Electric Working Machines and Methods for Controlling Noise Generated by Electric Working Machines
By optimizing the noise control system in the electric work platform and reducing the amount of computation, the computational requirements for noise suppression of the electric work platform were solved, and a highly efficient noise control effect was achieved.
Patent Information
- Application Number
- CN202111635417.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2021-01-27
- Filing Date
- 2021-12-29
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2041-12-29
Smart Images

Figure CN114822471B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to noise control in an electric work machine. Background Technology
[0002] Patent document 1 describes the use of active noise control (ANC) to suppress the noise of electric work machines.
[0003] ANC (Adaptive Noise Cancellation) is a technique that uses sound measured by a microphone on the sound source side (hereinafter, the reference signal) to generate an inverse-phase sound at the desired cancellation location, thereby canceling the sound. This inverse-phase sound is generated using an adaptive filter. The characteristics of the filter are calculated step-by-step using adaptive algorithms such as Filtered-X or NLMS.
[0004] In ANC, unlike adaptive filters, a second-order filter is used, modeled by the characteristics of the path from the speaker generating the control tone to the microphone detecting the control error (hereinafter, a second-order system). Both adaptive and second-order filters use digital filters. The number of taps in the digital filter is set to, for example, several hundred, based on the time required for the impulse response to fully converge.
[0005] Existing technical documents
[0006] Patent documents
[0007] Patent Document 1: Japanese Patent Publication No. 6-508695 Summary of the Invention
[0008] The computation of adaptive algorithms for setting the coefficients of digital filters requires a massive amount of processing power, and it needs to be completed sufficiently early compared to the propagation time from noise detection to reaching the desired cancellation point. Therefore, when high-performance computers are unavailable for various reasons, there is a significant challenge in applying ANC.
[0009] One aspect of the present invention is to reduce the amount of computational processing required for noise suppression of electric work machines.
[0010] An electric work machine according to one aspect of the present invention includes: a motor, a reference acquisition unit, a noise control filter, a control sound source, an error sensor, a second-order filter, and a characteristic adjustment unit.
[0011] The motor is configured to generate the driving force required to perform predetermined on-site operations. The reference acquisition unit is configured to acquire a reference signal correlated with the noise caused by the motor's operation, i.e., the object noise. The noise control filter is a digital filter including a series of taps and adjustable characteristics. The noise control filter is configured to receive the reference signal from the reference acquisition unit and generate a control signal for eliminating object noise. The control sound source is configured to generate suspected noise according to the control signal. The error sensor is configured to convert the synthesized sound of suspected noise and object noise at a predetermined noise cancellation point into an error signal. The second-order filter is a digital filter including N taps, obtained by pre-modeling the path from the control sound source to the error sensor, i.e., the second-order system. The second-order filter is configured to receive the reference signal from the reference acquisition unit and generate a filtered reference signal. The characteristic adjustment unit is configured to adjust the characteristics of the noise control filter using the error signal and the filtered reference signal according to an adaptive algorithm. Furthermore, the characteristic adjustment unit is configured to update the coefficients of M taps in the series of taps. The M taps correspond to some or all of the series of taps. M and N are positive integers that satisfy M < N.
[0012] Based on this configuration, the divergence of the noise control filter can be suppressed, and the number of taps in the noise control filter can be reduced, which in turn reduces the computational workload of the adaptive algorithm.
[0013] The M taps can correspond to all of a series of taps. The characteristic adjustment unit can be configured to update the coefficients of all M taps.
[0014] The aforementioned electric work machine can further include a divergence determination unit configured to determine whether the noise control filter tends to diverge. The characteristic adjustment unit can be configured to update the coefficients of L taps in a series of taps before determining, using the divergence determination unit, that the noise control filter tends to diverge. The characteristic adjustment unit can be configured to, after determining, using the divergence determination unit, set the coefficients of the (M+1)th tap and beyond in the series of taps to predetermined values, and update the coefficients of the 1st to Mth taps in the series of taps. The L taps can correspond to all taps in the series. L can be a positive integer satisfying L≥N.
[0015] The given value can be: the path from the motor to the error sensor, which is the impulse response of the first-order system multiplied by the inverse filtering of the second-order filter.
[0016] The divergence determination unit can be configured to determine whether the noise control filter tends to diverge based on the strength of the error signal and / or the trend of its change.
[0017] The divergence determination unit can be configured to determine whether the noise control filter tends to diverge based on the output intensity of each of the L taps, and / or the magnitude of the coefficients of each of the L taps, and / or the changing trend of the parameters used to update the coefficients.
[0018] The L taps can have a length corresponding to the path from the motor to the error sensor, that is, the time required for the impulse response of the first-order system to converge.
[0019] The N taps can be configured to have a length sufficient to terminate the processing required for generating suspected noise. The processing required for generating suspected noise may include: updating the coefficients of the characteristic adjustment unit according to the reference signal within the time required from detecting the reference signal to detecting the noise that is the source of the reference signal using an error sensor. The processing required for generating suspected noise may also include: generating a control signal for the noise control filter whose coefficients have been updated according to the reference signal.
[0020] The feature adjustment unit can be configured to stop updating the coefficients when the number of coefficient updates reaches a predetermined value.
[0021] The feature adjustment unit can be configured to update the coefficients by multiplying the update step size, which represents the degree of change of the coefficients, by the update value, which represents the direction of change of the coefficients.
[0022] The feature adjustment unit can be configured to calculate the updated value using the average of the gradient vectors calculated in multiple processing loops.
[0023] Based on this configuration, it is possible to suppress the influence of sudden changes in the external environment on the coefficient update of the noise control filter.
[0024] The characteristic adjustment unit can be configured to change the update step size according to the convergence status of the noise control filter.
[0025] Based on this configuration, the divergence of the noise control filter can be further suppressed.
[0026] The reference acquisition unit may include a reference sensor configured to detect object noise and generate a reference signal.
[0027] The reference sensor can be configured such that the distance between the reference sensor and the control sound source is longer than the distance between the control sound source and the error sensor.
[0028] The reference acquisition unit can be configured to acquire the drive signal that drives the motor as a reference signal.
[0029] The reference acquisition unit may include an arrival filter and an addition unit. The arrival filter is configured to have the same characteristics as a second-order filter, receiving a control signal from a noise control filter to generate an arrival signal representing the suspected noise of the arrival error sensor. The addition unit may be configured to add the arrival signal and the error signal to generate a reference signal.
[0030] The aforementioned electric work machine may include: a fan, a flow path, and an outlet. The fan may be configured to be driven by a motor to generate airflow. The flow path may be configured to allow airflow generated by the fan to pass through. The outlet may be configured to discharge airflow from the flow path. The control sound source and error sensor may be configured such that the outlet becomes a noise reduction point.
[0031] At least a portion of the wall of the flow path may be provided with a sound-absorbing material capable of suppressing wind noise generated by airflow impact. The error sensor may be configured at a position opposite to the flow path, with the sound-absorbing material sandwiched between it.
[0032] Based on this configuration, it is possible to suppress noise other than object noise by using an error sensor, thereby improving the suppression accuracy of object noise by suppressing suspected noise.
[0033] Another aspect of the present invention is a method for controlling noise generated by an electric work machine, comprising: obtaining a reference signal correlated with noise caused by the operation of a motor in the electric work machine, i.e., object noise; generating a control signal for eliminating object noise based on the reference signal using a digital filter, i.e., a noise control filter, comprising a series of taps and adjustable characteristics; generating a suspected noise using a control sound source according to the control signal; converting a synthesized sound of the suspected noise and object noise at a predetermined noise reduction point into an error signal using an error sensor; generating a filtered reference signal based on the reference signal using a digital filter, i.e., a second-order filter, comprising N taps obtained by pre-modeling the path from the control sound source to the error sensor, i.e., a second-order system; and updating the coefficients of M taps in the series of taps according to an adaptive algorithm using the error signal and the filtered reference signal. The M taps correspond to some or all of the series of taps. M and N are positive integers satisfying M < N. Attached Figure Description
[0034] Figure 1 This is a perspective view showing the appearance of the dust collector according to the first to third embodiments.
[0035] Figure 2 This is a bottom view of the main body of the dust collector.
[0036] Figure 3It is a three-dimensional view showing the internal state of the dust collector after the upper outer shell has been removed from the main body.
[0037] Figure 4 This is a perspective view of the lower casing from the joint surface between the upper casing and the lower casing, with all components removed.
[0038] Figure 5 This is a perspective view of the upper casing from the joint surface between the upper and lower casings, with all components removed.
[0039] Figure 6 It is a three-dimensional diagram showing the position of some components on the upper outer casing.
[0040] Figure 7 This is a top view of the lower outer shell.
[0041] Figure 8 This is a block diagram showing the electrical structure of a dust collector.
[0042] Figure 9 This is a block diagram representing a model of a feedforward ANC system.
[0043] Figure 10 This is a flowchart representing the noise control process.
[0044] Figure 11 This is a flowchart representing the coefficient update process.
[0045] Figure 12 This is a flowchart representing the divergence determination process.
[0046] Figure 13 This is a flowchart illustrating the coefficient update process in the second embodiment.
[0047] Figure 14 This is a flowchart illustrating the coefficient update process in the third embodiment.
[0048] Figure 15 This is a flowchart representing the step size update process.
[0049] Figure 16 This is a flowchart illustrating the divergence determination process in the third embodiment.
[0050] Figure 17 This is a perspective view showing the appearance of the portable vacuum cleaner according to the fourth embodiment.
[0051] Figure 18 This is a cross-sectional view of a portable vacuum cleaner.
[0052] Figure 19It is a perspective view showing the state of a portable vacuum cleaner with its outer casing and some components removed, and also showing the installation state of the error detection microphone and the control speaker.
[0053] Figure 20 This is a block diagram representing a model of a feedback-type ANC system.
[0054] Figure 21 This is a block diagram of an ANC used for error research of a second-order filter.
[0055] Figure 22 It is Figure 21 The equivalent block diagram obtained by deforming the block diagram shown.
[0056] Figure 23 It is Figure 22 The equivalent block diagram is obtained by further transforming the block diagram shown.
[0057] Symbol Explanation
[0058] 1…Dust collector, 3…Main body, 6…Operating device, 7…Assemblies, 8…Portable vacuum cleaner, 10…Noise control device, 30…Outer shell, 31, 81…Suction port, 32…Dust collection chamber, 33…First flow path, 34…Motor chamber, 35…Second flow path, 36…Exhaust chamber, 37A…First battery storage section, 37B…Second battery storage section, 38…Component configuration section, 41…Dust bag, 42…Filter, 43, 90…Drive unit, 44…Drive controller, 45A, 86A …First battery pack, 45B, 86B…Second battery pack, 46, 87…Sound-absorbing material, 52…Connector, 53…Reference microphone, 54, 88…Control speaker, 55, 89…Error microphone, 61…Wire, 71A…First shoulder strap, 71B…Second shoulder strap, 72…Waist belt, 80…Main body shell, 80a…Suction port side space, 80b…Exhaust port side space, 80c…Flow path, 82, 361…Exhaust port, 83…Holding part, 85A…First battery assembly part, 85B …Second battery assembly section, 84…electronic switch, 91…control circuit board, 301…lower housing, 302…upper housing, 303…plate, 304, 305, 344…mounting holes, 341…inlet, 342…outlet, 343…wall, 371A…first battery mounting port, 371B…second battery mounting port, 381…longitudinal portion, 382…horizontal portion, 431, 903…fan, 432, 902…motor, 433…muffler, 441…control Circuit, 442…Dust collection circuit group, 443…Noise circuit group, 444…First A / D converter, 445…Second A / D converter, 446…D / A converter, 447…Power supply circuit, 901…Motor housing, 911…Motor control unit, 912…Noise control unit, M1…Reference sensor, M2…Control sound source, M3…Error sensor, M4…Noise control filter, M5…Second-order filter, M6…Coefficient update unit, M7…Arrival filter, M8…Adder. Detailed Implementation
[0059] Hereinafter, exemplary embodiments of the present invention will be described with reference to the accompanying drawings.
[0060] [1. First Implementation Method]
[0061] [1-1. Composition of a dust collector]
[0062] The dust collector 1, as an example of an electric work machine, will be described. The dust collector 1 is used in a manner carried on the back of the operator. In this embodiment, for ease of explanation, the dust collector 1 will be described as follows: Figures 1 to 7 As shown, the terms front, back, top, bottom, left, and right are defined.
[0063] like Figures 1 to 7As shown, the dust collector 1 includes: a main body 3, an operating device 6, and accessories 7.
[0064] The accessory 7 includes a first shoulder strap 71A, a second shoulder strap 71B, and a waist belt 72, all mounted on the rear surface of the main body 3. The first shoulder strap 71A and the second shoulder strap 71B extend from the upper part of the main body 3 and are positioned near its left and right ends, respectively, and are mounted on the operator's left and right shoulders. The waist belt 72 extends from near the lower end of the main body 3 and is mounted on the operator's waist. The main body 3 is carried by the operator using the accessory 7.
[0065] The operating device 6 includes a switch for starting and stopping the dust collector 1, which is operated by the operator. The operating device 6 is connected to the lower center of the main body 3 via a wire 61.
[0066] The main body 3 has an outer shell 30.
[0067] The outer casing 30 includes a lower outer casing 301, an upper outer casing 302, and a plate 303. The lower outer casing 301 is a bottomed container-shaped component with an opening on its front surface. The upper outer casing 302 is a frame-shaped component with openings on both its front and rear surfaces. The plate 303 is a plate-shaped component that seals the opening on the upper surface of the upper outer casing 302. The outer casing 30 is formed by injection molding, for example, a resin material.
[0068] The housing 30 includes: a suction port 31, a dust collection chamber 32, a first flow path 33, a motor chamber 34, a second flow path 35, an exhaust chamber 36, a first battery storage section 37A, a second battery storage section 37B, and a component configuration section 38.
[0069] A suction port 31 is located at the center of the upper end of the housing 30. Although not shown in the figure, a flexible hose is connected to the first end of the suction port 31. Additionally, a nozzle with a suction inlet is connected to the second end of the hose.
[0070] The dust collection chamber 32 is a rectangular internal space located in the upper part of the outer casing 30. The dust collection chamber 32 houses the dust collection bag 41 connected to the suction port 31. The dust collection bag 41, for example, is a paper bag that collects dust drawn in from the suction port 31.
[0071] The first flow path 33 is arranged along the right side of the dust collection chamber 32, and its lower end is connected to the motor chamber 34. A filter 42 is arranged at the boundary between the first flow path 33 and the dust collection chamber 32. The filter 42 is, for example, a high-efficiency particulate air (HEPA) filter.
[0072] The motor chamber 34 is a rectangular internal space located below the dust collection chamber 32. The motor chamber 34 is configured such that it has an inlet 341 at the center of its right end, connected to the first flow path 33, and an outlet 342 at the upper left end, connected to the second flow path 35. The drive unit 43 is housed within the motor chamber 34.
[0073] The drive unit 43 includes a fan 431, a motor 432, and a silencer 433. The fan 431 is connected to the rotating shaft of the motor 432, and drives the motor 432 to generate airflow from the inlet 341 of the motor chamber 34 towards the outlet 342. The silencer 433 is a ring-shaped component that covers the area around the motor 432, absorbing the sound emitted by the motor 432. The silencer 433 may be, for example, made of sponge. It should be noted that... Figure 3 , Figure 6 In the middle, the motor 432 is covered by the muffler 433, which, although not shown, is located in the center of the muffler 433.
[0074] The second flow path 35 connects the outlet 342 of the motor chamber 34 and the exhaust chamber 36. Sound-absorbing material 46 is disposed in the second flow path 35. The sound-absorbing material 46 is located on the wall separating the second flow path 35 and the dust collection chamber 32 within the lower housing 301. For example, sponge is used as the sound-absorbing material 46.
[0075] The exhaust chamber 36 is an internal space located on the left side of the motor chamber 34, with a slit-shaped exhaust port 361 on the lower surface of the housing 30.
[0076] In the main body 3 configured as described above, when the drive unit 43 is driven to generate airflow, external air is drawn from the external space of the housing 30 into the internal space of the housing 30 via the suction port 31 and the hose and nozzle connected to the suction port 31. The drawn external air first reaches the dust collection chamber 32, and then passes through the dust collection bag 41 installed at the suction port 31, thereby capturing dust contained in the external air. The air passing through the dust collection bag 41 passes through the filter 42 and reaches the first flow path 33. In the filter 42, finer dust that cannot be captured by the dust collection bag 41 is captured. The air reaching the first flow path 33 passes through the motor chamber 34 and the second flow path 35 to reach the exhaust chamber 36, and is discharged into the external space of the housing 30 via the exhaust port 361.
[0077] The first battery storage section 37A is a space for storing the first battery pack 45A, located near the lower end of the outer casing 30, and has a first battery mounting port 371A that opens near the lower left end of the outer casing 30. The second battery storage section 37B is a space for storing the second battery pack 45B, located near the lower end of the outer casing 30, and has a second battery mounting port 371B that opens near the lower right end of the outer casing 30. The first and second battery packs 45A and 45B are inserted into the first and second battery storage sections 37A and 37B respectively through the first and second battery mounting ports 371A and 371B. The first and second battery packs 45A and 45B are universal batteries that can be used as power sources for various electric work machines.
[0078] The component configuration section 38 is an internal space located between the motor chamber 34, the second flow path 35, the exhaust chamber 36, and the first and second battery storage sections 37A and 37B, in which various electrical components are configured. The component configuration section 38 has a longitudinal portion 381 and a transverse portion 382 communicating with the longitudinal portion 381. The longitudinal portion 381 is surrounded in three directions by the walls forming the motor chamber 34, the second flow path 35, and the exhaust port 361, respectively. The transverse portion 382 is sandwiched between the motor chamber 34 and the first and second battery storage sections 37A and 37B.
[0079] A connector 52 and a reference microphone (hereinafter, reference microphone) 53 are provided in the horizontal part 382.
[0080] The connector 52 is disposed between the first battery storage section 37A and the second battery storage section 37B. The purpose of providing the connector 52 is to connect the wire 61 of the operating device 6 to the internal circuit.
[0081] The reference microphone 53 is mounted using a first mounting hole 344 formed in the wall 343, such that the directivity of the reference microphone 53 faces inward toward the motor chamber 34, wherein the wall 343 forms the boundary between the reference microphone 53 and the motor chamber 34. The reference microphone 53 detects object noise generated within the housing 30. Specifically, the object noise includes noise generated by the motor 432 of the drive unit 43, the fan 431, and noise generated by the airflow generated by the drive unit 43.
[0082] The longitudinal portion 381 houses a control speaker 54, an error microphone (hereinafter, error microphone) 55, and a drive controller 44. The control speaker 54 and error microphone 55 are mounted using a second mounting hole 304 and a third mounting hole 305 formed on the lower surface of the lower housing 301, such that the control speaker 54 and error microphone 55 are directionally oriented towards the outside of the housing 30. Figure 7 As shown, the drive controller 44 is mounted on the wall that forms the boundary between the longitudinal portion 381 and the motor chamber 34. The drive controller 44 is a circuit board that performs power control, motor control, noise control, etc., and its details will be described below.
[0083] It should be noted that the error microphone 55 is positioned near the outlet 361, which serves as the silencing point; that is, the error microphone 55 is positioned at a location where the silencing point is considered to exist, and where the airflow generated by the drive unit 43 will not directly impact it. The control speaker 54 is positioned where the phase of the suspected noise emitted by the control speaker 54 is the same as the phase at the silencing point and the position of the error microphone 55. Furthermore, the reference microphone 53, control speaker 54, and error microphone 55 are configured such that the time required for the suspected noise radiated by the control speaker 54 to reach the silencing point is shorter than the time required for the actual noise to directly reach the silencing point. That is, during this time difference, the process of generating suspected noise is performed.
[0084] The control speaker 54 radiates suspected noise outwards from the housing 30, and the error microphone 55 detects the sound synthesized from the actual noise and suspected noise emitted from the outlet 361. The control speaker 54 has the ability to emit a sound sufficiently loud compared to the actual noise. The error microphone 55 has the ability to receive the synthesized sound of the actual noise and suspected noise without distortion.
[0085] [1-2. Drive Controller]
[0086] like Figure 8 As shown, the drive controller 44 includes: a control circuit 441, a dust collection circuit group 442, a noise reduction circuit group 443, and a power supply circuit 447.
[0087] The power supply circuit 447 distributes the power supplied from the first and second battery packs 45A and 45B to each part at an appropriate voltage.
[0088] The control circuit 441 of this first embodiment is in the form of a microcomputer. Therefore, the control circuit 441 includes a CPU 441A and a memory 441B. Instead of a microcomputer, the control circuit 441 may include a combination of electronic components such as discrete components, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an application-specific general-purpose component (ASSP), a field-programmable gate array (FPGA), or a combination thereof; or, in addition to a microcomputer, the control circuit 441 may include a combination of electronic components such as discrete components, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an application-specific general-purpose component (ASSP), a field-programmable gate array (FPGA), or a combination thereof.
[0089] The dust collection circuit group 442 comprises various circuits necessary for functioning as a dust collector. The dust collection circuit group 442 includes a motor drive circuit, a battery switching circuit, and an anomaly detection circuit. The motor drive circuit drives the motor 432. The battery switching circuit appropriately switches the power supply to either the first or second battery group 45A or 45B based on the charging capacity of the first and second battery groups 45A and 45B. The anomaly detection circuit detects various anomalies related to the driving of the motor 432.
[0090] The driving circuit group 443 comprises various circuits necessary to perform the function of a noise control device. The noise circuit group 443 includes a first analog-to-digital (A / D) converter 444, a second A / D converter 445, and a digital-to-analog (D / A) converter 446.
[0091] The first A / D converter 444 performs A / D conversion on the detection signal from the reference microphone 53 and supplies it to the control circuit 441. The second A / D converter 445 performs A / D conversion on the detection signal from the error microphone 55 and supplies it to the control circuit 441. The D / A converter 446 performs D / A conversion on the control data output from the control circuit 441 to generate a control signal supplied to the control speaker 54.
[0092] The control circuit 441 performs the following processes: controlling the dust collection circuit group 442 to realize the function of the dust collector 1, and suppressing the noise emitted by the dust collector 1 to realize the function of the noise control device 10, i.e., noise suppression processing.
[0093] That is, the noise control device 10 includes: a control circuit 441 for performing noise suppression processing, a noise circuit group 443, a reference microphone 53, a control speaker 54, and an error microphone 55.
[0094] The control circuit 441 achieves feedforward active noise control (ANC) through noise control processing.
[0095] [1-3. ANC Model]
[0096] The model of the feedforward ANC implemented by the noise control device 10 is explained.
[0097] like Figure 9 As shown, the feedforward ANC model includes: a reference sensor M1, a control sound source M2, an error sensor M3, a noise control filter M4, a second-order filter M5, and a coefficient update unit M6. It should be noted that the reference sensor M1 corresponds to the reference microphone 53 and the first A / D converter 444. The control sound source M2 corresponds to the D / A converter 446 and the control speaker 54. The error sensor M3 corresponds to the error microphone 55 and the second A / D converter 445. Furthermore, the noise control filter M4, the second-order filter M5, and the coefficient update unit M6 can all be implemented through the control circuit 441. However, some or all of the noise control filter M4, the second-order filter M5, and the coefficient update unit M6 can also be implemented in hardware.
[0098] Reference sensor M1 generates a reference signal x by detecting object noise. n n represents the nth sampled data.
[0099] The noise control filter M4 is an FIR filter with L taps, where L is a positive integer. The noise control filter M4 is based on the L most recently detected reference signals {x}. n x n-1 ... x n-L+1} is an L-dimensional reference vector x(n) for the elements, used to generate the control signal u. n .
[0100] Control sound source M2 according to control signal u n This could produce what appears to be noise.
[0101] Error sensor M3 generates an error signal e by detecting the synthesized sound of object noise and suspected noise. n .
[0102] Hereinafter, the sound propagation path from the reference sensor M1 to the error sensor M3 will be referred to as the first-order system, and the sound propagation path from the control sound source M2 to the error sensor M3 will be referred to as the second-order system.
[0103] The second-order filter M5 is an FIR filter with N taps. N is a positive integer. The second-order filter M5 is based on the N most recently detected reference signals {x}. n x n-1 ... x n-N+1 Given an N-dimensional reference vector x(n) for the element, generate a filtered reference signal r. n The second-order filter M5 is a filter obtained by modeling the transfer characteristics of a second-order system, with fixed values for the coefficients of each tap. The filter reference signal r... n Yes: The second-order system influence component of the suspected noise, which is added to the suspected noise when it reaches the error sensor M3, is provided to the reference signal x. n The signal obtained.
[0104] The coefficient update unit M6 updates the coefficients {w1, w2, ..., w...} of the L taps in the noise control filter M4 at the position of the error sensor M3 (i.e., the silence point). L The update is performed so that the object noise and suspected noise cancel each other out, and the error signal e n Minimum. The following will use coefficients {w1, w2, ..., w...} L The L-dimensional vector of elements is called the coefficient vector w(n).
[0105] The coefficients of the noise control filter M4 can be updated using, for example, the Filtered-xNLMS algorithm, which is one of the adaptive algorithms.
[0106] In the noise control device 10, the number of taps L of the noise control filter M4 is set to approximately several hundred to several thousand, which is comparable to the time required for the impulse response of a first-order system to fully converge. The number of taps N of the second-order filter M5 is set based on the requirement that L ≥ N and that the noise control processing can be completed within the processing time allocated by the control circuit 441 for noise control processing; that is, the processing capability of the control circuit 441. Furthermore, the number of taps N of the second-order filter M5 is set to a value smaller than the number corresponding to the time required for the impulse response of a second-order system to fully converge. In other words, the second-order filter M5 reproduces an approximate characteristic after removing the latter half of the impulse response.
[0107] In the noise control filter M4 and the second-order filter M5, the tap processing the most recent signal is designated as the first tap, and the tap number for processing the most recent signal is increased. Furthermore, the tap with the smaller number is called the higher-order tap, and the tap with the larger number is called the lower-order tap.
[0108] Furthermore, the number of taps of the noise control filter M4 used when the coefficients of the noise control filter M4 updated by the adaptive algorithm tend to diverge is set to M. M is a positive integer satisfying M≤N, and is determined experimentally in advance using methods such as simulation.
[0109] For example, the transfer characteristic (hereinafter, theoretical characteristic) P / C calculated based on the pre-estimated impulse response P of the first-order system and the impulse response C of the second-order system is assumed to be the true value of the noise control filter M4. Furthermore, when a tendency towards divergence is detected, the number of remaining taps, excluding the lower taps of the noise control filter M4 that deviate from the true value by a predetermined value, can be set as M.
[0110] Additionally, for example, regarding the coefficients w1 to w of each tap of the noise control filter M4... L Starting from the lower tap, calculate the sum of squares sequentially and compare it with the sum of squares of the remaining upper taps. Furthermore, the number of remaining taps after removing the lower taps can be set as M until the sum of squares of the lower taps becomes less than or equal to a predetermined value (e.g., 1 / 2) of the sum of squares of the upper taps.
[0111] [1-4. Treatment]
[0112] [1-4-1. Noise Suppression Process]
[0113] use Figure 10 The flowchart below illustrates the noise control processing performed by the control circuit 441. Noise control processing is performed within the same period as the sampling period of the first A / D converter 444 and the second A / D converter 445. The sampling period is set to a period corresponding to twice the maximum frequency of the target noise. Furthermore, the period in which the noise control processing is performed is referred to as the processing cycle.
[0114] Control circuit 441 uses L reference signals x for the calculation of noise control filter M4. n A shift register that sequentially stores the L-dimensional reference vector x(n). Control circuit 441 uses N reference signals x for the calculation of the second-order filter M5. n ~x n-N+1 A shift register that sequentially stores (i.e., N-dimensional reference vectors x(n)). Control circuit 441 uses L filtered reference signals r for the calculation of coefficient update unit M6. n ~r n-L+1 A shift register that sequentially stores the filter reference vector r(n). Additionally, the control circuit 441 uses L registers, where the L registers are the processing results of the coefficient update unit M6 and the L coefficients w1 to w2 supplied to the noise control filter M4.L (That is, the coefficient vector w(n)) is stored. i Let be the coefficient of the i-th tap.
[0115] When control circuit 441 is started, the divergence flag FL is initialized to: FL = 0 (indicating disconnection), and the coefficients of the L taps in noise control filter M4, i.e., the coefficient vector w(n), are initialized to preset initial values. The divergence flag FL is set to FL = 1 (indicating connection) when the updated value of coefficient vector w(n) does not converge and tends to diverge. The initial value of coefficient vector w(n) can be, for example, a value representing the theoretical characteristic P / C, or an arbitrarily set value (e.g., all 1s).
[0116] When noise suppression processing begins, such as Figure 10 As shown, in S110, the control circuit 441 obtains the reference signal x. n The obtained reference signal x n They are stored in the shift registers used for noise control filter M4 and second-order filter M5, respectively.
[0117] In the next step, S120, the control circuit 441 performs the processing of the noise control filter M4. Specifically, the control circuit 441 uses the reference vector x(n) stored in the shift register for the noise control filter M4 and the coefficient vector w(n) stored in the output register of the coefficient update unit M6 to calculate the control signal u according to equation (1). n .
[0118]
Mathematical Formula 1
[0119]
[0120] In the next step, S130, the control circuit 441 performs coefficient update processing to update the coefficient vector w(n) of the noise control filter M4 according to the adaptive algorithm.
[0121] In the next step S140, the control circuit 441 performs a divergence determination process: adjusting the configuration of the noise control filter M4 based on whether the coefficient vector w(n) calculated through the coefficient update process tends to diverge.
[0122] In the next step, S150, the control circuit 441 will use the control signal u calculated in S120. n Output to D / A converter 446, ending processing. The control signal u, converted to an analog signal by D / A converter 446, is then processed. n Supply to control speaker 54, which generates control signal u nCorresponding suspected noise.
[0123] [1-4-2. Coefficient Update Processing]
[0124] use Figure 11 The flowchart illustrates the coefficient update process performed by the control circuit 441 in the previous S130.
[0125] When the coefficient update process begins, in S210, the control circuit 441 performs the processing for the second-order filter M5. Specifically, the control circuit 441 uses the N-dimensional reference vector x(n) stored in the shift register for the second-order filter M5, and the N pre-set coefficients c1 to c2 in the taps of the second-order filter M5. N According to equation (2), the filtered reference signal r is calculated. n The control circuit 441 uses the calculated filtered reference signal r n The shift register used by the coefficient update unit M6 is stored.
[0126]
Mathematical Formula 2
[0127]
[0128] In the next step, S220, the control circuit 441 calculates: R, the sum of squares of the L elements contained in the filter reference vector r(n) stored in the shift register used by the coefficient update unit M6. n The filter reference vector r(n) is: based on the L most recently calculated filter reference signals {r n r n-1 ... r n-L+1 Let} be an L-dimensional vector of the feature. The following will use the sum of squares R... n This is called the standardized value.
[0129] In the next step, S230, the control circuit 441 obtains the error signal e. n Error signal e n This is a scalar value. In the next step, S240, the control circuit 441 calculates the gradient vector S(n).
[0130] As shown in equation (3), the gradient vector S(n) is: the filter reference vector r(n) stored in the shift register used by the coefficient update unit M6, and the error signal e obtained in S230. n The product of the gradient vector S(n) and the gradient vector S(n) represent the direction of change of the coefficient vector w(n) in the L-dimensional coordinate space representing the coefficient vector w(n).
[0131]
Mathematical Expression 3
[0132] S(n)=e n ×r(n) (3)
[0133] In the next step S250, the control circuit 441 updates the coefficient vector w(n) using equation (4), stores the updated coefficient vector w(n+1) in the output register of the coefficient update unit M6, and supplies it to the noise control filter M4 to end the processing. μ is a scalar value used to adjust the convergence speed and estimation accuracy of the adaptive action, and is called the update step size.
[0134]
Mathematical Expression 4
[0135]
[0136] [1-4-3. Divergence Detection and Handling]
[0137] use Figure 12 The flowchart illustrates the divergence determination process performed by the control circuit 441 in the previous S140.
[0138] When the divergence determination process begins, in S310, the control circuit 441 determines whether the divergence flag FL is set to ON (i.e., FL = 1). If the control circuit 441 receives a positive determination, the process ends; if it receives a negative determination, the process proceeds to S320.
[0139] In S320, control circuit 441 acquires divergence determination parameters. The divergence determination parameters are parameters used to determine whether the updated coefficient vector w(n) in S130 tends to diverge. At least one of the first to third determination parameters shown below is used as the divergence determination parameters.
[0140] The first decision parameter is: the L taps in the noise control filter M4 are divided into upper taps and lower taps, and the value obtained by subtracting the value of the coefficient of the lower tap from the value of the coefficient of the upper tap. Here, the upper taps are the M taps from the 1st to the Mth tap, and the lower taps are the L-M taps from the (M+1)th to the Lth tap. The value of the coefficients can be, for example, the sum of the squares of the coefficients, or the sum of the absolute values of the coefficients. Here, the number of upper taps is set to M, however, the number of upper taps is not limited to M.
[0141] The second determination parameter is the error signal e. n The absolute value of.
[0142] The third determining parameter is the magnitude of the difference vector obtained by subtracting the updated coefficient vector w(n+1) from the original coefficient vector w(n). The magnitude of the difference vector can be the sum of the squares of the elements contained in the difference vector, or the sum of the absolute values of the elements.
[0143] In the next step S330, the control circuit 441 uses the divergence determination parameters obtained in S320 to determine whether the coefficients of the noise control filter M4 tend to diverge. If a positive determination is obtained, the process proceeds to S340; if a negative determination is obtained, the process ends.
[0144] Specifically, when the first determination parameter is used as the parameter for divergence determination, if the first determination parameter is negative, that is, the lower tap is greater than the upper tap, it can be determined that it tends to diverge.
[0145] When a second determination parameter is used as a parameter for divergence determination, if the second determination parameter is a certain value (e.g., twice) or more relative to the maximum noise value allowed by the device, it can be determined that the device is tending towards divergence. Furthermore, if the second determination parameter monotonically increases in multiple (e.g., three) processing cycles, it can be determined that the device is tending towards divergence.
[0146] If a third determination parameter is used as a parameter for divergence determination, and the third determination parameter monotonically increases in multiple (e.g., 3) processing loops, it can be determined that the process tends to diverge.
[0147] It should be noted that in determining the divergence trend, any one of the first to third determination parameters can be used, or multiple parameters can be combined. When multiple parameters are combined, if any one of the parameters is satisfied, it can be determined that the trend is diverging; if any multiple parameters are satisfied, it can also be determined that the trend is diverging.
[0148] In S340, control circuit 441 limits the noise control filter M4. Specifically, it limits the coefficients w of the L-M lower taps from the (M+1)th to the Lth tap of the noise control filter M4. M+1 ~w L Set a fixed value. The fixed value can be, for example, a coefficient representing the lower tap of the theoretical characteristic P / C. The fixed value can be set to any value. Alternatively, the L-M lower taps of the noise control filter M4 can be set as invalid taps, and the M upper taps from the 1st to the Mth can be set as valid taps. Using the valid taps, the control signal u can be calculated. n In S340, with the noise control filter M4 constrained, the control circuit 441, in a subsequent processing loop, uses the constrained noise control filter M4 to generate the control signal u when S120 is executed. n .
[0149] In the next step S350, the control circuit 441 sets the divergence flag FL to indicate that a divergence trend has been detected (i.e., FL = 1), and ends the process.
[0150] That is, when it is determined that the coefficient vector w(n) of the noise control filter M4 tends to diverge, the noise control filter M4 is restricted. After that, there is no need to remove the restriction, and the control signal u can continue to be calculated using the restricted noise control filter M4. n .
[0151] [1-5. Correspondence of Terms]
[0152] In this embodiment, the reference microphone 53 and the first A / D converter 444 (i.e., the reference sensor M1) correspond to an example of the reference acquisition unit in this invention, the processing of S220 to S250 and S340 corresponds to an example of the characteristic adjustment unit in this invention, and the processing of S320 to S330 corresponds to an example of the divergence determination unit in this invention.
[0153] [1-6. Effects]
[0154] According to the first embodiment described above, the following effects are achieved.
[0155] (1a) In this embodiment, the number of taps N of the second-order filter M5 is set to be shorter than the length of time that the impulse response fully converges, thus reducing the amount of computation required for noise control.
[0156] (1b) In this embodiment, before detecting the divergence trend, the control signal u is... n The noise control filter M4 used for generation contains all the coefficients w1 to w1 of the L taps. L Update accordingly. Additionally, after detecting the divergence trend, adjust the coefficient w of the lower tap. M+1 ~w L Set to a fixed value or invalidate the coefficients w1 to w of the upper tap. M An update is performed. Therefore, even if the number of taps N of the second-order filter M5 is set to an insufficient length, thus including a non-negligible estimation error, the coefficient divergence of the noise control filter M4 can be suppressed.
[0157] (1c) In this embodiment, the error microphone 55 is positioned to suppress noise other than object noise, such as wind noise caused by direct impact from the airflow generated by the drive unit 43, from mixing into the error microphone 55. Furthermore, the error microphone 55 is positioned at a location considered a noise cancellation point. The control speaker 54 is configured such that the phase of the suspected noise generated by the control speaker 54 is the same at the noise cancellation point and the position of the error microphone 55. As a result, the reference signal x detected by the reference microphone 53 can be improved. n The error signal e detected by error microphone 55 nThe coherence between them enables the control precision of suppressing object noise to be improved.
[0158] [1-7. The impact of second-order filter errors on the coefficients of noise control filters]
[0159] When a divergence trend is detected, the noise control filter M4 is limited, thereby suppressing the divergence. The principle behind this is explained below.
[0160] The error of the second-order filter M5 (hereinafter, the second-order system estimation error) represents the difference between the actual characteristics of the second-order system and the characteristics exhibited by the second-order filter M5. Here, it is first shown that the second-order system estimation error also has a significant impact on the coefficients of the noise control filter M4.
[0161] Figures 21-23 In this diagram, P(z) represents the z-transform of the characteristics of the first-order system from the reference sensor M1 to the error sensor M3. H(z) represents the z-transform of the characteristics of the noise control filter M4. C(z) represents the z-transform of the characteristics of the second-order system from the output of the noise control filter M4 to the error sensor M3 via the control sound source M2. CH(z) represents the z-transform of the characteristics of the second-order filter M5 after estimating the characteristics C(z) of the second-order system before the ANC operation begins and fixing the coefficients. N(z) represents the z-transform of the object noise. NLMS represents the coefficient update unit M6 based on the learning discrimination method. In this study, the return system B(z) from the control sound source M2 to the reference sensor M1 can be ignored. In this case, the structure of the feedforward ANC is... Figure 21 express.
[0162] When the coefficients of the noise control filter M4 are slowly updated using the coefficient update unit M6, the noise control filter M4 can be assumed to be a linear filter. Theoretically, as... Figure 22 As shown, the connection order of the second-order system and the noise control filter M4 can be interchanged.
[0163] Among them, the characteristic P(z) of the first-order system is replaced by PD(z) as shown in equation (5).
[0164]
Mathematical Expression 5
[0165]
[0166] Figure 22In the structure shown, if the characteristic C(z) of the second-order system is predetermined with zero error, then the input signals of the first-order system and the noise control filter M4, as well as the signals applied in the coefficient update unit M6, are all identical. In this case, when the characteristic H(z) of the noise control filter converges to the characteristic PD(z) of the first-order system and is adjusted to Hopt(z) as shown in equation (6), theoretically, the object noise is completely canceled. Hereinafter, Hopt(z) will be referred to as the characteristic of the optimal noise control filter.
[0167]
Mathematical Expression 6
[0168]
[0169] In practice, the estimated value of the second-order system (i.e., the characteristic of the second-order filter M5) CH(z) includes the error Δ(z) relative to the actual characteristic C(z) of the second-order system. Therefore, the characteristic CH(z) of the second-order filter M5 is expressed by equation (7).
[0170]
Mathematical Expression 7
[0171] CH(z)=C(z)+Δ(z) (7)
[0172] To examine the effect of error Δ(z), Figure 22 In this case, the characteristic C(z) of the second-order system connected to the first-order system at the input of the noise control filter M4 is replaced by the characteristic CH(z) of the second-order filter M5. In this case, as... Figure 23 As shown, the characteristic PD(z) of the first-order system becomes the characteristic PH(z) shown in equation (8).
[0173]
Mathematical Expression 8
[0174]
[0175] Depend on Figure 23 It can be seen that when the characteristic H(z) of the noise control filter M4 is adjusted to satisfy equation (9), that is, when the characteristic H(z) of the noise control filter M4 converges to equation (10), theoretically, the object noise is completely canceled.
[0176]
Mathematical Expression 9
[0177]
[0178]
[0179] However, in practice, since it is impossible to separate C(z) / CH(z) and the characteristic PH(z) of the noise control filter M4, the coefficients are updated according to the characteristics of the noise control filter M4 so that they converge to HD(z) as shown in equation (11). Here, ΔD(z) is represented by equation (12).
[0180]
Mathematical Formula 10
[0181]
[0182]
[0183] As can be seen from equations (11) and (12), the estimation error Δ(z) of the second-order system is not a simple addition operation, but is introduced into the characteristic HD(z) of the noise control filter M4 in the form of a product with the characteristic Hopt(z) of the optimal noise control filter, thereby reducing the noise reduction effect.
[0184] In order to clarify the degree of noise reduction effect degradation caused by the decomposition, the difference between the characteristic Hopt(z) of the optimal noise control filter and the characteristic HD(z) of the converged noise control filter M4 is shown in equation (13).
[0185]
Mathematical Expression 11
[0186]
[0187] As can be seen from equation (13), when ΔD(z) is sufficiently small, the difference between the characteristic HD(z) of the converged noise control filter M4 and the characteristic Hopt(z) of the optimal noise control filter is the product of Hopt(z) and ΔD(z).
[0188] If equation (11) is transformed, equation (14) is obtained.
[0189]
Mathematical Expression 12
[0190]
[0191] In this case, when the characteristic CH(z) of the second-order filter M5 contains errors relative to the characteristic C(z) of the actual second-order system, the noise reduction effect brought about by ANC will produce a reduction corresponding to the terms after the second term shown on the right side of equation (14).
[0192] Next, we will explain the case where the number of taps for the second-order filter M5 cannot be adequately prepared. Insufficiently prepared taps mean that the number of taps is less than the number required for the impulse response to fully converge.
[0193] When the number of taps of the second-order filter M5 cannot be adequately prepared, the error Δ(z) increases. When the number of taps of the second-order filter M5 is set to d, the characteristic CH(z) of the second-order filter M5 is expressed by equation (15).
[0194]
Mathematical Expression 13
[0195] CH(z)=C(z)-Δ(z)·z -d =C(z){1-ΔD(z)z -d} (15)
[0196] Δ(z)z -d This represents the z-transform of the cutoff portion of the impulse response of a second-order system. For convenience, it is assumed that the tap coefficients up to d⁻¹ are estimated with zero error.
[0197] In this case, the characteristic HD(z) of the converged noise control filter is expressed by equation (16), and the difference between the characteristic Hopt(z) of the optimal noise control filter and the case of equation (14) is expressed by the terms after the second term on the right.
[0198]
Mathematical Expression 14
[0199]
[0200] Equation (16) shows that the effect of the cutoff is limited to the dth and beyond coefficients of the noise control filter M4. This result means that if it is not possible to prepare a sufficient number of taps for the second-order filter M5, noise reduction can be achieved by making the number of taps of the noise control filter M4 the same as or less than the number of taps d of the second-order filter M5.
[0201] [2. Second Implementation]
[0202] [2-1. Differences from the first embodiment]
[0203] Regarding the second embodiment, the basic structure is the same as that of the first embodiment; therefore, the differences will be explained below. It should be noted that the same symbols as in the first embodiment denote the same configuration, as described in the existing explanation.
[0204] In the first embodiment described above, the coefficients of the noise control filter M4 are continuously updated in each processing loop. In contrast, in the second embodiment, the coefficients of the noise control filter M4 are updated only once per multiple processing loops (hereinafter, intervals), and the number of updates is limited, which differs from the first embodiment.
[0205] [2-2. Processing]
[0206] use Figure 13 The flowchart shows the noise control device 10 in the second embodiment. Figure 11 The coefficient update process performed in accordance with the coefficient update process of the first embodiment shown will be described. It should be noted that, regarding the coefficient update process of the first embodiment, the coefficient update process performed in accordance with the first embodiment shown will be described. Figure 11 The same processing labels and step numbers are used for the same processes shown, and descriptions are omitted. Additionally, when the noise control device 10 is activated, the count values kc and jc, as well as the cumulative values SR and SS, are all initialized to zero.
[0207] When the coefficient update process begins, in S200, the control circuit 441 determines whether the count value kc of the update count counter is less than the update limit Kmax. If the determination is positive, the process proceeds to S210; if the determination is negative, the process ends.
[0208] The processing in S210 is the same as in the first embodiment.
[0209] In the following step S222, control circuit 441 calculates the standardized value R in the same way as in S220. n Then, the standardized value R is calculated. n The cumulative value SR.
[0210] The subsequent processing in S230 is the same as in the first embodiment.
[0211] In the next step S242, the control circuit 441 calculates the gradient vector S(n) in the same way as in S240, and then calculates the cumulative value SS of the gradient vector S(n).
[0212] In the following S244, the control circuit 441 increases the count value jc of the interval calculation counter by 1.
[0213] In the next step, S246, the control circuit 441 determines whether the count value jc is the interval calculation count J. BK If a positive judgment is obtained, the process proceeds to S250; if a negative judgment is obtained, the process ends.
[0214] The processing of S250 is the same as in the first embodiment. However, in equation (5), the cumulative value SR is used instead of R. n The cumulative value RS is used instead of S(n).
[0215] In the next step S260, the control circuit 441 clears the count value jc of the interval calculation counter and increments the count value kc of the update count counter by 1, thus ending the process.
[0216] That is, in each J BKIn each processing cycle (below, within the interval), the standardized value R calculated and accumulated during this period is used. n The noise control filter M4 is updated using the accumulated value SR and the accumulated value SS of the gradient vector S(n). The update process for the noise control filter M4 is repeated Kmax times before ending.
[0217] J BK Kmax is set based on simulation and other experimental results to ensure that the error signal e n Simply reduce the number of times.
[0218] [2-3. Correspondence of Terms]
[0219] In this embodiment, Figure 13 The coefficient update process shown, excluding S210, is equivalent to an example of the characteristic adjustment unit in this invention.
[0220] [2-4. Effects]
[0221] According to the second embodiment detailed above, the effects of the first embodiment described above (1a)(1b)(1c) are achieved, and in addition, the following effects are achieved.
[0222] (2a) In this embodiment, the update of the coefficient vector w(n) of the noise control filter M4 uses the normalized value R obtained by interval calculation. n The cumulative value SR and the cumulative value SS of the gradient vector S(n). Therefore, even the standardized value R n Furthermore, it can suppress the impact of transient outliers in the gradient vector S(n) caused by disturbances, thereby improving the stability of the updated coefficient vector w(n).
[0223] (2b) In this embodiment, the number of updates to the coefficient vector w(n) of the noise control filter M4 is limited to Kmax. Therefore, it is possible to prevent the unnecessary updating of the coefficient vector w(n) after the value of the coefficient vector w(n) has converged. It should be noted that the number of updates to the coefficient vector w(n) of the noise control filter M4 can also be unlimited.
[0224] [3. Third Implementation Method]
[0225] [3-1. Differences from the first embodiment]
[0226] Regarding the third embodiment, the basic structure is the same as that of the second embodiment; therefore, the differences will be explained below. It should be noted that the same symbols as in the second embodiment denote the same configuration, as described in the existing explanation.
[0227] In the first and second embodiments described above, the update step size μ used for updating the coefficient vector w(n) of the noise control filter M4 is a fixed value. In contrast, in the third embodiment, the update step size μ is set variably, and the divergence trend is determined based on the deviation of the update step size μ, which differs from the first and second embodiments.
[0228] [3-2. Processing]
[0229] [3-2-1. Coefficient Update Processing]
[0230] use Figure 14 The flowchart shows the noise control device 10 in the third embodiment, which is used instead of the previous one. Figure 13 The coefficient update process performed in the second embodiment shown will be described.
[0231] In the coefficient update process of this embodiment, if a positive determination is obtained in S246, the coefficient / step size update process of S252 is performed instead of S250.
[0232] [3-2-2. Coefficient / Step Size Coefficient Update Processing]
[0233] use Figure 15 The flowchart shown illustrates the coefficient / step size update process performed by the control circuit 441 in S252.
[0234] When the coefficient / step update process begins, in S410, the control circuit 441 calculates, in interval units, the interference power Q, which is the signal power of noise other than the object noise. kc The signal power P of the object noise kc Interference power Q kc The error signal e was used n The cumulative squared value of the signal power P over the interval. kc Multiple reference signals x contained in the reference vector x(n) were used. n The sum of squares is the cumulative value over the interval.
[0235] In the next step S420, the control circuit 441 updates the step size μ according to equation (17). kc Update. TP J is the number of taps in the noise control filter M4. BK Let G be the interval length and G be the required value of the estimation error. The estimation error is the error Δ(z) between the characteristic CH(z) of the second-order filter M5 and the characteristic C(z) of the actual second-order system. The required value G is a parameter that determines the minimum accuracy of the estimation error, and is set to a value where 1 > G > 0. The initial value G0 is arbitrary and set to a value close to 1, for example, 0.9.
[0236]
Mathematical Expression 15
[0237]
[0238] In the next step, S430, the control circuit 441 updates the coefficient vector w(kc) of the noise control filter M4. In this process, μ calculated in S420 is used as the update step size. kc It replaces the fixed value μ, otherwise the process is the same as in S250.
[0239] In the following S440, the control circuit 441 calculates the change in the magnitude of the coefficient vector w(kc) of the updated noise control filter M4 in S430 (hereinafter, coefficient change) D. kc The magnitude of the coefficient vector w(kc) can be, for example, the sum of squares of all features, or the sum of the absolute values of all features.
[0240] In the following S450, control circuit 441 determines: the coefficient change D kc Whether it exceeds a preset judgment threshold TH. If the control circuit 441 receives a positive judgment, it will proceed to S460; if it receives a negative judgment, it will end the processing.
[0241] In S460, the control circuit 441 updates the step size μ according to equations (18) and (19). kc The required values of the parameters GX and the decision threshold TH are updated, and the process ends.
[0242]
Mathematical Expression 16
[0243] GX←a GX ·GX (18)
[0244] TH←b TH ·TH (19)
[0245] a GX Adjust the coefficient to the desired value, setting it to 1 > a. GX >0. b TH The threshold adjustment coefficient is set to 1 > b. TH >0. Required value adjustment factor a GX and threshold adjustment coefficient b TH They are all set to values close to 1, such as 0.9, to minimize the variation in the required value GX and the decision threshold TH.
[0246] As explained above, in the coefficient / step size update process, the update step size μ used for calculating the coefficient vector w(kc) of the noise control filter M4 is...kc The update is also performed during the coefficient / step size update process. Additionally, the step size μ is updated during the coefficient / step size update process. kc The required value GX and the decision threshold TH are adjusted.
[0247] The change in coefficient vector w(kc) D kc If the value is greater than the decision threshold TH (i.e., tending towards divergence), repeat the process: making the update step size μ... kc The required value GX and the decision threshold TH are adjusted by decreasing the required value GX and the decision threshold TH in stages.
[0248] If the desired value GX is reduced, the calculated update step size μ kc It is also relatively small, resulting in a smaller variation D in the calculated coefficient vector w(kc). kc It is also relatively small, thus the divergence is suppressed. However, when the calculated change D... kc When the value is small, it is difficult to detect the divergence trend, and it may go undetected regardless of whether the divergence trend continues. Therefore, the threshold TH used for determining the divergence trend is also adjusted to decrease according to the desired value GX.
[0249] [3-2-3. Divergence Detection and Handling]
[0250] use Figure 16 The flowchart illustrates the divergence determination process performed by the control circuit 441 in S140. This divergence determination process is performed instead of the process performed by the control circuit 441 in the first and second embodiments. Figure 12 The divergence determination process is shown below.
[0251] like Figure 16 As shown, when the divergence determination process begins, in S510, the control circuit 441 determines: the number of effective taps I of the noise control filter M4. TP If the number of valid taps (TPmin) exceeds the lower limit, the process proceeds to step S520 if a positive result is obtained; otherwise, the process ends. Valid tap count I TP Here, TPmin is the number of taps of the noise control filter M4, which is the target of the coefficient update unit M6. The lower effective number TPmin is set to, for example, less than or equal to the number of taps N of the second-order filter M5. However, the lower effective number TPmin can be set to a value greater than the number of taps N of the second-order filter M5.
[0252] In S520, the control circuit 441, similar to S330, determines whether the noise control filter M4 tends to diverge. However, in determining the divergence tendency, in addition to using the first to third determination parameters, the coefficient change D calculated in S440 can also be used.kc and the update step size μ calculated in S420 kc The deviation (e.g., the absolute value of the difference from the previous value |μ) kc -μ kc-1 |) etc. Using the coefficient change D kc and update step size μ kc If these values are greater than the allowable value in the case of deviation, it is determined that the trend is diverging.
[0253] In S530, control circuit 441 sets the effective tap number I of noise control filter M4 to... TP Reduced adjustment amount ΔI TP (For example, 50), then the process ends.
[0254] It should be noted that the noise control filter M4 is configured as follows: the first to the first... TP I TP The first tap is a valid tap, the Ith tap. TP +1 up to the Lth L-I TP The tap is invalid.
[0255] In the noise control filter coefficient vector update process of S430, the coefficients of valid taps are updated, while the coefficients of invalid taps are set to fixed values. Additionally, in the noise control filter processing of S120, invalid taps can be used to control the signal u. n The calculation can also be used without being used for the control signal u. n The calculation.
[0256] That is, update step size μ kc If the deviation exceeds the allowable value, it is determined that the coefficient vector w(kc) of the noise control filter M4 tends to diverge, and the effective taps of the noise control filter M4 are reduced by an adjustment amount ΔI starting from the lower position. TP .
[0257] If a large value is calculated in a certain processing loop, the update step size μ is... kc Then the coefficient vector w(kc) of the noise control filter M4 tends to diverge, which is the error signal e output by the error sensor M3. kc Increase. When a large error signal e is detected. kc In the next processing loop, as shown in equation (17), the update step size μ of the smaller value is calculated. kc The divergence trend is thus suppressed. The same operation is then repeated. That is, a cycle of divergence and suppression is generated, which causes the update step size μ to change. kc The system changes, and therefore, based on this change, it can be determined whether the system tends to diverge.
[0258] [3-3. Terminology Correspondence]
[0259] In this embodiment, Figure 14 The coefficient update process shown, excluding S210, and S530 are examples of the characteristic adjustment unit in this invention, while S520 are examples of the divergence determination unit in this invention.
[0260] [3-4. Effects]
[0261] According to the third embodiment detailed above, the effects of the first embodiment (1a), (1b), and (1c) and the effects of the second embodiment (2a) and (2b) are achieved, and the following effects are also achieved.
[0262] (3a) In this embodiment, the update step size μ is adjusted according to the convergence of the coefficient vector w(kc) of the noise control filter M4. kc A variable, so-called step-size control method. Therefore, even with disturbance power Q... kc With the change, the estimation error of the coefficient vector w(kc) of the noise control filter M4 can also be reduced to below the required value GX.
[0263] (3b) In this embodiment, the update step size μ kc If the deviation exceeds the allowable value, continue the following process until the step size μ is updated. kc Until stable, that is: starting from the lower side, adjust the taps contained in the noise control filter M4 by each adjustment amount ΔI. TP The interference is rendered ineffective. Therefore, even at the interference power Q... kc The changing conditions also allow for a stable update of the coefficient vector w(kc) of the noise control filter M4.
[0264] (3c) In this embodiment, the following is repeated: the update step size μ is adjusted. kc The required value GX and the decision threshold TH are gradually decreased until the change in the coefficient vector w(kc) caused by the update converges and stabilizes. Therefore, it is possible to make the required value GX and even the update step size μ... kc Converging to: the appropriate size for the noise control filter M4 to operate stably.
[0265] [4. Fourth Implementation Method]
[0266] [4-1. Differences from the first embodiment]
[0267] In the fourth embodiment, the work machine using the noise control device 10 and the control model of the ANC used in the noise control device 10 are different from those in the first embodiment.
[0268] [4-2. Components of a Vacuum Cleaner]
[0269] Another example of an electric work machine using the noise control device 10 will be described, namely a portable vacuum cleaner 8. The portable vacuum cleaner 8 is a rechargeable electric vacuum cleaner used by the operator in a handheld state.
[0270] In this embodiment, for ease of explanation, the portable vacuum cleaner 8 is described as follows: Figures 17-19 As shown, the terms front, back, top, bottom, left, and right are defined.
[0271] like Figures 17-19 As shown, the portable vacuum cleaner 8 has a main body shell 80. The main body shell 80 includes: a suction port 81, an exhaust port 82, a handle 83, and first and second battery assembly parts 85A and 85B.
[0272] The suction port 81 is a cylindrical portion located at the front of the main body casing 80, used to draw in external atmosphere. The discharge port 82 is a slit-like portion located at the lower rear of the main body casing 80, used to discharge air after dust removal. The grip portion 83 is a portion located on the upper surface of the main body casing 80 for the operator to hold. An electronic switch 84 is provided on the grip portion 83 so that it can be operated while the user is holding it. The first and second battery assembly portions 85A and 85B are located on the upper side of the rear surface of the main body casing 80. A first battery pack 86A is installed in the first battery assembly portion 85A, and a second battery pack 86B is installed in the second battery assembly portion 85B.
[0273] The portable vacuum cleaner 8 has a drive unit 90 and a control circuit board 91 inside the main body shell 80.
[0274] A drive unit 90 is disposed between the suction port 81 and the discharge port 82 within the housing, and is configured to divide the internal space of the main housing 80 into a suction port side space 80a and a discharge port side space 80b. A dust collection bag is disposed in the suction port side space 80a. A sound-absorbing material 87 is disposed on the inner wall of the main housing 80 forming the discharge port side space 80b. For example, a sponge is used as the sound-absorbing material 87. In addition, a control speaker 88 and an error microphone 89 are disposed on the upper part of the sound-absorbing material 87.
[0275] The control speaker 88 and the error microphone 89 are mounted so that their directionality is directed toward the exhaust port side space 80b and are sandwiched with sound-absorbing material 87, and are positioned opposite to the exhaust port side space 80b. That is, they are configured so that the airflow flowing within the main body housing 80 does not directly impact the error microphone 89.
[0276] Furthermore, the error microphone 89 is positioned at the location of the standing wave antinodes of the noise generated within the storage space of the main housing 80. The location where the error microphone 89 is positioned becomes a noise cancellation point. The cross-sectional area of the airflow flowing within the main housing 80 changes discontinuously at the outlet 82 (the cross-sectional area rapidly expands), causing sound to be reflected at the outlet 82, thereby generating a standing wave of noise.
[0277] It should be noted that the control speaker 88 and error microphone 89 essentially have the same capabilities as the control speaker 54 and error microphone 55 described in the first embodiment, and are configured with the same idea.
[0278] like Figure 18 As shown, the drive unit 90 includes: a motor housing 901, a motor 902, and a fan 903.
[0279] The motor housing 901 is a cylindrical component that contacts the inner wall of the main housing 80. By holding the motor 902 at the center of the cylinder, a flow path 80c with an annular cross-section is formed between the motor 902 and the inner wall of the motor housing 901.
[0280] Motor 902 is a DC motor that operates according to instructions from the control circuit board 91.
[0281] Fan 903 is fixed to the rotating shaft of motor 902.
[0282] The drive unit 90 is installed inside the main housing 80 with the motor 902 facing the suction port side space 80a and the fan 903 facing the discharge port side space 80b. The drive unit 90 is configured such that the motor 902 drives the fan 903 to rotate, thereby generating an airflow from the suction port 81 toward the discharge port 82 inside the main housing 80.
[0283] When the drive unit 90 generates airflow, it draws external air from the external space of the main body housing 80 into the internal space of the main body housing 80 via the suction port 81. As the drawn external air passes through the suction port side space 80a, dust contained in the external air is captured by the dust collection bag. The air passing through the dust collection bag reaches the discharge port side space 80b via the flow path 80c and is discharged from the discharge port 82.
[0284] [4-3. Control Circuit Board]
[0285] The control circuit board 91 operates by receiving power from the battery pack 86. The control circuit board 91 includes a motor control unit 911 that drives the motor 902 according to the operation of the electronic switch 85, and a noise control unit 912 that suppresses the noise (hereinafter, object noise) generated by the operation of the motor 902.
[0286] The motor control unit 911 is equipped with: Figure 8 The dust collection circuit group 442 and the control circuit 441 shown are equivalent in configuration.
[0287] The noise control unit 912 is equipped with: Figure 8 The second A / D converter 445, D / A converter 446, and control circuit 441 shown are configured similarly. The noise control unit 912, together with the control speaker 88 and the error microphone 89, can implement feedback-type ANC. That is, the noise control unit 912, the control speaker 88, and the error microphone 89 form a noise control device.
[0288] [4-4. ANC Model]
[0289] use Figure 20 This paper describes a model of a feedback-type ANC implemented using a noise control device. Specifically, it discusses the use of... Figure 9 Compared to the feedforward ANC model described above, only a portion of the components differ; therefore, the same symbols are used for the same components and the descriptions are omitted.
[0290] like Figure 20 As shown, the feedback-type ANC has the following structure: compared to the feedforward-type ANC, the reference sensor M1 is omitted, and an arrival filter M7 and an adder M8 are added. It should be noted that controlling the sound source M2 is equivalent to controlling the speaker 88 and... Figure 8 The D / A converter 446 and the error sensor M3 are equivalent to the error microphone 89. Figure 8 The second A / D converter 445 in the system.
[0291] When the noise control unit 912 includes a microcomputer (hereinafter, microcomputer), the noise control filter M4, the second-order filter M5, the coefficient update unit M6, the arrival filter M7, and the summing arithmetic unit M8 can be implemented by the microcomputer, and some or all of them can be implemented by hardware.
[0292] The newly added arrival filter M7 has the same configuration as the second-order filter M5, based on the recently calculated N control signals u. n (i.e., an N-dimensional control vector u(n)), presumably representing the arrival suspected signal a of suspected noise from the control sound source M2 to the error sensor M3. n .
[0293] Adder M8 reaches the suspected signal a n Subtract the error signal e n To deduce the reference signal x representing the noise of the object. n .
[0294] That is, in a feedback-type ANC, the reference signal xn Using control signal u n and error signal e n The estimated result is used instead of the detection result of the reference sensor M1, which is different from the feedforward type ANC.
[0295] [4-5. Processing]
[0296] The processing performed by the noise control unit 912 is basically the same as that described in any of the first to third embodiments. However, Figure 10 The noise suppression process S110 shown differs from the previous one in that it uses the arrival filter M7 to generate the arrival suspected signal a. n Furthermore, by reaching the suspected signal a n Subtract the error signal e n To generate the reference signal x n This replaces the reference signal x generated by the reference sensor M1. n .
[0297] [4-6. Correspondence of Terms]
[0298] In this embodiment, the arrival filter M7 corresponds to an example of the arrival filter in this invention, and the adder M8 corresponds to an example of the adder unit in this invention.
[0299] [4-7. Effects]
[0300] According to the fourth embodiment detailed above, the effects of the first embodiment (1a)(1b)(1c), the second embodiment (2a)(2b), and the third embodiment (3a)(3b)(3c) are achieved, and the following effects are also achieved.
[0301] (4a) In this embodiment, a feedback-type ANC is used, so the reference sensor M1 can be omitted, thereby simplifying the device configuration.
[0302] [5. Other Implementation Methods]
[0303] (5-1) In the above embodiment, the number of taps L of the noise control filter M4 is set to be greater than the number of taps N of the second-order filter M5. When a tendency towards divergence is detected, the number of taps in the noise control filter M4 that are the targets for coefficient updates is limited to the upper limit of M. The technology of the present invention is not limited to changing the number of taps of the noise control filter M4 from L to M; the number of taps of the noise control filter M4 can be fixed to M in advance. In this case, the process of detecting divergence and limiting the number of effective taps can also be omitted. In this case, the noise control process performed by the control circuit 441 is omitted. Figure 10The divergence determination process of S140 is shown.
[0304] (5-2) In the above embodiment, in the processing of S110, the reference signal x n The detection signal is used with reference microphone 53; however, the reference signal x n The reference signal x can be any signal that is correlated with the noise of the object, and is not limited to the detection signal of the reference microphone 53. For example, the signal that drives the motor 432 can be used as the reference signal x. n .
[0305] (5-3) In the above embodiments, the control speaker 54 and error microphone 55, as well as the control speaker 88 and error microphone 89 are arranged side by side in the same direction, however, they can be arranged opposite each other.
[0306] (5-4) In the third embodiment described above, the interval length J is... BK Constant, however, assuming signal power P kc Under the condition of change, the interval length J can be made BK The interval length J is variable. BK Extended control until signal power P kc Until it reaches a constant value.
[0307] (5-5) In the above embodiments, the dust collector 1 and the portable vacuum cleaner 8 are examples of electric work machines to which the technology of the present invention is applied. However, the electric work machines to which the technology is applied are not limited to the dust collector 1 and the portable vacuum cleaner 8. For example, they can be used in DIY, manufacturing, gardening, engineering and other work sites, and are applicable to various field electrical equipment that utilizes airflow generated by using a fan. Specifically, the technology of the present invention can be applied to various electric work machines such as gardening work machines and devices for tidying up the work site environment. More specifically, the present invention can be applied to various electric work machines such as electric lawnmowers, electric lawn shears, electric harvesters, electric vacuum cleaners, electric blowers, electric sprayers, electric spreaders, electric dust collectors and so on.
[0308] (5-6) In the above embodiments, the multiple functions of one constituent element can be implemented by multiple constituent elements, or the single function of one constituent element can be implemented by multiple constituent elements. Furthermore, the multiple functions of multiple constituent elements can be implemented by one constituent element, or the single function implemented by multiple constituent elements can be implemented by one constituent element. Additionally, a portion of the configuration of the above embodiments can be omitted. Furthermore, at least a portion of the configuration of the above embodiments can be added to or substituted relative to the configurations of other above embodiments.
Claims
1. An electric work machine, characterized in that, This electric work machine has the following features: A motor configured to generate the driving force required to perform a predetermined field operation; The reference acquisition unit is configured to acquire a reference signal that is correlated with the noise caused by the operation of the motor, i.e., the object noise. A noise control filter, which is a digital filter including a series of taps and capable of adjusting characteristics, and configured to receive the reference signal from the reference acquisition unit to generate a control signal for eliminating noise in the object; A control sound source is configured to generate a simulated noise according to the control signal; An error sensor is configured to convert the combined sound of the suspected noise and the object noise at a predetermined silence point into an error signal; The second-order filter is a digital filter with N taps obtained by modeling the path from the control sound source to the error sensor, i.e., the second-order system, in advance. The second-order filter is configured to receive the reference signal from the reference acquisition unit to generate a filtered reference signal. as well as The characteristic adjustment unit is configured to adjust the characteristics of the noise control filter using the error signal and the filter reference signal according to an adaptive algorithm. The characteristic adjustment unit is configured to update the coefficients of M taps in the series of taps. The M taps correspond to some or all of the series of taps. M and N are positive integers that satisfy M < N. The feature adjustment unit is configured to update the coefficients by multiplying an update step size, which represents the degree of change of the coefficients, by an update value, which represents the direction of change of the coefficients, and to calculate the update value using the cumulative value of the gradient vector calculated in multiple processing loops.
2. The electric work machine according to claim 1, characterized in that, The M taps correspond to all of the series of taps. The characteristic adjustment unit is configured to update all coefficients of the M taps.
3. The electric work machine according to claim 1, characterized in that, The electric work machine also includes a divergence determination unit, which is configured to determine whether the noise control filter tends to diverge. The characteristic adjustment unit is configured such that, before determining that the noise control filter tends to diverge using the divergence determination unit, the coefficients of L taps in the series of taps are updated; after determining that the noise control filter tends to diverge, the coefficients of the (M+1)th tap and beyond in the series of taps are set to predetermined values, and the coefficients of the first to the Mth taps in the series of taps are updated. The L taps correspond to all of the series of taps. L is a positive integer that satisfies L≥N.
4. The electric work machine according to claim 3, characterized in that, The predetermined value is: the path from the motor to the error sensor, which is the impulse response of the first-order system multiplied by the inverse filtering of the second-order filter.
5. The electric work machine according to claim 3 or 4, characterized in that, The divergence determination unit is configured to determine whether the noise control filter tends to diverge based on the intensity of the error signal and / or the trend of its change.
6. The electric work machine according to claim 3 or 4, characterized in that, The divergence determination unit is configured to determine whether the noise control filter tends to diverge based on the output intensity of each of the L taps, and / or the magnitude of the coefficients of each of the L taps, and / or the changing trend of the parameters used to update the coefficients.
7. The electric work machine according to claim 3 or 4, characterized in that, The L taps have a length corresponding to the time required for the impulse response of the first-order system to converge, i.e., the path from the motor to the error sensor.
8. The electric work machine according to any one of claims 1 to 4, characterized in that, The N taps have a length sufficient to terminate the processing required to generate the suspected noise. The processing required to generate the suspected noise includes at least one of the following processes: the process in which the characteristic adjustment unit updates the coefficients according to the reference signal during the time required from the detection of the reference signal to the detection of noise that is the source of the reference signal by the error sensor; and the process in which the noise control filter, after updating the coefficients, generates the control signal according to the reference signal.
9. The electric work machine according to any one of claims 1 to 4, characterized in that, The feature adjustment unit is configured to stop updating the coefficients when the number of updates reaches a predetermined value.
10. The electric work machine according to any one of claims 1 to 4, characterized in that, The characteristic adjustment unit is configured to change the update step size according to the convergence status of the noise control filter.
11. The electric work machine according to any one of claims 1 to 4, characterized in that, The reference acquisition unit includes a reference sensor configured to generate the reference signal by detecting noise in the object.
12. The electric work machine according to claim 11, characterized in that, The reference sensor is configured such that the distance between the reference sensor and the control sound source is longer than the distance between the control sound source and the error sensor.
13. The electric work machine according to any one of claims 1 to 4, characterized in that, The reference acquisition unit is configured to acquire a drive signal that drives the motor, and use it as the reference signal.
14. The electric work machine according to any one of claims 1 to 4, characterized in that, The reference acquisition unit includes: An arrival filter, configured to have the same characteristics as the second-order filter, receives the control signal from the noise control filter to generate an arrival signal representing the suspected noise arriving at the error sensor; and The addition operation unit is configured to add the arrival signal and the error signal to generate the reference signal.
15. The electric work machine according to any one of claims 1 to 4, characterized in that, This electric work machine has the following features: A fan configured to be driven by the motor to generate airflow; A flow path configured to allow airflow generated by the fan; and An outlet, configured to discharge the airflow from the flow path. The control sound source and the error sensor are configured such that the outlet becomes the noise reduction point.
16. The electric work machine according to claim 15, characterized in that, At least a portion of the wall surface of the flow path is provided with sound-absorbing material for suppressing wind noise generated by the airflow impact. The error sensor is positioned opposite the flow path while the sound-absorbing material is sandwiched between it.
17. A method for controlling noise generated by an electric work machine, characterized in that, include: The step of obtaining a reference signal that is correlated with the noise caused by the operation of the motor in the electric work machine, i.e., the object noise; The step of generating a control signal for eliminating noise in the object based on the reference signal using a digital filter, i.e., a noise control filter, that includes a series of taps and adjustable characteristics; The step of generating suspected noise by controlling a sound source according to the control signal; The step of converting the synthesized sound of the suspected noise and the object noise at a predetermined silencing point into an error signal using an error sensor; The step of generating a filtered reference signal based on the reference signal is to use a digital filter with N taps, i.e., a second-order filter, obtained by modeling the path from the control sound source to the error sensor (i.e., a second-order system) in advance. as well as The step of updating the coefficients of M taps in the series of taps using the error signal and the filtered reference signal, according to an adaptive algorithm. The M taps correspond to some or all of the series of taps. M and N are positive integers that satisfy M < N. The coefficients are updated by multiplying an update step size, representing the degree of change in the coefficients, by an update value, representing the direction of change in the coefficients. The method further includes the step of calculating the updated value using the cumulative value of the gradient vector calculated in multiple processing loops.
