Generally, program units include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification may also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program units may reside in local and remote computer storage media, including storage devices.

[0077] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0078] The above description is merely an embodiment of this specification and is not intended to limit the scope of one or more embodiments of this specification. Various modifications and variations can be made to one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of one or more embodiments of this specification.

Claims

1. A field stabilization method based on temperature monitoring, characterized in that, Includes the following steps: Temperature data at multiple spatial locations on the permanent magnet are collected in real time, and the center frequency of the proton resonance signal is measured using a magnetic resonance probe. The temperature data and center frequency data are preprocessed, and feature vectors characterizing the spatiotemporal distribution of the temperature field and the frequency variation trend are extracted based on the preprocessed data. The feature vector is input into a pre-trained machine learning model, which outputs a predicted value of the center frequency drift in the future or the current compensation period; wherein, the machine learning model is trained based on historical temperature-frequency data and is used to establish a nonlinear mapping relationship between the spatiotemporal characteristics of the temperature field and the center frequency drift. The target compensation magnetic field strength is calculated based on the predicted center frequency drift value, and the Z0 compensation coil, which is coaxial with the main magnetic field, is driven to generate a compensation magnetic field to counteract the center frequency drift caused by temperature changes.

2. The field stabilization method based on temperature monitoring as described in claim 1, characterized in that, Temperature data is collected by an array of temperature sensors arranged at both ends of the permanent magnet in the axial direction and on the opposite side in the radial direction. The temperature sensor array includes at least four temperature sensors.

3. The field stabilization method based on temperature monitoring as described in claim 1, characterized in that, The preprocessing includes at least time alignment, noise reduction, and interpolation. The extraction of the feature vector includes constructing a sample sequence of temperature field time-series data and frequency change data based on a preset time window, and calculating the statistical characteristics and / or spatial distribution characteristics of the temperature data.

4. The field stabilization method based on temperature monitoring as described in claim 3, characterized in that, The step of preprocessing the temperature data and center frequency data, and extracting feature vectors characterizing the spatiotemporal distribution features and frequency variation trends of the temperature field based on the preprocessed data, further includes: The features of the temperature data are subjected to dimensionality reduction processing, and the principal components after dimensionality reduction are used as input features of the machine learning model.

5. The field stabilization method based on temperature monitoring as described in claim 1, characterized in that, The machine learning model is specifically a neural network model, which is a hybrid structure containing a temperature feature extraction neural network and a time series analysis neural network. The temperature feature extraction neural network is used to extract long-range dependency features of the temperature field sequence, and the time series analysis neural network is used to fuse and model multi-scale time series information.

6. The field stabilization method based on temperature monitoring as described in claim 5, characterized in that, The neural network model supports online updates: during system operation, the model is periodically or triggered for fine-tuning based on newly collected temperature-frequency data.

7. The field stabilization method based on temperature monitoring as described in claim 1, characterized in that, It also includes a closed-loop calibration step: After applying compensation, the updated center frequency is measured, and the driving current of the Z0 compensation coil is finely adjusted based on the deviation between the measured center frequency and the target center frequency.

8. The field stabilization method based on temperature monitoring as described in claim 1, characterized in that, In the step of calculating the target compensation magnetic field strength based on the predicted center frequency drift value, and driving the Z0 compensation coil coaxial with the main magnetic field to generate a compensation magnetic field to counteract the center frequency drift caused by temperature changes, the formula is used. The center frequency drift prediction value Converted to target magnetic field compensation quantity Where γ is the gyromagnetic ratio of the proton; then, based on the magnetic field-current coefficient of the Z0 compensation coil, Converted into the target drive current.

9. A field stabilization system based on temperature monitoring, implementing the method as described in any one of claims 1 to 8, characterized in that, include: A temperature sensor array is used to collect temperature data at multiple spatial locations on the permanent magnet in real time. A magnetic resonance probe is used to measure the center frequency of a proton resonance signal. The data processing and prediction module is used to preprocess the temperature data and center frequency data, and extract feature vectors representing the spatiotemporal distribution characteristics of the temperature field and the frequency change trend based on the preprocessed data; input the feature vectors into a pre-trained machine learning model, and output the predicted value of the center frequency drift in the future or the current compensation period; wherein, the machine learning model is trained based on historical temperature-frequency data, and is used to establish a nonlinear mapping relationship between the spatiotemporal characteristics of the temperature field and the center frequency drift; the compensation control module is used to generate compensation control commands according to the predicted value of the center frequency drift; A programmable current source is used to respond to the compensation control command and output a corresponding compensation current; The Z0 compensation coil is arranged in the uniform field region of the permanent magnet and is used to generate a compensation magnetic field coaxial with the main magnetic field under the drive of the compensation current.

10. A field stabilization system based on temperature monitoring as described in claim 9, characterized in that, The data processing and prediction module integrates a temperature acquisition circuit, a frequency measurement circuit, a data preprocessing unit, a feature extraction unit, and the machine learning model.

11. A field stabilization system based on temperature monitoring as described in claim 9, characterized in that, The Z0 compensation coil is fabricated using a printed circuit board or wire winding. Its coil shape, number of turns, and spatial position are optimized to generate a uniform axial compensation magnetic field in the target area.

12. A field stabilization system based on temperature monitoring as described in claim 9, characterized in that, It also includes a human-machine interface for setting system parameters and displaying temperature distribution, center frequency drift, and compensation status information.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1 to 8.

Images