A noise reduction device for signal filters
By employing a dual-domain processing architecture that combines time-domain adaptive filtering with frequency-domain wavelet packet decomposition, the problem of reduced noise reduction performance of existing filtering devices in complex power grid environments is solved. This architecture achieves efficient signal processing and rapid response, making it suitable for complex power grid scenarios such as smart meters and new energy grid connection monitoring.
Patent Information
- Application Number
- CN202522539260.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Utility models(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-28
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2035-11-28
AI Technical Summary
Existing filtering devices only use a single time-domain or frequency-domain processing method, which makes it difficult to effectively separate high-frequency noise from low-frequency drift. When faced with drastic changes in power grid load or fluctuations in ambient temperature, the noise reduction performance drops significantly, and the signal processing delay is high, making it difficult to meet the needs of real-time sampling and rapid protection of the power grid.
A dual-domain processing architecture combining time-domain adaptive filtering and frequency-domain wavelet packet decomposition is adopted. By combining FPGA parallel processing and LVDS high-speed feedback link, a dynamic error suppression mechanism is introduced. Through the integrated design of signal input module, time-domain adaptive filtering module, frequency-domain wavelet packet decomposition module, dynamic error suppression module, signal output module, feedback link and power supply module, the system can accurately separate and suppress periodic interference, high-frequency noise and low-frequency drift.
It significantly improves the overall suppression capability against high-frequency noise, low-frequency drift and periodic interference, and the signal-to-noise ratio is significantly improved, meeting the requirements of real-time metering and rapid response of the power grid. The integrated design is compact, has strong anti-electromagnetic interference capability, and is suitable for complex power grid environments.
Smart Images

Figure CN224987412U_ABST
Abstract
Description
Technical Field
[0001] This utility model relates to the field of information infrastructure transformation technology, and in particular to a noise reduction device for signal filters. Background Technology
[0002] Currently, in complex power grid application scenarios such as smart meters, new energy grid connection monitoring, and industrial power meters, the power metering signal is easily affected by multiple interferences simultaneously. After the superposition of compound interferences, the sampling signal will be severely distorted, directly causing the power metering error to increase, and even exceeding the accuracy requirement of 0.2 level.
[0003] Most existing filtering devices employ only single time-domain or frequency-domain processing methods, making it difficult to effectively separate high-frequency noise from low-frequency drift. Their noise reduction performance deteriorates significantly under drastic changes in grid load or ambient temperature fluctuations. Existing solutions are often discrete designs or use single-core serial DSP processing, resulting in high signal processing latency, which fails to meet the requirements of real-time grid sampling and rapid protection.
[0004] Therefore, a dual-domain processing architecture that can construct time-domain adaptive filtering and frequency-domain wavelet packet decomposition in tandem is needed. By introducing a dynamic error suppression mechanism and combining FPGA parallel processing with LVDS high-speed feedback link, a noise reduction device for signal filters that accurately separates and suppresses periodic interference, high-frequency noise, and low-frequency drift is realized. Utility Model Content
[0005] Therefore, the technical problem to be solved by this invention is that using only a single time-domain or frequency-domain processing method is insufficient to effectively separate high-frequency noise from low-frequency drift, and the noise reduction performance deteriorates significantly when faced with drastic changes in power grid load or fluctuations in ambient temperature. Existing solutions are mostly discrete designs or use single-core serial processing with DSPs, resulting in high signal processing delays, which are difficult to meet the requirements of real-time sampling and rapid protection of the power grid.
[0006] The above-mentioned technical problems are solved by the following technical solution: This utility model proposes a noise reduction device for signal filters, which includes, The housing integrates a signal input module, a time-domain adaptive filtering module, a frequency-domain wavelet packet decomposition module, a dynamic error suppression module, a signal output module, a feedback link, a power supply module, and an SPI bus. The signal input module, time-domain adaptive filtering module, frequency-domain wavelet packet decomposition module, and signal output module are electrically connected in sequence. The input terminal of the dynamic error suppression module is electrically connected to the output terminal of the signal input module, and the output terminal of the dynamic error suppression module is electrically connected to the control terminals of the time-domain adaptive filtering module and the frequency-domain wavelet packet decomposition module respectively via the SPI bus; The time-domain adaptive filtering module and the frequency-domain wavelet packet decomposition module are bidirectionally electrically connected via a feedback link, and the power supply module provides power input.
[0007] In a preferred embodiment of the noise reduction device for signal filters described in this utility model: The signal input module includes a differential signal interface and an EMC filter capacitor. The EMC filter capacitor is connected in series between the differential signal interface and the time-domain adaptive filter module to initially suppress electromagnetic interference.
[0008] In a preferred embodiment of the noise reduction device for signal filters described in this utility model: The time-domain adaptive filtering module includes a first FPGA chip, an error calculation circuit, and a variable step size control circuit; The error calculation circuit is electrically connected to the feedback link to compare the input signal with the reconstructed signal to generate an error value. The variable step size control circuit is electrically connected to the first FPGA chip to dynamically adjust the filtering step size according to the error value.
[0009] In a preferred embodiment of the noise reduction device for signal filters described in this utility model: The frequency domain wavelet packet decomposition module includes a second FPGA chip, a signal decomposition circuit, and a sub-band processing circuit. The signal decomposition circuit is electrically connected to the output of the time-domain adaptive filtering module to perform signal decomposition. The sub-band processing circuit is electrically connected to the second FPGA chip to separate and suppress noise signals, and transmits the reconstructed signal to the time-domain adaptive filtering module through a feedback link.
[0010] In a preferred embodiment of the noise reduction device for signal filters described in this utility model: The dynamic error suppression module includes an ADC sampling circuit, a temperature sensor, and a compensation signal generation circuit; The ADC sampling circuit is electrically connected to the output of the signal input module to acquire signals, and the compensation signal generation circuit is electrically connected to the ADC sampling circuit and the temperature sensor to generate compensation signals. The compensation signal generation circuit transmits the compensation signal to the control terminals of the time-domain adaptive filtering module and the frequency-domain wavelet packet decomposition module via the SPI bus.
[0011] In a preferred embodiment of the noise reduction device for signal filters described in this utility model: The feedback link is an LVDS differential signal link, which is bidirectionally electrically connected to the time-domain adaptive filtering module and the frequency-domain wavelet packet decomposition module to realize high-speed data interaction and signal reconstruction transmission between modules.
[0012] In a preferred embodiment of the noise reduction device for signal filters described in this utility model: The SPI bus is electrically connected to the output of the dynamic error suppression module, and is also electrically connected to the control terminals of the time-domain adaptive filtering module and the frequency-domain wavelet packet decomposition module, respectively, to transmit compensation signals in real time to correct the filtering parameters.
[0013] In a preferred embodiment of the noise reduction device for signal filters described in this utility model: The power supply module outputs multiple DC voltages and is electrically connected to the signal input module, the time-domain adaptive filtering module, the frequency-domain wavelet packet decomposition module, the dynamic error suppression module, and the signal output module to meet the power supply requirements of each module.
[0014] In a preferred embodiment of the noise reduction device for signal filters described in this utility model: The time-domain adaptive filtering module and the frequency-domain wavelet packet decomposition module are electrically connected through the parallel architecture of the first FPGA chip and the second FPGA chip to reduce signal processing delay and achieve real-time data interaction in conjunction with the feedback link and SPI bus.
[0015] The beneficial effects of this invention are as follows: the coordinated operation of time-domain adaptive control and frequency-domain wavelet packet decomposition significantly improves the overall suppression capability against high-frequency noise, low-frequency drift, and periodic interference, resulting in a significant improvement in the signal-to-noise ratio. The introduction of variable step-size control and dynamic error compensation mechanisms allows the filter parameters to be adjusted in real time according to interference intensity and temperature. The adoption of a dual-FPGA parallel architecture and LVDS high-speed feedback link meets the requirements of real-time metering and rapid response in the power grid. The highly integrated modular design, combined with an aluminum alloy shielding housing, results in a compact size, strong anti-electromagnetic interference capability, and direct embedding within the instrument without requiring additional installation space. Attached Figure Description
[0016] To more clearly illustrate the technical solutions of the embodiments of this utility model, the accompanying drawings of the embodiments of this utility model will be briefly described below. Obviously, the drawings described below only relate to some embodiments of this utility model, and are not intended to limit the utility model.
[0017] Figure 1 The hardware architecture diagram of this utility model is shown.
[0018] Figure 2 The diagram shows the frequency domain wavelet packet decomposition module and its surrounding circuitry of this invention.
[0019] Figure 3 The diagram shows the ADC sampling circuit and its peripheral circuitry of this invention.
[0020] Figure 4 The circuit diagram of the power supply module of this utility model is shown.
[0021] Figure 5 The diagram shows the signal output module, error calculation circuit, and peripheral circuitry of this utility model. Detailed Implementation
[0022] To enable those skilled in the art to better understand this utility model, the present utility model will be further described in detail below with reference to specific embodiments and accompanying drawings.
[0023] The terminology used in this invention refers to those general terms currently widely used in the art in consideration of the functionality of this invention; however, these terms may vary according to the intent, precedent, or new technology of those skilled in the art. Furthermore, specific terms may be chosen independently, and in such cases, their detailed meanings will be described in the detailed description of this invention. Therefore, the terminology used in this specification should not be construed as simple names, but rather based on the meaning of the terms and the overall description of this invention.
[0024] Reference Figures 1-2 This embodiment provides a noise reduction device for signal filters.
[0025] The metal casing 1 integrates a signal input module 2, a time-domain adaptive filtering module 5, a frequency-domain wavelet packet decomposition module 9, a dynamic error suppression module 13, a signal output module 17, a feedback link 18, a power supply module 19, and an SPI bus 20.
[0026] In terms of overall structure, the metal casing 1 serves as the external protection and shielding component, while the internal space is designed with a compact layout to accommodate all functional modules. The signal input module 2 is located at the entrance of the device and is used to receive external power grid signals; the time-domain adaptive filtering module 5 follows immediately after it, performing preliminary time-domain processing; the frequency-domain wavelet packet decomposition module 9 is connected after the time-domain module and is responsible for frequency-domain analysis and noise separation; the signal output module 17 is located at the end of the link and outputs the processed clean signal.
[0027] The dynamic error suppression module 13 is arranged in parallel with the main signal link to monitor and compensate for errors; the feedback link 18 and the SPI bus 20 span multiple modules to realize data interaction and control signal transmission; the power supply module 19 is located at the bottom of the housing to provide stable power support.
[0028] The output of signal input module 2 is directly electrically connected to the input of time-domain adaptive filtering module 5 to ensure smooth signal transmission to the time-domain processing stage. The output of time-domain adaptive filtering module 5 is electrically connected to the input of frequency-domain wavelet packet decomposition module 9, forming a continuous path for signal processing. The output of frequency-domain wavelet packet decomposition module 9 is electrically connected to the input of signal output module 17 to complete the final signal output.
[0029] Meanwhile, the input terminal of the dynamic error suppression module 13 is electrically connected in parallel with the output terminal of the signal input module 2 to acquire the original input signal in real time for error analysis. The output terminal of the dynamic error suppression module 13 is electrically connected to the control terminals of the time-domain adaptive filtering module 5 and the frequency-domain wavelet packet decomposition module 9 via the SPI bus 20, respectively, for sending compensation signals to adjust the filtering parameters.
[0030] The time-domain adaptive filtering module 5 and the frequency-domain wavelet packet decomposition module 9 are bidirectionally electrically connected via feedback link 18 to achieve mutual transmission of processing results and iterative optimization. The power supply module 19 provides multiple voltage outputs through internal wiring to ensure stable power supply for each module.
[0031] In the actual assembly process, the metal casing 1 is first processed into an aluminum alloy structure, and an insulating partition is installed inside to separate the power supply area from the circuit area to reduce electromagnetic interference. The power supply area mainly houses the power supply module 19, while the circuit area houses other functional modules. Each module adopts a high-density integrated circuit board design and is fixed at the connection points with lead-free solder.
[0032] Feedback link 18 uses differential signal cables to connect the time-domain and frequency-domain modules, ensuring high-speed data transmission. SPI bus 20 uses shielded cabling to connect the dynamic error suppression module 13 to the control terminals of other corresponding waveform modules.
[0033] As one embodiment provided, such as Figures 1-4 , This embodiment mainly demonstrates the internal implementation and collaborative working mechanism of the time-domain adaptive filtering module 5 and the frequency-domain wavelet packet decomposition module 9. As the core part of signal processing, the time-domain adaptive filtering module 5 includes a first FPGA chip 6, an error calculation circuit 7, and a variable step size control circuit 8; the frequency-domain wavelet packet decomposition module 9 includes a second FPGA chip 10, a signal decomposition circuit 11, and a sub-frequency band processing circuit 12.
[0034] The first FPGA chip 6 serves as the control core, electrically connected to the error calculation circuit 7 and the variable step size control circuit 8. The error calculation circuit 7 consists of a multiplier-adder, an adder, and a register. Its input terminal is electrically connected to the output terminal of the signal input module 2 and the feedback link 18. It is used to compare the input signal with the reconstructed signal from the frequency domain module to generate an error value.
[0035] Error calculation circuit 7 calculates the quantization error index, mean square error, or peak deviation by comparing the difference between two signals in real time, which reflects the degree of deviation of the current filtering effect. Variable step size control circuit 8 dynamically adjusts the filtering step size according to the generated error value. When the error is large, the step size is increased to accelerate convergence; when the error is small, the step size is decreased for fine adjustment, thereby achieving adaptive filtering.
[0036] The variable step size control circuit 8 works closely with the first FPGA chip 6, which is responsible for synchronously updating the filter parameters to ensure that the time domain module can effectively suppress periodic interference.
[0037] In the frequency domain wavelet packet decomposition module 9, the second FPGA chip 10 is electrically connected to the signal decomposition circuit 11 and the sub-frequency band processing circuit 12. The signal decomposition circuit 11 uses wavelet basis for multi-level decomposition, and its input is electrically connected to the output of the time domain adaptive filtering module 5, decomposing the preprocessed signal into multiple sub-frequency band levels.
[0038] Through step-by-step decomposition, the signal is broken down into high-frequency and low-frequency components, facilitating subsequent targeted processing. The sub-band processing circuit 12 includes an energy comparator and an address decoder, used to analyze the energy distribution of each sub-band, identify and suppress high-frequency noise bands and low-frequency drift bands. After processing, the second FPGA chip 10 is responsible for reconstructing the effective signal components to form a reconstructed signal, which is then transmitted to the time-domain module via feedback link 18 to support further error correction.
[0039] In the collaborative working mechanism, the time-domain adaptive filtering module 5 first performs preliminary filtering on the input signal, adaptively adjusts it to address periodic interference, and then transmits the output signal to the frequency-domain wavelet packet decomposition module 9 for in-depth frequency domain analysis. After decomposition and noise suppression are completed, the reconstructed signal is transmitted back to the error calculation circuit 7 of the time-domain module via the feedback link 18, triggering a new round of step size adjustment and parameter optimization.
[0040] The closed-loop iterative mechanism ensures the complementarity of dual-domain processing, while the parallel architecture of the dual FPGA chips further accelerates the computation process, ensuring that signal processing latency remains at a low level, making it suitable for real-time power grid monitoring scenarios. Through this synergy, the signal-to-noise ratio can be significantly improved in complex noise environments while maintaining system stability and adaptability.
[0041] As one embodiment provided, such as Figures 1-5 , As the key to error correction, the dynamic error suppression module 13 includes an ADC sampling circuit 14, a temperature sensor 15, and a compensation signal generation circuit 16.
[0042] In terms of compensation mechanism, the ADC sampling circuit 14 is electrically connected to the output terminal of the signal input module 2 to collect the original signal and the intermediate signal during the processing in real time, which is used to monitor quantization error and accuracy loss during ADC conversion.
[0043] Temperature sensor 15 is installed inside the metal housing 1 to continuously monitor the operating temperature of the device and capture temperature drift errors caused by ambient temperature fluctuations. Compensation signal generation circuit 16 is electrically connected to ADC sampling circuit 14 and temperature sensor 15, and calculates compensation values based on the collected data.
[0044] The entire calculation process is as follows: First, the magnitude of the quantization error and the offset of the temperature drift are analyzed. Then, a corresponding compensation signal is generated, and the filtering threshold or offset correction coefficient is adjusted. The compensation signal is sent to the control terminals of the time-domain adaptive filtering module 5 and the frequency-domain wavelet packet decomposition module 9 via the SPI bus 20 to achieve real-time parameter correction.
[0045] When the temperature rises and causes the drift to increase, the compensation mechanism will dynamically adjust the wavelet decomposition threshold or time-domain step size to ensure the stability of the overall measurement accuracy. This ensures that the error is controlled at a low level across the entire temperature range, thereby improving the reliability and accuracy of the device.
[0046] The signal input module 2 includes a differential signal interface 3 and an EMC filter capacitor 4. The two are connected in series to initially suppress external electromagnetic interference and ensure the cleanliness of the input signal. The power supply module 19 outputs multiple DC voltages, which are electrically connected to each module to adapt to the different power supply requirements of FPGA chips and analog devices.
[0047] The entire device adopts a modular integrated layout, with all components highly compact. The metal housing 1 provides electromagnetic shielding and heat dissipation support, while internal insulating partitions separate the power supply and circuit areas to prevent interference. Feedback link 18 and SPI bus 20, as key connection components, ensure high-speed and reliable data exchange.
[0048] In the application scenario of smart meters, the original power grid signal is input through the differential signal interface 3, processed by the time domain and frequency domain modules, and compensated by the dynamic error suppression module 13. Finally, the signal output module 17 outputs a clean signal for high-precision power metering.
[0049] In actual deployment, the device can be directly embedded inside the instrument and connected to the system ground wire through the grounding post. The working indicator light displays the operating status.
[0050] Finally, it should be noted that the methods and devices described in detail above are merely embodiments, and those skilled in the art can modify these embodiments in different ways as long as they do not depart from the scope of this utility model.
Claims
1. A noise reduction device for a signal filter, characterized by, include: The housing (1) integrates a signal input module (2), a time-domain adaptive filtering module (5), a frequency-domain wavelet packet decomposition module (9), a dynamic error suppression module (13), a signal output module (17), a feedback link (18), a power supply module (19), and an SPI bus (20). The signal input module (2), the time-domain adaptive filtering module (5), the frequency-domain wavelet packet decomposition module (9), and the signal output module (17) are electrically connected in sequence; The input terminal of the dynamic error suppression module (13) is electrically connected to the output terminal of the signal input module (2), and the output terminal of the dynamic error suppression module (13) is electrically connected to the control terminals of the time-domain adaptive filtering module (5) and the frequency-domain wavelet packet decomposition module (9) respectively via the SPI bus (20). The time-domain adaptive filtering module (5) and the frequency-domain wavelet packet decomposition module (9) are bidirectionally electrically connected through the feedback link (18), and the power supply module (19) provides power input.
2. The noise reduction device for signal filters according to claim 1, characterized in that: The signal input module (2) includes a differential signal interface (3) and an EMC filter capacitor (4). The EMC filter capacitor (4) is connected in series between the differential signal interface (3) and the time-domain adaptive filter module (5) to initially suppress electromagnetic interference.
3. The noise reduction device for signal filtering according to claim 1, characterized in that: The time-domain adaptive filtering module (5) includes a first FPGA chip (6), an error calculation circuit (7), and a variable step size control circuit (8); The error calculation circuit (7) is electrically connected to the feedback link (18) to compare the input signal with the reconstructed signal to generate an error value. The variable step size control circuit (8) is electrically connected to the first FPGA chip (6) to dynamically adjust the filtering step size according to the error value.
4. The noise reduction device for signal filtering according to claim 1, characterized in that: The frequency domain wavelet packet decomposition module (9) includes a second FPGA chip (10), a signal decomposition circuit (11), and a sub-band processing circuit (12). The signal decomposition circuit (11) is electrically connected to the output of the time-domain adaptive filtering module (5) to perform signal decomposition. The sub-band processing circuit (12) is electrically connected to the second FPGA chip (10) to separate and suppress noise signals, and transmits the reconstructed signal to the time-domain adaptive filtering module (5) through the feedback link (18).
5. The noise reduction device for a signal filter according to claim 1, characterized in that: The dynamic error suppression module (13) includes an ADC sampling circuit (14), a temperature sensor (15), and a compensation signal generation circuit (16). The ADC sampling circuit (14) is electrically connected to the output of the signal input module (2) to acquire signals, and the compensation signal generation circuit (16) is electrically connected to the ADC sampling circuit (14) and the temperature sensor (15) to generate compensation signals. The compensation signal generation circuit (16) transmits the compensation signal to the control terminals of the time-domain adaptive filtering module (5) and the frequency-domain wavelet packet decomposition module (9) via the SPI bus (20).
6. The noise reduction device for a signal filter according to claim 1, characterized in that: The feedback link (18) is an LVDS differential signal link, which is bidirectionally electrically connected to the time-domain adaptive filtering module (5) and the frequency-domain wavelet packet decomposition module (9) to realize high-speed data interaction and signal reconstruction transmission between modules.
7. The noise reduction device for a signal filter according to claim 1, characterized in that: The SPI bus (20) is electrically connected to the output of the dynamic error suppression module (13), and is electrically connected to the control terminals of the time-domain adaptive filtering module (5) and the frequency-domain wavelet packet decomposition module (9) respectively, to transmit compensation signals in real time to correct the filtering parameters.
8. The noise reduction device for a signal filter according to claim 1, characterized in that: The power supply module (19) outputs multiple DC voltages and is electrically connected to the signal input module (2), the time-domain adaptive filtering module (5), the frequency-domain wavelet packet decomposition module (9), the dynamic error suppression module (13), and the signal output module (17) to adapt to the power supply requirements of each module.
9. The noise reduction device for a signal filter according to claim 1, characterized in that: The time-domain adaptive filtering module (5) and the frequency-domain wavelet packet decomposition module (9) are electrically connected through the parallel architecture of the first FPGA chip (6) and the second FPGA chip (10) to reduce signal processing delay and work with the feedback link (18) and the SPI bus (20) to achieve real-time data interaction.
