Data filtering method for weighing system
The proposed data filtering method for weight systems employs sliding window polynomial filters and adaptive filters to address the challenges of signal distortion and adaptability, achieving robust and efficient noise suppression and responsiveness in weight signal processing.
Patent Information
- Application Number
- CN202411754128.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-02
- Publication Date
- 2025-03-25
AI Technical Summary
Existing weight signal processing technologies in industrial automation and smart manufacturing face challenges such as high complexity in designing FIR/IIR filter parameters, low robustness, signal distortion during periodic noise handling, and lack of adaptability to dynamic noise conditions, making it difficult to achieve accurate and reliable weight measurements.
A data filtering method for weight systems using a combination of sliding window polynomial filters and adaptive filters, where polynomial fitting is used to smooth signals and adaptive filter weights are dynamically adjusted based on noise characteristics to enhance noise suppression and responsiveness.
This approach effectively removes high-frequency noise from weight signals without significant delay, ensuring accuracy and reliability, and adapts to varying noise conditions, improving processing efficiency and system robustness.
Smart Images
Figure CN119696544A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of weighing signal processing, and in particular to a data filtering method for a weighing system. Background Art
[0002] In the field of modern industrial automation and intelligent manufacturing, weighing sensors play a vital role. They can accurately convert the weight of an object into an electrical signal and are widely used in logistics, manufacturing, medical and other industries. However, these electrical signals are easily interfered by noise from environmental factors such as power fluctuations, electromagnetic interference, and temperature changes during actual measurement. Effectively filtering out these noises and obtaining accurate weighing data is crucial to ensuring product quality and production efficiency.
[0003] Existing weighing signal processing technologies still have limitations. For example, sliding average filtering, which smoothes noise by calculating the average value of the signal over a period of time, is simple to operate, but may cause signal distortion when processing periodic noise. Finite impulse response (FIR) filters can be used to filter out noise of a specific frequency by designing a specific impulse response, but their design usually requires an accurate understanding of the noise characteristics, which is difficult to achieve in a variable actual application environment. In addition, existing technologies usually lack adaptability and it is difficult to dynamically adjust the filter parameters according to different noise conditions.
[0004] Therefore, a new technical solution is urgently needed to solve the above technical problems. Summary of the invention
[0005] The purpose of the present invention is to overcome the above-mentioned problems of the prior art and provide a data filtering method for a weighing system to solve the technical problems in the prior art that the traditional FIR / IIR filter design parameters are difficult, the designed filter parameters have low robustness, signal distortion may occur when processing periodic noise, and there is a lack of adaptability, making it difficult to dynamically adjust the filter parameters according to different noise conditions.
[0006] The above objectives are achieved through the following technical solutions:
[0007] A data filtering method for a weighing system, comprising:
[0008] Step (1) collecting a weighing signal through a weighing sensor as an original signal;
[0009] Step (2) filtering the original signal through a sliding window polynomial filter to obtain a filtered signal; comprising: performing local polynomial fitting on the original signal to estimate a smoothing value to obtain the filtered signal;
[0010] Step (3) filters and suppresses noise on the filtered signal through an adaptive filter to obtain filtered data for use in subsequent processes.
[0011] Furthermore, the step (2) of performing local polynomial fitting on the original signal to estimate the smoothed value is specifically as follows:
[0012] Set the window length to 2M+1 and the data in the window to X n ={X -m ,X -m+1 ,…X0,…X m-1 ,X m}, perform polynomial fitting on the data in the window, and set the fitting polynomial to:
[0013]
[0014] In the formula, N is the polynomial order, n is the data index in the window, and a k is a multi-factor.
[0015] Furthermore, in step (3), the filtered signal is filtered and noise suppressed by an adaptive filter to obtain filtered data y(n), and the formula is as follows:
[0016] y(n)=W T (n)x(n)
[0017] Where W(n) is the filter weight coefficient, T is the matrix transpose, and x(n) is the original signal to be filtered.
[0018] Furthermore, an update step size μ is introduced to determine the coefficient weights, and then:
[0019] W(n)=f(μ,e)
[0020] Where f(μ, e) is the result of a linear function iteration.
[0021] Furthermore, the value of the update step size μ is as follows:
[0022]
[0023] Where L is the order of the adaptive filter.
[0024] The data filtering method for a weighing system provided by the present invention can effectively remove high-frequency noise in a weighing signal without introducing significant time delay by adopting a polynomial sliding filter, thereby ensuring the accuracy and reliability of the data; the step size of the filter can be dynamically adjusted according to the characteristics of the weighing signal by adopting an adaptive filtering step size learning strategy, which not only improves the response speed of the filter, but also enhances its adaptability to signal changes, thereby providing a stable filtering effect under various working conditions. This method can not only improve the efficiency of data processing, but also enhance the overall performance and robustness of the system, bringing significant progress to the field of weighing technology. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Figure 1 A flow chart of a data filtering method for a weighing system according to the present invention;
[0026] Figure 2 A schematic diagram of a data filtering method for a weighing system according to the present invention;
[0027] Figure 3 This is a schematic diagram of sliding average filtering in a data filtering method for a weighing system according to the present invention;
[0028] Figure 4 This is a noise reduction effect diagram of the N=4 local polynomial fitting method in the data filtering method for a weighing system described in the present invention;
[0029] Figure 5 The noise reduction effect of the N=1 local polynomial fitting method in the data filtering method for a weighing system described in the present invention;
[0030] Figure 6 It is a schematic diagram of the working of the adaptive filter in the data filtering method for a weighing system according to the present invention;
[0031] Figure 7 This is a diagram showing the impact suppression effect of the adaptive filter in the data filtering method for a weighing system described in the present invention;
[0032] Figure 8 This is a diagram showing the suppression effects of the sliding filter processing and the adaptive filter in the data filtering method for a weighing system described in the present invention. DETAILED DESCRIPTION
[0033] The present invention is further described in detail below based on the accompanying drawings and embodiments. The described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0034] like Figure 1 and Figure 2 As shown, this scheme provides a data filtering method for a weighing system, comprising the following steps:
[0035] Step (1) collecting a weighing signal through a weighing sensor as an original signal;
[0036] Step (2) filtering the original signal through a sliding window polynomial filter to obtain a filtered signal; comprising: performing local polynomial fitting on the original signal to estimate a smoothing value to obtain the filtered signal;
[0037] Step (3) filters and suppresses noise on the filtered signal through an adaptive filter to obtain filtered data for use in subsequent processes.
[0038] This scheme mainly includes two key components: data preprocessing filtering and adaptive filtering.
[0039] In the data preprocessing filtering stage, the main challenge is to effectively filter out the high-frequency noise in the signal while minimizing the time delay introduced by the filtering process. To meet this challenge, a polynomial fitting filtering algorithm is used, which ensures the real-time processing by maintaining the original characteristics and shape of the signal while reducing the noise.
[0040] Entering the adaptive filtering stage, the core difficulty lies in accurately determining the step size parameter of the filter. The setting of the step size parameter directly affects the response speed and learning effect of the filter.
[0041] If the step size is too large, the filter may respond too quickly, resulting in overfitting, making the processed signal amplitude abnormal;
[0042] If the step size is too small, the filter may respond slowly and fail to accurately reflect the signal characteristics in a timely manner.
[0043] Therefore, through the adaptive filtering strategy of dynamically adjusting the step size based on real-time signal changes provided by this solution, the step size of the filter can be automatically adjusted by comprehensively analyzing the characteristics and noise pattern of the weighing signal to achieve a fast and accurate response to the signal.
[0044] As a specific description of step (1) in this embodiment, the original signal generally comes from the voltage value of the voltage signal of the weighing sensor after AD conversion. In fact, similar digital signals with high-frequency noise can be processed by the method provided by the present invention.
[0045] As a specific description of step (2) in this embodiment, there are many stationary noises in the weighing signal, which are characterized by long duration, certain periodicity and regularity. Obvious features can be seen in the time domain analysis, such as the noise signal contained in the original signal is always odd-symmetrical.
[0046] Based on this feature, the sliding average filter can be used to preprocess the original signal. The window length of the sliding average method determines the smoothness of the filtered signal. The longer the window, the smoother it is. However, it will also lead to a decrease in the peak value of the weighing signal, such as Figure 3 shown.
[0047] Therefore, the smoothing value is estimated by fitting the local polynomial of the data points. The principle of this method is mainly as follows:
[0048] Assume that the length of a window is 2M+1 and the data in the window is X n ={X -m ,X -m+1 ,…X0,…X m-1 ,X m}, perform polynomial fitting on the data in the window, assuming that the fitted polynomial is:
[0049]
[0050] In the formula, N is the polynomial order, n is the data index in the window, and a k is a multi-factor.
[0051] It should be noted that the sliding window polynomial filter described in this embodiment is a signal processing technology. The filter is based on the idea of polynomial fitting, takes the data points in the sliding window as the input of the polynomial, and solves the smoothed function value through polynomial calculation. It uses the least squares method or other optimization methods to find the best fitting polynomial, thereby estimating the filtered value.
[0052] As a specific description of step (3) in this embodiment, the adaptive filter weights the input signal through a filter with adjustable coefficients to generate an output, which is then compared with a desired reference or training signal to form an error signal.
[0053] The error signal can be used to correct and update the weight coefficients of the filter, and eventually the error is gradually minimized, so that the output of the processor is closer to the reference or training signal. Its structure is as follows: Figure 6 As shown in the figure, x(n) is the original signal to be filtered, y(n) is the signal after filtering, d(n) is the expected signal, and e(n) is the error signal used by the adaptive algorithm to update the filter weights.
[0054] The present invention is based on the variant adaptive NLMS filter for improvement, mainly when the weight is updated, the specific process is as follows:
[0055] y(n)=W T (n)x(n)
[0056] Where W(n) is the filter weight coefficient, T is the matrix transpose, and x(n) is the original signal to be filtered.
[0057] This patent introduces an update step size μ for determining the coefficient weights and adopts an adaptive solution. The specific formula is as follows:
[0058] W(n)=f(μ,e)
[0059] Wherein, f(μ, e) can be determined by a variety of functions, and this patent uses a single function to iterate the calculation result.
[0060] This embodiment proposes to accurately define the quantization range for updating the step size μ, which can ensure that the system is stable and does not diverge. The specific range is as follows:
[0061]
[0062] Where L is the order of the adaptive filter.
[0063] In order to demonstrate the effect of the processing of this method, this embodiment is verified according to the following steps:
[0064] For the acquisition of the original signal in step (1)
[0065] The verification uses common resistance strain type weighing sensors to collect data, and uses Xinjie weighing dedicated modules for signal acquisition. (For example, the Xinjie weighing dedicated module with model number XD-E4WT-C is developed and produced for industrial sites that require dynamic weighing. The weighing module has the characteristics of small size, stable performance, simple and applicable operation. It can be widely used in: seed, chemical, grain, feed weight control and other occasions.)
[0066] Obtaining and applying the sliding window polynomial filter coefficients in step (2)
[0067] Through the local polynomial fitting method, a set of filter coefficients can be obtained under the conditions of setting the polynomial order N to 4 and the window length M to 11. These coefficients are applied to the finite impulse response (FIR) filter, which significantly suppresses the periodic noise in the data.
[0068] like Figure 4As shown in the figure, the dotted line represents the original noisy data, and the solid line represents the data after polynomial fitting. It can be seen that there is an obvious noise reduction effect. Nevertheless, the processed data still closely follows the original curve, indicating that the smoothness can be further improved.
[0069] When the polynomial order N is reduced to 1 and the window length M is kept at 11, another set of filter coefficients can be obtained and applied to the FIR filter.
[0070] like Figure 5 As shown, reducing the order of the polynomial makes the processed data more stable and better meets the stability requirements of the weighing display.
[0071] Verification of the adaptive filter in step (3)
[0072] For weighing, the noise sources of the system are mainly: self-mechanical vibration, external natural environment interference, object rapid weighing interference, and external force collision interference. Self-mechanical vibration and external natural environment interference exist throughout the entire life cycle of the weighing signal. Object rapid weighing interference and external force collision interference can be approximately regarded as instantaneous noise. Therefore, when the adaptive algorithm begins to adaptively converge on the weighing signal, the smaller the error, the closer it is to the tracked signal.
[0073] During the verification process, the test found that the adaptive filter’s suppression effect on impulse noise under a small step size setting is close to that of a fixed-parameter FIR filter.
[0074] like Figure 7 As shown in the figure, the solid line represents the output value of the adaptive filter, the dotted line represents the original analog-to-digital (AD) conversion value, and the dashed line represents the output of the fixed parameter FIR filter. The results show that the adaptive filter can effectively suppress the impact.
[0075] also, Figure 8 The signal graph of the sliding filter (solid line) and the signal graph after the adaptive filter processing (dashed line) are shown. The signal processed by the adaptive filter has weakened peaks and valleys, and the delay between the signal and the original signal is only 5ms, which is significantly better than the traditional filtering result. This proves that the adaptive filter can achieve excellent noise suppression while maintaining a high response speed.
[0076] The above description is only for illustrating the implementation mode of the present invention and is not intended to limit the present invention. For those skilled in the art, any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A data filtering method for a weighing system, characterized in that: include: Step (1) collecting a weighing signal through a weighing sensor as an original signal; Step (2) filtering the original signal through a sliding window polynomial filter to obtain a filtered signal; comprising: performing local polynomial fitting on the original signal to estimate a smoothing value to obtain the filtered signal; Step (3) filters and suppresses noise on the filtered signal through an adaptive filter to obtain filtered data for use in subsequent processes.
2. A data filtering method for a weighing system according to claim 1, characterized in that: The step (2) of performing local polynomial fitting on the original signal to estimate the smoothed value is specifically as follows: Set the window length to 2M+1 and the data in the window to X n ={X -m ,X -m+1 ,…X0,…X m-1 ,X m }, perform polynomial fitting on the data in the window, and set the fitting polynomial to: In the formula, N is the polynomial order, n is the data index in the window, and a k is a multi-factor.
3. The data filtering method for a weighing system according to claim 1, characterized in that: In step (3), the filtered signal is filtered and noise suppressed by an adaptive filter to obtain filtered data y(n), and the formula is as follows: y(n)=W T (n)x(n) Where W(n) is the filter weight coefficient, T is the matrix transpose, and x(n) is the original signal to be filtered.
4. A data filtering method for a weighing system according to claim 3, characterized in that: Introducing an update step size μ to determine the coefficient weights, we have: W(n)=f(μ,e) Where f(μ, e) is the result of a linear function iteration.
5. A data filtering method for a weighing system according to claim 4, characterized in that: The update step size μ is as follows: Where L is the order of the adaptive filter.