Grouped Simplified Particle Swarm Optimization And Its Application In Sound Power Measurement

Sound power is the total sound energy radiated outward from the sound source in unit time.The accurate measurement of sound power is conductive to the reliable analysis of noise characteristics and causes,which promotes the development of research on effective noise control.In most cases,sound power measurement is based on sound pressure,sound intensity or vibration velocity on the surface of the device under test.Devices with sound power measurement functions include sound level meters and sound intensity meters.A sound level meter can measure noise in real time and display the analysis results by LCD,but it has low working efficiency and has limited display and storage capabilities.A sound intensity meter has higher measurement accuracy in complex sound environment than a sound level meter,while it is not popular because of its high price,insufficient measurement accuracy and limited measurement range.There are two key technologies for accurate measurement and analysis of sound power: frequency weighting and spectrum analysis.This thesis studies the design errors and their sources of the traditional bilinear transformation method for designing frequency weighting.After the comparision of advantages and disadvantages of Particle Swarm Optimization,Improved Chicken Swarm Optimization and Modified Imperialist Competitive Algorithm,a grouped simplified particle swarm optimization is proposed combining the grouping idea from Frog Leaping Algorithm.Tested with four standard testing functions,Sphere,Ackley,Griewank,and Alpine,the proposed algorithm is used to design frequency weighting network.This thesis adopts the precision method for reverberation rooms,which belongs to the sound pressure method,and builds a multi-channel real-time noise measurement and analysis system to achieve accurate measurement,analysis,display and management of related information on noise.The hardware part of the system consists of a microphone,a signal conditioning circuit and a data acquisition card.The software part is designed and implemented by a virtual instrument based on the Lab VIEW platform.The sound power measurement and analysis system built in this research is intelligent,networked and virtual.It can realize the simultaneous acquisition and processing of multiple signals,complete spectrum analysis and frequency weighting,and calculate sound power level and uncertainty,etc.The system has other functions such as chart display and report generation.The measurement process and accuracy meet the design requirements of the international standard ISO 3747-2010 for the precision method for reverberation rooms.The guided interface provided by the system makes the operation easy and the human-computer interaction friendly.The optimized frequency weighting network is applied to the sound power measurement system built in this study.Four pure tone signals with sound pressure levels of 78 d B,98 d B,118 d B,and 128 d B were used to test the system.The test results show that the grouped simplified particle swarm optimization proposed in this research has the advantages of high optimization accuracy and strong algorithm stability.Especially in the high-frequency test section,the signal weighting error is small and the relevant data is quite close to standard weighting values.