A linear motor position signal detection method based on adaptive stochastic resonance

Abstract

The present invention discloses a linear motor position signal detection method based on adaptive random resonance. Based on the principle of magnetic field superposition and the combination of time-grid technology, the equally divided winding slots already possessed by the linear motor are used as part of the magnetic field type time-grid, and a linear motor embedded time-grid position detection system model is constructed to realize embedded synchronous linear motor position signal detection without destroying the motor body structure. The method combines genetic algorithm with parameter random resonance, takes the signal-to-noise ratio of the random resonance system as a measurement index, uses the monostable random resonance technology to realize the detection of the weak signal induced and output by the time-grid sensor, uses the genetic algorithm to realize the optimal selection of the structural parameters of the monostable random resonance system, completes the adaptive selection of the random resonance parameters a and b, and solves the problem of difficulty in presenting weak signals and low detection accuracy in the process of permanent magnet linear motor position detection.

Description

Technical Field

[0001] The invention belongs to the technical field of motor control, and in particular relates to a linear motor position signal detection method based on adaptive stochastic resonance. Background Art

[0002] Linear motors, which work on the electromagnetic principle, are the main actuators that provide high power and high thrust for mechanical products. Compared with rotary motors, linear motors do not require any conversion device to generate linear thrust, which shortens the system response time, effectively reduces the noise and loss during the system operation process, and improves the control accuracy. However, in the signal extraction of electromechanical systems, position detection technology has never been able to achieve effective integration with synchronous linear motors and form a structured and modular detection system. At present, the commonly used external position sensors of synchronous linear motors mainly include gratings, magnetic gratings and Hall sensors, but they not only destroy the original motor structure during the position detection process, but also greatly restrict the control performance of the motor, and have never been able to achieve effective integration of detection technology and linear motors. The detection accuracy of emerging sensorless detection methods, such as back-electromotive force position estimation method, high-frequency signal injection method, model reference adaptive method, etc., is seriously dependent on the operating parameters of the motor, and the reliability of its control effect is difficult to guarantee. It can only be used as an auxiliary means of traditional position detection technology. Aiming at the position detection technology of synchronous linear motors, a linear motor position signal detection method based on adaptive stochastic resonance is proposed.

[0003] In the detection of position signals, effectively extracting weak characteristic signals from the background of strong interference noise has always been the focus of people's attention. Traditional signal detection mainly includes correlation detection, empirical mode decomposition, wavelet transform and spectrum analysis, which realize weak signal detection by eliminating or filtering noise. Spectral analysis is a frequency domain characteristic method based on Fourier integral, which is suitable for weak periodic signal detection under background noise, but it is difficult to realize composite fault signal detection based on the time series requirement of the number of sampling points; the correlation detection method can effectively extract useful signals from noise by using the temporal correlation of signals, but the detection results are seriously dependent on the selection of reference signals; empirical mode decomposition is based on Fourier transform and has the advantage of high signal-to-noise ratio, but there are endpoint effects and mode aliasing problems in the signal detection process; wavelet transform can adaptively meet the analysis requirements of time-frequency signals through scaling and translation operations, making up for the defects of Fourier local transform, but its detection accuracy is limited by Heisenberg uncertainty principle.

[0004] Looking back at traditional signal detection methods, whether implemented in hardware or software, they are based on using various measures to suppress noise to extract useful signals. The stochastic resonance method is different. It uses the mechanism of converting part of the noise energy into signal energy to detect the weak characteristic signal of time-grating induction to realize the extraction of motor position information. In the monostable stochastic resonance system, the increase of noise not only does not make the output signal worse, but also greatly increases the local signal-to-noise ratio of the output, thereby enhancing the appearance of the time-grating characteristic signal. Therefore, compared with the commonly used linear system technology, the stochastic resonance technology can obtain a higher signal-to-noise ratio than the optimal linear filter with the help of nonlinear systems. In this way, the monostable stochastic resonance technology can be used to solve the weak signal problem in motor position detection that cannot be solved by traditional linear analysis methods. At the same time, the stochastic resonance system of the adaptive genetic algorithm has a global search capability, which improves the accuracy of characteristic signal recognition and improves the position detection accuracy of the permanent magnet synchronous linear motor. Summary of the invention

[0005] The motor is a typical electromagnetic field coupling component, and the rotation position signal is in a spatially varying magnetic field. Since the magnetic field time grating structure is flexible and changeable, the magnetic field time grating can be integrated with the synchronous linear motor to achieve embedded position detection. In order to detect the motor position characteristic signal, the present invention proposes a linear motor position signal detection method based on adaptive stochastic resonance.

[0006] A linear motor position signal detection method based on adaptive stochastic resonance comprises the following steps:

[0007] S1: Generate corresponding signals through the signal generation module

[0008] The first signal generating module provides low-frequency motor drive signals A, B, and C with a phase difference of 120° on the stator of the linear motor, and the second signal generating module provides high-frequency time-grating excitation signals S1, S2, and S3 with a phase difference of 120° on the stator winding. The signal loading module then combines the above signals to form signals X, Y, and Z, and based on the principle of magnetic field superposition, generates a high-voltage and low-frequency magnetic field generated by the three-phase motor drive signal and a low-voltage and high-frequency magnetic field generated by the magnetic field time-grating signal at the air gap of the permanent magnet synchronous linear motor.

[0009] S2: Obtaining time-gate sensing signal

[0010] According to the law of electromagnetic induction, the dynamic measuring head will generate a corresponding superimposed magnetic field signal containing spatial position information. The stator structure of the linear motor is used as the time-grid fixed measuring head, and the added winding coil is used as the time-grid dynamic measuring head to form an embedded time-grid motor. The time-grid induction signal reflecting the rotor position sensed by the winding coil is extracted to obtain the corresponding detection signal.

[0011] S3: Process the acquired time-grid induction signal to obtain the linear motor time-grid induction weak signal containing noise interference

[0012] The obtained time-grid induction signal is processed by a signal processing module, and the signal processing module includes a signal conditioning module and a winding coil. The signal conditioning module includes a sampling circuit, a subtraction circuit, an instrument amplifier circuit and a filter circuit, which are used to realize the conversion of the position signal with phase difference of the time-grid dynamic probe from analog to digital. The subtraction circuit is used to remove the bias component of the time-grid induction signal and convert the two-way differential sampling output into a single-way state output. While superimposing the traveling wave signal, the time-grid induction signal is amplified and processed by an instrument amplifier circuit to improve the suppression ratio of the common mode signal. The low-frequency motor drive signal is filtered out by the filter circuit to obtain the linear motor time-grid induction weak signal containing noise interference.

[0013] S4: Extract the weak signal induced by the time grating;

[0014] S5: After the extracted linear motor time-grating induced weak signal is sent to the stochastic resonance system, the monostable stochastic resonance is used as the nonlinear system model to adjust the noise intensity, generate the periodic transmission phenomenon of the time-grating induced signal caused by noise enhancement, and output the time-grating induced enhanced signal under the optimal stochastic resonance state;

[0015] S6: Output the signal through the output module

[0016] In order to effectively improve the signal-to-noise ratio of the output signal, based on the combination of genetic algorithm and stochastic resonance, the genetic algorithm can be used to improve the parameter search capability of the stochastic resonance system and modulate an output signal with more obvious periodic components. The modulated triangular waveform output signal is shaped to obtain a square wave position signal. Similarly, after the input signal of the time grating excitation is shaped, the two triangular wave signals will be sent to the time measurement module for phase comparison processing, and the position data converted by high-frequency pulse interpolation technology will be displayed in the output module.

[0017] Furthermore, step S4 is specifically as follows:

[0018] S41. Based on the twin nature of the structures and principles of magnetic field time grating and linear motors, and on the basis of combining the space-time coordinate conversion theory with the time grating technology, the stator structure of the linear motor is regarded as the stator structure of the magnetic field time grating. The time grating sensor is used to detect the changes in the motor rotor magnetic field, and a permanent magnet synchronous linear motor embedded time grating position detection system model is constructed.

[0019] S42. In the method of using the time-grid sensor to sense the spatial position information, through the sampling, amplification, filtering and other means of the signal processing module, a signal waveform reflecting the position information in the form of a triangular wave is displayed in the output module, so as to realize the extraction and separation of the weak signal of the time-grid induction of the synchronous linear motor based on the random resonance system and the adaptive genetic algorithm.

[0020] S43, encoding: Set the search range and precision α of the stochastic resonance parameters, and then calculate the corresponding encoding length to obtain the individual encoding e in the group. i e i-1 e i-2 …e1f j f j-1 f j-2 …f1. The formula used for parameter binary encoding is:

[0021]

[0022]

[0023] Where m and n are the encoding lengths of the two parameters, c1 is a constant, which is 2, and c is also a constant, which is 1; Δl = l max -l min , Δk=k max -k min ,[l min ,l max ]、[k min ,k max ] is the corresponding search range interval.

[0024] S44. Initialize the population: According to the population size, randomly select a certain number of individuals to initialize the population P0.

[0025] S45. Decoding: Use the following individual decoding formula to solve the stochastic resonance system parameters M and N.

[0026]

[0027]

[0028] S46, fitness function evaluation: select the system output signal-to-noise ratio Q as the index function of stochastic resonance, and the expression of the output signal-to-noise ratio is:

[0029]

[0030] Where Y(b) is the Fourier transform of the output signal of the stochastic resonance system at point b.

[0031] S47. Select crossover and mutation: According to the adaptive evaluation function, select excellent individuals for crossover, mutation and inversion operations to achieve the recombination of the new generation of population.

[0032] S48, termination: repeat the above population generation process until the set conditions are met, terminate the adaptive genetic algorithm process, and output the structural parameters of the optimal stochastic resonance system.

[0033] The present invention has the following beneficial effects:

[0034] 1. For the weak signal induced by the time-grating of the linear motor, part of the noise energy is converted into the characteristic signal through the synergistic effect of the nonlinear stochastic resonance system. While effectively suppressing the noise signal related to the operation of the motor, the signal-to-noise ratio of the original signal is enhanced, and the characteristic frequency components contained in the stochastic resonance output signal are extracted, and the time-grating induced signal is further enhanced. In addition, the genetic algorithm program is used to process its input time domain waveform. Compared with before and after algorithm processing, the periodic signal of a specific frequency is retained, which will significantly suppress the interference around the characteristic signal frequency, more accurately capture the characteristics of the weak signal, and its spectrum is steeper, which greatly reduces the difficulty of identifying weak signals submerged in strong noise.

[0035] 2. The overall idea of ​​embedded detection of time-grid induction coils is to load three-phase symmetrical, high-frequency, low-amplitude time-grid excitation signals and low-frequency, high-amplitude motor winding drive signals on the three-phase symmetrical stator windings of the synchronous linear motor at the same time. The two signals with large amplitude-frequency differences will generate superimposed magnetic fields in the motor air gap. As the linear motor starts, the rotor winding structure induces two superimposed induced electromotive forces, which are used in two parts. One part provides the required energy for the normal operation of the motor, and the other part provides the required motor rotor spatial position feature information for time-grid measurement. The weak induction signal of the magnetic field time-grid measurement can be further separated and extracted with the help of the principle of adaptive stochastic resonance, so as to realize embedded time-grid position information detection without affecting the normal operation of the linear motor. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a principle block diagram of the embedded time-grid type motor induced weak signal detection system of the present invention;

[0037] Figure 2 It is the structural block diagram of the signal processing module;

[0038] Figure 3 Genetic algorithm optimization system parameter flow chart;

[0039] Figure 4 Flowchart of the adaptive stochastic resonance method. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of 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.

[0041] The coil-based sensor unit is the earliest method used in traditional detection technology. The coil can sense the change of the permanent magnet's magnetic field and output an electrical signal without power. The magnetic field time grating with coils wound around the stator or rotor has a twin relationship with the synchronous linear motor in structure and appearance. Both have stator and rotor structures and use the magnetic field generated by the winding to achieve their own functions. The magnetic field time grating uses the time and space duality of the magnetic field to convert the spatial position into the count of time pulses; the linear motor uses the spatiality of the magnetic field to use the changing magnetic field to provide the energy required for the rotor to rotate. This makes it possible to replace the encoder with the magnetic field time grating displacement sensor embedded in the synchronous linear motor, and realize the detection of weak signals induced by the motor time grating.

[0042] A linear motor position signal detection method based on adaptive stochastic resonance comprises the following steps:

[0043] S1: Generate corresponding signals through the signal generation module

[0044] The first signal generating module provides low-frequency motor drive signals A, B, and C with a phase difference of 120° on the stator of the linear motor, and the second signal generating module provides high-frequency time-grating excitation signals S1, S2, and S3 with a phase difference of 120° on the stator winding. The signal loading module then combines the above signals to form signals X, Y, and Z, and based on the principle of magnetic field superposition, generates a high-voltage and low-frequency magnetic field generated by the three-phase motor drive signal and a low-voltage and high-frequency magnetic field generated by the magnetic field time-grating signal at the air gap of the permanent magnet synchronous linear motor.

[0045] S2: Obtaining time-gate sensing signal

[0046] According to the law of electromagnetic induction, the moving probe will generate a corresponding superimposed magnetic field signal containing spatial position information. The linear motor stator structure is used as the time-grid fixed probe, and the added winding coil is used as the time-grid moving probe, which is composed as follows: Figure 1 The embedded time-grid motor shown extracts the time-grid induction signal reflecting the rotor position sensed by the winding coil to obtain a corresponding detection signal.

[0047] S3: Process the acquired time-grid induction signal to obtain the linear motor time-grid induction weak signal containing noise interference

[0048] The obtained time-grid sensing signal is processed by a signal processing module, and the signal processing module includes: Figure 2 The signal conditioning module and winding coil shown. The signal conditioning module includes a sampling circuit, a subtraction circuit, an instrument amplifier circuit and a filter circuit, which are used to realize the conversion of the position signal with phase difference of the time-grid dynamic probe from analog to digital. The subtraction circuit is used to remove the bias component of the time-grid induction signal and convert the two-way differential sampling output into a single-way state output. Because the amplitude of the time-grid standing wave signal processed by the subtraction circuit is lower than the amplitude of the original time-grid induction signal, it is necessary to use an instrument amplifier circuit to amplify the time-grid induction signal while superimposing it into a traveling wave signal to improve the suppression ratio of the common-mode signal. Since the frequency of the motor drive signal is much smaller than the frequency of the time-grid excitation signal, the filter circuit is used to filter out the low-frequency motor drive signal to obtain a linear motor time-grid induction weak signal containing noise interference.

[0049] S4: Extract the weak signal induced by the time grating;

[0050] S5: After the extracted linear motor time-grating induced weak signal is sent to the stochastic resonance system, the monostable stochastic resonance is used as the nonlinear system model to adjust the noise intensity, generate the periodic transmission phenomenon of the time-grating induced signal caused by noise enhancement, and output the time-grating induced enhanced signal under the optimal stochastic resonance state;

[0051] S6: Output the signal through the output module

[0052] In order to effectively improve the signal-to-noise ratio of the output signal, based on the combination of genetic algorithm and stochastic resonance, the genetic algorithm can be used to improve the parameter search capability of the stochastic resonance system and modulate an output signal with more obvious periodic components. The modulated triangular waveform output signal is shaped to obtain a square wave position signal. Similarly, after the input signal of the time grating excitation is shaped, the two triangular wave signals will be sent to the time measurement module for phase comparison processing, and the position data converted by high-frequency pulse interpolation technology will be displayed in the output module.

[0053] Preferably, step S4 is specifically:

[0054] S41. Based on the twin nature of the structures and principles of magnetic field time grating and linear motors, and on the basis of combining the space-time coordinate conversion theory with the time grating technology, the stator structure of the linear motor is regarded as the stator structure of the magnetic field time grating. The time grating sensor is used to detect the changes in the motor rotor magnetic field, and a permanent magnet synchronous linear motor embedded time grating position detection system model is constructed.

[0055] S42. In the method of using the time-grid sensor to sense the spatial position information, through the sampling, amplification, filtering and other means of the signal processing module, a signal waveform reflecting the position information in the form of a triangular wave is displayed in the output module, so as to realize the extraction and separation of the weak signal of the time-grid induction of the synchronous linear motor based on the random resonance system and the adaptive genetic algorithm.

[0056] S43, encoding: Set the search range and precision α of the stochastic resonance parameters, and then calculate the corresponding encoding length to obtain the individual encoding e in the group. i e i-1 e i-2 …e1f j f j-1 f j-2 …f1. The formula used for parameter binary encoding is:

[0057]

[0058]

[0059] Where m and n are the encoding lengths of the two parameters, c1 is a constant, which is 2, and c is also a constant, which is 1; Δl = l max -l min , Δk=k max -k min ,[l min ,l max ]、[k min ,k max ] is the corresponding search range.

[0060] S44. Initialize the population: According to the population size, randomly select a certain number of individuals to initialize the population P0.

[0061] S45. Decoding: Use the following individual decoding formula to solve the stochastic resonance system parameters M and N.

[0062]

[0063]

[0064] S46, fitness function evaluation: select the system output signal-to-noise ratio Q as the index function of stochastic resonance, and the expression of its output signal-to-noise ratio is:

[0065]

[0066] Where Y(b) is the Fourier transform of the output signal of the stochastic resonance system at point b.

[0067] S47. Select crossover and mutation: According to the adaptive evaluation function, select excellent individuals for crossover, mutation and inversion operations to achieve the recombination of the new generation of population.

[0068] S48, termination: repeat the above population generation process until the set conditions are met, terminate the adaptive genetic algorithm process, and output the structural parameters of the optimal stochastic resonance system.

[0069] According to the characteristics of the stochastic resonance system of the adaptive genetic algorithm, the amplitude of the induced signal after genetic algorithm processing is greatly reduced, which is expected to further improve the signal-to-noise ratio. This will help to effectively improve the signal-to-noise ratio and better extract weak signals submerged in noise. It has certain engineering application prospects for the extraction of system periodic characteristic signals.

Claims

1. A linear motor position signal detection method based on adaptive stochastic resonance, characterized in that: The method comprises the following steps: S1: Generate a corresponding signal through a signal generating module. Step S1 is specifically as follows: A first signal generating module is used to provide low-frequency motor drive signals A, B, and C with a phase difference of 120° on the stator of the linear motor, and a second signal generating module is used to provide high-frequency time-grid excitation signals S1, S2, and S3 with a phase difference of 120° on the stator winding. Then, the signal loading module combines the above signals to form signals X, Y, and Z, and generates a high-voltage low-frequency magnetic field generated by the three-phase motor drive signal and a low-voltage high-frequency magnetic field generated by the magnetic field time-grid signal at the air gap of the permanent magnet synchronous linear motor according to the magnetic field superposition principle; S2: Obtaining time-grid sensing signal; S3: Processing the acquired time-grid induction signal to obtain a linear motor time-grid induction weak signal containing noise interference; S4: Extract the weak signal induced by the time grating, and use the fitness function evaluation in the extraction. The fitness function evaluation is: select the system output signal-to-noise ratio Q as the indicator function of stochastic resonance, and the expression of its output signal-to-noise ratio is: In the formula, is the output signal of the stochastic resonance system b The Fourier transform at S5: After the extracted linear motor time-grating induced weak signal is sent to the stochastic resonance system, the monostable stochastic resonance is used as the nonlinear system model to adjust the noise intensity, generate the periodic transmission phenomenon of the time-grating induced signal caused by noise enhancement, and output the time-grating induced enhanced signal under the optimal stochastic resonance state; S6: Output the signal through the output module. Step S6 is specifically as follows: In order to effectively improve the signal-to-noise ratio of the output signal, on the basis of combining genetic algorithm and stochastic resonance, the genetic algorithm is used to improve the parameter search capability of the stochastic resonance system and modulate an output signal with a more obvious periodic component; the modulated triangular waveform output signal is shaped to obtain a square wave position signal; similarly, after the input signal of the time grating excitation is shaped, the two triangular wave signals will be sent to the time measurement module for phase comparison processing, and the position data converted by high-frequency pulse interpolation technology will be displayed on the output module.

2. The detection method according to claim 1, characterized in that Step S2 is specifically as follows: According to the law of electromagnetic induction, the dynamic measuring head will generate a corresponding superimposed magnetic field signal containing spatial position information. The stator structure of the linear motor is used as the time-grid fixed measuring head, and the added winding coil is used as the time-grid dynamic measuring head to form an embedded time-grid motor. The time-grid induction signal reflecting the rotor position sensed by the winding coil is extracted to obtain the corresponding detection signal.

3. The detection method according to claim 1, characterized in that Step S3 is specifically as follows: The obtained time-grid induction signal is processed by a signal processing module, and the signal processing module includes a signal conditioning module and a winding coil; the signal conditioning module includes a sampling circuit, a subtraction circuit, an instrument amplifier circuit and a filter circuit, which are used to realize the conversion of the position signal with phase difference of the time-grid dynamic probe from analog to digital; the subtraction circuit is used to remove the bias component of the time-grid induction signal and convert the two-way differential sampling output into a single-way state output; while superimposing a traveling wave signal, the time-grid induction signal is amplified and processed by an instrument amplifier circuit to improve the suppression ratio of the common-mode signal; the filter circuit is used to filter out the low-frequency motor drive signal to obtain the linear motor time-grid induction weak signal containing noise interference.

Images