GB2480440A - Ultra-high speed LDA-PIV with integrated intelligent algorithm and smart sensors - Google Patents

Abstract

The present invention relates to a Laser Doppler Anemeometry-Particle Image Velocimetry(LDA-PV) measurement system and to uses thereof. The invention comprises an ultra-high speed imaging sensor, and imaging intensifier for ultra sensitive and onboard data processing for data reduction. The methods and apparatus described may be used for the measurement of nano/micro particle's velocity, size and concentration in fluid mediums for a wide range of applications in chemical and pharmaceutical processes, combustion in engines and biological processes where solid particles and fluid drops are involved in fluid mediums.

Description

DESCRIPTION

U1tra-hih speed LDA-PIV with integrated intelligent algorithm and smart sensors

BACKGROUND OF THE INVENTION

The present invention relates to an ultra-high speed CMOS smart imaging system, to methods of making them and to uses thereof. In particular the invention concerns such a system, methods and uses for the application of this smart imaging system for nano-particle velocirnetry, sizing and concentration measurement.

Laser Doppler Anemometry (LDA) has been developed for about 3 decades and has been extended to measure particle size as a Phase Doppler Particle Analyser (PDPA or Phase LDA) and a Shadow Doppler Velocimetry (SDV). The lower limits of particle size for conventional configurations are about 0.3 for PDPA and several for SDV. But PDPA can only measure spherical particles.

Driven by the development of imaging technology, particle image velocimetry (PIV) has been developed to measure velocities as well as particle size. Micro-PIV system can measure the swimming velocity of algae whose size is typically in the order of lOU in conditions of stationary flows (Chen et al 1998). A Spectral Turbidimetry technique has been developed recently to measure particle size down to 0.1g. The turbidity senor has to be placed in the flows and so it is intrusive (Crawley et al 1997) which is an obvious disadvantage for many applications. As light or acoustic scattering used in spectral methods have a fundamental limitation: the size, shape and concentration information are mixed up in the receiving signals so complex calibrations are needed.

So far, there is no non-intrusive instrument available to measure general particles (spherical and non-spherical) down to the order of 0.1k, which is urgently needed for many industrial processes using nano/micro particles.

Ultra-High Speed LDV-PIV With a LDA, a fringe pattern is created at the intersection of the two incident beams. As a particle moves through the fringes, the photodetector records a single burst whose frequency is proportional to the flow velocity. For a typical configuration, a frequency of 1MHz will indicate a flow velocity of about 5mIs. Since the whole measurement volume of the fringe pattern is treated as a point and detected by a photodetector, the size of the particles and the direction of particle movement are unknown and extra laser beams have to be introduced to measure 2-D or 3-D flows. The ultra-high speed LDA-PIV in the present invention is the combination of LDA and Ply, which records the fringe pattern using a high-speed imaging device. The fringe pattern will define the location of the measurement volume and the direction of the flow precisely. The spaces between the black/white strips on the fringe pattern will provide an accurate reference for length scale. A standard LDA receiver can be used simultaneously to validate the PIV results. An ultra-high speed imaging system is needed for the invention.

High Speed Imaging Until the advent of electronic imaging devices, the frame rate of imaging cameras was limited by the speed at which film could be passed through a camera -thus the constraints were mechanical in nature. The development of silicon imaging devices has removed the moving parts from imaging systems and thus the constraints restricting high frame rates have changed.

A fast response from a silicon device is not difficult to achieve; silicon PIN diodes and APDs can operate at a repetition rate of lOs of MHz with sub-nanosecond rise times (Menlo Systems APD21O, FPD310). However achieving high spatial resolution -especially in 2 dimensions -is far more challenging (APDs typically have an active area of -0.2 -1 mm2 and are individually packaged).

There are two main obstructions to the development of ultra-high frame rate imaging devices: data readout and system memory. Both are related to the vast data rates necessary for full frame readout of MHz rate systems.

CCD Technology Dalsa have developed a high-rate line-scan CCD (IT-PI-1024). This 10-bit, 10 x 10 pixel device operates at 87k fps, but is one dimensional (1024 x 1 pixels). The fast frame rate is achieved by reading out at 4 points, giving a total data rate of 100 MHz. The four point readout is used in order to circumvent the major problem when using CCDs at high speeds: CCDs do not have parallel read-out. All data from a CCD is (usually) clocked out through a single read-node and thus cannot be read-out simultaneously, leading to a bottle neck which restricts the ultimate frame rate.

Various techniques have been utilised to create fast CCD imagers, all of which involve either consecutive imaging with multiple CCDs, segmented frames (e.g. four separate images on a frame, one in each quadrant), or on-chip image storage, either by buffering the intact frame in a store area or by selective pixel masking and single line-stepping. All of these techniques can produce extremely high frame rates, but limit the number of frames that can be obtained before imaging must stop and readout begins. The Imacon 200 (DRS Hadland) uses a pyramid beam splitter to focus light onto segmented photocathode gated image intensifiers coupled to CCDs. This allows operation at a frame rate of 200 MHz, but limits the number of frames that can be acquired to 16 before stopping to readout. The SVR3 (DRS Hadland) does not employ beam splitters, but masks all pixels but the seventeenth in each column. The system then clocks down by one pixel between consecutive frames to obtain 16 images at a frame rate of 100 MHz, with reduced resolution in one direction, which can be reconstructed by system software at readout.

CMOS Technology CMOS devices utilise parallel output structures, thus avoiding the single-node readout issues associated with CCD technologies. This means that data obtained at high data rates may be transferred off chip in real-time, allowing for potentially continuous imaging. Thus the most pressing aspect for fast imaging with CMOS becomes not data read-out, but data transfer and storage. Photron's Ultima APX-RS can operate in full frame mode (1024 x 1024 pixels) at 3000 frames per second, but the camera's 16 GB RAM means that only 4.1 seconds of data may be obtained before the memory is full; using a 128 x 16 pixel ROl at 250 kfps, the camera can obtain 25.2 seconds of data, equating to 6.3 M frames.

Vision Research recently launched the Phantom v.l2 -the fastest camera on the market -which has 32GB RAM. The camera can run at 6242 fps in full frame (1280 x 800 pixels) (5.1 seconds of data stored in RAM), 20978 fps with a 512 x 512 ROT (6 seconds of data), and 1 Mfps with a 128 x 8 ROT (22.5 seconds of data). The camera has a 256 GB additional memory magazine (CineMag) which can receive data at 1000 fps, thus allowing 256 seconds (4 minutes and 16 seconds) of full frame data to be obtained at this rate.

The Invention After putting all components together, the ultra-high speed LDA-PIV system is to enable simultaneous measurements of flow and particles (both spherical and non-spherical, down to nano/micro scales), for example, with a size range of 0.5 to 5tm and velocity of the order of lOcmIs. This instrument uses an ultra-high speed image sensor to replace the photodetector in the receiver of a conventional Laser Doppler Anemometry.

I

BRIEF SUMMARY OF THE INVENTION

It is an objective of the present invention to provide a use of the ultra-high speed LDA-PIV system or measure nano/micro particle's velocity, size and concentration. It is a further objective of the invention to provide such a system which is sufficiently robust to be useful in commercial applications. Another object of the invention is to provide a system which can be manufactured reproducibly, conveniently and without excessive expense.

According to the present invention there is provided a novel ultra-high speed LDA-PIV system comprising: 1) An intelligent PIV image-processing algorithm, specially designed for the LDA-PIV to accommodate the features of very few particles in each image and the requirement of high speed.

2) An integrated PIV algorithm and smart sensor system. A smart on-board processing CMOS active pixel sensor is used. The structures and the data readout system are designed according to the required information of the PIV algorithm. The onboard control filters out the background and only transmits the particle size and velocity information so that an ultra-high speed system can be achieved.

3) A laser and optics to provide the needed beams. A dedicated optical system provides adequate magnification up to 2000 times and maximise the light input. The reference beam in traditional LDA-PIV system is used to determine the location of the measurement point.

In the context of this document, "particles" is used to refer to small solid or liquid mass suspended in liquids or gases, for example petrol or diesel drops in an engine or sediments in a river.

The particles may aggregate or breakdown to yield variable sizes for which the present invention is able to get non-intrusive and in-situ measurements. The examples described later show the applications in environmental engineering and combustion processes.

The ultra-high speed imaging method of the invention have advantages over conventional engineering, scientific approaches used in particle velocimetry, sizing and concentration measurement. In addition to flow measurements, the process of particle aggregation and break-down, which are vitally important to many industrial processes, can be observed in a non-intrusive way. As the instrument of the invention can measure three different parameters (velocity, size and concentration) simultaneously in one instrument then the mass fluxes and other correlations of the different parameters, which are very difficult to measure with conventional instruments, can be easily obtained. The overall apparatus embodying the invention would be quite small in dimension, allowing transportation from site to site and deployment to form large sensor networks thus allowing large numbers of data to be acquired on site. The invention would also greatly reduce the costs of obtaining results, waiting times and instrument maintenance.

The shape, size and movement of nano/micro particles can be recorded by the LDA-PIV system.

The method of the invention therefore provides a novel technique which may find application in a wide variety of circumstances. For example, the method according to the invention may be used to measure the particle's size and concentration distributions in the following list (which is by no means exhaustive) of products and services: The aggregation of nano-particles in photo-catalytic water purification processes and many other chemical processes; In chemical and pharmaceutical processes, both solid particles and fluid drops are widely involved in the mixing and mass transport processes.

The variable size of fuel drops in the combustion process in engines of aircraft and automobile etc; The aggregation of cohesive sediments in fluids; The dusts of natural (vocalic) and industrial processes for environmental protection; The aggregation of platelet and other bio-cells in blood vessel; In addition there will be many others applications relevant to "particles" which are still unforeseeable at the moment.

Furthermore, existing technologies used in these areas could easily be replaced by the present invention. To our knowledge, there are no prior art methods or other techniques for the means described above. Many methods used for nano/micro particle analysis, mainly light scattering, tend to be non-ideal for many commercial applications as they are either intrusive or cannot be used for in-situ applications.

It will be apparent to the skilled person, that the method of the invention can be carried out using direct visualisation so there is no need for complex calibrations as used for other instruments.

BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

The invention will now be more particularly described with reference to the following Figures and

Examples in which:

FIGS. 1 & 2 show graphically the ultra-high speed LDV-PIV system;

DETAILED DESCRIPTION OF THE INVENTION

The proposed system will consist of a laser source, a microscopic optical system, a receiving optic, photo sensor and PIV data processing. The microscopic optical system will split the laser into two beams and then a fringe pattern will be generated at the intersection of the two beams. The micro volumes of the fringe pattern are from 1 to 100 cubic micrometres depending on the particle size to be investigated. The receiving optics and the ultra-high speed photo image sensor will pick up the photons reflected by particles passing through the fringe pattern. Data processing will be done as much as possible in the sensor and the output will be at low frequency: the intermediate products or final values of velocity and particle sizes.

The invention innovatively increases the image frame rate to 5 -10 MHz which is critically important. First of all, the CMOS process has such a high frequency response at each pixel. The restriction from the readout electronics is resolved and has been achieved by a much smarter control inside the sensor than Region of Interest (ROl). Such smart control relies on the knowledge gained from the requirements of the PIV data processing on the fringe pattern. Data reduction is achieved for the invention due to the following facts: I Only very few particles exist in the fringe at the same time Information about the edge of the particles is enough to determine its size and position.

Therefore, only very limited number of pixels can be involved.

For the rest of the area around the particles, there is no need to read out the data.

The measurement of particles of 0.5 to 5tm with velocity at the order of lOcrnls is beyond the widest imagination in the community at the present time. Recent development in sensor technology has improved the situation slightly and the novel idea of data reduction by integrating PIV algorithm and smart sensors provide a feasible way to solve the problems. The intelligent PIV image-processing algorithm and the dedicated optical system are specially designed for this purpose and so are novel too.

The continuous imaging of nanoparticles in a fast flowing liquid is a very challenging problem and one that cannot be solved using current available technology. The detection system for the invention is described here. The detection system consists of a fully-customized imaging camera. The sensor aims to acquire continuous image frames at high rates (up to few MHz) to make quantitative measurements on nanoparticles in a fast flowing fluid (of a few mis). Since the integration time due for obtaining the requested frame rate is of the order of s or less, the number of photons available for producing the image of the flow is relatively small. So a very high SNR is required. In addition, running a CCD at ultra high frame rates leads to problems due to the generation of huge volumes of data that need to be rapid'y read out.

In order to achieve the required properties and overcome the problems that would be encountered in standard systems the device has the following features: Intensified imaging -the level of light illumination on the camera from a nanoparticle moving at very high speed in a fluid is much below the sensitivity level of any camera. Therefore an intensifier stage is included to amplify the signal.

Sparse readout -continuously transferring and storing millions of full image frames per second to a PC is not possible. State of the art data transfer lines will soon make it a possibility but the problem of data storage will still be a problem because terabytes of data will be generated every minute.

Sparse readout is a technique that allows only certain pixels to be readout rather than every pixel.

The PlY technique typically shows an image of illuminated particles on a dark background. Only details of the particles are required therefore this technique lends itself to only reading out only the necessary information. In combination with on-line processing (next section) to extract only the relevant information from the particle this will substantially reduce the data being transferred to the PC and will enable continuous operation of the camera system.

On-line processing -only certain bits of information are required from each particle, for example, particle size, shape, coordinates, brightness, etc. On-pixel and FPGA based techniques can extract only that information from each particle and transfer that data via the sparse readout method. In this way we can transfer just a few numbers rather than image data, and consequently significantly reduce the amount of data being stored and enable rapid processing of the subsequent information.

On-line data corrections can also be applied using PGA technology to reconstruct a high quality image from the raw data that is partially corrupted due to the demands of imaging at ultra-high frame rates. Typically cameras are mechanically shuttered to avoid illuminating the sensor during the readout phase. At such high frame rates this is not feasible, so an algorithm is to be implemented to correct for smearing effects due to rapid readout of the data without a shutter system. Since a complete redesign of the read-out electronics has been proposed by the company for achieving the desired performances, corrections of some effects like the smearing will be colTected by digital signal processing performed via PGAs.

Innovative aspects of the intensified, gated CMOS sensor technology is further described below. A novel imaging sensor has been developed specifically for the purpose of high speed, micro-scale particle PlY. In this sense it is not compromised for generic applications but focused on a primary goal. The novel aspects of the sensor are: The sensor can be operated continuously at high speed. Fast sensors typically rely on a buffer to collect short bursts of frames in order to limit data transfer rates, and also suffer from generating extremely large data volumes over time. The solution for the NV sensor is to firstly benefit from the parallel readout of CMOS technology, but more importantly to employ sparse data readout and online data processing to significantly limit data transfer and storage. A typical PlY image shows an array of particles in which only certain parameters are required, for example, coordinates, intensity and size; a significant fraction of the pixels within an image frame do not contain useful information. Therefore, only reading out the useful pixels to an FPGA and then processing those pixels to sort the relevant information leads to a significant reduction in data transfer and storage.

The active pixel CMOS sensor employed is coupled directly to a gated image intensifier. The gating allows multiple images to be captured in a single frame (for subsequent reconstruction) thereby enabling data capture at even higher rates than the sensor frame rate. Whereas the intensification allows the capability of imaging very low illumination levels; this is necessary due to the minimal light intensity reflected from micro-scale particles within such short time windows. In addition the image intensifier has a photocathode that makes it responsive to light across the visible spectrum and into the UV in order to make it applicable to imaging very small particles.

The main aspect of the invention is the development of an application specific, smart imaging sensor to image a small area of a fast flowing fluid for velocimetry.

The sensor will: (i) enable imaging of nanoparticles moving at very high speeds due to an intensification stage that amplifies the light signal.

(ii) Allow continuous imaging of particles moving at very high speeds. This is made possible by on-line processing and a sparse readout mode of operation, which considerably reduces the data readout from the sensor to the PC and also enables rapid processing of the information.

(iii) The sensor will allow on-line image correction and signal processing via PGAs to remove the effects of imaging at such high frame rates.

The design of the smart sensor system guarantees the achievement of the goal to produce a very reliable and powerful system for the study of particle and aggregates dynamics in a flow in the micro scale.

Such a high frame rate, smart sensor with high sensitivity and on-line signal processing will be widely applicable to other areas.

EXAMPLE 1

In a specific semiconductor photocatalysis technique, UV-Ti02 photo-catalysis system, Ti02 particles acting as catalysts are suspended in the flow to speed up the reactions. The mass transfer at the Ti02 particle surface is the limiting factor when the pollutant concentration is relatively high. As the standard practice in research work of UV-Ti02 photo-catalysis systems, the P25 Ti02 powder has been widely used. This type of powder has a surface area of 50 *5 m2/g and an average particle diameter of approximately several tens of nanometres by a number count of 21 nm. 90% of the particles fall in the size range 9-38 nm. However, the particles do not exist in isolation, but rather as complex primary aggregates, typically roughly 100 -1000 nm in diameter, which is about 5-50 times larger than an individual particle, or approximately containing more than 50 particles.

The surface of particles, which is available for mass transfer, can be significantly reduced when small particles aggregate into big ones. The sizes of Ti02 particles and aggregates are typically smaller than the Kolmogorov microscales in flows of most common mixers or reactors. This means that the flows may not be fully turbulent and are governed by viscous phenomena. In addition, collisions caused by Brownian motion may become relatively more important. Therefore, nanoparticle aggregation is fundamentally different from conventional particle aggregation and it needs to be investigated to improve the efficiency of the industrial processes involved.

EXAMPLE 2

Fuel injects in petrol or diesel engines have recently become dominant. Fuel drops formed by high speed jets are in the nano/micro size range. The smaller the particles, the faster and more complete the combustion would be. The energy conversion efficient is largely depending on this process.

Therefore, extensive research has been carried out but the lack of a non-intrusive in-situ measurement technique is the major drawback. The invention would offer a new generation of instrument for research.

EXAMPLE 3

Nano/micro particles are also widely used in chemical and pharmaceutical processes as both solid particles and fluid drops are widely involved in the mixing and mass transport processes. The invention would offer a new generation of instrument for research and monitoring industrial processes.

EXAMPLE 4

Platelet and other biological cells flow in blood vessels and aggregate to form viable size particles.

The invention would offer a new generation of instrument for research.

Claims (10)

1. CLAIMSWhat is claimed is: 1. The measurement and use of the particle images for the measurement of nano/micro particle's velocity, size and concentration in fluid mediums comprising: a smart ultra-high speed imaging sensor; an ultra-high sensitivity image intensifier; an intelligent algorithm specially designed for the invention to achieve great data reduction.
2. 2. The smart ultra-high speed imaging sensor according to claim 1, wherein the applications for nano/micro particle's measurement thereof.
3. 3. The ultra-high sensitivity image intensifier according to claim 1, wherein the continuous operation and gating are specifically designed for the invention.
4. 4. The intelligent algorithm according to claim 1, wherein the knowledge of particles are used for data reduction thereof.
5. 5. The intelligent algorithm according to claim 1, wherein the LDA fringe patterns are used as thebackground for PlY velocity directions.
6. 6. The nano/micro particles according to claim 1, wherein the nano/micro particles comprise solid particles and fluid drops in the size of nano/micro meters range and in any shapes.
7. 7. The fluid medium according to claim 1, wherein the fluids comprises gas, liquid and plasma, for example, air, water, etc.
8. 8. The velocity according to claim 1, wherein the velocity comprises the movements of particles by all driving forces and can be in 1-dimension, 2-dimensions or 3-dimensions thereof.
9. 9. The size according to claim 1, wherein the size of particles comprises the maximum, minimum, the distribution, mean, median and standard deviation of the particle's outside boundaries.
10. 10. The concentration according to claim 1, wherein the concentration can be expressed by volumetric fraction, weight fraction or any other means.SHARE OF THE INTELLECTURAL PROPERTY RIGHTProfessor Daoyi Chen ---70% Dr Gary Royal ---20% Dr. Jennifer Ann Griffiths ---10%