Decoupling method of multi-degaussing power supply system

Abstract

The invention discloses a decoupling method of a multi-degaussing power supply system, which relates to the technical field of power supply systems and realizes decoupling control among degaussing power supply modules by combining a BP neural network and PID control. Proportion, integral and differential parameters of each PID controller are adjusted in real time through a BP neural network, so that the output current of each degaussing power supply module is converged to a target value, and the degaussing precision and stability of the system are improved; dynamic proportional adjustment and variable speed integration are adopted, and the response speed and the interference immunity of the system are enhanced; the method is suitable for a complex coupling multiple-input-multiple-output system, can effectively improve the synchronism and overall performance of a degaussing system, and is suitable for ship body degaussing requirements in different magnetic field environments.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply systems, and more particularly to a decoupling method for a multi-demagnetization power supply system. Background Art

[0002] With the continuous increase in the size of ships, the magnetic field distribution around ships becomes increasingly complex, which poses great difficulties to the design of demagnetization systems. In this case, if the traditional method of using a single demagnetization winding to demagnetize the hull is still adopted, it will be impossible to effectively eliminate the ship's magnetic field, and it is also difficult to adapt to the current production mode of ship hull block construction. To solve these problems, researchers have proposed a technical solution of zoned demagnetization, that is, dividing a single large coil into several zones, each zone having a set of demagnetization power supplies, and an independent loop is formed between the power supply and the coil. By using multiple sets of demagnetization power supplies to jointly demagnetize the hull, not only the demagnetization effect on the ship's magnetic field is improved, but also the requirements of ship hull block construction are met, and the reliability and maintainability of the demagnetization system are enhanced. However, when multiple sets of demagnetization power supplies are used for joint demagnetization, due to the mutual inductance coupling effect between the coils, the entire system behaves as a coupled multi-input multi-output system, which will affect the dynamic and steady-state performance and current synchronization performance of the demagnetization current output by each demagnetization power supply, and further affect the overall demagnetization quality of the system. Therefore, in order to ensure the demagnetization effect when the demagnetization power supplies are jointly demagnetized, it is necessary to perform decoupling control on the multi-demagnetization power supply system.

[0003] At present, there are few domestic and foreign literatures related to the decoupling control of multi-demagnetization power supply systems. However, for such complex coupled systems with multiple variables and non-linearity, domestic and foreign researchers have proposed many decoupling control methods, such as traditional decoupling control, adaptive decoupling control, fuzzy decoupling control, and neural network decoupling control. Among them, neural network decoupling control uses the non-linear mapping and adaptive adjustment capabilities of neural networks to model and learn the controlled system, and then realizes the decoupling control of the controlled system. However, this idea is only in its infancy. Usually, neural networks cannot be used alone as a controller to achieve decoupling control, and additional other controllers are required. At present, there is a lack of practical solutions. Summary of the Invention

[0004] The purpose of the present invention is to propose a decoupling method for a multi-demagnetization power supply system, which solves the problem that the dynamic and steady-state performance and current synchronization performance of the demagnetization current output by each demagnetization power supply are affected in the case of joint demagnetization of multiple sets of demagnetization power supplies, and improves the overall demagnetization quality of the system.

[0005] The technical solution of the present invention is to provide a decoupling method for a multi-demagnetization power supply system, which includes:

[0006] Design a decoupling controller composed of multiple BP-PID controllers in parallel. The decoupling controller dynamically adjusts the proportional, integral, and derivative coefficients of each PID controller through a BP neural network;

[0007] The decoupling method using the decoupling controller includes the following steps:

[0008] S1. Initialize the BP neural network structure of the system. Divide the neural network into an input layer, a hidden layer, and an output layer. The input layer receives the deviation signal data and transfers it to the hidden layer. After processing by the hidden layer, the data is transferred to the output layer, and the output layer generates the output signals for the proportional K p , integral K i and derivative K d parameters;

[0009] S2. Establish the system error function:

[0010]

[0011] where i nref (k) and i n (k) are respectively the current given value and the actual output current value at the k-th sampling moment of the n-th power supply module. Adjust the weights of the BP neural network through the gradient descent algorithm to minimize the error function E(k) and gradually converge the system deviation to zero;

[0012] S3. Use the BP neural network to generate the dynamically adjusted proportional coefficient K p , integral coefficient K i and derivative coefficient K d and input them into the PID controller to automatically update the parameters of the PID controller according to the deviation value of the system;

[0013] S4. Use the dynamic proportional coefficient adjustment formula in the PID controller:

[0014]

[0015] where is the updated proportional coefficient of the n-th PID controller, α1 is the adjustment factor of the proportional coefficient, and e n (k) represents the current deviation signal;

[0016] The degaussing power supply module increases the proportional coefficient when the deviation is large and decreases the proportional coefficient when the deviation is small, so as to achieve fast control of the current response and prevent overshoot of the output current;

[0017] S5. Introduce variable-speed integration in the integral link of the PID controller. The output formula of the integral control signal is:

[0018]

[0019] wherein, u in (k) and are respectively the voltage control signals output before and after the improvement of the integral link in the nth PID controller, α2 is the integral link suppression coefficient, α3 is the integral link adjustment coefficient, and f(e n (k)) is a function related to the current deviation, and the value-taking method of f(e n (k)) is as follows:

[0020]

[0021] Through this dynamic integral formula, the integral control signal is weakened when the deviation is large, and the integral control signal is strengthened when the deviation is small, so as to improve the error elimination speed of the system and smooth the current response;

[0022] S6. The control system gradually updates the connection weights of the neural network and the PID control parameters through the BP neural network in each control cycle, and iterates until the output currents of each degaussing power supply module converge to the given current values of each module.

[0023] In any of the above technical solutions, further, the network structure of the BP neural network includes three neural layers, including an input layer, a hidden layer, and an output layer;

[0024] Among them, the number of neurons in the input layer is 3, and the input layer receives the deviation signal of the system; the number of neurons in the hidden layer is 8, and the hidden layer performs nonlinear processing on the deviation signal of the system; the number of neurons in the output layer is 3, which respectively correspond to the proportional, integral, and differential parameters of the PID controller, and the output layer outputs the adjusted PID parameter values.

[0025] In any of the above technical solutions, further, the input deviation signals of the BP neural network controller are respectively generated by multiple degaussing power supply modules, and a parallel controller structure is adopted. The input layer and the hidden layer of the system are independent of each other, and the output layers are connected to form a decoupled control that is mutually related.

[0026] In any of the above technical solutions, further, each layer of neurons of the BP neural network is processed by an activation function, and the form of the activation function is the Sigmoid function, so as to improve the processing ability of the network for non-linear signals and ensure the stable operation of the degaussing power supply system under complex coupling conditions.

[0027] The beneficial effects of the present invention are:

[0028] The technical solution in the present invention combines the BP neural network with the PID control, and realizes the decoupled control between multiple degaussing power supply modules by adjusting the PID parameters in real time, significantly improving the synchronization of the current output and ensuring the degaussing accuracy.

[0029] Through the dynamic proportional adjustment formula, the system automatically increases the proportional coefficient when the deviation is large and decreases the proportional coefficient when the deviation is small, effectively preventing overshoot and improving the response speed.

[0030] Introduce a variable-speed integral mechanism to weaken the integral control signal when the deviation is large and strengthen it when the deviation is small, thereby accelerating error convergence and achieving stable current control.

[0031] This method can effectively handle complex multi-input multi-output coupling relationships, meet the hull degaussing requirements in complex magnetic field environments, and improve the overall reliability and anti-interference ability of the system. Brief Description of the Drawings

[0032] The above and additional advantages of the present invention will become apparent and easy to understand in conjunction with the description of the embodiments with the following drawings, where:

[0033] Figure 1 is a schematic flow chart of an improved BP-PID decoupling control algorithm for the decoupling method of a multi-degaussing power supply system according to an embodiment of the present invention;

[0034] Figure 2 is a schematic diagram of a BP-PID decoupling controller for a two-degaussing power supply system in the decoupling method of a multi-degaussing power supply system according to an embodiment of the present invention;

[0035] Figure 3 is a schematic diagram of the structure of a BP-PID single-variable controller for the decoupling method of a multi-degaussing power supply system according to an embodiment of the present invention;

[0036] Figure 4 is a schematic diagram of the closed-loop control of a two-degaussing power supply system in the decoupling method of a multi-degaussing power supply system according to an embodiment of the present invention. Detailed Embodiments

[0037] In order to more clearly understand the above objects, features, and advantages of the present invention, the present invention will be further described in detail below in conjunction with the drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0038] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0039] As Figure 1 shown, this embodiment provides a decoupling method for a multi-degaussing power supply system, and the method includes:

[0040] A decoupling controller is designed to be composed of two BP-PID controllers in parallel to achieve the control of each module in the multi-demagnetization power supply system. The essence of the BP neural network PID decoupling control is to make the outputs of the coupled modules converge quickly to the given values to achieve a decoupling control effect. The network structure of each BP-PID single-variable controller includes an input layer, a hidden layer, and an output layer, and each layer is connected by neurons.

[0041] Among them, the number of neurons in the input layer is 3, which receives the real-time deviation signal of the system; 8 neurons are set in the hidden layer to process complex coupling relationships; the output layer has 3 neurons, which will generate signals for adjusting the proportional K p , integral K i , and derivative K d parameters.

[0042] Each controller processes the output current control of a demagnetization power supply module. The input layers and hidden layers of the two controllers are independent of each other, and an association relationship is established between the output layers to ensure the decoupling control of the system; within each control cycle, the BP neural network adjusts the control signals of the output layer to generate two output values u1(k) and u2(k), and these two output values are transmitted to each PID controller to adjust the current output of the demagnetization power supply module to make it synchronous and stable.

[0043] The basic working process of the BP-PID controller is as follows:

[0044] Data input: Within each control cycle, the input layer receives the deviation signals e1(k) and e2(k) of the two demagnetization power supply modules and transmits them to the hidden layer;

[0045] Calculate and adjust the error: After the deviation signals are processed by the neurons in the hidden layer, they are transmitted to the output layer, and the weights are adjusted through the gradient descent algorithm of the BP neural network to make the control system gradually tend to the decoupled state;

[0046] Generate control signals: The output layer generates control signals u1(k) and u2(k), which are respectively transmitted to each PID controller and further adjust the output currents i1(k) and i2(k) of the two demagnetization power supply modules.

[0047] To make the output of the system converge quickly to the given value, the following error function E(k) is designed in this embodiment:

[0048]

[0049] In the formula, i nref (k) and i n (k) are respectively the current given value and the actual output current value at the kth sampling moment of the nth power supply module.

[0050] According to the error function E(k), the BP neural network updates the weights of the output layer and the hidden layer through the gradient descent method, so that the output of the system gradually approaches the target current value, thereby realizing decoupling control; when the BP neural network uses the gradient descent algorithm to update the weights, the weight adjustment amounts of the output layer and the hidden layer and the updated weights are as follows:

[0051]

[0052] In the formula, η is the learning rate, β is the momentum coefficient, and the value ranges of η and β are both between 0 and 1.

[0053] The essence of BP-PID decoupling control is to make the output values of the coupled modules reach the given values, so as to achieve the purpose of decoupling. However, for the degaussing power supply system, it not only requires that the output current of each module has a high current accuracy, but also has strict requirements on the current convergence rate. Therefore, in order to improve the current convergence rate of each module, this embodiment improves the BP-PID decoupling control algorithm.

[0054] For a load with inertia such as a degaussing coil, if it is desired that the current output by the power supply has a fast convergence rate and a small current overshoot. Usually, it is expected that the proportional coefficient K p in its PID controller is proportional to the input current error. Therefore, the input current deviation value can be used as a factor for updating the proportional coefficient, that is, a scheme for dynamically adjusting the proportional coefficient:

[0055]

[0056] In the formula, is the proportional coefficient updated by the nth PID controller, α1 is the adjustment factor of the proportional coefficient, and in this embodiment, α1 = 3, e n (k) represents the current deviation signal.

[0057] The greater the current deviation, the greater the updated proportional coefficient is, and the higher the sensitivity of the system to the deviation, which can accelerate the rate at which the current approaches the given value. When the current deviation is small, the updated proportional coefficient is small, which can effectively prevent the current from acting too large and causing overshoot, and improve the stability of the system.

[0058] The present invention introduces the concept of variable-speed integral regulation in the integral link of the PID controller to further optimize the performance of the control system, so that the integral link in the PID controller has the ability of dynamic regulation, that is, when the deviation of the system is large, the role of the integral link is reduced; when the deviation is small, the role of the integral link is enhanced. The integral output signal The calculation formula is:

[0059]

[0060] wherein, u in (k) and are respectively the voltage control signals output before and after the improvement of the integral link in the nth PID controller, α2 is the integral link suppression coefficient, α3 is the integral link adjustment coefficient, f(e n (k)) is a function related to the current deviation, and the value-taking method of f(e n (k)) is as follows:

[0061]

[0062] When the deviation e n (k) is greater than 1.5, by reasonably setting the value of α2, the voltage control signal output by the integral link can be weakened; while when the deviation e n (k) is less than 1.5, since the value of f(e n (k)) increases as e n (k) decreases, the voltage control signal output by the integral link gradually increases, and the ability of the system to eliminate the deviation is enhanced, and its enhancement effect is related to the value of α3. In the embodiments of this patent, the values of α2 and α3 are 1000 and 0.5 respectively.

[0063] The following is a comparison of the current response of the traditional PID control and the BP-PID decoupling control method of the present invention in the two degaussing power supply systems through simulation tests to prove the superior effects of the technical solution of the present invention in improving the current response convergence speed, suppressing overshoot and improving the steady-state performance.

[0064] The traditional PID control and the BP-PID decoupling control method are respectively debugged for the same set current of 100A, and a group of sampled currents are recorded at fixed intervals. The sampling results are as follows:

[0065] Sampling time k Traditional PID current (A) BP-PID decoupling control current (A) 0 0.0 0.0 1 38.5 52.0 2 68.4 79.3 3 106.1 96.5 4 100.7 99.8 5 100.1 100.0

[0066] In the initial stage of k = 0 to 2, for the traditional PID control, the current rapidly rises to 38.5A at k = 1, and although it reaches 68.4A at k = 2, due to the mutual inductance coupling effect between the degaussing modules, the overall response has hysteresis and deviation; for the BP-PID decoupling control, at k = 1, by using the online adjustment function of the BP neural network, the proportional parameter is reasonably amplified, and the current rises to 52.0A, and the response is more rapid and smooth; at k = 2, it has reached 79.3A, showing a strong ability to approach the target value.

[0067] In the convergence stage when k = 3, the current of the traditional PID control reaches 106.1 A at k = 3, showing an obvious overshoot phenomenon, which is caused by the lack of effective compensation under complex coupling conditions. The BP-PID decoupling control of the present invention uses the dynamic proportional regulation and variable-speed integral strategy, and the current is only slightly lower than the set value at k = 3, which is 96.5 A, significantly reducing the overshoot risk and shortening the error convergence time.

[0068] In the stable response stage when k = 4 - 5, after continuous adjustment, both control methods achieve stable output from k = 4 to k = 5. However, the BP-PID scheme is smoother during the adjustment process. The parameter adaptive regulation enables the system to more accurately maintain the target current at the steady state, avoiding the risk of steady-state drift.

[0069] This simulation experiment proves that in a multi-demagnetization power supply system, introducing a decoupling control method combining BP neural network and PID control can effectively overcome the response lag and overshoot problems existing in traditional PID under complex coupling environments. After adopting the dynamic proportional regulation and variable-speed integral mechanism, the system can quickly respond, converge rapidly, and accurately maintain the set current value, thus significantly improving the demagnetization accuracy and system stability.

[0070] In summary, the present invention provides a decoupling method for a multi-demagnetization power supply system, and the method includes:

[0071] Designing a decoupling controller composed of multiple BP-PID controllers in parallel, and the decoupling controller dynamically adjusts the proportional, integral, and differential coefficients of each PID controller through a BP neural network;

[0072] The decoupling method using the decoupling controller includes the following steps:

[0073] S1. Initializing the BP neural network structure of the system, dividing the neural network into an input layer, a hidden layer, and an output layer. The input layer receives deviation signal data and transmits it to the hidden layer. After being processed by the hidden layer, the data is transmitted to the output layer, and the output layer generates output signals for the proportional K p , integral K i and differential K d parameters of the PID controller;

[0074] S2. Establishing a system error function:

[0075]

[0076] where i nref (k) and i n(k) are the current set value and the actual output current value at the k-th sampling moment of the n-th power supply module respectively; the weight values of the BP neural network are adjusted by the gradient descent algorithm to minimize the error function E(k) so that the system deviation gradually converges to zero;

[0077] S3. Use the BP neural network to generate a dynamically adjusted proportionality coefficient K p , integral coefficient K i and derivative coefficient K d and input them into the PID controller, and automatically update the parameters of the PID controller according to the deviation value of the system;

[0078] S4. Use the dynamic proportionality coefficient adjustment formula in the PID controller:

[0079]

[0080] In the formula, is the updated proportionality coefficient of the n-th PID controller, α1 is the adjustment factor of the proportionality coefficient, and e n (k) represents the current deviation signal;

[0081] The degaussing power supply module increases the proportionality coefficient when the deviation is large and decreases the proportionality coefficient when the deviation is small, so as to achieve fast control of the current response and prevent overshoot of the output current;

[0082] S5. Introduce variable-speed integration in the integral link of the PID controller, and the output formula of the integral control signal is:

[0083]

[0084] In the formula, u in (k) and are the voltage control signals output before and after the improvement of the integral link in the n-th PID controller respectively, α2 is the integral link suppression coefficient, α3 is the integral link adjustment coefficient, and f(e n (k)) is a function related to the current deviation, and the value-taking method of f(e n (k)) is:

[0085]

[0086] Through this dynamic integral formula, the integral control signal is weakened when the deviation is large, and the integral control signal is strengthened when the deviation is small, so as to improve the system error elimination speed and smooth the current response;

[0087] S6. The control system gradually updates the connection weights of the neural network and the PID control parameters through the BP neural network in each control cycle, and iterates until the output currents of each degaussing power supply module converge to the given current values of each module.

[0088] The steps in the present invention can be adjusted in sequence, combined, and deleted according to actual requirements.

[0089] The units in the device of the present invention can be combined, divided, and deleted according to actual requirements.

[0090] Although the present invention has been disclosed in detail with reference to the accompanying drawings, it should be understood that these descriptions are merely exemplary and are not intended to limit the application of the present invention. The protection scope of the present invention is defined by the appended claims and may include various variations, modifications, and equivalent solutions made to the invention without departing from the protection scope and spirit of the present invention.

Claims

1. A decoupling method for a multi-demagnetization power supply system, characterized in that, The method includes: Designing a decoupling controller composed of multiple BP-PID controllers in parallel, and the decoupling controller dynamically adjusts the proportional, integral, and differential coefficients of each PID controller through a BP neural network; The decoupling method using the decoupling controller includes the following steps: S1. Initialize the BP neural network structure of the system. Divide the neural network into an input layer, a hidden layer, and an output layer. The input layer receives deviation signal data and transmits it to the hidden layer. After processing by the hidden layer, the data is transmitted to the output layer, and the output layer generates output signals for the proportional K p , integral K i , and derivative K d parameters; S2. Establishing a system error function: where, i nref (k) and i n (k) are respectively the current set value and the actual output current value at the k-th sampling moment of the n-th power supply module; the weight of the BP neural network is adjusted by the gradient descent algorithm to minimize the error function E(k) so that the system deviation gradually converges to zero; S3. Generate a dynamically adjusted proportional coefficient K p , integral coefficient K i and derivative coefficient K d and input them into the PID controller to automatically update the parameters of the PID controller according to the deviation value of the system; S4. Using a dynamic proportional coefficient adjustment formula in the PID controller: Wherein, is the proportionality coefficient after the update of the nth PID controller, α1 is the adjustment factor of the proportionality coefficient, and e n (k) represents the current deviation signal; The degaussing power supply module increases the proportional coefficient when the deviation is large and decreases the proportional coefficient when the deviation is small, so as to achieve fast control of the current response and prevent overshoot of the output current; S5. Introducing variable-speed integration in the integral link of the PID controller, and the output formula of the integral control signal is: where u in (k) and are the voltage control signals output before and after the improvement of the integral link in the nth PID controller respectively, α2 is the integral link suppression coefficient, α3 is the integral link adjustment coefficient, f(e n (k)) is a function related to the current deviation, and the value-taking method of f(e n (k)) is as follows: Through this dynamic integral formula, the integral control signal is weakened when the deviation is large, and the integral control signal is strengthened when the deviation is small, so as to improve the system error elimination speed and smooth the current response; S6. The control system gradually updates the connection weights of the neural network and the PID control parameters through the BP neural network in each control cycle, and iterates until the output currents of each degaussing power supply module converge to the given current values of each module.

2. The decoupling method of the multi-demagnetization power supply system according to claim 1, characterized in that, The network structure of the BP neural network includes three neural layers, including an input layer, a hidden layer, and an output layer; Among them, the number of neurons in the input layer is 3, and the input layer receives the deviation signal of the system; the number of neurons in the hidden layer is 8, and the hidden layer performs nonlinear processing on the deviation signal of the system; the number of neurons in the output layer is 3, corresponding to the proportional, integral, and differential parameters of the PID controller respectively, and the output layer outputs the adjusted PID parameter values.

3. The decoupling method of the multi-demagnetization power supply system according to claim 2, characterized in that The input deviation signal of the BP neural network controller is generated by multiple degaussing power supply modules respectively, and a parallel controller structure is adopted. The input layer and the hidden layer of the system are independent of each other, and the output layers are connected to form an interrelated decoupling control.

4. The decoupling method of the multi-demagnetization power supply system according to claim 1, characterized in that Each layer of neurons of the BP neural network is processed by an activation function, and the form of the activation function is the Sigmoid function to improve the network's processing ability for nonlinear signals and ensure the degaussing power supply system to operate stably under complex coupling conditions.

Images