LC filter circuit based on artificial neural network
By introducing artificial neural networks and flexible configuration capacitive arrays and inductor arrays into the LC filtering circuit, the problem of limited filtering effect and flexibility in complex RF environments in traditional LC filtering circuits is solved, and precise filtering of signals of different frequency and system flexibility and configurability are achieved.
Patent Information
- Application Number
- CN202510049763.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-13
- Publication Date
- 2025-05-06
AI Technical Summary
The filtering effect and flexibility of traditional LC filtering circuits in complex and variable RF environments are limited, resulting in the need to replace printed boards in high-power, multi-frequency RF power systems, which increases costs and causes waste of resources.
The LC filtering circuit based on artificial neural network is adopted. Through the flexible configuration of capacitor arrays and inductive arrays and the precise control of power switch tubes, combined with the adaptive capabilities of artificial neural networks, the dynamic adjustment of the filter circuit can be realized, and signals can be accurately filtered for different frequency.
It realizes accurate filtering of frequency points in different modes, improves the filtering effect and flexibility of the system, reduces design costs and material costs, and avoids waste of resources.
Smart Images
Figure CN119945377A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of radio frequency microwave technology, and in particular to an LC filter circuit based on artificial neural network. Background Art
[0002] In the RF power supply system, the filter circuit plays a vital role. For RF power supplies of different frequencies, the way to generate the RF drive signal that controls the operation of the power amplifier module is the same. The main control module first generates a square wave signal of the corresponding frequency and contains K (K ≥ 2) harmonics and transmits it to the RF drive module. The LC filter circuit in the RF drive module filters out the unnecessary harmonics, leaving the fundamental wave, and then filters the square wave signal into a single-frequency sine wave signal. Then, through several RF drive signal output ports, the single sine wave RF drive signal is transmitted to the power amplifier module, so that the power amplifier module enters the working state and amplifies the RF drive signal.
[0003] Although the traditional LC filter circuit can meet the filtering requirements to a certain extent, its filtering effect and flexibility are often limited when facing complex and changeable RF environments. Especially in high-power, multi-frequency RF power supply systems, the design of traditional filter circuits becomes particularly complicated. When the frequency of the RF power supply changes, it is often necessary to replace the printed circuit board containing different LC filter circuits, which increases the cost, and the printed circuit board cannot be reused, resulting in a waste of resources. For example, the operating frequency range of a 400KHz RF power supply is 330KHz~440KHz, while the operating frequency range of a 200KHz RF power supply is 150KHz~250KHz. If a 200KHz RF power supply uses a 400KHz LC filter circuit, severe third harmonics will appear at 150KHz, which will have an unpredictable impact on the power amplifier module of the equipment. Therefore, it is usually replaced with an RF driver module that can match the 150KHz~250KHz LC filter circuit, resulting in material waste and increased costs. Summary of the invention
[0004] The present application provides an LC filter circuit based on an artificial neural network, which solves the problem that the traditional LC filter circuit has a narrow filtering bandwidth and cannot accurately filter specific frequency points, and the cost of replacing printed circuit boards increases and wastes resources. It realizes that the frequency points can be accurately filtered in different modes.
[0005] The embodiment of the present application provides an LC filter circuit based on an artificial neural network, comprising: A capacitor array, wherein each capacitor in the capacitor array is connected to an independent power switch tube, and the on and off states of these power switch tubes are precisely controlled to achieve variable adjustment of the equivalent capacitance value of the capacitor array; An inductor array, wherein each inductor in the inductor array is connected to an independent power switch tube, and the equivalent inductance value of the inductor array can be variably adjusted by precisely controlling the on and off states of these power switch tubes; A control unit includes an artificial neural network (ANN), an input layer of the artificial neural network receives input parameters related to input signal characteristics, and the input layer of the artificial neural network is connected to the control end of each power switch tube.
[0006] The beneficial effects of the above embodiments are as follows: the capacitor array and the inductor array are connected according to the basic topological structure of the LC filter circuit to form an initial LC filter circuit, the input parameters are output as control parameters after the artificial neural network, each power switch tube is controlled, and dynamic adjustment of the filter circuit is realized, which can achieve accurate filtering of signals of different frequencies and improve the filtering effect of the system.
[0007] Based on the above embodiments, the present application can be further improved as follows: In one embodiment of the present application, the capacitors in the capacitor array are connected in parallel. According to the expected frequency coverage and bandwidth adjustment capability of the filter circuit, a series of capacitor elements are selected to construct the capacitor array, and the capacitance values of these capacitors can be the same or increase or decrease in a specific sequence to achieve continuous variable adjustment of the equivalent capacitance value of the capacitor array.
[0008] In one embodiment of the present application, each capacitor in the capacitor array is controlled by one power switch tube, so as to reduce the number of switch tubes used and reduce the power loss caused by the conduction of the switch tubes.
[0009] In one embodiment of the present application, the inductors in the inductor array are connected in series. According to the inductance range required for filtering and the requirement for dynamic adjustment of the inductance value, multiple inductor elements with corresponding inductance values are selected to form an inductor array to achieve continuous variable adjustment of the equivalent inductance value of the inductor array. These inductors can be selected to have the same or different inductance values, but the key is that their step value must be small enough to ensure that when the inductors are connected in series, unnecessary harmonic components at each frequency point can be accurately filtered out.
[0010] In one embodiment of the present application, each inductor in the inductor array is controlled by a pair of power switch tubes. By regulating the on and off states of these switch tubes, the series combination of inductors can be flexibly changed to achieve the purpose of adjusting the equivalent inductance value.
[0011] In one embodiment of the present application, the control unit includes an acquisition module and a feedback module, the acquisition module is used to obtain the corresponding input parameters according to the input signal, and the feedback module is used to detect the output signal of the LC filter circuit and feed it back to the control unit. Through the cooperation of the detection circuit and the algorithm, the sine wave output by the LC filter circuit is monitored. Once it is found that the detection result does not conform to the sine waveform of the expected frequency, the feedback mechanism will be triggered. At this time, the algorithm will recalculate and adjust the opening state of the power switch tube to accurately control the number of capacitors and inductors connected to the LC filter circuit, so as to make corrections.
[0012] In one embodiment of the present application, the input parameters include the frequency, amplitude, and harmonic order of the signal, and this information can be obtained by real-time sampling, analysis, and feature extraction of the input signal.
[0013] In one of the embodiments of the present application, the input layer of the artificial neural network is provided with a plurality of neurons which respectively receive the input parameters, and after complex calculations and feature extraction in the hidden layer, the control signals corresponding to the power switch tubes in the capacitor array and the inductor array are output through the output layer.
[0014] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. This LC filter circuit uses capacitors, inductors and power switch tubes to form capacitor arrays and inductor arrays, which are integrated into a printed circuit board. The on and off of the power switch tubes are controlled by an artificial neural network algorithm to achieve dynamic adjustment of the filter circuit, ensuring the stability and reliability of the filter circuit in high-frequency and high-power environments; 2. This LC filter circuit can achieve accurate filtering of signals of different frequencies through flexible configuration of capacitor arrays and inductor arrays, as well as precise control of power switch tubes, thereby improving the filtering effect of the system and enhancing the flexibility and configurability of the system; 3. This LC filter circuit reduces the workload of manual debugging and optimization and reduces the design cost through an intelligent automatic control system; at the same time, it can broaden the bandwidth of the LC filter circuit, reuse the printed circuit board, reduce material costs and avoid waste of resources. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.
[0016] Figure 1This is a structural block diagram of an LC filter circuit based on an artificial neural network in an embodiment of the present application; Figure 2 Schematic diagram of a circuit of a capacitor array in an embodiment of the present application; Figure 3 is a circuit diagram of an inductor array in an embodiment of the present application; Figure 4 This is a flow chart of the artificial neural network controlling the operation of the LC filter circuit in an embodiment of the present application. DETAILED DESCRIPTION
[0017] The present invention is further explained below in conjunction with specific implementation methods. It should be understood that these implementation methods are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, various equivalent forms of modifications to the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.
[0018] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.
[0019] In the description of the present invention, it should be noted that the terms "disposed", "connected" and "connected" should be understood in a broad sense, for example, it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0020] In the description of the present invention, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in the present invention and the features of different embodiments or examples without contradiction.
[0021] The embodiment of the present application provides an LC filter circuit based on an artificial neural network, which solves the problem that the traditional LC filter circuit has a narrow filtering bandwidth and cannot accurately filter specific frequency points, and the cost of replacing printed circuit boards increases and wastes resources. It achieves accurate filtering of frequency points in different modes.
[0022] The technical solution in the embodiment of the present application is to solve the above problems, and the overall idea is as follows: Example: like Figure 1As shown, an LC filter circuit based on an artificial neural network includes: a capacitor array, an inductor array and a control unit.
[0023] Capacitor array, each capacitor in the capacitor array is connected to an independent power switch tube. By accurately controlling the on and off states of these power switch tubes, variable adjustment of the equivalent capacitance value of the capacitor array can be achieved; for example, when a smaller equivalent capacitance value is required, the power switch tubes corresponding to more capacitors with higher capacitance values can be turned off; conversely, turning on a specific combination of power switch tubes can increase the equivalent capacitance value, thereby flexibly adapting to the filtering requirements of signals of different frequencies.
[0024] Inductor array, each inductor in the inductor array is connected to an independent power switch tube. By precisely controlling the on and off states of these power switches, the connection structure inside the inductor array is changed, thereby achieving flexible adjustment of the equivalent inductance value within a large range. This helps to work with the capacitor array in different filtering scenarios to optimize the filtering effect.
[0025] The control unit includes an artificial neural network (ANN). The input layer of the artificial neural network receives input parameters related to the characteristics of the input signal, such as the frequency, amplitude, harmonic number and other information of the signal, which can be obtained by real-time sampling, analysis and feature extraction of the input signal. The input layer of the artificial neural network is connected to the control end of each power switch tube in the capacitor array and the inductor array, and directly outputs the RF drive signal for controlling the on or off of the power switch tube, and controls the on and off of the corresponding power switch tube through the RF drive signal.
[0026] Among them, artificial neural network (ANN), as a powerful intelligent algorithm, has the ability of self-learning, self-adaptation and parallel processing, and is very suitable for complex system control and optimization. Therefore, introducing ANN into the design of LC filter circuit can realize automatic adjustment and optimization of filter parameters and improve the overall performance of the system. In the training stage, the artificial neural network uses a large number of representative input signal samples and the corresponding power switch control signals under the ideal filtering effect as training data, and uses training methods such as back propagation algorithm to train the artificial neural network, so that it can learn the complex nonlinear mapping relationship between the input signal characteristics and the optimal control strategy of the power switch.
[0027] Furthermore, if Figure 2 As shown, according to the expected frequency coverage and bandwidth adjustment capability of the filter circuit, a series of capacitor elements are selected to construct a capacitor array, and the capacitors in the capacitor array are connected in parallel. Figure 2 In, S MN is the power switch tube, C MNThe capacitance of these capacitors can be the same or can increase or decrease in a specific sequence to achieve continuous variable adjustment of the equivalent capacitance value of the capacitor array. But it is important that the change is kept within a small step range to ensure that the harmonic order can be accurately filtered out. The artificial neural network analyzes the RF signal component output by the LC filter circuit, calculates the capacitance that needs to be adjusted (i.e., increase or decrease), and then sends an RF drive signal to the corresponding RF power switch tube. These signals determine whether the capacitor controlled by the switch tube is connected to the bus for parallel connection or disconnected from the bus. Given that the RF power switch tube will generate power loss in the on state, thereby increasing the power consumption of the overall circuit, by allowing each capacitor to be controlled by only one power switch tube, the number of switches used can be effectively reduced, thereby reducing the power loss caused by the switch tube being turned on.
[0028] Furthermore, if Figure 3 As shown, according to the inductance range required for filtering and the requirement for dynamic adjustment of the inductance value, multiple inductance elements with corresponding inductance values are selected to form an inductance array. The inductors in the inductance array are connected in series to achieve continuous variable adjustment of the equivalent inductance value of the inductance array. Figure 3 In, S MN is the power switch tube, L M is a capacitor. In this circuit, a pair of power switches are installed at both ends of each inductor. These inductors can be selected to have the same or different inductance values, but the key is that their step values must be small enough to ensure that when the inductors are connected in series, the unnecessary harmonic components at each frequency point can be accurately filtered out. Specifically, if only the inductance value L1 is required in the circuit, the switch S 11 and S 12 will be in the on state; if the inductance value of L1 and L2 needs to be connected in series, then S 11 , S 21 and S 22 The switches must be turned on at the same time; if the inductance of L1, L2 and L3 needs to be further connected in series, S 11 , S 21 , S 31 and S 32 The switch is turned on, and so on. In this way, the inductors can be connected in series to the circuit in sequence as needed.
[0029] Furthermore, the control unit includes an acquisition module and a feedback module. The acquisition module is used to obtain an input signal and obtain corresponding input parameters based on the input signal. The feedback module is used to detect the output signal of the LC filter circuit and feed it back to the control unit. The feedback module includes a detection circuit.
[0030] It is necessary to further explain that: The hardware construction process of this LC filter circuit is as follows: based on the requirements for the expected frequency coverage range and bandwidth adjustment capability of the filter circuit, a series of capacitor elements with different capacitance values are selected to construct a capacitor array to achieve continuous variable adjustment of the equivalent capacitance value of the capacitor array; similarly, based on the inductance range required for filtering and the requirements for dynamic adjustment of the inductance value, multiple inductance elements with different inductance values are selected to form an inductor array; power switching tubes with fast response speed and suitable for high frequency are selected to ensure the stability and reliability of the filter circuit in high frequency and high power environments; finally, some resistors, diodes and other components are selected according to the actual circuit to construct the switching circuit and the control circuit.
[0031] The automatic control implementation method of this LC filter circuit is as follows: The designed capacitor array and inductor array are connected according to the basic topology of the LC filter circuit to form an initial LC filter circuit. In the actual working process, the input signal first enters the circuit system, and the artificial neural network quickly calculates and outputs the corresponding power switch control signal based on the input signal characteristic information collected in real time. These control signals drive the power switch tubes in the capacitor array and the inductor array to change the working state, thereby dynamically adjusting the equivalent capacitance and inductance values of the LC filter circuit to achieve adaptive filtering of the input signal. By continuously adjusting the filter parameters according to the changes in the input signal, the LC filter circuit can always maintain the best filtering performance, effectively filter out the noise, clutter and unnecessary frequency components in the signal, and output high-quality filtered signals.
[0032] Figure 4 This is a flowchart of the artificial neural network algorithm controlling the operation of the LC filter circuit. The LC filter circuit is simulated according to the operating frequency of the RF power supply to obtain the approximate value range of the capacitor and inductor. The artificial neural network algorithm is used to dynamically monitor the frequency change of the input signal sent by the main control module. According to the signal frequency sent by the main control module, the algorithm will intelligently adjust the on-off state of the power switch tube, so as to accurately connect the capacitor and inductor to the LC filter circuit to achieve accurate matching filtering for this specific frequency point. The sine wave output by the LC filter circuit is monitored through the cooperation of the detection circuit and the algorithm. Once it is found that the detection result does not match the sine waveform of the expected frequency, the feedback mechanism will be triggered. At this time, the algorithm will recalculate and adjust the on state of the power switch tube to accurately control the number of capacitors and inductors connected to the LC filter circuit for correction.
[0033] The optimization and testing of this LC filter circuit are as follows: A comprehensive performance evaluation is conducted on the LC filter circuit system that has been designed and put into operation. By inputting test signals of various types, frequency ranges, and amplitude changes, professional signal analysis instruments (such as spectrum analyzers, oscilloscopes, etc.) are used to accurately measure and analyze the signals before and after filtering to obtain various performance indicators of the filter circuit, such as filter efficiency, passband flatness, stopband attenuation, etc. According to the performance evaluation results, the feedback mechanism is used to further optimize the structure of the artificial neural network, the training algorithm, and the component parameter selection of the capacitor array and the inductor array, so as to continuously improve the overall performance and stability of the filter circuit system, so that it can better meet various complex filtering requirements in practical applications.
[0034] The specific implementation example of this LC filter circuit is as follows: In the capacitor array construction phase, we assume that the expected filtering frequency range is [F1, F2] and the bandwidth adjustment range is [B1, B2]. After calculation and analysis, the capacitor values are C 11 , C 12 , ..., C MN Each capacitor C MN With power switch tube S MN Connected, when S MN When conducting, C MN For example, for filtering of lower frequency signals, a larger equivalent capacitance value may be required. At this time, the artificial neural network controls multiple power switch tubes to conduct according to the frequency characteristics of the input signal, so that multiple capacitors are connected in parallel to the circuit to increase the equivalent capacitance.
[0035] When constructing the inductor array, we select inductance values as L1, L2, ..., L m The inductors form an array, each inductor L m Both with a pair of power switch tubes S m1 and S m2 If L1, L2, ..., L m The series connection only needs to turn on the power switch tube S at the same time. 11 , S 21 ,...,S m1 and S m2 By regulating the on and off states of these switch tubes S, we can flexibly change the series combination of inductors to achieve the purpose of adjusting the equivalent inductance value. For example, in situations where high-frequency signals need to be filtered out efficiently, artificial neural networks may intelligently command the inductor array to form a smaller equivalent inductance value and make it work in conjunction with the capacitor array to achieve the best filtering effect.
[0036] In the design of artificial neural network control system, the input layer can set multiple neurons to receive the frequency f, amplitude A, harmonic order k and other parameters of the input signal respectively. After complex calculations and feature extraction in the hidden layer (which can contain multiple hidden layers, each with several neurons), the output layer outputs the control signal corresponding to the power switch tube in the capacitor array and the inductor array. During the training process, a large number of signal samples with different frequency components and amplitude changes are collected, such as sine waves, square waves, triangle waves and complex signals containing multiple harmonic orders. The artificial neural network is repeatedly trained to continuously optimize the internal weights and thresholds to accurately generate the best power switch tube control strategy based on the input signal. For the actual collected data, the best method is generally to analyze the harmonic order. For example, when the harmonic suppression is better than 30dBm, the LC filter is judged to be better.
[0037] In the integration and adaptive control stage of the LC filter circuit, the trained artificial neural network is integrated with the capacitor array and the inductor array in the same circuit system. The input signal is input to the artificial neural network through the sampling circuit, and the artificial neural network quickly calculates and outputs the control signal to the power switch tube, realizing real-time adjustment of the equivalent parameters of the capacitor array and the inductor array, ensuring that the filter circuit is always in the best working state.
[0038] In the performance evaluation and optimization stage, a signal generator is used to generate various test signals, such as single-frequency signals of different frequencies, swept-frequency signals, and complex signal waveforms in practical applications, which are input into the LC filter circuit system. A spectrum analyzer is used to measure the spectrum characteristics of the signal before and after filtering, and an oscilloscope is used to observe the time domain waveform changes of the signal to obtain performance indicators such as filtering efficiency, passband fluctuations, and stopband attenuation. Based on these indicators, the structure of the artificial neural network (such as adding or reducing hidden layers and the number of neurons), the training algorithm (such as adjusting the learning rate and optimizing the function), and the component parameters of the capacitor array and the inductor array (such as replacing the capacitor and inductor values and adjusting the array scale) are optimized and adjusted, and then the performance test is performed again, and this process is repeated until the performance of the filter circuit system is optimized.
[0039] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages: 1. This LC filter circuit uses capacitors, inductors and power switch tubes to form capacitor arrays and inductor arrays, which are integrated into a printed circuit board; and uses power switch tubes as the core components of the switch circuit to control the connection and disconnection of each component in the capacitor array and inductor array. The power switch tube is controlled by an artificial neural network algorithm to achieve dynamic adjustment of the filter circuit. The selection and control strategy of the power switch tube ensures the stability and reliability of the filter circuit in a high-frequency, high-power environment.
[0040] 2. This LC filter circuit can achieve accurate filtering of signals of different frequencies and improve the filtering effect of the system through the flexible configuration of capacitor arrays and inductor arrays and the precise control of power switch tubes; the design of capacitor arrays and inductor arrays enables the filter circuit to be dynamically adjusted according to different needs, enhancing the flexibility and configurability of the system; the adaptive ability of the artificial neural network enables the filter circuit to cope with environmental changes and changes in system requirements and maintain the stability of the system; the rapid learning and adjustment capabilities of the artificial neural network enable the filter circuit to quickly respond to changes in system requirements and improve the response speed.
[0041] 3. This LC filter circuit reduces the workload of manual debugging and optimization and reduces the design cost through an intelligent automatic control system; at the same time, it can broaden the bandwidth of the LC filter circuit, reuse the printed circuit board, reduce material costs and avoid waste of resources.
[0042] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.
Claims
1. An LC filter circuit based on artificial neural network, characterized in that: include: A capacitor array, each capacitor in the capacitor array is connected to an independent power switch tube; An inductor array, wherein each inductor in the inductor array is respectively connected to an independent power switch tube; A control unit, the control unit includes an artificial neural network, the input layer of the artificial neural network receives input parameters related to input signal characteristics, and the input layer of the artificial neural network is connected to the control end of each power switch tube.
2. The LC filter circuit according to claim 1, characterized in that: The capacitors in the capacitor array are connected in parallel.
3. The LC filter circuit according to claim 2, characterized in that: Each of the capacitors in the capacitor array is controlled by one of the power switch tubes.
4. The LC filter circuit according to claim 1, characterized in that: The inductors in the inductor array are connected in series.
5. The LC filter circuit according to claim 4, characterized in that: Each of the inductors in the inductor array is controlled by a pair of power switch tubes.
6. The LC filter circuit according to claim 1, characterized in that: The control unit includes an acquisition module and a feedback module. The acquisition module is used to obtain the corresponding input parameter according to the input signal, and the feedback module is used to detect the output signal of the LC filter circuit and feed it back to the control unit.
7. The LC filter circuit according to claim 1, characterized in that: The input parameters include the frequency, amplitude and harmonic order of the signal.
8. The LC filter circuit according to claim 7, characterized in that: The input layer of the artificial neural network is provided with a plurality of neurons which respectively receive the input parameters, and after complex calculations and feature extraction in the hidden layer, the control signals corresponding to the power switch tubes in the capacitor array and the inductor array are outputted through the output layer.