Divergence control of adaptive filter for interior noise cancellation

Abstract

An active noise cancellation (ANC) system may include an adaptive filter divergence detector for detecting divergence of one or more controllable filters as they adapt based on various time domain and frequency domain amplitude characteristics. Upon detection of controllable filter divergence, the ANC system may be disabled or certain speakers may be muted. Alternatively, the ANC system may modify the diverging controllable filters to restore proper operation of the noise cancellation system. This may include adjusting a leakage value of an adaptive filter controller.

Description

Technical Field

[0001] The present disclosure relates to active noise cancellation and, more particularly, to mitigating the effects of adaptive filter divergence in engine order noise cancellation systems and / or road noise cancellation systems. Background Art

[0002] Active noise control (ANC) systems attenuate undesirable noise using feed-forward and feedback structures to adaptively remove undesirable noise within a listening environment, such as a vehicle cabin. ANC systems typically cancel or reduce undesirable noise by generating canceling sound waves to destructively interfere with the undesirable audible noise. Destructive interference occurs when noise and "anti-noise" (which is substantially the same magnitude as the noise but opposite in phase to the noise) reduce the sound pressure level (SPL) at a certain location. In a vehicle cabin listening environment, potential sources of undesirable noise come from the engine, the interaction between the vehicle's tires and the road surface (on which the vehicle travels), and / or the sound radiated by vibrations of other parts of the vehicle. Therefore, undesirable noise varies with the speed, road conditions, and operating state of the vehicle.

[0003] A road noise cancellation (RNC) system is a specific ANC system implemented on a vehicle to minimize undesirable road noise inside the vehicle cabin. RNC systems use vibration sensors to sense road-induced vibrations generated from the tire-road interface, which can result in unwanted audible road noise. This unwanted road noise inside the vehicle cabin is then canceled or reduced in level by using speakers to generate sound waves that are ideally opposite in phase and equal in magnitude to the noise to be reduced at the ears of one or more listeners. Cancelling this road noise can result in a more pleasant ride for vehicle occupants and enables automakers to use lightweight materials, thereby reducing energy consumption and emissions.

[0004] An engine order noise cancellation (EOC) system is a specific ANC system implemented on a vehicle to minimize unwanted vehicle interior noise originating from narrowband acoustic and vibration emissions from the vehicle's engine and exhaust system. EOC systems use a non-acoustic signal, such as a revolutions per minute (RPM) sensor, which generates a reference signal representing the engine speed, as a reference. This reference signal is used to generate a sound wave that is opposite in phase to the engine noise that can be heard inside the vehicle. Because EOC systems use data from an RPM sensor, they do not require a vibration sensor.

[0005] RNC systems are typically designed to eliminate broadband signals, while EOC systems are designed and optimized to eliminate narrowband signals (such as individual engine orders). An ANC system within a vehicle may provide both RNC and EOC techniques. Such vehicle-based ANC systems are typically least mean square (LMS) adaptive feedforward systems that continuously adapt W filters based on noise inputs (e.g., acceleration inputs from vibration sensors in the RNC system) and signals from error microphones located in various locations inside the vehicle cabin. ANC systems are susceptible to instability or divergence of adaptive W filters. When the W filters are adapted by the LMS system, one or more of the W filters may diverge instead of converging to minimize stress at the location of the error microphone. Typically, the first tap in an adaptive W filter represented in the time domain has a high amplitude, and the amplitude decreases to zero in later taps. However, if the adaptive W filters diverge, they may not have this feature. Divergence of the adaptive filter may result in broadband or narrowband noise enhancement or other undesirable behavior of the ANC system. Summary of the invention

[0006] In one or more illustrative embodiments, a method for controlling stability in an active noise cancellation (ANC) system is provided. The method may include: receiving filter coefficients corresponding to at least one controllable filter from an adaptive filter controller. The method may also include: calculating a parameter based on an analysis of at least a portion of the filter coefficients; detecting divergence of the at least one controllable filter based on a comparison of the parameter to a threshold; and modifying a property of the at least one controllable filter that has diverged.

[0007] Implementations may include one or more of the following features. The controllable filter may include a plurality of coefficients. The parameter may be a sum of absolute values ​​of at least a portion of the coefficients in the at least one controllable filter. The parameter may be a maximum value of at least a portion of the coefficients in the at least one controllable filter.

[0008] Furthermore, detecting the divergence of the at least one controllable filter based on a comparison of the parameter with a threshold value may include detecting the divergence of the at least one controllable filter when the parameter exceeds the threshold value. The threshold value may be a dynamic threshold value calculated based on a statistical analysis of the parameter calculated from filter coefficients in one or more previous adaptations of the at least one controllable filter. The threshold value may be an average value of the parameter obtained from multiple previous adaptations of the at least one controllable filter multiplied by a gain factor.

[0009] In addition, detecting the divergence of the at least one controllable filter based on the comparison of the parameter with a threshold may include: comparing the currently adapted parameter from the at least one controllable filter with an average value of the same parameter from one or more previous adaptations of the at least one controllable filter; and detecting the divergence of the at least one controllable filter when the difference between the currently adapted parameter from the at least one controllable filter and the average value from the one or more previous adaptations of the at least one controllable filter exceeds a threshold.

[0010] Modifying the properties of the at least one controllable filter that has diverged may include deactivating at least one of the ANC system and the at least one controllable filter that has diverged. Alternatively, modifying the properties of the at least one controllable filter that has diverged may include resetting the filter coefficients of the at least one controllable filter to zero and allowing the at least one controllable filter to re-adapt, or it may include resetting the filter coefficients of the at least one controllable filter to a set of filter coefficient values ​​stored in a memory of the ANC system.

[0011] Additionally, modifying the property of the at least one controllable filter that has diverged may include increasing a leakage value of the adaptive filter controller in response to detecting the divergence of the at least one controllable filter. The leakage value of the adaptive filter controller may increase at the divergence frequency of the at least one controllable filter. Additionally, the method may further include decreasing the leakage value of the adaptive filter controller when the highest magnitude filter coefficient of the at least one controllable filter is below a predetermined threshold.

[0012] One or more additional embodiments may be directed to an ANC system comprising: at least one controllable filter configured to generate an anti-noise signal based on an adaptive transfer characteristic and a noise signal received from a sensor. The adaptive transfer characteristic of the at least one controllable filter may be characterized by a set of filter coefficients. The ANC system may also include: an adaptive filter controller and a divergence controller, the divergence controller communicating at least with the adaptive filter controller. The adaptive filter controller may include a processor and a memory programmed to adapt the set of filter coefficients based on the noise signal and an error signal received from a microphone located in a cabin of a vehicle. The divergence controller may include a processor and a memory programmed to: receive the set of filter coefficients corresponding to a current adaptation of the adaptive transfer characteristic of the at least one controllable filter; calculate a parameter based on an analysis of at least a portion of the set of filter coefficients; and detect a divergence of the at least one controllable filter based on a comparison of the parameter with a threshold.

[0013] Implementations may include one or more of the following features. The threshold may be a predetermined static threshold programmed for the ANC system. The divergence controller may be programmed to detect divergence of the at least one controllable filter when a difference between a parameter calculated from the current adaptation of the at least one controllable filter and an average of the same parameter from one or more previous adaptations of the at least one controllable filter exceeds the threshold. The divergence controller may be further programmed to increase a leakage value of the adaptive filter controller in response to detecting divergence of the at least one controllable filter.

[0014] One or more additional embodiments may be directed to a computer program product embodied in a non-transitory computer-readable medium, the computer program product being programmed for active noise cancellation (ANC). The computer program product may include instructions for: receiving a set of filter coefficients corresponding to a current adaptation of at least one controllable filter from an adaptive filter controller; calculating a parameter based on an analysis of at least a portion of the filter coefficients; detecting a divergence of the at least one controllable filter based on a comparison of the parameter to a threshold; and in response to detecting a divergence of the at least one controllable filter, modifying an adaptive transfer characteristic of the at least one controllable filter during the current adaptation.

[0015] Implementations may include one or more of the following features. The computer program product, wherein the instructions for detecting the divergence of the at least one controllable filter may include: detecting the divergence frequency of the at least one controllable filter in the time domain; and wherein the instructions for modifying the adaptive transfer characteristic may include: resetting the divergence frequency of the at least one controllable filter to zero, attenuating the filter coefficients at the divergence frequency, or increasing a leakage value of the adaptive filter controller at the divergence frequency in the time domain. Furthermore, wherein the instructions for detecting the divergence of the at least one controllable filter may include: detecting the divergence frequency of the at least one controllable filter in the frequency domain; and wherein the instructions for modifying the adaptive transfer characteristic may include: notching the divergence frequency in the frequency domain using an error signal received from a microphone and previously adapted filter coefficients from the one controllable filter stored in a memory. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is an environmental block diagram of a vehicle having an active noise control (ANC) system including road noise cancellation (RNC) according to one or more embodiments of the present disclosure;

[0017] Figure 2is a sample schematic diagram showing relevant portions of an RNC system expanded to include R accelerometer signals and L speaker signals;

[0018] Figure 3 is a sample schematic diagram of an ANC system including an engine order noise cancellation (EOC) system and an RNC system;

[0019] Figure 4 is a sample lookup table of the frequency of each engine order for a given RPM in the EOC system;

[0020] Figure 5 is a schematic block diagram representing an ANC system including a divergence controller according to one or more embodiments of the present disclosure;

[0021] Figure 6 is a flow chart depicting a method for detecting corrected divergence of an adaptive filter in an ANC system according to one or more embodiments of the present disclosure; and

[0022] Figure 7 is a graphical representation of an analysis of a controllable filter in the frequency domain using a threshold value in accordance with one or more embodiments of the present disclosure. DETAILED DESCRIPTION

[0023] As required, detailed embodiments of the present invention are disclosed herein; however, it should be understood that the disclosed embodiments are merely exemplary of the present invention that may be embodied in various forms and alternative forms. The drawings are not necessarily drawn to scale; some features may be exaggerated or minimized to show details of particular components. Therefore, the specific structural and functional details disclosed herein should not be interpreted as limiting, but merely as a representative basis for teaching one skilled in the art to employ the present invention in various ways.

[0024] Any one or more of the controllers or devices described herein include computer executable instructions that can be compiled or interpreted from a computer program generated using a variety of programming languages ​​and / or techniques. In general, a processor (such as a microprocessor) receives instructions, for example, from a memory, a computer readable medium, etc., and executes the instructions. The processing unit includes a non-transitory computer readable storage medium capable of executing the instructions of the software program. The computer readable storage medium can be, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof.

[0025] Figure 1A road noise cancellation (RNC) system 100 is shown for a vehicle 102 having one or more vibration sensors 108. Vibration sensors are positioned throughout the vehicle 102 to monitor the vibration behavior of the vehicle's suspension, subframe, and other axle and chassis components. The RNC system 100 may be integrated with a broadband feedforward and feedback active noise control (ANC) framework or system 104 that generates anti-noise by adaptively filtering signals from the vibration sensors 108 using one or more microphones 112. The anti-noise signal may then be played through one or more speakers 124. S(z) represents the transfer function between a single speaker 124 and a single microphone 112. While Figure 1 A single vibration sensor 108, microphone 112, and speaker 124 are shown for simplicity only, but it should be noted that typical RNC systems use multiple vibration sensors 108 (e.g., 10 or more), microphones 112 (e.g., 4 to 6), and speakers 124 (e.g., 4 to 8).

[0026] The vibration sensor 108 may include, but is not limited to, an accelerometer, a dynamometer, a seismometer, a linear variable differential transformer, a strain gauge, and a load cell. For example, an accelerometer is a device whose output signal amplitude is proportional to acceleration. A wide variety of accelerometers can be used in RNC systems. These include accelerometers that are sensitive to vibrations in one, two, and three generally orthogonal directions. These multi-axis accelerometers typically have separate electrical outputs (or channels) for vibrations sensed in their X, Y, and Z directions. Therefore, single-axis and multi-axis accelerometers can be used as vibration sensors 108 to detect the magnitude and phase of acceleration, and can also be used to sense orientation, motion, and vibration.

[0027] Noise and vibration originating from the wheels 106 moving on the road surface 150 may be sensed by one or more of the vibration sensors 108 mechanically coupled to the suspension device 110 or chassis components of the vehicle 102. The vibration sensors 108 may output a noise signal X(n), which is a vibration signal representing the detected road-induced vibrations. It should be noted that multiple vibration sensors are possible, and their signals may be used individually or may be combined in various ways known to those skilled in the art. In certain embodiments, a microphone may be used in place of a vibration sensor to output a noise signal X(n), which indicates the noise generated from the interaction of the wheels 106 with the road surface 150. The noise signal X(n) may be filtered by the secondary path filter 122 using the modeled transfer characteristic S'(z), which estimates the secondary path (i.e., the transfer function between the anti-noise speaker 124 and the error microphone 112).

[0028] Road noise originating from the interaction of the wheels 106 with the road surface 150 is also mechanically and / or acoustically transmitted into the passenger compartment and received by one or more microphones 112 within the vehicle 102. The one or more microphones 112 may be located, for example, in a headrest 114 of a seat 116, such as a Figure 1 Alternatively, one or more microphones 112 may be located in the roof trim of the vehicle 102 or in some other suitable location to sense the acoustic noise field heard by occupants inside the vehicle 102. Road noise originating from the interaction of the road surface 150 with the wheels 106 is transferred to the microphone 112 according to a transfer characteristic P(z) representing the primary path (i.e., the transfer function between the actual noise source and the error microphone).

[0029] The microphone 112 may output an error signal e(n) representing the noise present in the cabin of the vehicle 102 detected by the microphone 112. In the RNC system 100, the adaptive transfer characteristic W(z) of the controllable filter 118 may be controlled by an adaptive filter controller 120, which may operate according to a known least mean square (LMS) algorithm based on the error signal e(n) and the noise signal X(n) filtered by the filter 122 using the modeled transfer characteristic S'(z). The controllable filter 118 is generally referred to as a W filter. The LMS adaptive filter controller 120 may provide a summed cross spectrum configured to update the filter coefficients of the transfer characteristic W(z) based on the error signal e(n). The process of adapting or updating W(z) that results in improved noise cancellation is called convergence. Convergence refers to the generation of a W filter that minimizes the error signal e(n), which is controlled by the step size that governs the adaptation rate for a given input signal. The step size is a scaling factor that determines how quickly the algorithm converges to minimize e(n) and is accomplished by limiting the magnitude of the change in the W filter coefficients upon each update of the controllable W filter 118 .

[0030] The anti-noise signal Y(n) may be generated by an adaptive filter formed by the controllable filter 118 and the adaptive filter controller 120 based on the identified transfer characteristic W(z) and the vibration signal or combination of vibration signals X(n). Ideally, the anti-noise signal Y(n) has a waveform such that when played through the speaker 124, anti-noise is generated near the ear of the occupant and the microphone 112 that is substantially opposite in phase and the same in magnitude as the road noise audible to the occupants of the vehicle cabin. The anti-noise from the speaker 124 may combine with the road noise in the vehicle cabin near the microphone 112, resulting in a reduction in the road noise induced sound pressure level (SPL) at this location. In certain embodiments, the RNC system 100 may receive sensor signals from other acoustic sensors (such as acoustic energy sensors, acoustic intensity sensors, or acoustic particle velocity or acceleration sensors) in the passenger cabin to generate the error signal e(n).

[0031] When the vehicle 102 is running, the processor 128 can collect and optionally process data from the vibration sensor 108 and the microphone 112 to build a database or map containing data and / or parameters to be used by the vehicle 102. The collected data can be stored locally at the storage device 130 or in the cloud for future use by the vehicle 102. Examples of data types related to the RNC system 100 that can be used to be stored locally at the storage device 130 include, but are not limited to, optimal W filters, W filter thresholds, initial W filters, W filter gain factors, leakage increments and decrements, accelerometer or microphone spectra or time-related signals, and engine SPL to torque and RPM. In one or more embodiments, the processor 128 and the storage device 130 can be integrated with one or more RNC system controllers such as the adaptive filter controller 120.

[0032] As previously described, a typical RNC system may use several vibration sensors, microphones, and speakers to sense the structurally transmitted vibration behavior of the vehicle and generate anti-noise. The vibration sensor may be a multi-axis accelerometer with multiple output channels. For example, a three-axis accelerometer typically has separate electrical outputs for vibrations sensed in its X-direction, Y-direction, and Z-direction. A typical configuration of an RNC system may have, for example, 6 error microphones, 6 speakers, and 12 acceleration signal channels from 4 three-axis accelerometers or 6 two-axis accelerometers. Therefore, the RNC system will also include multiple S'(z) filters (i.e., secondary path filters 122) and multiple W(z) filters (i.e., controllable filters 118).

[0033] Figure 1The simplified RNC system schematically depicted in FIG. 1 shows one secondary path (denoted by S(z)) between each speaker 124 and each microphone 112. As previously mentioned, an RNC system typically has multiple speakers, microphones, and vibration sensors. Thus, a 6-speaker, 6-microphone RNC system would have a total of 36 secondary paths (i.e., 6×6). Accordingly, a 6-speaker, 6-microphone RNC system may also have 36 S'(z) filters (i.e., stored secondary path filters 122) that estimate the transfer function of each secondary path. As shown in FIG. Figure 1 As shown, the RNC system will also have one W(z) filter (i.e., controllable filter 118) between each noise signal X(n) from the vibration sensor (i.e., accelerometer) 108 and each speaker 224. Thus, an RNC system with 12 accelerometer signals and 6 speakers may have 72 W(z) filters. The relationship between the number of accelerometer signals, speakers, and W(z) filters is given by Figure 2 Shown in.

[0034] Figure 2 The display is expanded to include R accelerometer signals [X 1 (n), X 2 (n),…X R (n)] and the L speaker signals [Y 1 (n), Y 2 (n),…Y L (n)] is a sample schematic diagram of relevant portions of an RNC system 200 for controlling an accelerometer signal. Thus, the RNC system 200 may include R*L controllable filters (or W filters) 218 ​​located between each of the accelerometer signals and each of the speakers. As an example, an RNC system with 12 accelerometer outputs (i.e., R=12) may employ 6 dual-axis accelerometers or 4 tri-axis accelerometers. In the same example, a vehicle with 6 speakers (i.e., L=6) for reproducing anti-noise may therefore use a total of 72 W filters. At each of the L speakers, the R W filter outputs are summed to produce an anti-noise signal Y(n) for the speaker. Each of the L speakers may include an amplifier (not shown). In one or more embodiments, the R accelerometer signals filtered by the R W filters are summed to produce an electrical anti-noise signal y(n), which is fed to an amplifier to generate an amplified anti-noise signal Y(n) that is sent to the speaker.

[0035] Figure 1The illustrated ANC system 104 may also include an engine order noise cancellation (EOC) system. As mentioned above, EOC technology uses a non-acoustic signal (such as an RPM signal representing the engine speed) as a reference in order to generate a sound that is opposite in phase to the engine noise audible inside the vehicle. Common EOC systems utilize a narrowband feed-forward ANC framework to generate anti-noise using the RPM signal to guide the generation of an engine order signal that is the same frequency as the engine order to be cancelled, and adaptively filter the engine order signal to produce the anti-noise signal. After being transmitted from the anti-noise source to the listening position or error microphone through a secondary path, the anti-noise ideally has the same amplitude but opposite phase as the combined sound generated by the engine and exhaust pipe and filtered by the primary path, which extends from the engine to the listening position and from the exhaust pipe outlet to the listening position. Therefore, at the location in the vehicle cabin where the error microphone is located (i.e., most likely at or near the listening position), the superposition of engine order noise and anti-noise will ideally become zero, so that the acoustic error signal received by the error microphone will only record sounds other than one or more engine orders (ideally cancelled) generated by the engine and exhaust.

[0036] Typically, a non-acoustic sensor (e.g., an RPM sensor) is used as a reference. The RPM sensor may be, for example, a Hall Effect sensor placed adjacent to a rotating steel disk. Other detection principles may be employed, such as an optical sensor or an inductive sensor. The signal from the RPM sensor may be used as a pilot signal for generating any number of reference engine order signals corresponding to each of the engine orders. The reference engine orders form the basis for the noise cancellation signal generated by one or more narrowband adaptive feed-forward LMS blocks forming the EOC system.

[0037] Figure 3304 is a schematic block diagram showing an example of an ANC system 304 including both an RNC system 300 and an EOC system 340. Similar to the RNC system 100, the RNC system 300 may include elements 308, 312, 318, 320, 322, and 324 that are consistent with the operation of the elements 108, 112, 118, 120, 122, and 124 discussed above, respectively. The EOC system 340 may include an RPM sensor 342 that may provide an RPM signal 344 (e.g., a square wave signal) indicating the rotation of an engine drive shaft or other rotating shaft (which indicates engine speed). In some embodiments, the RPM signal 344 may be obtained from a vehicle network bus (not shown). Because the radiated engine order is proportional to the drive shaft RPM, the RPM signal 344 represents the frequency generated by the engine and the exhaust system. Therefore, the signal from the RPM sensor 342 may be used to generate a reference engine order signal corresponding to each of the engine orders of the vehicle. Thus, the RPM signal 344 may be used in conjunction with a lookup table 346 of RPM versus engine order frequency that provides a list of engine orders radiated at each engine RPM.

[0038] Figure 4 An exemplary EOC cancellation adjustment table 400 that can be used to generate the lookup table 346 is shown. The exemplary table 400 lists the frequency (in cycles per second) of each engine order for a given RPM. In the example shown, four engine orders are shown. The LMS algorithm takes the RPM as an input and generates a sine wave for each order based on this lookup table 400. As previously described, the associated RPM for the table 400 can be the drive shaft RPM.

[0039] Reference again Figure 3, the frequency of a given engine order at the sensed RPM retrieved from the lookup table 346 can be supplied to a frequency generator 348, thereby generating a sine wave at a given frequency. This sine wave represents a noise signal X(n) indicating the engine order noise of a given engine order. Similar to the RNC system 300, this noise signal X(n) from the frequency generator 348 can be sent to an adaptive controllable filter 318 or a W filter, which provides a corresponding anti-noise signal Y(n) to a speaker 324. As shown, the various components of this narrowband EOC system 340 can be the same as the broadband RNC system 300, including an error microphone 312, an adaptive filter controller 320, and a secondary path filter 322. The anti-noise signal Y(n) broadcast by the speaker 324 generates anti-noise that is substantially out of phase with the actual engine order noise but the same in magnitude at the location of the listener's ear (which may be very close to the error microphone 312), thereby reducing the sound amplitude of the engine order. Because the engine order noise is narrowband, the error microphone signal e(n) may be filtered by the bandpass filters 350, 352 and then passed into the LMS-based adaptive filter controller 320. In one embodiment, proper operation of the LMS adaptive filter controller 320 is achieved when the noise signal X(n) output by the frequency generator 348 is bandpass filtered using the same bandpass filter parameters.

[0040] To reduce the amplitude of multiple engine orders simultaneously, the EOC system 340 may include multiple frequency generators 348 for generating a noise signal X(n) for each engine order based on the RPM signal 344. As an example, Figure 3 A second order EOC system is shown with a method for generating a unique noise signal (eg, X) for each engine order based on engine speed. 1 (n), X 2 (n), etc.). Because the frequencies of the two engine orders are different, the bandpass filters 350, 352 (labeled BPF and BPF2, respectively) have different high pass and low pass filter corner frequencies. The number of frequency generators and corresponding noise cancellation components will ultimately vary based on the number of engine orders for the particular engine of the vehicle. When the second order EOC system 340 is combined with the RNC system 300 to form the ANC system 304, the anti-noise signals Y(n) output from the three controllable filters 318 are summed and sent to the speaker 324 as the speaker signal S(n). Similarly, the error signal e(n) from the error microphone 312 can be sent to the three LMS adaptive filter controllers 320.

[0041] A dominant factor that can cause instability or reduced noise cancellation performance in an ANC system occurs when the adaptive W filter diverges during adaptation by a feedforward LMS system. When the adaptive W filter converges correctly, the sound pressure level (and the associated error signal e(n)) at the location of the error microphone is minimized. However, when one or more of these adaptive W filters diverge, noise enhancement may occur instead. Typically, the first tap in an adaptive W filter has a high amplitude, and the amplitude decreases to zero in later taps. However, if the LMS ANC system diverges, one or more W filters may not have this feature. Therefore, a system and method can be used to detect and control the divergence of the adaptive filter to maintain ANC system performance and stability. Simply put, the W filter value (i.e., the adaptive filter coefficient) can be compared with a predetermined threshold in the time domain or frequency domain. If the value of the W filter exceeds these thresholds, divergence mitigation measures can be used to prevent noise enhancement or other undesirable behavior. Divergence mitigation measures may include, for example, muting the ANC system, resetting the divergent W filter to a zero state or some other stored state, adding leakage at frequencies including the divergent frequencies, etc.

[0042] Figure 5 is a schematic block diagram of a vehicle-based ANC system 500 illustrating a number of key ANC system parameters that may be used to detect divergence of an adaptive W filter and optimize ANC system performance. For ease of explanation, Figure 5 The illustrated ANC system 500 is shown as having components and features of an RNC system such as RNC system 100. However, the ANC system 500 may include, for example, a combination of Figure 3 Thus, the ANC system 500 is an RNC and / or EOC system that features additional system components (such as a combination of Figures 1 to 3 1 and 124). Similar components may be numbered using similar conventions. For example, similar to RNC system 100, ANC system 500 may include elements 508, 510, 512, 518, 520, 522, and 524 that operate consistent with elements 108, 110, 112, 118, 120, 122, and 124 discussed above, respectively.

[0043] As shown, the ANC system 500 may also include a divergence controller 562 disposed along a path between the controllable filter 518 and the adaptive filter controller 520. The divergence controller 562 may include a processor and memory (not shown) programmed to detect the divergence of the controllable filter 518. This may include calculating parameters by analyzing samples from the adaptive filter values ​​(e.g., filter coefficients) in either or both the time domain or the frequency domain. To this end, Figure 5Fast Fourier Transform (FFT) blocks 564, 566 and Inverse Fast Fourier Transform (IFFT) block 568 are explicitly shown for transforming the signal between the time domain and the frequency domain. Figure 5 The variable names in Figures 1 to 3 The ones shown are slightly different. Uppercase variables represent signals in the frequency domain, while lowercase variables represent signals in the time domain. The letter "n" represents a sample in the time domain, while the letter "k" represents a bin in the frequency domain. Figure 5 The presence of multiple signals is further illustrated in the diagram of FIG, which shows R reference signals, L loudspeaker signals, and M error signals. The following table provides Figure 5 Detailed explanation of the various symbols and variables in .

[0044]

[0045]

[0046] Similar to Figure 1 , the noise signal x from the noise input (such as the vibration sensor 508) r [n] can be transformed and used by the secondary path filter 522 using the modeled transfer characteristic Filtering is performed (using stored estimates of the secondary path as previously described). In addition, the adaptive transfer characteristic w of the controllable filter 518 (eg, W filter) is r，l [n] can be controlled by an LMS adaptive filter controller (or simply LMS controller) 520 to provide an adaptive filter. The noise signal filtered by the secondary path filter 522 and the error signal from the microphone 512 are m [n] is the input to the LMS adaptive filter controller 520. The anti-noise signal y l [n] can be obtained by LMS controller 520 and noise signal x r [n] A controllable filter 518 is generated to perform adaptation.

[0047] Divergence controller 562 may receive the time domain filter coefficients W generated by LMS adaptive filter controller 520 for each adaptation of controllable filter 518 r，l [n] and / or frequency domain filter coefficients W r，l[k]. In addition, the divergence controller 562 may calculate one or more parameters by analyzing the filter coefficients. If a divergence of one or more controllable filters is detected, the divergence controller 562 may send a signal, such as an adjustment signal, back to the adaptive filter controller 520, which commands the adaptive filter controller to modify the properties of at least one controllable filter 518 that has diverged. For example, in an RNC system or an EOC system, a response to detecting a divergence of a controllable W filter 518 may be that the divergence controller 562 replaces the diverged W filter value with, for example, an adjusted W filter that has been previously stored. Other responses of the divergence controller 562 to detecting a W filter divergence may include replacing the controllable filter 518 with a filter consisting of zeros, which effectively resets the controllable filter. Other divergence mitigation measures performed by the divergence controller 562 may include adding leakage at frequencies including the divergence frequency, attenuating some or all of the W filter coefficients, or reducing the step size to reduce the risk of future divergence events.

[0048] The divergence controller 562 may be a dedicated controller for the controllable W filter for detecting divergence, or may be integrated with another controller or processor in the ANC system, such as the LMS controller 520. Alternatively, the divergence controller 562 may be integrated into another controller or processor within the vehicle 102 that is separate from other components in the ANC system 500.

[0049] although Figure 5 Specifically shown is an ANC system that processes in both the time domain and the frequency domain, but it is possible to implement an ANC system that processes only in the time domain. In this case, the secondary path estimate is stored in the time domain and the LMS update also occurs in the time domain. In one embodiment, the divergence detection performed by the divergence controller 562 may also occur in the time domain. In another embodiment, the FFT of the time domain W filter may allow for divergence detection by calculating parameters from the frequency domain W filter.

[0050] Figure 6 is a flow chart depicting a method 600 for mitigating the effects of a divergent or mis-adapted controllable W filter in the ANC system 500. The various steps of the disclosed method may be performed by the divergence controller 562 alone or in conjunction with other components of the ANC system.

[0051] At step 610, the divergence controller 562 may receive the time domain (ie, w r，l [n]) and / or frequency domain (i.e., W r，l[k]) indicates the input of one or more controllable filters 518. To this end, the divergence controller 562 can receive a set of samples of the time domain or frequency domain filter coefficients output from the adaptive filter controller 520. In one embodiment, the controllable W filter can be composed of 128 taps in the time domain. In alternative embodiments, more or less filter taps are possible. The filter value or coefficient can be received from the LMS adaptive filter controller 520 and can represent the current adaptation of the controllable filter 518. As described above, each controllable filter 518 is continuously adapted by the adaptive filter controller 520, and its rate of change is limited by the step size. The update rate of the controllable filter 518 can be controlled by the incoming X r [k, n]) and E m The divergence controller 562 may receive these updated W filter coefficients for each controllable filter.

[0052] At step 620, analysis of the W filter data may be performed, and one or more parameters may be calculated in the time domain or frequency domain. There are several methods to detect divergence or misadaptation in the time domain version of the controllable W filter based on analysis of the filter coefficients. In one embodiment, the parameter calculated by the divergence controller 562 may be the sum of the absolute values ​​of the taps in the entire controllable W filter. In another embodiment, the parameter calculated by the divergence controller 562 may be the sum of the absolute values ​​of the taps in the rear portion of the controllable W filter (such as the rear half or the last quarter of the controllable filter coefficients). In yet another embodiment, the parameter may be the maximum value of the individual tap values ​​in at least a portion of the controllable filter (such as the rear half (or the last quarter)) to determine whether any tap value exceeds a predetermined amplitude. Controllable filter property parameters may also be calculated in the frequency domain. Parameters calculated in the frequency domain may include, for example, phase deviations within a frequency range. In another embodiment, the parameter calculated by the divergence controller 562 may be the sum of all or part of the W filter coefficients. In yet another embodiment, the parameter may be a maximum value of the W filter coefficients in at least a portion of the frequency range of the controllable W filter. In various embodiments, these sums or maximum values ​​may be calculated using real, imaginary or magnitude values ​​of the W filter coefficients.

[0053] Step 620 may also include storing one or more parameters and / or current W filter values ​​for performing future W filter analysis. In one embodiment, one or more parameters or W filter data from the W filter immediately preceding the current W filter may be stored. In another embodiment, a statistical analysis (e.g., to determine a threshold) may be performed on the parameters obtained from multiple previous W filters. For example, a short-term or long-term average of the parameters obtained from multiple previous W filters may be calculated and the average may be stored as its own parameter for use in step 630, whether as a threshold or to obtain a difference with the current W filter for comparison with a threshold. In some of these embodiments, a predetermined gain margin may be added to the average (or other statistical value) calculated from multiple previous W filters to form a threshold. This may include adding a gain margin of 20%, 50%, or 100% to the average or other statistical value. Thus, the average value from multiple previous W filters may be multiplied by a gain factor (e.g., 120%, 150%, 200%, etc.) to obtain a threshold. In other embodiments, other gain factors are possible. In addition, one or more controllable W filters may be stored for future use to mitigate W filter divergence.

[0054] At step 630, the parameters calculated from the current controllable W filter may be directly compared to the corresponding thresholds. If the parameters from the current W filter exceed the threshold, the divergence controller 562 may conclude that divergence or misadaptation has been detected. If the parameters from the current W filter do not exceed the threshold, the noise controller 562 may conclude that divergence or misadaptation has not been detected. For example, the divergence controller 562 may calculate the average of the absolute values ​​in the last tenth of the highest magnitude frequency domain W filter coefficient values ​​or the time domain filter taps of the W filter, and compare the peak amplitude or sum to the corresponding threshold to determine whether a divergence or misadaptation event has occurred. As another example, divergence may be detected if the phase difference between the beginning and the end of the frequency range exceeds a threshold.

[0055] Alternatively, the parameters calculated from the current W filter may be compared to statistics (e.g., averages) of the same parameters from one or more previous W filters, as previously described. The difference between the parameters of the current W filter and the statistics may then be compared to a threshold, which may be referred to as a W threshold. If the difference exceeds the threshold, the divergence controller 562 may conclude that divergence or misadaptation has been detected. If the difference does not exceed the threshold, the divergence controller 562 may conclude that divergence has not occurred. For example, in one embodiment, the divergence controller 562 may calculate the average of the absolute values ​​in the last sixth of the time domain filter taps of the W filter and compare it to the average of the previous W filter of the absolute values ​​in the last tenth of the time domain filter taps of the W filter, noting that any difference exceeding a predetermined threshold may indicate divergence of the W filter.

[0056] In one or more embodiments, the threshold value may be a predetermined static threshold value that is set and programmed by a trained engineer during the tuning of the ANC system and its corresponding algorithm. In an alternative embodiment, the threshold value may be a dynamic threshold value calculated based on a statistical analysis of parameters obtained in one or more previous W filters (as discussed above with respect to step 620). For example, the threshold value may be a short-term or long-term average of the parameters obtained from multiple previous W filters. In addition, as previously discussed, the average value may be enhanced by a gain factor to establish a dynamic threshold value. In yet another embodiment, the threshold value may simply be the value of the parameter from the previous W filter, which value may also be multiplied by a gain factor.

[0057] Referring to step 640, when the threshold has been exceeded (which indicates divergence of the controllable filter), the method may proceed to step 650. At step 650, mitigation measures may be applied to the diverging controllable W filter to minimize the noise enhancement or ANC effect reduction in the cabin caused by the W filter divergence. However, when W filter divergence is not detected, the method may skip any mitigation measures and return to step 610 so that the process may be repeated with new W filter coefficients corresponding to the next filter adaptation.

[0058] At step 650, mitigation measures may be applied to either or both of the time domain or frequency domain W filters that have diverged or misadapted. In general, this may involve modifying the properties of at least one controllable filter 518 in which divergence has been detected. Such properties may be modified based in part on or in response to an adjustment signal sent from the divergence controller 562 to the adaptive filter controller 520. In certain embodiments, countermeasures may be applied to the entire W filter or only to specific frequencies of the frequency domain W filter. Mitigation methods that may be applied to the entire controllable W filter (in the time domain or frequency domain) may include resetting the filter coefficients of one or more W filters to zero to allow them to be re-adapted, or setting the filter coefficients to a set of filter coefficient values ​​stored in the memory of the ANC system. The set of filter coefficient values ​​stored in the memory may include W filters from a known good state (such as a W filter that has been adjusted by a trained engineer), or obtained from a controllable filter before the divergence is detected. For example, the controllable filter may be reset using the filter coefficients of, for example, 10 seconds or 1 minute before the divergence. Alternatively, the controllable W filter may be reset to an initial condition, such as when the ANC system 500 is powered on. Another mitigation technique may be to simply deactivate or mute the ANC system when divergence has been detected. In one embodiment, when divergence has been detected, only the W filter that has diverged may be deactivated or set to zero and not allowed to adapt. In one embodiment, when divergence has been detected, the amplitude of all filter taps or the magnitude of all frequency domain filter coefficients may be reduced. The properties of the controllable W filter may be modified directly, such as by setting the filter coefficients to specific values. Alternatively, the properties of the controllable W filter 518 may be modified indirectly. For example, when divergence has been detected, the leakage values ​​at all frequencies may be increased by the adaptive filter controller 520 in response to an adjustment signal from the divergence controller 562.

[0059] Countermeasures applied only to frequency domain methods may include attenuating the W filter coefficients at or near the divergence frequency and adding or increasing the leakage value at or near the divergence frequency. r [n] and e m [n] or their frequency domain counterparts, the divergence controller 562 may adaptively notch the unstable divergence frequencies identified in step 630. This may prevent the adaptive filter controller 520 from increasing the magnitude of the W filter in the problematic frequency range in future operations of the ANC system 500. This may optionally be accompanied by a reset of the W filter as outlined above, or the use of leakage at these unstable divergence frequencies or all frequencies.

[0060] As previously mentioned, in one or more additional embodiments, when divergence has been detected (such as when the highest magnitude W filter coefficient exceeds a predetermined threshold), the leakage value may be increased at the LMS adaptive filter controller 520. Increasing the leakage value of the adaptive filter controller 520 may reduce the magnitude of the controllable W filter 518. As long as the highest magnitude W filter coefficient still exceeds the predetermined threshold, the leakage value may be increased by Figure 6 The processing flow shown continuously increases this leakage value by a predetermined amount at each iteration. Once the highest magnitude W filter coefficient no longer exceeds the predetermined threshold, the leakage value can be increased by a predetermined amount as long as the highest magnitude W filter coefficient no longer exceeds the predetermined threshold. Figure 6 The illustrated process flow reduces the leakage value by a predetermined amount during subsequent iterations. Reducing the leakage value of the adaptive filter controller 520 may increase the magnitude of the controllable W filter 518. In this way, the leakage value of the adaptive filter controller may be continuously adjusted up and down based on the magnitude of the filter coefficient relative to a threshold.

[0061] In one embodiment, when the highest magnitude W filter coefficient of any of the W filters exceeds a predetermined threshold, leakage is increased for all W filters in the ANC system 500. In another embodiment, when the highest magnitude W filter coefficient of any of the W filters associated with a particular speaker exceeds a predetermined threshold, leakage is increased on all W filters for that speaker. In response to receiving an adjustment signal from the divergence controller 562, the LMS controller 520 may be commanded to increase or decrease the leakage value. As previously described, adjusting the leakage value for one or more of the controllable filters 518 may directly affect the magnitude of the W filter coefficients. For example, increasing leakage may generally decrease the magnitude of the filter coefficients, while decreasing leakage may generally increase the magnitude of the filter coefficients.

[0062] As previously described, for each combination of speaker 512 and noise input (e.g., each engine order or vibration sensor), there is one controllable W filter. Thus, a 12 accelerometer, 6 speaker RNC system would have 72 W filters (i.e., 12×6=72), and a 5 engine order, 6 speaker EOC system would have 30 W filters (i.e., 5×6=30). This may be performed after calculating each new set of W filters or at a lower frequency. Figure 6 The method 600 is shown so as to reduce the required computing power, thereby saving CPU cycles.

[0063] Figure 7Depicts an exemplary analysis of frequency domain threshold comparisons. The ANC system 500 may store a set of threshold limits (i.e., W thresholds) for each controllable filter. Under normal operating conditions, all controllable W filter points are less than the W threshold. Under divergent or misadaptive operating conditions, one or more coefficients of the W filter exceed the W threshold. The divergence controller 562 may detect and indicate which W filter or which frequency bins of the W filter have exceeded the W threshold so that the adaptive filter controller 520 or the divergence controller 562 may apply countermeasures.

[0064] although Figure 1 , Figure 3 and Figure 5 LMS-based adaptive filter controllers 120, 320, and 520 are shown, respectively, but other methods and devices for adapting or generating optimal controllable W filters 118, 318, and 518 are possible. For example, in one or more embodiments, a neural network can be used instead of an LMS adaptive filter controller to generate and optimize the W filter. In other embodiments, machine learning or artificial intelligence can be used instead of an LMS adaptive filter controller to generate an optimal W filter.

[0065] In the foregoing description, the inventive subject matter has been described with reference to specific exemplary embodiments. However, various modifications and changes may be made without departing from the scope of the inventive subject matter as set forth in the claims. The present description and drawings are illustrative rather than restrictive, and modifications are intended to be included within the scope of the inventive subject matter. Therefore, the scope of the inventive subject matter should be determined by the claims and their legal equivalents, rather than by the examples alone.

[0066] For example, the steps recited in any method or process claim may be performed in any order and are not limited to the specific order presented in the claim. The equations may be implemented with filters to minimize the effects of signal noise. Additionally, the components and / or elements recited in any device claim may be assembled or otherwise operably configured in a variety of arrangements and are therefore not limited to the specific configuration recited in the claim.

[0067] It will be appreciated by those skilled in the art that functionally equivalent processing steps may be performed in the time domain or the frequency domain. Figure 3 ) are not explicitly stated for each signal processing block, but signal processing can occur in the time domain, frequency domain, or a combination thereof. In addition, although various processing steps are explained in typical terms of digital signal processing, equivalent steps can be performed using analog signal processing without departing from the scope of the present disclosure.

[0068] The benefits, advantages and solutions to problems have been described above with respect to specific embodiments. However, any benefit, advantage, solution to a problem, or any element that may make any particular benefit, advantage or solution to a problem appear or become more obvious, should not be construed as a critical, required or essential feature or component of any or all claims.

[0069] The terms "comprise," "comprises," "comprising," "having," "including," "includes," or any variations thereof, are intended to refer to a non-exclusive inclusion, such that a process, method, article, composition, or apparatus that includes a list of elements includes not only those elements that are enumerated, but may include other elements not expressly listed or inherent to such process, method, article, composition, or apparatus. Other combinations and / or modifications of the above-described structures, arrangements, applications, proportions, elements, materials, or components for practicing the present subject matter, except those not specifically enumerated, may be varied or otherwise specially adapted to a particular environment, manufacturing specification, design parameters, or other operating requirements without departing from the general principles of the present subject matter.

Claims

1. A method for controlling stability in an active noise cancellation (ANC) system, the method comprising: receiving filter coefficients corresponding to at least one controllable filter from an adaptive filter controller; calculating parameters based on analysis of at least a portion of the filter coefficients; detecting a divergence of the at least one controllable filter based on a comparison of the parameter with a threshold value, wherein the threshold value is a dynamic threshold value calculated based on a statistical analysis of the parameter calculated from filter coefficients in one or more previous adaptations of the at least one controllable filter; and A property of the at least one controllable filter that has diverged is modified.

2. The method of claim 1, wherein the controllable filter comprises a plurality of coefficients, the parameter being the sum of absolute values ​​of at least a portion of the coefficients in the at least one controllable filter.

3. The method of claim 1, wherein the controllable filter comprises a plurality of coefficients, the parameter being a maximum value of at least a portion of the coefficients in the at least one controllable filter.

4. The method of claim 1 , wherein detecting a divergence of the at least one controllable filter based on a comparison of the parameter with a threshold comprises: When the parameter exceeds the threshold, a divergence of the at least one controllable filter is detected.

5. The method of claim 1, wherein the threshold is an average of parameters obtained from a plurality of previous adaptations of the at least one controllable filter multiplied by a gain factor.

6. The method of claim 1 , wherein detecting a divergence of the at least one controllable filter based on a comparison of the parameter with a threshold comprises: comparing the parameter from a current adaptation of the at least one controllable filter with an average of the same parameter from one or more previous adaptations of the at least one controllable filter; as well as Divergence of the at least one controllable filter is detected when a difference between the parameter from the current adaptation of the at least one controllable filter and the average value from the one or more previous adaptations of the at least one controllable filter exceeds a threshold.

7. The method of claim 1 , wherein modifying the property of the at least one controllable filter that has diverged comprises: At least one of the ANC system and the at least one controllable filter that has diverged is deactivated.

8. The method of claim 1 , wherein modifying the property of the at least one controllable filter that has diverged comprises: The filter coefficients of the at least one controllable filter are reset to zero and the at least one controllable filter is allowed to readapt.

9. The method of claim 1 , wherein modifying the property of the at least one controllable filter that has diverged comprises: The filter coefficients of the at least one controllable filter are reset to a set of filter coefficient values ​​stored in a memory of the ANC system.

10. The method of claim 1, wherein modifying the property of the at least one controllable filter that has diverged comprises: In response to detecting divergence of the at least one controllable filter, a leakage value of the adaptive filter controller is increased.

11. The method of claim 10, wherein the leakage value of the adaptive filter controller is increased at a divergence frequency of the at least one controllable filter.

12. The method of claim 10, further comprising: The leakage value of the adaptive filter controller is reduced when a highest magnitude filter coefficient of the at least one controllable filter is below a predetermined threshold.

13. An active noise cancellation (ANC) system, comprising: at least one controllable filter configured to generate an anti-noise signal based on an adaptive transfer characteristic and a noise signal received from a sensor, the adaptive transfer characteristic of the at least one controllable filter being characterized by a set of filter coefficients; an adaptive filter controller comprising a processor and a memory, the adaptive filter controller being programmed to adapt the set of filter coefficients based on the noise signal and an error signal received from a microphone located in a cabin of the vehicle; as well as a divergence controller, the divergence controller being in communication with at least the adaptive filter controller, the divergence controller comprising a processor and a memory, the processor and the memory being programmed to: receiving the set of filter coefficients corresponding to a current adaptation of the adaptive transfer characteristic of the at least one controllable filter; calculating a parameter based on an analysis of at least a portion of the set of filter coefficients; and Divergence of the at least one controllable filter is detected based on a comparison of the parameter with a threshold value, wherein the threshold value is a dynamic threshold value calculated based on a statistical analysis of the parameter calculated from filter coefficients in one or more previous adaptations of the at least one controllable filter.

14. The ANC system of claim 13, wherein the threshold is a predetermined static threshold programmed for the ANC system.

15. An ANC system as claimed in claim 13, wherein the divergence controller is programmed to detect a divergence of the at least one controllable filter when a difference between the parameter calculated based on the current adaptation of the at least one controllable filter and an average of the same parameter from one or more previous adaptations of the at least one controllable filter exceeds the threshold.

16. The ANC system of claim 13, wherein the divergence controller is further programmed to increase a leakage value of the adaptive filter controller in response to detecting divergence of the at least one controllable filter.

17. A computer program product embodied in a non-transitory computer readable medium, the computer program product being programmed for active noise cancellation (ANC), the computer program product comprising instructions for: receiving from an adaptive filter controller a set of filter coefficients corresponding to a current adaptation of at least one controllable filter; calculating parameters based on analysis of at least a portion of the filter coefficients; detecting a divergence of the at least one controllable filter based on a comparison of the parameter with a threshold value, wherein the threshold value is a dynamic threshold value calculated based on a statistical analysis of the parameter calculated from filter coefficients in one or more previous adaptations of the at least one controllable filter; as well as In response to detecting a divergence of the at least one controllable filter, an adaptive transfer characteristic of the at least one controllable filter is modified during the current adaptation.

18. The computer program product of claim 17, wherein the instructions for detecting divergence of the at least one controllable filter comprise: detecting a divergence frequency of the at least one controllable filter in the time domain; and The instructions for modifying the adaptive transfer characteristic include: in the time domain, resetting the divergence frequency of the at least one controllable filter to zero, attenuating the filter coefficients at the divergence frequency, or increasing a leakage value of the adaptive filter controller at the divergence frequency.

19. The computer program product of claim 17, wherein the instructions for detecting divergence of the at least one controllable filter comprise: detecting a divergence frequency of the at least one controllable filter in the frequency domain; and Wherein the instructions for modifying the adaptive transfer characteristic include notching the divergent frequencies in the frequency domain using an error signal received from a microphone and previously adapted filter coefficients from the at least one controllable filter stored in a memory.

Images