Modeling And Optimization Of Local Scheduling For Large-Element Ultrasonic Phased Array Multi-group Scanning System

Ultrasound phased array instrument has the characteristics of complete coverage of space acoustic field(without moving probe),high accuracy,strong anti-interference ab ility,omni-directional real-time ultrasound imaging and so on.It is an important scient ific instrument for scientific research work such as material parameter characterization,material failure mechanism analysis and life prediction.The large array ultrasonic phas ed array multi-group scanning system(LE-UPA-MGSS)can increase the scanning flex ibility,image contrast,diversity of focusing rules and realize distributed scanning,but it will produce a large number of echo signals of different arrival time,length and fo rm,which need to consume a lot of hardware resources of the instrument.In order to improve resource utilization,real-time performance,reduce hardware consumption and avoid conflict of signal processing,it is necessary to study resource reuse and data pr ocessing cooperative scheduling.This paper focuses on the local scheduling modeling and optimization of the large array ultrasonic phased array multi scanning system,foc using on the local scheduling theory and key technologies of multi groups of delay an d focus,multi group homogeneous flow signal processing,heterogeneous hierarchical multi DAG signal processing,and so on,and develops experimental prototype with G uangzhou Doppler Electronics Technology Co.,Ltd.The research has been funded by the National Key Foundation for Exploring Scientific Instrument "development and app lication of the new ultrasonic phased array instrument"(2013YQ230575)and Guangzh ou Science and Technology Plan Project(201509010008).In this paper,the current research situations is discussed from the aspects of mult i-group delay and focus scheduling algorithm,multi-group homogeneous flow signal pr ocessing scheduling algorithm,heterogeneous hierarchical multi-DAG signal processing scheduling algorithm and so on,and the research content is determined.The main wor k of this paper includes:Research on Delay and Focus Scheduling and Optimizing Algorithms for Large-element Ultrasound Phased Array Multi-group Scanning System.In order to solve the contradiction between delay accuracy and resource utilization,a ultrasonic phased array multi-group scanning architecture based on time division multiplexing is proposed,which abstracts the scheduling problem of delay and focus equal identical parallel machine,and analyses its mathematical model and time complexity.An improved Bound Fit(IBF)algorithm based on the BF algorithm of the identical parallel machine problem is proposed to reduce the scheduling of Total-focus method(TFM)frame task data.Compared with SPT,BF and IBF algorithms,simulation results show that the large array ultrasonic phased array multi-group scanning delay and focus scheduling architecture and IBF algorithm shorten the maximum completion time by 8.76-21.48%,reduce hardware resources by 30-40%,and increase the maximum operating frequency by 6.72-11.22%.The simulation results of FPGA show that compared with LIST algorithm,the frame task time is saved by 12.26%.Research on Homogeneous Flow Signal Processing Scheduling and Optimizing Algorithms for Large-element Ultrasound Phased Array Multi-group Scanning System.The delay and focus of ultrasonic phased array are analyzed,and the start time of frame task signal is deduced.A multi-group scan signal processing architecture based on time division multiplexing is proposed to abstract the distributed flow shop scheduling problem,and the homogeneous simplified model with start time,i.e.the job shop scheduling problem with start time,is studied.According to this problem,the start time and processing time are combined.And The sum of start time and processing time adjacent algorithm(SSPA)is proposed;through simulation algorithm experiment,the maximum completion time is increased by 4.00-36.84%,and the resource utilization is increased by 1.01-21.67%.Compared with FCFS and SPT algorithm,the virtual logic analyzer experiment shows that SSPA algorithm improves the resource utilization by 9.72%,shortens the maximum completion time by 11%,and increases the flexibility of element grouping through architecture.Research on signal processing scheduling and optimization algorithm of heterogeneous hierarchical multi-DAG in large array ultrasonic phased array multi-group scanning system.In order to increase the diversity of distributed scanning modes,a heterogeneous multi-group scanning signal processing architecture based on on on-chip bus is proposed.Based on this architecture,a heterogeneous hierarchical DAG workflow problem(HHMDP)is abstracted for different signal processing workflows generated by different grouping scanning modes in multigroup scanning phased array systems and their hierarchical processing in specific functional signal processing modules.Based on Heterogeneous Earliest Completion Time(HEFT)algorithm,a Heterogeneous Precedent Earlist Finish Time(HPEFT)algorithm with precedent pointer adjustment is proposed.Compared with SPT,Round-Robin,HEFT and RHEFT(Reverse Heterogeneous Precedent Earlist Finish Time),the maximum completion time,idle time slot ratio(RITS),Relative Laxity with Deadline(RLD)and missed deadline rate(MDR)were compared.The HPEFT algorithm's maximum completion time was shortened by 3.87-57.68%,RITS decreased by 4.32-6.82%,RLD increased by 2.27-8.58%,MDR is reduced by 25%-58% under severe load conditions.Compared with HEFT algorithm in pre-simulation of FPGA,the frame rate of specified signal processing module increases by 3.56%.Experimental verification and analysis of large-element ultrasonic phased array multigroup scanning instruments.Based on the self-designed scheduling verification platform for ultrasonic phased array signal processing,under the given boundary conditions,LIST algorithm and IBF algorithm are used for multi-group delay and focus,SPT algorithm and SSPA algorithm are used for multi-group homogeneous flow signal processing,and R-R and HPEFT algorithm are used for heterogeneous hierarchical multi-DAG signal processing,the overall frame rate is increased by 2.95-9.36%.The new instument is used for imaging,which blind area is narrower than the original one.The image has better penetration and resolution;the contrast of 64-element A-scan imaging is 2.30 times of that of 32-element imaging by the contrast experiment of array number imaging;the maximum signal intensity of 64-element TFM scanning is 22.09% higher than that of 32-element scanning;the higher the number of array elements,the better the signal-to-noise ratio;The experiment result show that the multigroup scanning image is not significantly different from that of single group scanning,and multi-group scan can ensure real-time frame rate,which verifies the feasibility and effectiveness of multi-group scan scheduling.