Encyclopedia of Continuum Mechanics

Living Edition
| Editors: Holm Altenbach, Andreas Öchsner

Active Control of Sound, Applications of

  • Malte MisolEmail author
  • Stephan Algermissen
  • Thomas Haase
Living reference work entry
DOI: https://doi.org/10.1007/978-3-662-53605-6_282-1



The active control of sound is realized with active noise reduction systems. These systems are made up of actuators, sensors and digital signal processing. Unlike classical passive sound control the active control of sound is especially effective at low frequencies. Active noise reduction systems are able to reduce unwanted sound with a minimum of added mass and volume which makes them useful for lightweight applications for example in the automotive or aerospace industry.





Active noise control


Active noise vibration control


Active structural acoustic control


Active vibration control


Carbon fiber reinforced plastics


Counter-rotating open rotor




Electromagnetic interference


Finite impulse response filter


Frequency response function


Infinite impulse response filter


Loudspeaker array


Printed circuit board


Polyvinylidene fluoride


Radiation modal expansion


Structural health monitoring


Scanning laser Doppler vibrometer


Sound pressure level


Signal processing unit


Turbulent boundary layer


Transmission loss


Due to the fact that the methods to achieve active control of sound are highly application dependent, it is reasonable to describe the most prominent approaches using dedicated applications in mind.

Sound Control

Sound control is an important issue for many lightweight structures because they are prone to structural vibration and sound radiation. Especially at frequencies below 500 Hz, passive sound control is generally not compatible with lightweight construction. This is due to the fact that low-frequency sound transmission loss (TL) is determined by mass. According to the so-called mass law, an increase in TL by 6 dB (factor of four) requires a doubling of mass per unit area. If certain frequencies are to be addressed, tuned vibration absorbers or Helmholtz resonators might be an option. These systems are usually not able to track frequency changes, and their mass and volume also scale with the desired amount of disturbance rejection. Alternatively, active sound control methods can be applied to achieve high performance with a minimum of added mass and volume. These methods use actuators, sensors, and control to reduce the sound emission of structural parts. Generally, active sound control can be subdivided into active noise control (ANC), active noise vibration control (ANVC), and active structural acoustic control (ASAC). In addition, also active vibration control (AVC) is often used to attenuate unwanted noise even if this principle is not especially tailored to acoustics. AVC is limited to cases, where the occurring vibrations directly induce the unwanted noise, and is most effective if the correlation of the vibration patterns and the radiated sound is high. In this case, elaborate models of the sound radiation process can be avoided leading to simpler and sometimes more robust active systems. As AVC, ASAC systems are built from structures with embedded actuators and sensors. In contrast to AVC, the objective function however is based on acoustic measures. In the case of ANVC and ANC, the sensor information is usually obtained from sound pressure-based devices like microphones, and the latter principle uses loudspeakers as actuators for the active control. Microphones and loudspeakers can be positioned independently of the walls and sound-emitting parts to optimize the performance of the active system.

Whereas ANC and ANVC generally lead to local sound control at the error sensor positions, ASAC aims at a reduction of the radiated sound power which is a global metric. This difference is caused by and has implications on the error sensor scheme. Generally, a prerequisite for global control is the observability of the global system behavior which can either be achieved by filtering local sensor data through an observer (a global system model) or by a global physical sensor scheme. A performance metric for ASAC is obtained from the so-called radiation modes of a structure. These modes couple independently into the sound power, and a reduction of their modal amplitude is directly linked to a sound power reduction. This is a key difference to the suppression of structural modes, which may even lead to an increase in radiated sound power. Since radiation modes are frequency dependent, they must be modeled by a dynamic system. The radiation modal expansion (RME) technique provides a trade-off between accuracy and efficiency. Special filtering of sensor signals facilitates the incorporation of psychoacoustic metrics to achieve higher sound quality or a comfortable acoustic environment. Psychoacoustic metrics are, however, not easily included in real-time control applications. A typical simple example is the postprocessing of microphone data with an A-weighting filter to account for human sound perception. Regarding the actuator schemes, ANC uses loudspeakers, and ANVC and ASAC directly actuate the radiating structure. For ASAC this is a requirement, but for ANVC the actuated and the noise radiating structural parts must not necessarily coincide. In the latter case, the structure acts as a loudspeaker, and ANVC is similar to ANC. Usually an ANVC system acts on the vibration of the radiating structural parts aiming at a reduction of sound pressure or sound power.

Transmission Loss

Specifies in general the accumulated decrease in intensity of a sound wave as it propagates through a certain type of structure. Measures of TL are expressed in decibel and directly quantify the success in sound attenuation through windows, walls, or other separating constructions.

Free-Field Radiation

Describes the situation that a vibrating structure emits sound into a region with no or insignificant reflections. The sound can ideally radiate indefinitely without hitting any obstacles. This can approximately be the case in large rooms with sufficient damping. Sound radiated under free-field condition almost keeps its energy while traversing the media if the damping of the media is neglected. A sound field can be roughly divided into a near field and a far field. The former is in the region of the emitting structure and exhibits interference effects, while the latter is present in the outer region with almost plane waves. Measuring the sound energy on spherical surfaces in the far field, so-called radiation modes can be determined in the case of free-field radiation, which can be used to determine the radiated sound energy of the given vibration pattern of the radiating structure. This can significantly simplify the construction of ASAC systems.

Cabin Noise

If a structure emits sound into a cabin, usually the free-field condition does not hold, because significant parts of the radiated sound waves are reflected at the boundary of the cavity and contribute to the resulting sound field. In addition to the structural resonance frequencies, the cavity itself also has acoustic resonances, which must be considered as well. Adequate modeling needs elaborate models based on volumetric finite element models of the cavity or surface-based boundary element models. In general the reflections of the boundary are described by complex impedances leading to inherent complex system models with real and imaginary parts.

Active Noise Control (ANC)

If the noise sources are distributed or the sound-emitting structural parts are otherwise inaccessible to active measures, ANC is the principle of choice to attenuate the noise or change its patterns as desired. It is most effective if the controlled region is local to the microphones in use or if the sound path is restricted. Successful applications contain the attenuation of machine-induced noise passing through industrial chimneys to the environment or the reduction of sound waves travelling through air conditioner ducts. More generally, if good reference signals are available, feedforward-based ANC systems can be very effective. Noise-cancelling headphones are also based on ANC and provide great possibilities to filter certain noise patterns from the human ear perception due to the fact that the target region is well determined and sound waves travel mainly through the auditory canal.

Placement of Actuators and Sensors

The efficiency of an active system is directly connected to the controllability and observability of the used actuators and sensors. Besides the properties of the devices itself, their location naturally has major influence. Poorly chosen positions might fully prohibit the detection or stimulation of a certain structural vibration mode. Therefore carefully chosen positions are an important step in the design of a reasonable adaptive structure.

Active vibration control (AVC) systems are widely used in order to achieve global vibration reduction of thin plates and beams. In order to enable a sound power reduction with an AVC system, the uncontrolled vibration patterns have to be inefficient sound radiators. In order to design an AVC system for a fuselage section of a Dornier 728 experimental aircraft, a feedforward control system with structural sensors and actuators is optimized. The AVC system is designed for a reduction of counter-rotating open rotor (CROR) noise in the cabin interior. Figure 1 shows an interior view on the Dornier 728 active control system and the loudspeaker array used for the CROR sound field generation. In order to reduce the interior noise induced by the synthesized CROR pressure field, two sections are equipped with sensors and actuators. The actuators are placed in the areas of the highest velocities of the overlayed vibration patterns of the CROR frequencies. The velocities of the vibration patterns are measured with a scanning laser Doppler vibrometer (SLDV). The sensors of the active control system are placed by a optimization tool, and its working principle is shown in Fig. 2.
Fig. 1

Active control system on the fuselage of the Dornier 728 (left) and loudspeaker array (right)

Fig. 2

Optimization tool for the optimized sensor placement

The global system model is also measured with a SLDV. Therefore, each of the 12 secondary actuators (Visaton EX 60s) is excited sequentially with a broadband white noise signal, and a grid of 505 sensor points are scanned for each actuator. The result is a frequency response function (FRF) matrix which describes the transfer paths of the 12 actuators to the possible 505 sensor locations. Yet, the placement of the sensors can be optimized by using a genetic algorithm with the radiated sound power of the control section as its cost function. In order to reduce the sound power, a local AVC system with 24 of the possible 505 sensor locations is used. After the calculation of the local AVC system, the impact of the local system is calculated for the global system model. Therefore, the vibration of the 505 sensor locations is postprocessed with the radiation resistance matrix (Fahy and Gardonio, 2007) which results in an estimate of the radiated sound power. The optimized sensor placements are realized on the Dornier 728, and an active feedforward controller is used to reduce the vibrations at the error sensors. The implementation is based on Elliott (2001). In summary, a collocated system with 12 sensors and 12 actuators (Col 12 × 12), an optimized system with 12 sensors and 12 actuators (Opt 12 × 12), an empirical system with 25 sensors and 12 actuators (Emp 12 × 25), a regular system with 24 sensors and 12 actuators (Reg 12 × 24), and an optimized system with 24 sensors and 12 actuators (Opt 12 × 24) are investigated. The achieved sound power reductions are presented in Fig. 3.
Fig. 3

Sound power reduction of the different feedforward control systems

It is clearly shown that the optimized system with 24 error sensors shows the largest reduction averaged over the 5 addressed CROR frequencies. It is also shown that no amplification is present for the optimized system because only negative values indicate a sound power reduction in this study. Therefore, it can be stated that the AVC feedforward control systems optimized with the optimization tool indeed reduce the radiated sound power.

Active Windshield of a Passenger Car

In normal driving operation, the windshield of a passenger car is forced to structural vibration due to different disturbance sources. One important disturbance source is the engine which is able to excite structural resonances of the windshield. The importance of the windshield’s second and third structural resonance for the interior noise level was identified during road trials and roller test bench experiments. For window application, passive damping treatment is limited to the viscous layer of the laminated glass panel. In order to achieve higher damping and sound pressure level reduction in the compartment, active control is applied. The active control of structural vibration leads to a sound pressure level reduction in the interior. Here, an active vibration control (AVC) approach is implemented which aims at reducing the mean structural vibration level of the windshield in a frequency band containing the second and third eigenfrequency (EF). This method is especially useful in a reactive environment where the control points (here: the passenger’s ears) are located in the acoustic near field. The control of active sound power by means of active structural acoustic control (ASAC) is not reasonable for the present application. To achieve global vibration reduction, the windshield is equipped with actuators and sensors which are connected to a digital signal processing unit. Figure 4 gives an overview of the active windshield system. An electrodynamic force exciter (shaker) is connected to the roof brace near the A-pillar to reproduce the space-frame structural vibration occurring in normal driving operation. Road trial measurement data of the windshield vibration is used to determine the required shaker force amplitude (F1). Three pairs of piezoelectric d31-patch transducers (A1 to A6) are used to counteract the disturbance excitation of the shaker. Real-time sensor data is captured from six accelerometers (S1 to S6). The positions of actuators and sensors are indicated in Fig. 4. To avoid line-of-sight obstruction, the accelerometers have to be relocated or replaced by strain-based sensors for the final implementation. The information required for global vibration control is obtained from a least-squares fit of the identified discrete-time state-space model to the frequency response functions measured from all actuators to 101 SLDV points. More details on the calculation of the augmented state-space model and the design of the feedback controller can be found in Misol et al. (2013). A microphone is used for the evaluation of control performance (not used for real-time control) at different locations in the passenger compartment. The reductions of structural vibration and interior sound pressure level (SPL) due to active control are shown in Figs. 5 and 6. Figure 5 contrasts the contour lines of the measured normal surface velocity amplitudes of the windshield at the second and third eigenfrequency in the uncontrolled and controlled case. The positions of the accelerometers S1 to S4 are indicated by black stars. The x- and y-coordinates are relative to the coordinate system defined in Fig. 4. The global feedback control minimizes the overall vibration level averaged over the 101 SLDV points. Therefore, the minimum amplitudes must not coincide with the accelerometer positions. The mean vibration level reductions are 7.5 dB at the second and 4.4 dB at the third eigenfrequency. The limited control authority of the piezo actuators is considered as a major constraint on control performance. It can therefore be expected that the application of more powerful actuators would permit higher vibration and sound pressure level reductions.
Fig. 4

Block diagram of the active windshield system

Fig. 5

Normalized measured vibration amplitudes at second (left) and third (right) eigenfrequency without (top) and with (bottom) active control

Fig. 6

Sound pressure level reduction due to active control measured at different locations in the passenger compartment

The effect of active global vibration control on the interior SPL can be seen in Fig. 6. The SPL is monitored at six positions in the front part of the cabin on the driver and co-driver side. A maximum reduction of 16 dB was achieved at the third EF at the driver side. It can be concluded that the active system is able to reduce the interior SPL induced by the windshield. Regarding the acoustic performance in real driving operation, it must be noted that other external and internal disturbance sources will contribute to the interior SPL which must also be controlled by passive or active means. Nevertheless, the windshield is a critical part because it has a large vibrating surface in the vicinity of the (co-)driver and the applicability of passive damping or absorption is limited.

Active Fuselage Panels

Turbulent boundary layer (TBL) induced noise is one of the dominant noise sources in high subsonic aircrafts. The fuselage is excited by the TBL, and noise is transmitted into the cabin. In order to enhance the passenger comfort, the fuselage is equipped with an ASAC system that increases the transmission loss of the structure using the radiated sound energy as objective function. For the analysis of an active system under TBL excitation, a panel made of carbon fiber-reinforced plastics (CFRP) is mounted in an aeroacoustic wind tunnel (see Fig. 7). Flow speeds up to Mach 0.16 excite the panel and enable experiments under realistic disturbances. In the experiments robust H control in combination with an extended plant is applied. By including additional performance outputs (Fig. 9), it is possible to gain an enhanced control of the noise transmission. The active panel is made of CFRP and has the size of 500 × 800 × 1.3 mm3. Four L-stringers and an aluminum frame for fast mounting are bonded to the panel. To generate a TBL excitation, the panel is mounted in a closed test section of the aeroacoustic wind tunnel. While wind speeds of up to 60 m/s generate a TBL with 41 mm thickness, the panel is excited and radiates sound into the anechoic chamber of the wind tunnel. For control actuating and sensing, the panel is equipped with five DuraAct P-876.A15 piezo patch actuators and ten PCB 352A24 accelerometers. The actuator positions were calculated with an optimization tool in order to achieve maximum noise reduction. The sensor positions are chosen for good observability of all mode shapes in the control bandwidth from 100 to 500 Hz.
Fig. 7

Aeroacoustic wind tunnel with closed test section

For model-based control, an accurate description of the controlled plant must be available. The controlled plant includes the transfer functions from all actuators to all sensors. Since models from finite element calculations lack precise parameters, the model is generated by system identification of the real panel. Its result is a discrete state-space model that includes a description of the relevant dynamic properties of the structure in the chosen control bandwidth. Furthermore, the model is extended by additional structural velocity outputs. A grid of 13 × 20 scan points is applied to the panel; see Fig. 8. For the system identification of the extended plant, a SLDV measures the point velocities while the panel is excited by one actuator at a time. Finally, a least-squares fit of the measured frequency response functions (FRF) to the controlled plant gives the extended plant G with actuator inputs u and sensor y and structural velocity p outputs (Fig. 9). Since the dynamic of the extended plant is known, only the accelerometer signals have to be measured in real time during controller operation. The velocities of the scan points are estimated by the controller using an included observer. These estimated velocities can be used in ASAC as well as in AVC.
Fig. 8

Active panel with actuators (A1–A5), sensors (S1–S10) and scan points (white)

Fig. 9

Closed control loop

Due to the special experimental setup with a closed test section, additional wind noise was present in the wind tunnel. As a consequence, acoustic measurements with a sound intensity probe in front of the panel showed poor results. Therefore, the radiated sound power of the panel is estimated with the well-known approach of the radiation resistance matrix (Fahy and Gardonio, 2007). The panel is subdivided into elementary radiators, and the normal velocity of the center point is needed for the calculation of the sound radiation. For this panel, the grid of 260 scan points is used to determine these radiators. The procedure has been validated at the experimental setup using shaker excitation and acoustic measurement equipment in the inactive wind tunnel.

The TBL-induced disturbances are modeled as process noise d entering the closed loop between controller R and plant G. The control objective is the reduction of the influence of the disturbances onto the velocity outputs p to achieve global control of the vibrations of the entire panel (Fig. 9). The final H controller is synthesized as discrete state-space model in Matlab/SIMULINK with a sampling time of 1 kHz. Afterward, it is implemented in a dSPACE rapid prototyping system.

The experiments in the wind tunnel are conducted at a wind speed of 55 m/s (=Mach 0.16). During operation of the tunnel, the normal velocities of the 260 scan points are measured with the LSV for the open and closed loop case. The calculation of the radiated sound power bases on these measurements and uses the frequency-dependent 260 × 260 radiation resistance matrix mentioned above. Figure 10 shows the results of the open and closed loop experiments. The velocity spectra for all scan points are summed up to generate an overview of the entire structure. At certain eigenfrequencies, a reduction of up to 11.8 dB could be realized. Otherwise, it is obvious that the controller has no effect in the frequency range below 110 Hz. The excitation in this range is a kind of low-frequency rumbling that has its source in the wind tunnel itself and is not caused by the TBL. The controller needs no authority there, because the first eigenfrequency of the panel, and therefore of the extended plant, is located at 122 Hz. In Fig. 11 the results for the acoustic analysis are summarized. The open loop case is taken as base line for the closed loop data. As expected from the vibration reduction, a sound power reduction can only be achieved in the third-octave bands above 100 Hz. All third-octave bands from 125 to 400 Hz show reductions up to 6 dB(A).
Fig. 10

Summed vibration spectra of all scan points

Fig. 11

Difference in sound power compared to open loop case

The results from the presented experiments show that an active control system is able to reduce the transmission of TBL noise below 500 Hz significantly. The usage of the extended plant in combination with H control leads to global vibration reductions and finally to less noise radiation of the panel.

Active Aircraft Sidewall Panels

Aircraft interior noise is caused by internal and external disturbance sources. Important external sources are the turbulent boundary layer and the engines. External disturbances propagate through the fuselage and the attached sidewall panels (linings) as structure and airborne sound into the cabin. The interior sound pressure level (SPL) depends on the TL of the structural systems and on the amplitudes of the external sources. Novel, energy-efficient aircraft configurations with carbon fiber-reinforced plastics (CFRP) fuselage structures and counter-rotating open rotor (CROR) engines are critical because they combine a low TL with high external sound pressure levels. Furthermore, the highest SPL occur at frequencies below 500 Hz where the TL is determined by mass. Therefore, alternative acoustic solutions are required which do not obey the mass law. In the present approach, ASAC and ANVC (depending on the sensors used to control the active structure) are applied to a lining structure to reduce the SPL at the passenger seats in front of the lining. As shown in Fig. 12, a passive lining is equipped with actuators, sensors, and a signal processing unit (SPU). The actuators transform the control signals of the SPU into structural vibration, and the sensors provide feedback to the SPU. The system will be described in more detail in the subsequent section. A test setup is required to design, implement, and evaluate the active lining. In order to avoid flight and wind tunnel tests at this early development stage, the setup is realized in a sound transmission loss facility. As shown in Fig. 13, the test specimen is mounted in the test opening between the reverberant sending and the semi-anechoic receiving room. A loudspeaker array (LSA) is placed directly in front of the fuselage panel to simulate the effect of the engine. As in real aircraft, the induced structural vibration propagates from the fuselage to the lining and radiates sound into the receiving room. This configuration permits the evaluation of the transmission loss either of parts or of the whole double panel system. Based on this information, the acoustic performance of active linings can be assessed. A simplified CFRP fuselage panel was designed which approximates the vibro-acoustic properties of an Airbus-A350 fuselage part. The lining is coupled to the fuselage panel by means of the original secondary structural links. Glass wool blankets are mounted on the fuselage and on the lining to account for mass loading and cavity damping. More details on the structural parts can be found in Misol et al. (2018). Figure 14 shows the connectivity and the signal flow of the active lining. An adaptive feedforward controller with finite impulse response (FIR) filter and filtered-x least mean squares (lms) algorithm is implemented on the SPU. More information on active control can be found in Kuo and Morgan (1995). It is assumed that correlated reference signals are available from the rotor engine of the aircraft. Therefore, in the lab, the reference signal is taken from the LSA. Depending on the control law, either accelerometer or microphone signals are used for the calculation or adaptation of the controller. The accelerometers are mounted on the lining at optimized positions, and the microphones are located at the headrests of the seats. The control signals are amplified and fed to the actuators. Electrodynamic inertial actuators are used because of their high performance at low frequencies. The actuators are mounted on the lining at optimized positions. In one preferred configuration, two actuators and one microphone per headrest are used. A replacement of the microphones by structural sensors and appropriate filtering is considered a desirable final implementation stage. Further information can be found in Misol (2019). The main functionality of the active lining is the reduction of low-frequency sound in the cabin. The adaptive feedforward control with two actuators achieves mean SPL reductions at the three error microphones of 25 dB at 74 Hz and 19 dB at 124 Hz. Although an ANVC is implemented, the effect is also global. This is due to the relatively long acoustic wavelengths in the considered frequency range. The capability of broadband sound reduction was tested with the same test setup, and a mean broadband SPL reduction of 6.6 dB and broadband sound power reduction of 5.9 dB were obtained in the bandwidth 50–500 Hz. It must, however, be noted that the reference signal configuration used in the lab is not realizable for real broadband disturbance sources in aircraft. The restrictions imposed in this case by coherence and causality are discussed in more detail in the references Misol (2014).
Fig. 12

Passive lining (left) and active lining with integrated actuators (red squares), sensors (green circles) and signal processing unit (SPU) (right)

Fig. 13

Test setup in a transmission loss facility with a loudspeaker array in the reverberation room and the fuselage-lining-system mounted in the test opening between reverberation and semi-anechoic room

Fig. 14

Connectivity and signal flow of the test setup in the transmission loss facility



  1. Elliott S (2001) Signal processing for active control. Academic Press, San DiegoGoogle Scholar
  2. Fahy F, Gardonio P (2007) Sound and structural vibration: radiation, transmission and response, 2nd edn. Academic Press, OxfordGoogle Scholar
  3. Kuo S, Morgan D (1995) Active noise control systems: algorithms and DSP implementations. Wiley, New YorkGoogle Scholar
  4. Misol M (2014) Aktive Steuerung des Transmissionsverhaltens stochastischer Störquellen durch flächige Leichtbaustrukturen [Active feedforward control of the transmission of stochastic disturbances through lightweight panel structures]. DLR-Forschungsbericht 2014-38, Technische Universität Carolo-Wilhelmina zu Braunschweig, Köln. http://elib.dlr.de/99510/
  5. Misol M (2019) Active sidewall panels with remote microphone technique for aircraft interior noise reduction. In: 26th international congress on sound and vibrationGoogle Scholar
  6. Misol M, Algermissen S, Monner HP (2013) Experimental study of an active window for silent and comfortable vehicle cabins. In: Sinapius M, Wiedemann M (eds) Adaptive, tolerant and efficient composite structures, research topics in aerospace, chap 36. Springer, Berlin/Heidelberg, pp 439–447. https://doi.org/10.1007/978-3-642-29190-6 CrossRefGoogle Scholar
  7. Misol M, Algermissen S, Rose M, Monner HP (2018) Aircraft lining panels with low-cost hardware for active noise reduction. In: Joint conference acoustics 2018. https://elib.dlr.de/122049/ Google Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2019

Authors and Affiliations

  • Malte Misol
    • 1
    Email author
  • Stephan Algermissen
    • 1
  • Thomas Haase
    • 1
  1. 1.Institute of Composite Structures and Adaptive SystemsGerman Aerospace Center (DLR)BrunswickGermany

Section editors and affiliations

  • Hans Peter Monner
    • 1
  1. 1.Institut für Faserverbundleichtbau und Adaptronik, Abteilung AdaptronikDeutsches Zentrum für Luft- und Raumfahrt e.V. (DLR)BraunschweigGermany