All-weather mobile magnetic resonance permanent magnet temperature control device and method

Abstract

The invention discloses an all-weather mobile magnetic resonance permanent magnet temperature control device and method, and relates to the technical field of magnetic resonance permanent magnet temperature control. The all-weather mobile magnetic resonance permanent magnet temperature control device comprises a control assembly, an amplification assembly, a heating assembly, a measurement assembly, a first permanent magnet and a second permanent magnet; the control assembly amplifies voltage through the first amplification assembly and the second amplification assembly and is used for controlling the first heating assembly and the second heating assembly, and the first heating assembly and the second heating assembly adjust the temperature of the first permanent magnet and the temperature of the second permanent magnet respectively. According to the all-weather mobile magnetic resonance permanent magnet temperature control device and method, the temperature fluctuation of a magnetic resonance permanent magnet is smaller than 0.05 DEG C under the condition that the environment temperature fluctuates by more than ten degrees, and the magnetic resonance permanent magnet can still keep the high-stability temperature condition under the natural condition that the environment temperature fluctuates greatly and rapidly.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature control of magnetic resonance permanent magnets, and in particular to an all-weather mobile magnetic resonance permanent magnet temperature control device and method. Background Art

[0002] As a high-precision analysis tool, magnetic resonance technology has been widely used in many fields such as chemistry, physics, biology, and medicine for molecular structure analysis, substance composition detection, and disease diagnosis. However, the magnetic properties of the permanent magnet in the magnetic resonance system are extremely sensitive to temperature changes. Temperature fluctuations will cause changes in the magnetic field strength of the magnet, thereby affecting the stability of the magnetic resonance signal and the accuracy of measurement. Therefore, ensuring the temperature stability of the permanent magnet is the key to improving the performance of the magnetic resonance system and ensuring the reliability of experimental results.

[0003] Traditional magnetic resonance systems usually operate in a constant temperature environment. For example, in the article "Design and Implementation of a Precision Temperature Controller for NMR Permanent Magnets", the ambient temperature fluctuation of the magnet operation is only about 2°C, and the magnet is placed in a heat insulation chamber. The inner loop control circuit in the dual-loop temperature control algorithm is used to control and adjust the air temperature to further improve the working environment of the magnetic resonance permanent magnet. In the patent application with the patent application number CN201710626836.6 and the name "A Temperature Control Device for an Industrial Nuclear Magnetic Resonance Permanent Magnet System", a three-stage temperature control method is adopted. The first stage stabilizes the temperature at a certain temperature close to room temperature, the second stage controls the temperature at a certain constant temperature near the working temperature of the magnet, and the third stage controls the magnet temperature to the target working temperature. These actually regulate the ambient temperature or increase the microenvironment, directly resulting in an increase in equipment cost and a too complex system. Another example is the invention patent with the patent application number CN202111388986. and the name "An Event-Triggered Fuzzy Neural Network Temperature Control System and Method", which adopts a cascade control loop combining an outer loop control circuit and an inner loop control circuit. Rapid changes in the external temperature are likely to cause self-excitation of the cascaded multi-loop system, and the environmental adaptability is poor.

[0004] In order to adapt to the all-weather environment with large and rapid temperature fluctuations and realize an extremely low-cost mobile magnetic resonance main magnet, the present application will greatly simplify the system structure and achieve a high-stability temperature control system for the magnetic resonance main magnet without an external air conditioner by designing a better control algorithm. Summary of the Invention

[0005] The purpose of the present invention is to provide an all-weather mobile magnetic resonance permanent magnet temperature control device and method, which solves the problem that the existing temperature control technology is difficult to meet the high-stability requirements of the magnet temperature when the magnetic resonance equipment operates in the natural environment, so that when the ambient temperature fluctuates by more than ten degrees, the magnet temperature fluctuation is less than 0.05 degrees Celsius.

[0006] To achieve the above object, the present invention provides an all-weather mobile magnetic resonance permanent magnet temperature control device, which includes a control component, an amplification component, a heating component, a measurement component, a first permanent magnet and a second permanent magnet. The temperature of the second permanent magnet and the first permanent magnet is measured by the measurement component, and the control component amplifies the voltage through the first amplification component and the second amplification component respectively to control the first heating component and the second heating component. The first heating component and the second heating component adjust the temperatures of the first permanent magnet and the second permanent magnet respectively.

[0007] Preferably, the control component controls the voltage through a neural network, and the neural network includes an input layer, a hidden layer and an output layer;

[0008] The inputs of the input layer include the target temperature, the ambient temperature, the permanent magnet temperature, and the bias;

[0009] The outputs of the output layer include multiple groups of PIDs. The number of PIDs is twice the number of temperature control objects, and the number of PIDs is at least the number of temperature control objects plus 1.

[0010] Preferably, the heating component includes a first heating component and a second heating component. Both the second heating component and the first heating component are heating wires. The heating wires are placed inside the first permanent magnet and the second permanent magnet, and adiabatic materials are wrapped around the outsides of the second permanent magnet and the first permanent magnet.

[0011] Preferably, the amplification device includes a first amplification component and a second amplification component. Both the second amplification component and the first amplification component are adjustable power supplies.

[0012] Preferably, the measurement component includes a first temperature probe for measuring the temperature of the first permanent magnet, a second temperature probe for measuring the temperature of the second permanent magnet, and an ambient temperature probe for measuring the ambient temperature.

[0013] A method for an all-weather mobile magnetic resonance permanent magnet temperature control device includes the following steps:

[0014] Step S1, set the target temperatures x(k) of the first permanent magnet and the second permanent magnet;

[0015] Step S2, use the target temperature x(k), the ambient temperature s(k) measured by the ambient temperature probe, the actual temperature y1(k) of the first permanent magnet measured by the first temperature probe, and the actual temperature y2(k) of the second permanent magnet measured by the second temperature probe as the inputs of the input layer of the neural network;

[0016] Step S3, after the input layer of the neural network receives the inputs, perform calculations to obtain n groups of PID parameters and output them through the output layer;

[0017] Step S4, update the weights of the neural network for the next iterative calculation;

[0018] Step S5: The control component uses the n sets of PID parameters output by the neural network in Step S3 for the fluctuation of the environmental temperature and the deviation between the temperature of the permanent magnet and the target temperature, and jointly calculates the control voltage;

[0019] Step S6: The two control voltages obtained in Step S5 are respectively passed through the first amplification component and the second amplification component to enhance the driving ability of the control component;

[0020] Step S7: The amplified control voltages in Step S6 are used to control the first heating component and the second heating component respectively to adjust the temperatures of the first permanent magnet and the second permanent magnet;

[0021] Step S8: Repeat Step S2 to Step S7 to make the temperatures of the first permanent magnet and the second permanent magnet stable within the error range.

[0022] Preferably, the input of the input layer of the neural network in Step S2 is as shown in the following formula:

[0023] I(k) = [x(k), y1(k), y2(k), s(k), ……, 1] (1)

[0024] In the formula, I(k) represents the input of the input layer of the neural network, which is a 1*j-dimensional row vector, j represents the number of neurons at the input end of the neural network, x(k) represents the target temperature, y1(k) represents the temperature of the first permanent magnet, y2(k) represents the temperature of the second permanent magnet, s(k) represents the environmental temperature, 1 is used as a bias, and k represents the number of steps of iterative calculation, always corresponding to the current moment.

[0025] Preferably, the output of the output layer of the neural network in Step S3 is as shown in the following formula:

[0026]

[0027] In the formula, O(k) represents the output of the output layer, and the output result is n sets of PID parameters. n is equal to twice the number of temperature control objects. w1 represents the input weight, w2 represents the output weight, a and b can take real numbers greater than 1, usually taking the natural constant e. I(k) represents the input of the input layer of the neural network, and k represents the number of steps of iterative calculation, always corresponding to the current moment.

[0028] Preferably, the calculation formula for updating the weight in Step S4 is as follows:

[0029]

[0030] In the formula, w m (k - 1) represents the weight of the previous iteration, w m (k - 2) represents the weight of the iteration before the previous step, represents the loss function with respect to w mThe partial derivative, η represents the learning rate, α represents the inertia coefficient, m taking 1 and m taking 2 respectively represent the input weight and the output weight;

[0031] The expression of the loss function is as follows:

[0032]

[0033] In the formula, loss represents the loss function, x(k) represents the target value, y1(k) represents the temperature of the first permanent magnet, and y2(k) represents the temperature of the second permanent magnet.

[0034] Preferably, the calculation formula of the control voltage in the step S5 is as follows:

[0035]

[0036] In the formula, U i represents the control voltage of the i-th magnet, error i represents the deviation between the magnet temperature and the target temperature, i takes 1, 2; kpid i,1 corresponds to the PID coefficient of the magnet temperature deviation, kpid i,2 corresponds to the PID coefficient of the ambient temperature deviation, error3 represents the fluctuation of the ambient temperature, ∫error represents the integral of the temperature deviation or fluctuation, represents the differential of the temperature deviation or fluctuation.

[0037] Therefore, the present invention adopts the above-mentioned all-weather mobile magnetic resonance permanent magnet temperature control device and method, so that when the ambient temperature fluctuates by more than ten degrees, the temperature fluctuation of the magnetic resonance permanent magnet is less than 0.05 degrees Celsius, and the magnetic resonance permanent magnet can still maintain a highly stable temperature condition under the natural condition of large and rapid fluctuations in the ambient temperature.

[0038] The following will further describe the technical solutions of the present invention in detail through the drawings and embodiments. Brief Description of the Drawings

[0039] Figure 1 is a schematic structural diagram of an embodiment of an all-weather mobile magnetic resonance permanent magnet temperature control device and method of the present invention;

[0040] Figure 2 is a schematic structural diagram of a neural network of an embodiment of an all-weather mobile magnetic resonance permanent magnet temperature control device and method of the present invention;

[0041] Figure 3 is a measured experimental data diagram of an embodiment of an all-weather mobile magnetic resonance permanent magnet temperature control device and method of the present invention;

[0042] Reference numerals: 1, measurement component; 2, first permanent magnet; 3, first temperature probe; 4, first heating component; 5, ambient temperature probe; 6, second permanent magnet; 7, second temperature probe; 8, second heating component; 9, first amplification component; 10, second amplification component; 11, control component; 12, input layer; 13, hidden layer; 14, output layer; 15, first temperature PID controller; 16, first ambient temperature fluctuation PID controller; 17, second ambient temperature fluctuation PID controller; 18, second temperature PID controller; 19, control voltage; 191, first control voltage; 192, second control voltage; 20, neural network; 21, target temperature; 22, first permanent magnet temperature; 23, ambient temperature; 24, second permanent magnet temperature; 25, bias. Detailed implementation

[0043] The technical solution of the present invention will be further described below with reference to the drawings and embodiments.

[0044] Unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meaning understood by those with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. The terms such as "including" or "comprising" mean that the elements or objects appearing before this word cover the elements or objects listed after this word and their equivalents, without excluding other elements or objects. The terms such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left", "right" are only used to represent relative position relationships, and when the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0045] Embodiment

[0046] Please refer to Figures 1-3 , the present invention provides an all-weather mobile magnetic resonance permanent magnet temperature control device, including a control component 11, an amplification component, a heating component, a measurement component 1, a first permanent magnet 2 and a second permanent magnet 6. The temperature of the second permanent magnet 6 and the first permanent magnet 2 is measured by the measurement component 1, and the control component 11 amplifies the voltage through the first amplification component 9 and the second amplification component 10 respectively to control the first heating component 4 and the second heating component 8. The first heating component 4 and the second heating component 8 respectively adjust the temperatures of the first permanent magnet 2 and the second permanent magnet 6.

[0047] The control component 11 controls the voltage through the neural network 20, and the neural network 20 includes an input layer 12, a hidden layer 13 and an output layer 14. As Figure 1As shown, the inputs of the input layer 12 include the target temperature 21, the temperature of the first permanent magnet 22, the ambient temperature 23, the temperature of the second permanent magnet 24, and the bias 25. The outputs of the output layer 14 include multiple sets of PIDs. The number of PIDs is twice the number of temperature control objects, and the number of PIDs is at least the number of temperature control objects plus 1.

[0048] The heating assembly includes a first heating assembly 4 and a second heating assembly 8. Both the second heating assembly 8 and the first heating assembly 4 are heating wires. The heating wires are placed inside the first permanent magnet 2 and the second permanent magnet 6, and both the outside of the second permanent magnet 6 and the first permanent magnet 2 are wrapped with a certain thickness of heat insulation material to reduce the energy consumption of temperature control.

[0049] The amplification device includes a first amplification assembly 9 and a second amplification assembly 10. Both the second amplification assembly 10 and the first amplification assembly 9 are adjustable power supplies.

[0050] The measurement assembly 1 includes a first temperature probe 3 for measuring the temperature of the first permanent magnet 2, a second temperature probe 7 for measuring the temperature of the second permanent magnet 6, and an ambient temperature probe 5 for measuring the ambient temperature.

[0051] The method for the above high-stability magnetic resonance permanent magnet constant temperature control device includes the following steps:

[0052] Step S1, as Figure 1 shown, set the target temperatures x(k) of the first permanent magnet 2 and the second permanent magnet 6;

[0053] Step S2, use the target temperature x(k), the ambient temperature s(k) measured by the ambient temperature probe 5, the actual temperature y1(k) of the first permanent magnet 2 measured by the first temperature probe 3, and the actual temperature y2(k) of the second permanent magnet 6 measured by the second temperature probe 7 as the inputs of the input layer 12 of the neural network 20;

[0054] The inputs of the input layer 12 of the neural network 20 are shown as follows:

[0055] I1(k) = [x(k), y1(k), y2(k), s(k), ……, 1] (1)

[0056] In the formula, I1(k) represents the input of the input layer 12 of the neural network 20. It is a 1*j-dimensional row vector, where j represents the number of neurons at the input end of the neural network 20. x(k) represents the target temperature, y1(k) represents the temperature of the first permanent magnet 2, y2(k) represents the temperature of the second permanent magnet 6, s(k) represents the ambient temperature, 1 serves as the bias 25, and k represents the number of steps of iterative calculation, always corresponding to the current moment.

[0057] Step S3, after the input layer 12 of the neural network 20 receives the input, it performs calculations to obtain n sets of PID parameters and outputs them through the output layer 14.

[0058] The output of the output layer of the neural network in step S3 is shown as follows:

[0059]

[0060] In the formula, O(k) represents the output of the output layer 14, and the output result is n groups of PID parameters. n is equal to twice the number of temperature control objects. w1 represents the input weight, w2 represents the output weight. a and b can take real numbers greater than 1, usually taking the natural constant e. I(k) represents the input of the input layer of the neural network, and k represents the number of steps of iterative calculation, always corresponding to the current moment.

[0061] Step S4: Update the weights of the neural network 20 for the next iterative calculation;

[0062] The calculation formula for updating the weights is as follows:

[0063]

[0064] In the formula, w m (k - 1) represents the weight of the previous iteration, w m (k - 2) represents the weight of the iteration before the previous one, represents the partial derivative of the loss function with respect to w m , η represents the learning rate, α represents the inertia coefficient, m taking 1 and m taking 2 respectively represent the input weight and the output weight;

[0065] The expression of the loss function is as follows:

[0066]

[0067] In the formula, loss represents the loss function, x(k) represents the target value, y1(k) represents the temperature of the first permanent magnet 2, and y2(k) represents the temperature of the second permanent magnet 6.

[0068] Step S5: The control component 11 uses the n groups of PID parameters output by the neural network 20 in step S3 for the fluctuations of the ambient temperature 23, the deviation between the temperature of the first permanent magnet 22, the temperature of the second permanent magnet 24 and the target temperature 21, and jointly calculates the control voltage 19;

[0069] The calculation formula for the control voltage 19 is as follows:

[0070]

[0071] In the formula, U i represents the control voltage 19 of the i-th magnet, error i represents the deviation between the magnet temperature and the target temperature, i takes 1, 2; kpid i,1 corresponds to the PID coefficient of the magnet temperature deviation, kpid i,2The PID coefficients corresponding to the ambient temperature deviation, error3 represents the fluctuation of the ambient temperature, and ∫error represents the integral of the temperature deviation or fluctuation. represents the differential of the temperature deviation or fluctuation.

[0072] The first temperature PID controller 15 of the permanent magnet 1-2 in this embodiment is expressed as:

[0073]

[0074] In the formula, Val1 represents the first temperature PID controller 15 of the permanent magnet 1-2, error1 represents the deviation between the temperature of the permanent magnet 1-2 and the target temperature, and ∫error1 represents the integral of the temperature deviation of the permanent magnet 1-2. represents the differential of the temperature deviation of the permanent magnet 1-2.

[0075] The first ambient temperature fluctuation PID controller 16 and the second ambient temperature fluctuation PID controller 17 can both be expressed as:

[0076]

[0077] Among them, Val2 represents the first ambient temperature fluctuation PID controller 16, Val3 represents the second ambient temperature fluctuation PID controller 17, error3 represents the fluctuation of the ambient temperature, and ∫error3 represents the integral of the ambient temperature fluctuation. represents the differential of the ambient temperature fluctuation.

[0078] The second temperature PID controller 18 of the permanent magnet 2-6 is expressed as:

[0079]

[0080] Among them, Val4 represents the second temperature PID controller 18 of the permanent magnet 2-6, error2 represents the deviation between the temperature of the permanent magnet 2-6 and the target temperature, and ∫error2 represents the integral of the temperature deviation of the permanent magnet 2-6. represents the differential of the temperature deviation of the permanent magnet 2-6.

[0081] U1 = kpid 1,1 *Val1 + kpid 1,2 *Val2 (9)

[0082] Among them, U1 represents the first control voltage 191 of the permanent magnet 1-2 calculated, and kpid 1,1 represents the first set of PID parameters output by the neural network 20, and kpid 1,2 represents the second set of PID parameters output by the neural network 20, and Val1 and Val2 are respectively formulas (6) and (7).

[0083] U2 = kpid 2，2 *Val3 + kpid 2，1 *Val4 (10)

[0084] Among them, U2 represents the second control voltage 192 of the permanent magnet two 6 obtained by calculation, and kpid 2,2 represents the third set of PID parameters output by the neural network 20, and kpid 2,1 represents the fourth set of PID parameters output by the neural network 20. Val3 and Val4 are respectively formula (7) and formula (8).

[0085] Step S6: Respectively pass the two-way control voltages obtained in step S5 through the first amplification component 9 and the second amplification component 10 to enhance the driving ability of the control component 11;

[0086] Step S7: Use the control voltages amplified in step S6 to control the first heating component 4 and the second heating component 8 respectively to adjust the temperatures of the permanent magnet one 2 and the permanent magnet two 6;

[0087] Step S8: Repeat step S2 to step S7 to make the temperatures of the permanent magnet one 2 and the permanent magnet two 6 stable within the error range.

[0088] The number of PID groups n output by the neural network 20 is determined according to the number of temperature control objects. The algorithm of the neural network 20 in this embodiment is only for illustrative purposes.

[0089] The actual measurement results of this embodiment are as Figure 3 shown. When the ambient temperature fluctuates by more than ten degrees, the temperature fluctuation of the magnet is less than 0.05 degrees Celsius.

[0090] Therefore, the present invention adopts the above-mentioned all-weather mobile magnetic resonance permanent magnet temperature control device and method, so that when the ambient temperature of the magnetic resonance permanent magnet fluctuates by more than ten degrees, the temperature fluctuation of the magnet is less than 0.05 degrees Celsius, enabling the magnetic resonance permanent magnet to maintain a highly stable temperature condition under the natural condition of large and rapid fluctuations in the ambient temperature.

[0091] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that: they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. An all-weather mobile magnetic resonance permanent magnet temperature control device, characterized in that: It includes a control component, an amplification component, a heating component, a measurement component, a first permanent magnet and a second permanent magnet. The temperature of the second permanent magnet and the first permanent magnet is measured by the measurement component. The control component amplifies the voltage through a first amplification component and a second amplification component respectively to control the first heating component and the second heating component. The first heating component and the second heating component adjust the temperatures of the first permanent magnet and the second permanent magnet respectively.

2. The all-weather mobile magnetic resonance permanent magnet temperature control device according to claim 1, characterized in that: The control component controls the voltage through a neural network, and the neural network includes an input layer, a hidden layer and an output layer; The inputs of the input layer include the target temperature, the ambient temperature, the permanent magnet temperature, and the bias; The outputs of the output layer include multiple groups of PIDs. The number of PIDs is twice the number of temperature control objects, and the number of PIDs is at least the number of temperature control objects plus 1.

3. An all-weather mobile magnetic resonance permanent magnet temperature control device according to claim 2, characterized in that: The heating component includes a first heating component and a second heating component. Both the second heating component and the first heating component are heating wires. The heating wires are placed inside the first permanent magnet and the second permanent magnet, and the outsides of the second permanent magnet and the first permanent magnet are both wrapped with heat-insulating materials.

4. The all-weather mobile magnetic resonance permanent magnet temperature control device according to claim 3, characterized in that: The amplification device includes a first amplification component and a second amplification component. Both the second amplification component and the first amplification component are adjustable power supplies.

5. The all-weather mobile magnetic resonance permanent magnet temperature control device according to claim 4, characterized in that: The measurement component includes a first temperature probe for measuring the temperature of the first permanent magnet, a second temperature probe for measuring the temperature of the second permanent magnet, and an ambient temperature probe for measuring the ambient temperature.

6. A method for applying an all-weather mobile magnetic resonance permanent magnet temperature control device according to any one of claims 1-5, characterized in that, It includes the following steps: Step S1: Set the target temperatures x(k) of the first permanent magnet and the second permanent magnet; Step S2: Take the target temperature x(k), the ambient temperature s(k) measured by the ambient temperature probe, the actual temperature y1(k) of the first permanent magnet measured by the first temperature probe, and the actual temperature y2(k) of the second permanent magnet measured by the second temperature probe as the inputs of the input layer of the neural network; Step S3: After the input layer of the neural network receives the inputs, perform calculations to obtain n groups of PID parameters and output them through the output layer; Step S4: Update the weights of the neural network for the next iterative calculation; Step S5: The control component uses the n groups of PID parameters output by the neural network in Step S3 for the fluctuations of the ambient temperature and the deviations between the permanent magnet temperatures and the target temperatures respectively, and jointly calculates the control voltage; Step S6: Pass the two paths of control voltages obtained in Step S5 through the first amplification component and the second amplification component respectively to enhance the driving ability of the control component; Step S7: Use the control voltages amplified in Step S6 to control the first heating component and the second heating component respectively to adjust the temperatures of the first permanent magnet and the second permanent magnet; Step S8: Repeat Step S2 to Step S7 to make the temperatures of the first permanent magnet and the second permanent magnet stable within the error range.

7. A method for an all-weather mobile magnetic resonance permanent magnet temperature control device according to claim 6, characterized in that, The inputs of the input layer of the neural network in Step S2 are shown as follows: I(k) = [x(k), y1(k), y2(k), s(k), ……, 1] (1) In the formula, I(k) represents the input of the input layer of the neural network, which is a 1*j-dimensional row vector. j represents the number of neurons at the input end of the neural network. x(k) represents the target temperature, y1(k) represents the temperature of the first permanent magnet, y2(k) represents the temperature of the second permanent magnet, s(k) represents the ambient temperature, 1 is used as the bias, and k represents the number of steps of iterative calculation, always corresponding to the current moment.

8. A method for an all-weather mobile magnetic resonance permanent magnet temperature control device according to claim 6, characterized in that, The output of the output layer of the neural network in step S3 is shown as follows: In the formula, O(k) represents the output of the output layer, and the output result is n groups of PID parameters. n is equal to twice the number of temperature control objects. w1 represents the input weight, w2 represents the output weight. a and b can take real numbers greater than 1, usually taking the natural constant e. I(k) represents the input of the input layer of the neural network, and k represents the number of steps of iterative calculation, always corresponding to the current moment.

9. The method of an all-weather mobile magnetic resonance permanent magnet temperature control device according to claim 6, characterized in that, The calculation formula for updating the weight in step S4 is as follows: where, w m (k - 1) represents the weight of the previous iteration, w m (k - 2) represents the weight of the iteration before the previous one, represents the partial derivative of the loss function with respect to w m , η represents the learning rate, α represents the inertia coefficient, m taking 1 and m taking 2 respectively represent the input weight and the output weight; The expression of the loss function is as follows: In the formula, loss represents the loss function, x(k) represents the target value, y1(k) represents the temperature of the first permanent magnet, and y2(k) represents the temperature of the second permanent magnet.

10. A method for an all-weather mobile magnetic resonance permanent magnet temperature control device according to claim 6, characterized in that, The calculation formula for the control voltage in step S5 is as follows: Wherein, U i represents the control voltage of the i-th magnet, error i represents the deviation between the magnet temperature and the target temperature, and i takes 1, 2; kpid i,1 corresponds to the PID coefficient of the magnet temperature deviation, kpid i,2 corresponds to the PID coefficient of the ambient temperature deviation, error3 represents the fluctuation of the ambient temperature, and ∫error represents the integral of the temperature deviation or fluctuation, represents the differential of the temperature deviation or fluctuation.

Images