阅读说明：本技术 消除加速度传感器直流分量和趋势项的信号处理方法 (Signal processing method for eliminating direct current component and trend term of acceleration sensor ) 是由 李晓龙 刘栋材 吴先瑜 于 2020-05-26 设计创作，主要内容包括：一种消除加速度传感器直流分量和趋势项的信号处理方法,在加速度传感器的前端信号分发设备中嵌入信号处理算法,利用信号处理算法对待处理信号进行信号处理,利用信号处理算法得到去直流分量和趋势项的信号,实时消除加速度传感器中的直流分量和趋势项。本发明通过在加速度传感器的前端信号分发设备中嵌入信号处理算法,并通过对加速度传感器的信号进行处理,得到去直流分量和趋势项的信号,能够实时消除加速度传感器直流分量和趋势项,较传统解决方案在实时性、准确性和经济性上均具有明显优势。(A signal processing method for eliminating direct current components and trend items of an acceleration sensor is characterized in that a signal processing algorithm is embedded in front-end signal distribution equipment of the acceleration sensor, signals to be processed are processed through the signal processing algorithm, signals without the direct current components and the trend items are obtained through the signal processing algorithm, and the direct current components and the trend items in the acceleration sensor are eliminated in real time. According to the invention, the signal processing algorithm is embedded in the front-end signal distribution equipment of the acceleration sensor, and the signals of the acceleration sensor are processed to obtain the signals without the direct-current components and the trend items, so that the direct-current components and the trend items of the acceleration sensor can be eliminated in real time, and the method has obvious advantages in real time, accuracy and economy compared with the traditional solution.)

1. A signal processing method for eliminating direct current components and trend terms of an acceleration sensor is characterized by comprising the following steps: embedding a signal processing algorithm in front-end signal distribution equipment of the acceleration sensor, performing signal processing on a signal to be processed by using the signal processing algorithm, obtaining a signal without a direct-current component and a trend term by using the signal processing algorithm, and eliminating the direct-current component and the trend term in the acceleration sensor in real time.

2. The signal processing method of eliminating a direct current component and a trend term of an acceleration sensor according to claim 1, characterized in that: the signal processing algorithm comprises the steps of firstly obtaining a signal to be processed through sampling, then removing a direct current component in the signal to be processed by utilizing Fourier transform and inverse Fourier transform, and then removing a trend term from the signal without the direct current component through operation to obtain a signal without the direct current component and the trend term; and sequentially intercepting the signals of the window size from the processed signal y4 as an acceleration sensor output signal y.

3. The signal processing method of eliminating a direct current component and a trend term of an acceleration sensor according to claim 2, characterized in that: the signal to be processed is obtained by sampling, wherein a sliding window signal y0 is determined according to the sampling frequency fs and the vibration characteristics of the measured object, then the sliding window signal y0 is used as the first half of the signal to be processed, the last output signal y is used as the second half of the signal to be processed, and the two sections of signals are spliced to obtain a signal to be processed y 1.

4. The signal processing method of eliminating a direct current component and a trend term of an acceleration sensor according to claim 3, characterized in that: the vibration characteristics of the measured object refer to the vibration frequency and the vibration amplitude of the measured object; the sliding window signal y0 is typically 10-20 times fs window length.

5. The signal processing method of eliminating a direct current component and a trend term of an acceleration sensor according to claim 2, characterized in that: the signal to be processed is subjected to fast Fourier transform and then high-pass filtering, and the filtered signal is subjected to inverse Fourier transform to obtain a signal y2 subjected to direct current removal.

6. The signal processing method of eliminating a direct current component and a trend term of an acceleration sensor according to claim 5, characterized in that: the high-pass filtering after the fast Fourier transform of the signal y1 to be processed is the high-pass filtering after the fast Fourier transform of the signal y1 to be processed, and is characterized in that the cut-off frequency of the high-pass filtering is generally 0.1Hz, so that the low-frequency component of the original signal is not influenced on the premise of filtering out the direct-current component.

7. The signal processing method of eliminating a direct current component and a trend term of an acceleration sensor according to claim 2, characterized in that: the signal without the direct current component is subjected to least square polynomial fitting by calculating a detrending item, namely the signal without the direct current component y2 is subjected to least square polynomial fitting to obtain a fitting expression y 3; and subtracting a least square polynomial y3 from the signal y1 to be processed to obtain a signal y4 with direct-current components and trend terms removed.

8. The signal processing method of eliminating a direct current component and a trend term of an acceleration sensor according to claim 7, characterized in that: the signal to be processed y1 minus the least squares polynomial y3 is:

the least squares polynomial fit is a quadratic polynomial, and the fitting expression y3 is expressed as:

y3＝ax²+bx+c (1)

the expression of the signal y4 of the direct current component and the trend term is as follows:

y4＝y1-y3 (2)

9. the signal processing method of eliminating a direct current component and a trend term of an acceleration sensor according to claim 2, characterized in that: the signals with the window size are intercepted from the processed signals y4 in sequence to serve as output signals y, and the intercepted sequence is consistent with the sliding window updating sequence of the signals y1 to be processed; the front-end signal distribution equipment can receive signals acquired by the sensor in real time, integrates a signal processing algorithm, and has a storage function and an operation function. The signals collected by the sensor are processed by the front-end signal distribution equipment, and then the processed signals are packaged and sent to the receiving end.

10. The signal processing method of eliminating a direct current component and a trend term of an acceleration sensor according to claim 1, characterized in that: the specific implementation steps for eliminating the direct current component and the trend item in the acceleration sensor are as follows:

1) determining a sliding window signal y0 according to the sampling frequency fs and the vibration characteristics of the measured object;

2) taking the sliding window signal y0 as the first half part of a signal to be processed, taking the last output signal y as the second half part of the signal to be processed, and splicing the two sections of signals to obtain a signal to be processed y 1;

3) performing fast Fourier transform on the signal y1 to be processed, performing high-pass filtering, and performing inverse Fourier transform on the filtered signal to obtain a signal y2 subjected to direct current removal;

4) performing least square polynomial fitting on the direct current signal removed y2 to obtain a fitting expression y 3;

5) subtracting a least square polynomial y3 from a signal y1 to be processed to obtain a signal y4 from which direct-current components and trend terms are removed;

6) sequentially intercepting window-sized signals from the processed signal y4 as output signals y;

7) a1.1-1.6 signal processing algorithm is embedded into front-end signal distribution equipment, and a direct-current component and a trend item of a sensor are eliminated in real time.

Technical Field

The invention relates to a signal processing method of a sensor, in particular to a signal processing method for eliminating a direct current component and a trend item of an acceleration sensor, which can effectively eliminate the direct current component of the sensor in real time and eliminate a signal trend direction caused by the environment in real time; belongs to the technical field of sensor signal processing.

Background

An acceleration sensor is a type of sensor that measures the force acting on an object during its vibration. At present, the acceleration sensor is widely applied to various fields of automobile vibration monitoring, rail transit pantograph-catenary relation monitoring, earthquake monitoring, game control, military industry and the like.

In the application of the acceleration sensor, firstly, zero marking processing needs to be carried out on the measured value of the acceleration sensor, and in practical use, due to the influence of human factors or other complex environment working conditions, the acceleration sensor often has the condition that zero is not completely marked due to direct current components when a measured object is static. In addition, because the application range of the acceleration sensor is extremely wide, the working environment of the acceleration sensor can be quite complex, and environmental factors such as temperature, air pressure and the like can cause the drift of the measurement value of the acceleration sensor, thereby causing an obvious signal trend term. In a measurement environment with relatively high requirements, it is necessary to eliminate the direct current component and the trend term of the acceleration sensor.

The traditional method generally calibrates the cut-off time once to ensure that the interval time is calibrated to zero; for trend terms caused by environmental factors such as temperature, acceleration sensors are generally tested in laboratories, sensor signal acquisition is performed once for each temperature segment, and then trend compensation at the corresponding temperature is obtained through fitting and interpolation. There are three main problems with the conventional approach. The method has the problems that the separation time calibration wastes manpower and material resources once, and the economic benefit is poor; secondly, the accidental zero value jump of the acceleration sensor cannot be corrected in time; the third problem is that the environment simulated by the laboratory is limited, and the given compensation expression of the sensor may have large errors in practical application.

Patent documents in which the same technology as that of the present invention is not found through patent search are reported, and the following patents which have a certain relationship with the present invention are mainly included:

1. the patent number is CN201710273858.9, entitled "a fully differential force balance mode MEMS acceleration sensor signal processing circuit", and the applicant is: the invention patent of Hunan Tan university discloses a signal processing circuit of a fully differential force balance mode MEMS acceleration sensor, which comprises a fully differential switched capacitor circuit, a sample-and-hold circuit, an instrument amplifier, a multiphase clock circuit and a reference source circuit, wherein the fully differential switched capacitor circuit detects MEMS variable capacitance, outputs differential signals, is connected to the instrument amplifier through the sample-and-hold circuit, and converts the differential signals into single-ended output signals through the instrument amplifier. The acceleration sensor signal processing circuit can realize high-precision and low-noise voltage output.

2. The patent number is CN201310346010.6, the name is "an inertial acceleration sensor frequency control force feedback signal processing circuit", the applicant is: the invention patent of the university of Fuzhou discloses a frequency control force feedback signal processing circuit of an inertial acceleration sensor, which comprises a weak signal amplifying circuit, a filtering and integrating circuit, a current mode controllable oscillator and a digital differential circuit. The input end and the output end of the weak signal amplifying circuit are respectively connected with the acceleration sensor and the filtering integration circuit so as to amplify the acquired weak signals; the input end and the output end of the filtering integration circuit are respectively connected with the weak signal amplification circuit and the current mode controllable oscillator so as to amplify the error of the integration signal; the input end and the output end of the current mode controllable oscillator are respectively connected with the filtering integration circuit and the digital differential circuit to output signals by the phase integration system; the input end and the output end of the digital differential circuit are respectively connected with the current mode controllable oscillator and the acceleration sensor so as to output a quantization noise shaping signal containing feedback information to be fed back to the feedback input end of the acceleration sensor.

3. The patent number is CN201811109023.0 entitled "a method for compensating acceleration sensor signal for thrombelastogram apparatus", applicant is: the invention patent of Shenzhen Warden Life technologies, Inc., discloses an acceleration sensor signal compensation method for a thromboelastogram instrument, and eliminates the interference of a sampling noise signal by adopting N-point mean filtering; meanwhile, a piezoelectric sensor technology is adopted, a vibration signal digital model is analyzed, vibration noise caused by vibration is eliminated, and effective dynamic signals are reserved; because the frequency and the amplitude of the vibration have periodicity and randomness and are superposed with effective signals, the good effect cannot be achieved under special conditions completely through a digital signal processing technology, and therefore, under the condition that a feedback signal of the piezoelectric sensor has strong correlation with vibration noise, a dynamic signal is compensated through the feedback signal input of the piezoelectric sensor, and a better filtering effect can be achieved.

Through careful analysis of the above patents, although all of them relate to an acceleration sensor, some improved technical solutions for the acceleration sensor are proposed, some of them also relate to a signal processing circuit of the acceleration sensor; however, through careful analysis, none of the patents deals with the direct current component and signal trend of the sensor of the acceleration sensor, so the problems mentioned above still exist, and further research and improvement are still needed.

Disclosure of Invention

The invention aims to provide a signal processing method for eliminating direct current components and trend items of an acceleration sensor aiming at three problems of the existing acceleration sensor processing method, wherein the direct current components and the trend items of the acceleration sensor are eliminated in real time without calibrating the sensor; the sensor does not need to be subjected to a laboratory environment compensation experiment, and the signal trend direction caused by the environment is eliminated in real time.

In order to achieve the purpose, the invention provides a signal processing method for eliminating a direct current component and a trend item of an acceleration sensor.

Further, the signal processing algorithm is that a signal to be processed is obtained through sampling, then a direct current component in the signal is removed from the signal to be processed through Fourier transform and inverse Fourier transform, and then the signal without the direct current component is subjected to trend term operation to obtain a signal without the direct current component and the trend term; and sequentially intercepting the signals of the window size from the processed signal y4 as an acceleration sensor output signal y.

Further, the step of obtaining the signal to be processed through sampling is to determine a sliding window signal y0 according to the sampling frequency fs and the vibration characteristics of the measured object, use the sliding window signal y0 as the first half of the signal to be processed, use the last output signal y as the second half of the signal to be processed, and splice the two sections of signals to obtain the signal to be processed y 1.

Further, the vibration characteristics of the object to be measured refer to the vibration frequency and the vibration amplitude of the object to be measured; the sliding window signal y0 is typically 10-20 times fs window length.

Further, the step of removing the direct current component in the signal to be processed by using fourier transform and inverse fourier transform is to perform high-pass filtering after performing fast fourier transform on the signal to be processed y1, and perform inverse fourier transform on the filtered signal to obtain a signal y2 after direct current removal.

Furthermore, the high-pass filtering after the fast fourier transform of the signal to be processed y1 is the high-pass filtering after the fast fourier transform of the signal to be processed y1, and the method is characterized in that the cutoff frequency of the high-pass filtering is generally 0.1Hz, so that the low-frequency component of the original signal is not affected on the premise of ensuring the filtering of the direct-current component.

Further, the step of performing operation detrending on the signal without the direct current component is to perform least square polynomial fitting on the detrended signal y2 to obtain a fitting expression y 3; and subtracting a least square polynomial y3 from the signal y1 to be processed to obtain a signal y4 with direct-current components and trend terms removed.

Further, the subtraction of the least squares polynomial y3 from the signal to be processed y1 is:

the least squares polynomial fit is a quadratic polynomial, and the fitting expression y3 is expressed as:

y3＝ax²+bx+c (1)

the expression of the signal y4 of the direct current component and the trend term is as follows:

y4＝y1-y3 (2)

further, the signals of the window sizes are sequentially intercepted from the processed signal y4 to serve as the output signal y, and the intercepting sequence is consistent with the sliding window updating sequence of the signal to be processed y 1.

Furthermore, the front-end signal distribution equipment can receive signals acquired by the sensor in real time, integrates a signal processing algorithm, and has a storage function and an operation function. The signals collected by the sensor are processed by the front-end signal distribution equipment, and then the processed signals are packaged and sent to the receiving end.

Further, the specific implementation steps for eliminating the direct current component and the trend term in the acceleration sensor are as follows:

1) determining a sliding window signal y0 according to the sampling frequency fs and the vibration characteristics of the measured object;

2) taking the sliding window signal y0 as the first half part of a signal to be processed, taking the last output signal y as the second half part of the signal to be processed, and splicing the two sections of signals to obtain a signal to be processed y 1;

3) performing fast Fourier transform on the signal y1 to be processed, performing high-pass filtering, and performing inverse Fourier transform on the filtered signal to obtain a signal y2 subjected to direct current removal;

4) performing least square polynomial fitting on the direct current signal removed y2 to obtain a fitting expression y 3;

5) subtracting a least square polynomial y3 from a signal y1 to be processed to obtain a signal y4 from which direct-current components and trend terms are removed;

6) sequentially intercepting window-sized signals from the processed signal y4 as output signals y;

7) a1.1-1.6 signal processing algorithm is embedded into front-end signal distribution equipment, and a direct-current component and a trend item of a sensor are eliminated in real time.

The invention has the advantages that:

according to the invention, the signal processing algorithm is embedded in the front-end signal distribution equipment of the acceleration sensor, and the signals of the acceleration sensor are processed to obtain the signals without the direct-current components and the trend items, so that the direct-current components and the trend items of the acceleration sensor can be eliminated in real time, and the method has obvious advantages in real time, accuracy and economy compared with the traditional solution.

Drawings

FIG. 1 is a flow chart of a signal processing method for eliminating a direct current component and a trend term of an acceleration sensor according to the present invention;

FIG. 2 is a diagram of signals to be processed according to one embodiment;

FIG. 3 is a signal diagram of an embodiment with a Fourier removed DC component;

FIG. 4 is a plot of the signal after detrending and a sliding window least squares fit trend for the example;

fig. 5 is a signal graph of eliminating the acceleration sensor dc component and the trend term.

Detailed Description

The invention is further illustrated with reference to the following figures and specific examples.