Adaptive Filtering-Based Method for Extracting Active Movements of Tremor Patients and Electronic Device

Abstract

The present invention provides a method for extracting the active movement of tremor patients based on adaptive filtering and an electronic device, aiming to improve the filtering ability of the filter for tremor signals and reduce the lag of the active movement signals. It includes: proposing a first-order filter structure with adjustable cut-off frequency; subtracting the original measured signal from the filtered signal and inputting the result into a Hamming window, and obtaining the main frequency of the error signal through fast Fourier transform; in view of the frequency difference between the active movement signal and the tremor signal, using the main frequency of the error signal to set an adaptive law to adjust the filter cut-off frequency in real time. In view of the problem that it is difficult to extract the active movement signal of wrist tremor patients, the present invention proposes a method for extracting the active movement of tremor patients based on adaptive filtering, realizing high-quality extraction of the active movement signal from the measured signal, providing technical guidance for solving the control of the actual exoskeleton system and the problem of human-computer interaction, and having important theoretical and application values.

Description

Technical Field

[0001] The present invention relates to the technical field of rehabilitation robot control systems, and particularly to a method for extracting active movement of tremor patients based on adaptive filtering and an electronic device. Background Art

[0002] Wrist tremor is the main symptom of Parkinson's patients. It is an involuntary and irregular oscillatory movement, whose amplitude varies with time and frequency is relatively stable. Existing tremor treatment methods mainly include drug treatment and deep brain electrical stimulation, etc., which have problems such as drug resistance and large side effects. In recent years, the development of robot technology has been rapid, and the anti-tremor exoskeleton set by using robot technology provides a new solution for traditional medicine. The working mechanism of the anti-tremor exoskeleton is to adjust the damping and inertia of the interactive coupling system composed of the patient's upper limb and the exoskeleton mechanical structure by controlling the output of the actuator installed on the exoskeleton structure, and then change the frequency response of the system to achieve anti-tremor. Compared with traditional treatment methods, the non-invasive anti-tremor exoskeleton has the advantages of low risk and good wearability.

[0003] As a non-subjective intentional movement, tremor often appears together with the human intentional movement. Selecting sensors such as pressure sensors, gyroscopes or accelerometers to measure the human body signal contains two parts: tremor signal and active movement signal. Scholars' research shows that the frequency band of wrist tremor of Parkinson's patients is 3.5 - 7.7 Hz, while the frequency band of active movement is 0 - 2 Hz. Since the frequency bands of the two are very close, traditional linear filters cannot completely separate the two. Therefore, how to separate tremor movement and active movement from the measured signal is the key to realizing the control of the anti-tremor exoskeleton, laying a foundation for subsequent high-precision tracking of human active movement, and having important theoretical and application values. Summary of the Invention

[0004] The main technical problem to be solved by the present invention is to separate the tremor signal and the active movement signal from the original measured signal of tremor patients and achieve high-quality extraction of the active movement signal.

[0005] The technical solution adopted by the present invention to solve the above problems is: providing a method for extracting active movement of tremor patients based on adaptive filtering, including the following steps:

[0006] Obtain the original measured signal, which includes a tremor signal and an active movement signal;

[0007] Filter the original measured signal using a first-order low-pass filter to obtain a filtered signal;

[0008] Take the difference between the original measured signal and the filtered signal to obtain an error signal, input the error signal into a Hamming window, and perform a fast Fourier transform on the windowed error signal to obtain the main frequency of the error signal;

[0009] Set an adaptive law to adjust the cut-off frequency of the first-order low-pass filter in real time according to the magnitude of the main frequency, so as to filter out the tremor signal by reducing the cut-off frequency when the wrist trembles; when the wrist does not tremble, increase the cut-off frequency to reduce the lag of the first-order low-pass filter in estimating the active movement signal.

[0010] Furthermore, the first-order low-pass filter is a first-order system with an adjustable cut-off frequency, and its specific expression is as follows:

[0011] T(s) = F a (s) / F(s) = ω c / (s + ω c )

[0012] Its time-domain state equation is where f(t) is the original measurement signal, and f a (t) represents the estimated active movement signal obtained after filtering, is the derivative of f a (t), s represents the Laplace operator, and ω c represents the cut-off frequency of the first-order low-pass filter, and F(s) and F a (s) are the results of the Laplace transforms of f(t) and f a (t) respectively.

[0013] Furthermore, when filtering the original measurement signal using a first-order low-pass filter, the cut-off frequency is taken as 25 Hz.

[0014] Furthermore, the Hamming window size N needs to satisfy the following constraint conditions:

[0015]

[0016] where k is a positive integer, R f is the resolution of the main frequency, and R f = 1 / (NT s ), and T s is the system sampling period.

[0017] Furthermore, determine whether the wrist trembles according to the magnitude of the main frequency, and set an adaptive law to adjust the cut-off frequency of the first-order low-pass filter in real time. The adaptive law is set as follows:

[0018] If f w > 3.5 Hz, then ω c = 2 Hz; otherwise ω c = 25 Hz

[0019] where f w represents the main frequency, and ω cRepresents the cut-off frequency of a first-order low-pass filter.

[0020] In addition, to solve the above technical problems, the present invention also provides an electronic device, which includes a memory and a processor. A computer program is stored on the memory, and when the computer program is executed by the processor, the active movement extraction method for tremor patients is executed.

[0021] The beneficial effects brought by the technical solution provided by the present invention are as follows: Aiming at the problem that it is difficult to extract the active movement signals of wrist tremor patients widely existing, a method for extracting the active movement of tremor patients based on adaptive filtering is proposed, aiming to improve the filtering ability of the filter for tremor signals and reduce the lag of the active movement signals, realizing the high-quality extraction of the active movement signals from the measured signals, providing technical guidance for solving the control of the actual exoskeleton system and the problem of human-computer interaction, and having important theoretical and application values. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The specific effects of the present invention will be further described below in conjunction with the drawings. In the drawings:

[0023] Figure 1 is a flowchart of a method for extracting the active movement of tremor patients based on adaptive filtering according to the present invention;

[0024] Figure 2 is a system structure block diagram of a method for extracting the active movement of tremor patients based on adaptive filtering according to the present invention;

[0025] Figure 3 is a filtering effect diagram using a first-order low-pass filter with a fixed cut-off frequency;

[0026] Figure 4 is a filtering effect diagram of a method for extracting the active movement of tremor patients based on adaptive filtering according to the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0027] In order to have a clearer understanding of the technical features, objectives, and effects of the present invention, the specific implementation manners of the present invention will now be described in detail with reference to the drawings.

[0028] Please refer to Figure 1 , the embodiment of the present invention provides a flowchart of a method for extracting the active movement of tremor patients based on adaptive filtering, which specifically includes:

[0029] S1: Obtain the original measured signal, which includes tremor signals and active movement signals;

[0030] Specifically, sensors such as pressure sensors, gyroscopes, or accelerometers can be selected to measure the signals of tremor patients, which include two parts: tremor signals and active movement signals;

[0031] S2: Filter the original measurement signal using a first-order low-pass filter Since the frequency bands of the tremor signal and the active movement signal are very close, a fixed cut-off frequency ω c cannot meet the requirement of completely separating the two parts of the signal. Therefore, continue to set an adaptive law through the following steps to adjust ω c in real time.

[0032] In this embodiment, the original measurement signal is filtered using a first-order low-pass filter, and the initial cut-off frequency is set to 25 Hz.

[0033] S3: Subtract the original measurement signal from the filtered signal to obtain an error signal e(t). Input e(t) into a moving Hamming window (with N points) to prevent frequency leakage. Perform a fast Fourier transform (FFT) on the windowed signal to obtain its main frequency f w .

[0034] S4: Set the adaptive law and adjust the cut-off frequency ω w of the low-pass filter in real time according to the magnitude of f c , so that when the wrist trembles, ω c is adjusted smaller to increase the ability of the filter to filter out the tremor signal; when there is no tremor, ω c is increased to reduce the filtering lag of the filter for the active movement signal.

[0035] Please refer to Figure 2 , Figure 2 which is the system structure block diagram of a method for extracting active movement of tremor patients based on adaptive filtering according to the present invention; it includes the fixed structure of the low-pass filter and the adaptive part;

[0036] The first-order low-pass filter adopted in step S2 is a first-order system with an adjustable cut-off frequency, specifically as follows:

[0037] The first-order low-pass filter is T(s) = F a (s) / F(s) = ω c / (s + ω c ), and its time-domain state equation is where f(t) is the original measurement signal, and f a (t) represents the estimated active movement signal obtained after filtering, is the derivative of f a (t), s represents the Laplace operator, and ω c represents the cut-off frequency of the low-pass filter, and F(s) and F a (s) are the results of the Laplace transforms of f(t) and f a (t) respectively.

[0038] A first-order low-pass filter with a fixed cut-off frequency is difficult to meet the requirements for separating tremor signals and active movement signals, as shown in Table 1 specifically:

[0039] Table 1: Amplitude ratio and phase of active movement and tremor movement at two fixed cut-off frequencies

[0040]

[0041] It can be seen that when ω c = 2 Hz, 92% of the tremor signal can be filtered out, but the lag of the active movement signal is very serious (72°); if ω c = 25 Hz, its lag for the active movement signal is smaller (14°), but it loses the filtering ability for the tremor signal (71% of the tremor movement remains after filtering). Therefore, it is very necessary to adjust ω c in real time by detecting whether there is wrist tremor.

[0042] Furthermore, step S3 introduces how to judge whether there is wrist tremor, in which a Hamming window is introduced and a method for selecting the size N of the Hamming window is given, as follows:

[0043] S301: Subtract the filtered signal f a (t) from the original measurement signal f(t) to obtain the error signal e(t) = f(t) - f a (t). Since T(s) is a low-pass filter, the main components in e(t) are measurement noise and the filtered tremor signal. Assuming that the measurement noise is Gaussian white noise and its power spectral density is uniformly distributed at all frequencies, the main frequency component in the spectrum obtained by performing a Fourier transform on e(t) must depend on the main frequency of the tremor signal.

[0044] S302: Since directly truncating the signal (with a rectangular window) will cause frequency leakage, this problem is improved by using a Hamming window. Perform a fast Fourier transform (FFT) on the data within the Hamming window in real time, and define the frequency corresponding to the maximum amplitude (main frequency) in the obtained power spectrum as f w .

[0045] S303: Calculate that the resolution of f w is R f = 1 / (NT s ), where T s is the system sampling period. The method for selecting the size N of the Hamming window is to select the smallest N that satisfies the following constraint (according to the properties of the fast Fourier transform, N needs to be selected as an integer power of 2):

[0046]

[0047] where k is a positive integer.

[0048] Further, in step S4, the main frequency f w is used to determine whether the wrist trembles, and the cut-off frequency ω of the low-pass filter T(s) is adjusted in real time using the adaptation law c as follows:

[0049] S401: Establish the adaptation law:

[0050] If f w > 3.5 Hz, then ω c = 2 Hz; otherwise ω c = 25 Hz

[0051] Realize the automatic switching of the filter, so as to realize the high-quality extraction of the active motion signal in the measurement signal.

[0052] In this embodiment, the signal measured by a common pressure sensor on the exoskeleton is used as the original measurement signal, and the active motion signal is extracted from the original measurement signal using a low-pass filter with a fixed cut-off frequency and the adaptive filter proposed in the present invention respectively. The sampling period of the pressure sensor is selected as T s = 0.01 s. According to the selection strategy in S303, N needs to be selected as an integer power of 2. If N is selected as 32, then

[0053] R f = 1 / (NT s ) = 3.125

[0054] It is impossible to find a positive integer k that satisfies the constraint

[0055]

[0056] If N = 64 is selected, then R f = 1 / (NT s ) = 1.5625. When k = 1 is taken

[0057]

[0058] it is satisfied. Therefore, the size of the Hamming window N = 64 is selected. In the simulation, the original measurement signal is set as:

[0059]

[0060] Assume that the wrist trembles when 10 s ≤ t < 15 s, and the rest of the time is in active motion. Using the above parameters for the simulation experiment, the experimental results are as shown in Figure 3 and Figure 4 shown. Figure 3 Shows the filtering effect when using a low-pass filter with a fixed cut-off frequency. It can be clearly seen that ωc When the frequency is 25 Hz, the filter cannot effectively filter out the tremor signal, while when ω c is 2 Hz, although the tremor signal is filtered out, there is also a large attenuation and delay for the active movement signal. Figure 4 The effect of the adaptive filter is shown. Figure 4 (a) is the result of f w . Figure 4 (b) is the comparison between the original measured signal and the filtered result. It can be clearly seen that the method of the present invention can judge whether the wrist trembles and adjust the cut-off frequency of the filter in real time, meeting the requirement of accurately extracting the active movement signal from the original measured signal.

[0061] In other embodiments of the present invention, an electronic device is further provided. The electronic device includes a memory and a processor. A computer program is stored on the memory. When the computer program is executed by the processor, the method for extracting the active movement of a tremor patient is executed.

[0062] The embodiments of the present invention provide a method for extracting the active movement of a tremor patient based on adaptive filtering and an electronic device, aiming to improve the filtering ability of the filter for tremor signals and reduce the lag of the active movement signal. The method includes: First, a first-order filter structure with an adjustable cut-off frequency is proposed; then, the difference between the original measured signal and the filtered signal is input into a Hamming window, and the main frequency of the error signal is obtained through a fast Fourier transform; finally, in view of the frequency difference between the active movement signal and the tremor signal, the main frequency of the error signal is used to set an adaptive law to adjust the cut-off frequency of the filter in real time. The embodiments of the present invention propose a method for extracting the active movement of a tremor patient based on adaptive filtering for the problem that it is difficult to extract the active movement signal of patients with widespread wrist tremors, realizing the high-quality extraction of the active movement signal from the original measured signal, providing technical guidance for solving the control of actual exoskeleton systems and human-computer interaction problems, and having important theoretical and application values.

[0063] It should be noted that in this article, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, article or system. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, article or system including that element.

[0064] The serial numbers of the embodiments of the present invention above are only for description and do not represent the advantages or disadvantages of the embodiments. Among the several apparatus claims listing several apparatuses, several of these apparatuses may be embodied by the same hardware item. The use of the terms first, second, and third, etc. does not denote any order and these terms may be construed as identifiers.

[0065] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be similarly included in the patent protection scope of the present invention.

Claims

1. An active movement extraction method for tremor patients based on adaptive filtering, characterized in that, it includes the following steps: Obtain the original measurement signal, which includes a tremor signal and an active movement signal; Filter the original measurement signal using a first-order low-pass filter to obtain the filtered signal; Subtract the filtered signal from the original measurement signal to obtain an error signal, input the error signal into a Hamming window, perform a fast Fourier transform on the windowed error signal, and obtain the main frequency of the error signal; Set an adaptive law, and adjust the cut-off frequency of the first-order low-pass filter in real time according to the magnitude of the main frequency, so as to filter out the tremor signal by reducing the cut-off frequency when the wrist trembles; when the wrist does not tremble, increase the cut-off frequency to reduce the lag of the first-order low-pass filter in estimating the active movement signal; The size N of the Hamming window needs to satisfy the following constraint conditions: where k is a positive integer, R f is the resolution of the main frequency, Rf = 1 / (NTs), T s is the system sampling period.

2. The active movement extraction method for tremor patients according to claim 1, characterized in that, the first-order low-pass filter is a first-order system with an adjustable cut-off frequency, and the specific expression is as follows: T(s) = F a (s) / F(s) = ω c / (s + ω c ) Its time-domain state equation is where f(t) is the original measurement signal, and f a (t) represents the active motion estimation signal obtained after filtering, is the derivative of f a (t), s represents the Laplace operator, and ω c represents the cut-off frequency of the first-order low-pass filter. F(s) and Fa(s) are the results of the Laplace transforms of f(t) and fa(t) respectively.

3. The active movement extraction method for tremor patients according to claim 1, characterized in that, when filtering the original measurement signal using a first-order low-pass filter, the cut-off frequency is set to 25 Hz.

4. The active movement extraction method for tremor patients according to claim 1, characterized in that, judge whether the wrist trembles according to the magnitude of the main frequency, and set an adaptive law to adjust the cut-off frequency of the first-order low-pass filter in real time. The adaptive law is set as follows: If f w > 3.5 Hz, then ω c = 2 Hz; otherwise ω c = 25 Hz where f w represents the main frequency, and ω c represents the cut-off frequency of the first-order low-pass filter.

5. An electronic device, characterized in that, the electronic device includes a memory and a processor, and a computer program is stored on the memory. When the computer program is executed by the processor, it executes the active movement extraction method for tremor patients according to any one of claims 1-4.

Images