3rd Technical Meeting - WP4

Project SLOPE

WP 4 – Multi-sensor model-based quality control of mountain forest production

Work Package 4: Multi-sensor model-based

quality control of mountain forest production

The goals of this WP are:• to develop an automated and real-time grading (optimization) system for the forest production, in order to improve log/biomass segregation and to help develop a more efficient supply chain of mountain forest products• to design software solutions for continuous update the pre-harvest inventory procedures in the mountain areas • to provide data to refine stand growth and yield models for long-term silvicultural management

Work Package 4: work to be done T4.1

Quality rules & specificationsCNR, TRE:

Develop tool Harvest Simulator TRE:

Develop models of treesGRA, TRE:

Compare models with real dataTRE, GRA, TRE:

Link automatic system with visualTRE,CNR:

Develop 3D quality indexTRE, CNR:

Measurement of standing treesCNR, TRE:

Measurement of felled treesCNR:

T4.1 3D quality

D03.01

D01.04

D04.07

TRE

D04.02

TRE

D01.04

Determine optimal protocolCNR:

Calibration transferBOK, CNR:

Develop models for labCNR, BOK:

Measure NIR on standing treesTRE, CNR, FLY:

Measure NIR on felled treesCNR, GRE:

Measure NIR on processor headCNR, COM:

Measure NIR on pale of logsCNR, BOK:

Develop models for in fieldCNR, BOK:

Compare models with lab dataCNR, BOK:

Develop NIR quality indexCNR, BOK:

Develop provenance NIR modelsCNR, BOK:

Design data base of NIR spectraBOK, CNR:

T4.2 NIR quality

D04.03

CNR

D04.08

CNR

Determine usabilityCNR:



Imaging standing trees BOK, FLY, TRE:

Imaging fallen trees BOK, GRE:

Imaging on processor headBOK, COM:

Imaging on pale of logsBOK, CNR:



Develop hyperspectral indexCNR, BOK:

Design data base of hyperspectraBOK, CNR:

T4.3 hyperspectral quality

D04.04

D04.09

BOK

BOK

Determine optimal set-up for the hyperspectral camera, illumination, and sample holdingBOK, CNR:

D01.04

Determine optimal set-up for the NIR sensor, illumination, and sample holdingCNR, BOK:

Develop report on using SWCNR:

Develop models for SW qualityCNR:

Test on standing trees CNR, GRE:

Tests on fallen trees CNR, GRE:

Tests on processor headCNR, COM:

Imaging on pale of logsCNR:

Develop SW quality indexCNR:

Define quality thresholdsCNR:

Analyze material dependant factorsCNR:

T4.4 stress wave quality

D04.05

D04.10

CNR

CNR

Determine optimal set-up for the stress wave measurement, including time of flight and free vibrations sensorCNR:

D01.04

Determine quality requirements for high-end assortments

CNR:

Laboratory scale tests for delimbing energy needs

CNR:

Develop CP quality indexCNR:

T4.5 cutting power quality

D04.11

D04.06

CNR

CNR

Determine optimal set-up for the measurement of cutting forces on the processor headCNR:

D01.04Laboratory scale tests for chain saw energy needs

CNR:

Develop models linking CP in delimbing and quality

CNR:

Develop models linking CP in chain sawing and quality

CNR:

Develop report on using CPCNR:

Link in-field data with cloud database

CNR:

Compare automatic and visual grading resultsBOK, CNR:

Determine threshold valuesCNR:

Develop grading expert systemCNR:

Develop algorithm for data fusionCNR, COM, TRE:

In field visual quality assessment CNR, BOK:

Develop data base for prices of woody commodities

CNR, BOK:

Reliability studiesBOK:

Economic advantage studiesBOK, CNR:

T4.6 quality implementation

D04.01

CNR

D04.12

CNR

Identify grading rules for standard and niche productsCNR:

Prepare state-of-the-art report on grading rulesCNR:

T4.1: Data mining and model integration of stand quality indicators from on-field survey









T4.1 3D quality

D03.01

D01.04

D04.07

TRE

D04.02

TRE

draft: October 2014

accepted: July 2015

OCtober 2015

the resources planned: 9 M/Mthe resources utilized:PROBLEMS: Not reported










T4.1 3D quality

D03.01

D01.04

D04.07

TRE

D04.02

TRE

D01.04













T4.2 NIR quality

D04.03

CNR

D04.08

CNR













D04.04

D04.09

BOK

BOK


D01.04












D04.05

D04.10

CNR

CNR


D01.04


CNR:


CNR:



D04.11

D04.06

CNR

CNR



CNR:


CNR:


CNR:



CNR:







CNR, BOK:




D04.01

CNR

D04.12

CNR



T4.2: Evaluation of NIRS as a tool for determination of log/biomass quality index

D01.04













T4.2 NIR quality

D04.03

CNR

D04.08

CNR


the resources planned: 13 M/Mthe resources utilized:PROBLEMS: Delay in access to sensor (sensor arrived Oct 2015, software Dec 2015)SOLUTIONS: intensify efforts, working meetings with BOKU and COMPOLAB

draft: Dec 2015

draft: October 2014

accepted: July 2015










T4.1 3D quality

D03.01

D01.04

D04.07

TRE

D04.02

TRE

D01.04













T4.2 NIR quality

D04.03

CNR

D04.08

CNR













D04.04

D04.09

BOK

BOK


D01.04












D04.05

D04.10

CNR

CNR


D01.04


CNR:


CNR:



D04.11

D04.06

CNR

CNR



CNR:


CNR:


CNR:



CNR:







CNR, BOK:




D04.01

CNR

D04.12

CNR



T4.3: Evaluation of hyperspectral imaging for the determination of log/biomass quality index













D04.04

D04.09

BOK

BOK


D01.04

the resources planned: 17 M/Mthe resources utilized:PROBLEMS: Delay in access to sensor (sensor arrived Oct 2015)SOLUTIONS: intensify efforts, working meetings with BOKU and COMPOLAB

Jan 2016

draft: May 2014

accepted: July 2015










T4.1 3D quality

D03.01

D01.04

D04.07

TRE

D04.02

TRE

D01.04













T4.2 NIR quality

D04.03

CNR

D04.08

CNR













D04.04

D04.09

BOK

BOK


D01.04












D04.05

D04.10

CNR

CNR


D01.04


CNR:


CNR:



D04.11

D04.06

CNR

CNR



CNR:


CNR:


CNR:



CNR:







CNR, BOK:




D04.01

CNR

D04.12

CNR



T4.4: Data mining and model integration of log/biomass quality indicators from stress-wave











D04.05

D04.10

CNR

CNR


D01.04


CNR:

the resources planned: 5.5 M/Mthe resources utilized:PROBLEMS: delay with access to sensors (arrived January 2016), change of Task LeaderSOLUTIONS: intensify efforts, change in staff involved

draft: Jan 2016

draft: December 2014

accepted: July 2015










T4.1 3D quality

D03.01

D01.04

D04.07

TRE

D04.02

TRE

D01.04













T4.2 NIR quality

D04.03

CNR

D04.08

CNR













D04.04

D04.09

BOK

BOK


D01.04












D04.05

D04.10

CNR

CNR


D01.04


CNR:


CNR:



D04.11

D04.06

CNR

CNR



CNR:


CNR:


CNR:



CNR:







CNR, BOK:




D04.01

CNR

D04.12

CNR



T4.5: Evaluation of cutting process (CP) for the determination of log/biomass CP quality index


CNR:



D04.11

D04.06

CNR

CNR



CNR:


CNR:


CNR:


the resources planned: 6.0 M/Mthe resources utilized:PROBLEMS: delay with access to sensors (arrived January 2016)SOLUTIONS: intensify work, close collaboration with COMPOLAB

draft: Jan 2016

draft: January 2014

accepted: July 2015










T4.1 3D quality

D03.01

D01.04

D04.07

TRE

D04.02

TRE

D01.04













T4.2 NIR quality

D04.03

CNR

D04.08

CNR













D04.04

D04.09

BOK

BOK


D01.04












D04.05

D04.10

CNR

CNR


D01.04


CNR:


CNR:



D04.11

D04.06

CNR

CNR



CNR:


CNR:


CNR:



CNR:







CNR, BOK:




D04.01

CNR

D04.12

CNR



T4.6: Implementation of the log/biomass grading system


CNR:







CNR, BOK:




D04.01

CNR

D04.12

CNR



the resources planned: 8.0 M/Mthe resources utilized:PROBLEMS: Delay related to other tasks – difficulties with implementationSOLUTIONS: LAB scanner + prototype software developed in lab + algorithms ready

31.06.2016

draft: October 2014

accepted: July 2015

fulfillment of the project work plan:related deliverables (M25)

WP4 M17

task deliverable title type of

deliverable

lead participant

due date foreseen or actual delivery date comment

T4.1D4.2 on field survay data for tree characterization report TRE 31.10.2014 31.10.2014 accepted

D4.7 estimation of log/biomass quality by external tree shape analysis software tool TRE 31.05.2015 18.12.2015 Waiting for final

approval

T4.2D4.3 establisghing NIR measurement protocol report CNR 31.10.2014 31.10.2014 accepted

D4.8 estimation of log/biomass quality by NIR software tool CNR 30.09.2015 March 2016

T4.3D4.4 establisghing hyperspectral imaging measurement protocol report BOK 30.11.2014 05.05.2015 accepted

D4.9 estimation of log/biomass quality by hyperspectral imaging software tool BOK 31.10.2015 April 2016

T4.4D4.5 establishing acoustic-based measurement protocol report CNR 31.12.2014 05.05.2015 accepted

D4.10 estimation of log/biomass quality by acoustic methods software tool CNR 31.11.2015 May 2016

T4.5D4.6 establisghing cutting power measurement protocol report CNR 31.01.2015 31.01.2015 accepted

D4.11 estimation of log/biomass quality by cutting power analysis software tool CNR 30.12.2015 April 2016

T4.6D4.1 existing grading rules for log/biomass report CNR 31.10.2014 31.10.2014 accepted

D4.12 implementatio and callibration of prediction models for log/biomass quality classes software tool CNR 31.06.2016 June 2016 NO



Planning actions for all activities and deliverables to be executed in M25-30:

Finalize + close: D04.8, D04.9, D04.10, D04.11Deliver + finalize + close: D04.12 Initiate + deliver: -

Assemble sensors + control systemInstall sensors in the processor headContinue field tests with portable instrumentsCalibrate system in the lab (“model tree”)Collaborate with WP3 (and others) in hardware development



Risks and mitigating actions:

Significant delay related to DoW amandment:• the purchase and delivery of sensors delayed set-up of the system in the lab (laboratory scanner) as well as on the processor head; intensify efforts for all involved partners, direct collaboration and working group meetings, involve additional staff for developments, testing and implementation

Technologies provided will not be appreciated by “conservative” forest users; demonstrate financial (and other) SLOPE advantages

Limited reliability of some sensors when implemented on the forest machinery; careful planning, collaboration with SLOPE (+outside) engineers

Sensors and electronics (WP3 & WP4) in progress

MicroNIR

Hamamatsu C11708

Hamamatsu C12666

Accelerometers time of flight

Mechanical excitator

Accelerometers free vib ration LDS correction

Laser Displacement Sensor

AE sensor + amplifier

Tensionmeters 1/4 bridge

Dynamic load cell

Hydraulic pressure sensor

Hydraulic flow sensor

Absolute encoders

Hamamatsu C11351

NI 9234

NI 9223

NI 9235/NI 9236

NI 9220

Port #8

CompactDaq SENSORS

Port #7 Port #6 Port #5 Port #4 Port #3 Port #2 Port #1

LAN port #2

Industrial PC

LAN port #1 Port #6 Port #5

Video output + USB port #4 USB port #3 USB port #2 USB port #1

NI 9403 (Dig ital I/O)

Custom line scan camera

Port #8

CRio (real t ime?) MACHINE CONTROL

Port #7 Port #6 Port #5 Port #4 Port #3 Port #2 Port #1

SEA 9744 (GSM + GPS)

Joystic(s)

RFID reader

Hydraulic actuators

???

???

???

???

LAN port #5 LAN port #4 LAN port #3

Touch screen

T4.2

+T4

.3

T4.4

T4

.5

T4.5

T4

.4

WP3

WP3

W

P3

NI 9220

Temperatures of oil and air



Thank you! – Grazie!

4.2 Deliverables status

Deliverable D.4.03 Establishing NIR measurement protocolevaluating the usability of NIR spectroscopy for characterization of bio-resources along the harvesting chain, providing guidelines for proper collection and analysis of NIR spectra.

Delivery Date M10, October 2014 - acceptedEstimated person Month = 5

Deliverable D.4.08 Estimation of log/biomass quality by NIRSet of chemometric models for characterization of different “quality indicators” by means of NIR and definition of “NIR quality index”

Delivery Date M21, September 2015 – draft presenting protocol validation uploaded in dropbox, the deliverable are the models currently improvedEstimated Man/Month = 8

Detailed procedure related to measurement of NIR along the whole harvesting scenario

Forest modelingNIR quality index #1 will be related directly to the health status, stress status and to the productivity capabilities of the tree(s) foreseen for harvest –images from FLYBY to be analyzed

Tree markingDirect measurement of the NIR spectra by means of portable instruments will be performed in parallel to the tree marking operation (NIR quality index #2) – first trials done in December

Cutting of treetesting the possibility of collecting sample of wood in a form of the triangular slice being a part of the chock cut-out from the bottom of the log (NIR quality index #3) – first trials done in December

(optional) prepare samples #1

measurement of infrared spectra (wet state)

prepare samples #2

condition samples

chemometric models for wet wood and/or in field

chemometric models for dry/conditioned wood (lab)

measurement of infrared spectra

collect sample #1: chip of axe

collect sample #2: core ~30mm deep

collect sample #3: chips after drilling core

collect sample #4: triangular slices

measurement NIR profile or hyperspectral image

measurement profile of infrared spectra

consider approach: max slope, pith position, WSEN

compute NIR quality index#2



measurement profile of infrared spectra

consider approach: pith position, defects


tree marking

cutting tree

processor head

pile of logs

expert system & data base

refresh sample surface

measurement of infrared spectra (dry state)

compute dry wood NIR quality index#6 compute the log quality

class (optimize cross-cut)

estimated tree quality

forest models

update the forest database

compare results of wet and dry woods

combine all available char-acteristics of the log

lab

Calibration transfer f(MC, surface_quality)

3D tree quality index

hyperspectral HI quality index

stress wave SW quality index

cutting force CF quality index


Detailed procedure related to measurement of NIR along the whole harvesting scenario

Processor headNIR sensors will be integrated with the processor head (NIR quality index #4). The first trials are foreseen for mid January on the lab scanner during measurement of the model tree

Pile of logsThe cross section of logs stored in piles is easily accessible for direct measurement. Such measurements will be repeated periodically in order to monitor the quality depreciation and to determine the most optimal scanning frequency. The result of measuring NIR spectra of logs stored in piles will be NIR quality index #5 – first trials done on 60 logs

LaboratorySamples collected in the forest will be measured instantaneously after arrival in the laboratory (at the wet state and with rough surface) by using the bench equipment (NIR quality index #6). Campaign done by BOKU with FT instrument, recently parallel measurement at IVALSA with MicroNIR

Current activity

Software developmnet and measurement campaign

First models – defects detection

knots

compression wood

fungi

compression wood

wood

resin pocket

knots

resin pocket

fungi

Defects detection with lab equipment

(data source: BOKU, PCA models: CNR)

Defects detection with MicroNIR

● - knots ▲- stain ▼- wood (model tree) ■ bark ♦ resin.

Scheduled activity

activity responsible status schedule

Determination of measurement conditions of MicroNIR CNR On going

Measurement discs from BOKU CNR On going

Calibration transfer BOKU February/March 2016

Measurement trees in field CNR On going

Data base of spectra for QI CNR/BOKU On going

Report from in field measurement CNR done

Chemometric models in PLS toolbox CNR/BOKU February 2016

Installation of MicroNIR on processor head COMPOLAB March 2016

Implementation of the software in the system CNR May 2016

Report of NIR traceability CNR March 2016

Use existing models for prediction of calorific value CNR February/March 2016

Project SLOPE

WP4: Multi-sensor model-based quality control of mountain forest production

T4.3– Evaluation of hyperspectral imaging (HI) for the determination of log/biomass “HI quality index”

Cork, January 19th-21st, 2016

Andreas Zitek, Katharina Böhm, Jakub Sandak, Anna Sandak, Barbara Hinterstoisser

BOKU & CNR

Technical Meeting, Cork 19.01.2016

Mid-term Review 2/Jul/15

Task 4.3 – Output

D4.04 Establishing hyperspectral measurement protocol• Methodology, laboratory setup and field transfer

D4.09 Estimation of log quality by hyperspectral imaging• Labscale investigations ((visible)/near infrared hyperspectral cameras)

• Validation by NIR measurements• Application of chemometric approaches for data evaluation and

multivariate image analysis• Identification of most relevant spectral information

• Development of transfer options to (harsh) field conditions• Development of the “HI quality index” for quality grading• Technological implementation on prototype

Fulfillment of the project work plan:related deliverables (M25)

WP4

task deliverable title type of

deliverable

lead participant

due date foreseen or actual delivery date comment

T4.1D4.2 on field survay data for tree characterization report TRE 31.10.2014 31.10.2014 accepted

D4.7 estimation of log/biomass quality by external tree shape analysis software tool TRE 31.05.2015 18.12.2015 Waiting for final

approval

T4.2D4.3 establisghing NIR measurement protocol report CNR 31.10.2014 31.10.2014 accepted

D4.8 estimation of log/biomass quality by NIR software tool CNR 30.09.2015 March 2016

T4.3D4.4 Establishing hyperspectral imaging measurement

protocol report BOK 30.11.2014 05.05.2015 accepted

D4.9 Estimation of log/biomass quality by hyperspectral imaging software tool BOK 31.10.2015 April 2016

T4.4D4.5 establishing acoustic-based measurement protocol report CNR 31.12.2014 05.05.2015 accepted

D4.10 estimation of log/biomass quality by acoustic methods software tool CNR 31.11.2015 May 2016

T4.5D4.6 establisghing cutting power measurement protocol report CNR 31.01.2015 31.01.2015 accepted

D4.11 estimation of log/biomass quality by cutting power analysis software tool CNR 30.12.2015 April 2016

T4.6D4.1 existing grading rules for log/biomass report CNR 31.10.2014 31.10.2014 accepted

D4.12 implementatio and callibration of prediction models for log/biomass quality classes software tool CNR 31.06.2016 June 2016 NO

PROBLEMS: Delay in access to sensors, that produce the data to develop the model and implement the system (=D 4.09) SOLUTIONS: intensify efforts, working meetings with CNR, sharing and transfer of samples measured at BOKU with NIR and HSI to CNR for MicroNIR and Hamamatsu measurements, meeting in February at BOKU to produce models, implement system and finalize D4.09 in April 2016

Task 4.3 – Field transfer optionsImplementation of the hyperspectral imaging in the field:• Hyperspectral imaging using new technologies

Optimal accuracy and spatial resolution Rigidity of sensors (not suitable for harsh conditions) Relatively high cost

• Mono/multi spectral imaging the log cross-section Optimal spatial resolution Reasonable cost Poor spectral accuracy Challenges with implementation

• Several simple spectrometers installed on the scanning bar & measuring the log cross-section Optimal spectral accuracy and sufficient spatial resolution Reasonable cost Difficulties with implementation

Mid-term Review2/Jul/15

T3.4 Intelligent processor head

Task 4.3 HSI – general setups

• Whiskbroom imaging : During whiskbroom imaging the sample is scanned pixel per pixel in the x–y–spatial direction in a sequential manner.

• Staring (staredown) imaging: Staring imaging is done by a two-dimensional camera capturing the spectral information in each pixel x-, y-plane at once.

• Implementation of the Pushbroom imaging idea: as a line scanning system with multiple sensors acquiring the information for a reduced set of pixels in the line at once – subsequent interpolation planned and possible.

Project meeting19-22/jan/2015

From: BOLDRINI, B., KESSLER, W., REBNER, K. & KESSLER, R. W., 2012. Hyperspectral imaging: a review of best practice, performance and pitfalls for inline and online applications. Journal of Near Infrared Spectroscopy, 20 (5): 438-508.

Task 4.3 HSI – general setupsSoftware #1: simulation of the NIR sensor results on the scanning bar.It is possible to simulate the timing of scan by changing the integration time.Cycle time of software (including integration of signal and acquisition of data by USB + display on monitor.) = 0.25 s (4 Hz)


Task 4.3 HSI – general setupsSoftware #2: simulation of the hyperspectral sensor results on the scanning bar (only single wavelength). Image on the left is an input showing all points measured with hyperspectral system. Image on the right is reconstructed image by using simple interpolation (part of the code is shown, used for reconstruction only).


settings of scanning density – rotation (degree)

settings of scanning density – pixels on the scan bar

size of the probe / measured area (pixels, ROI)

resolution of interpolated image

LabView code for reconstruction

raw data from scanner 2D interpolation

Δ radius = 10 Δ angle = 2 probe_size = 10Δ reconstruction = 20


Δ radius = 50 Δ angle = 0.5 probe_size = 10Δ reconstruction = 20


Δ radius = 20 Δ angle = 1 probe_size = 10Δ reconstruction = 10 ( computation time…)

Task 4.3 Model development

Collection oftraining

samples withdifferent deficits

Measurementswith NIR and HSI

Laboratory equipment

Detection ofmost significant

wavelengthregions for

deficitsFirst models, lab

equipment

Measurements withNIR and HSI with

sensors that will beon Processor Head

MicroNIRHamamatsu

Model development and exportwith PLS model exporter

Models can be directly used fordata from scanning bar and the

Labview software installed on PC incl. preprocessing and statistical

methodsModels sensor arm equipment

WorkflowLab (scientific basis, calibration transfer)

Calibration & fieldtransfer

Task 4.3 Sensor wavelength range comparison

Visible & near infrared range (VNIR)

400 nm

• Visible wavelength range ~ 390 - 700 nm• Near IR wavelength range ~ 700 nm - 2500 μm

2500 nm

FT NIR (lab) 800 – 2400 nm

Hyperspectral (lab) 900 – 1700 nm

MicroNir (sensor)900 – 1700 nm

Hamamatsu C12666MA

340 – 780 nm

Hamamatsu C11708MA

640 – 1050 nm

Range covered by sensors on processor head340 – 1700 nm

Mid-term Review2/Jul/15

Task 4.3 – 25 samples (spruce, Picea abies) with defects

resin pockets

eccentric pith + compression wood + rot eccentric pith + rot + knot

shakes, checks, splitsknots

Measured with NIR and hyperspectral imaging at

BOKU, and MicroNIR andHamamatsu at CNR

NIR-Spectroscopic measurements –BOKU - laboratory

• 14 out of 25 samples wood discs were measured using a FT-NIR with a fibre optic probe at BOKU

Meeting 19/Jan/2016

NIR-Spectroscopic measurementsData evaluation

Meeting 19/Jan/2016

Scores Loadings

• Principal component analysis for wood and resin/ resin pockets

wood

resin pocket

Wavelength 800-1400 nm and 1st derivative + vector normalized

resin


Meeting 19/Jan/2016

Scores Loadings

• Principal component analysis for wood and fungi

woodfungi



Meeting 19/Jan/2016

Scores Loadings

Principal component analysis for wood, bark and fungi

bark

wood

fungi



Meeting 19/Jan/2016

Scores Loadings

Principal component analysis for wood, bark, compression wood and fungi

barkwood

fungi

compression wood



Meeting 19/Jan/2016

Scores Loadings

Principal component analysis for wood, bark, compression wood and fungi

bark

wood

fungi

compression wood



Meeting 19/Jan/2016

Scores Loadings

Principal component analysis for wood, bark, compression wood, knot and fungi

bark

wood

fungi

compression wood

knot



Meeting 19/Jan/2016

Scores Loadings

Principal component analysis for wood, bark, compression wood, knot and fungi

bark

wood

fungi

compression wood

knot



Meeting 19/Jan/2016

Scores Loadings

Principal component analysis for wood, bark, compression wood, knot, fungi and resin/ resin pockets

bark

wood

fungi

compression wood

knot

resin

resin pocket



Meeting 19/Jan/2016

Scores Loadings

Principal component analysis for wood, bark, compression wood, knot, fungi and resin pockets

bark

wood

fungi

compression wood

knot

resin pocket



Meeting 19/Jan/2016

Scores Loadings

Principal component analysis for wood, bark, compression wood, knot, fungi and resin pockets

bark

wood

fungi

compression wood

resin pocket


knot

Task 4.3 – Hyperspectral imaging of 23 logs – example resin pockets intensity slabs, final explorations ongoing

Brussels3/jul/2015

1190 nm 1377 nm

Task 4.3 Status of the sensor & model development & implementation (D 4.09)

NIR measurements of BOKU samples with MicroNIR

Prototype of sensor arm

HSI measurements of BOKU samples - Hamamatsu

Pototype of LabView software

Focus lenses mounted on Hamatsu sensors

Integration of sensors, soft- & hardware, models Model development & quality index

Implementation of full system on sensor arm withhard- and software

Unt

ilFe

brua

ry/M

arch

D4.0

9 in

Apr

il

NIR-Spectrocopic measurementsScientific publication in prep.Principal component analysis for wood and resin (resin pockets)

Scores Loadings

Meeting 19/Jan/2016

Böhm, Zitek et al., in prep, Assessing resin pockets on freshly cut wood logs of spruce by NIR and hyperspectral imaging, European Journal of Wood and Wood Products

Mid-term Review 2/Jul/15

Thank you for your attention!

The objectives of this task is to optimize testing procedures and prediction models for characterization of wood along the harvesting chain, using acoustic measurements (i.e. stress-wave tests).

A part of the activity will be dedicated to the definition of optimal procedures for the characterization of peculiar high-value assortments, typically produced in mountainous sites, such as resonance wood.

Task Leader: CNRTask Participants: Greifenberg, Compolab

WP4: T 4.4 Data mining and model integration of log/biomass quality indicators from stress-wave (SW) measurements, for the determination of the “SW quality index”

Objectives

WP4: T 4.4 Deliverables

D4.05) Establishing acoustic-based measurement protocol: This deliverable contains a report and protocol for the acoustic-based measurement procedureStarting Date: August 2014 - Delivery Date: December 2014

D4.10) Estimation of log quality by acoustic methods: Numerical procedure for determination of “SW quality index” on the base of optimized acoustic velocity conversion models.Starting Date: January 2015 - Delivery Date: August 2015

Estimated person Month= 6.00

T4.4: Data mining and model integration of log/biomass quality indicators from stress-wave











D04.05

D04.10

CNR

CNR


D01.04


CNR:

the resources planned: 5.5 M/Mthe resources utilized:PROBLEMS: delay with access to sensors (arrived January 2016), change of Task LeaderSOLUTIONS: intensify efforts, change in staff involved

draft: Jan 2016

draft: December 2014

accepted: July 2015

D: 4.5 Establishing acoustic-based measurement

protocol: hardware

Hardware design has been made by COMPOLAB in collaboration with CNR

Two approaches for measuring stress waves on the processor are considered:• ToF (time of flight)• FV (free vibrations)


protocol


protocol

Time of Flight in SLOPE

l1 l2

t0

t1

t2

01

110 tt

lv−

=−

02

2120 tt

llv−+

=−

12

221 tt

lv−

=−


protocol: ToF challenges

•Preliminary tests highlighted great problem with coupling of accelerometers and wood, especially due to bark•Wet wood attenuates a lot stress wave – hardly measurable, especially with ultrasound…•Several properties of log/wood are not known during test (such as MC, density)•What does the value of velocity means? (regarding quality)

Special design of hardware on the processor head

The QI is (may be) computed after processing of log

Experimental campaign is foreseen & self learning system on the base of historic data


protocol

Free vibrations

if:f1 = f2 - machine vibrations

f3 <> f1 - free vibrations of log,fundamental frequency

D1

l

D2

time

time

frequency

f2 f3

FFT

f1

frequencyFFT


protocol: FV challenges

•Laser displacement sensor’s spot is absorbed by rough surface •Are we measuring free vibrations of log or processor head?•What is the noise of signal?•Several properties of log/wood are not known during test (such as MC, density, diameters, length)•What does the value of frequency means? (regarding quality)

Special sensor with enlarged spot size (Keyence LK-G87)

The QI is (may be) computed after processing of log and related later by RFID identificationExperimental campaign is foreseen & self learning system on the base of historic data

Compensation of LDS results with additional acclerometer

Conclusions

Many factors influence SW propagation in wood.

Parameters measured with the other NDT methods will be incorporated in the SW prediction models

Multiple linear regression analysis will be implemented for the definition of the importance of the different parameters (regression t-values) for the model.

The further development of Task 4.4 is based on the implementation of the lab scanner (i.e. purchase of sensors)

For the implementation of the methodology in the real case scenario, some practical issues (e.g. coupling-decoupling of sensors, etc.) have to be considered in combination with activity of Task 3.4

Sensors arrived: work will be done… and is ongoing

Task 4.5: cutting process quality indexObjectives

The goals of this task are:• to develop a novel automatic system for measuring of the cutting resistance of wood processed during harvesting• to use this information for the determination of log/biomass quality index

Task 4.5: Cutting Process (CP) for the determination of

log/biomass “CP quality index”

Task Leader: CNRTask Partecipants: Compolab

Starting : October 2014Ending: January 2016Estimated person-month = 4.00 (CNR) + 2.00 (Compolab)

CNR : will coordinate the research necessary, develop the knowledge base linking process and wood properties, recommend the proper sensor, develop software tools for computation of the CP quality index

Compolab: will provide expertise in regard to sensor selection and integration with the processor head + extensive testing of the prototype

Task 4.5: cutting process quality index

Deliverables

D.4.06 Establishing cutting power measurement protocolReport: This deliverable will contain a report and recommended protocol for collection of data chainsaw and delimbing cutting process.

Delivery Date: January 2015 (M.13) DONE

D.4.11 Estimation of log quality by cutting power analysisPrototype: Numerical procedure for determination of “CP quality index” on the base of cutting processes monitoring

Delivery Date: January 2016 (M.25)

T4.5: Evaluation of cutting process (CP) for the determination of log/biomass CP quality index


CNR:



D04.11

D04.06

CNR

CNR



CNR:


CNR:


CNR:


the resources planned: 6.0 M/Mthe resources utilized:PROBLEMS: delay with access to sensors (arrived January 2016)SOLUTIONS: intensify work, close collaboration with COMPOLAB

31.01.2016

draft: January 2014

accepted: July 2015

working time of the cutting tools (knifes and chain): estimation of the tool wear and correction of the cutting forces

position of the saw bar while cross-cutting: monitoring of the cutting progress correction factors related to the determination of the cutting forces and material

characteristics

log diameter (combined with position of the saw bar): determination of the cutting length at each moment of the cross-cutting

position of the main hydraulic actuator while cutting-out branches: monitoring of the de-limbing progress determination/mapping of the detailed knot position

Task 4.5: cutting process quality indexother sources of information

sensor type sim

plic

ity

relia

bilit

y

info

rmat

ion

qual

ity

easy

inte

rpre

tatio

n

low

cos

t

suita

ble

for

SLO

PE

labo

rato

ry te

sts

suita

ble

for

SLO

PE

in-f

ield

app

licat

ion

load cell

strain gauge

electric multimeter

oil pressure

oil flow

AE

microphone

Task 4.5: cutting process quality indexcomparison of sensors

Task 4.5: cutting process quality indexworking plan

activity responsible status (end of task)

Assemble sensors and controllers in lab CNR Ongoing (Feb 2016)

Design solutions for sensors placement COMPOLAB Ongoing (Feb 2016)

Report from lab measurements CNR Ongoing (Mar 2016)

Installation of sensor on processor head COMPOLAB (Mar 2016)

Testing of sensors in the shop COMPOLAB (Mar 2016)

Implementation of the software for QI CNR (April 2016)

Final adjustments + callibrations CNR + COM (May 2016)

Processor ready for pilot: June 2016

hydraulic pressure sensors , hydraulic flow sensor , termometer , linear gauge

Task 4.5: cutting process quality indexschematic of the log cross-cutting system of the ARBRO1000

Task 4.5: cutting process quality indexcross-cutting with the chain saw

Hydraulic flow (l/min)

Oil pressure (MPa)

Oil temperature (°C)

Position of the saw (mm)+Total working time of tool (min)Log diameter (mm)

time of one sawing stroke/cycle

cutting resistance log diameter quality Index

“easy” “small” “low” (0,2)

“easy” “small” “very low” (0,0)

“difficult” “small” “very high” (1,0)

“difficult” “big” “high” (0,8)

hydraulic pressure sensor , load cell

Task 4.5: cutting process quality indexschematic of the instrumented de-branching system of the ARBRO1000

Task 4.5: cutting process quality indexde-branching

Load cell#1 (N)

Load cell#2 (N)

Oil pressure (MPa)

Oil temperature (°C)

Position of the feed piston (mm)+Total working time of tool (min)

time of one debranching stroke/cycle

map of knots

CF quality index#2


time of one debranching stroke/cycle


map of knots – displayed for operator

CF quality index#2

two quality indexes (numbers in the range from 0 to 1) associated to wood/log properties are determined:

CP quality index #1: reflects the estimation of the “wood density” as related to the cutting resistance during cross-cutting of log by chain saw. The quality index #1 value is unique for the whole log.

CP quality index #1 = f(wood moisture content, tool wear, cutting speed, feed speed, log diameter, ellipsoid shape, presence of defects)

CP quality index #2: reflects the “brancheness” of the log along its length and is estimated by means of signals associated with cutting out branches. The quality index #2 is spatially reolved.

CP quality index #2 = f(hydraulic pressure changes along the log length, changes of cutting forces in time, number of AE events or sound pressure level)

Task 4.5: cutting process quality indexalgorithms for data mining

Task 4.5: cutting process quality indexChallenges

Important delay with prototype developing: the equipment just now ready for testing

How to physically install sensors on the processor?

How reliable will be measurement of cutting forces in forest?

What is an effect of tool wear?

How to link cutting force (wood density) with recent quality sorting rules?

Delimbing or debarkining?

Thank you very much

Task 4.6: Implementation of the log/biomass grading system

Task Leader: CNRTask Participants: GRAPHITECH, COMPOLAB ,MHG, BOKU, GRE, TRE

Starting : June 2014Ending: July 2016Estimated person-month = 1.50 (GRAPHITECH) + 2.0 (CNR) + 1.00 (COMPOLAB) + 1.00 (MHG) + 1.00 (BOKU), 0.50 (GRE) + 1.00 (TRE)

CNR: will coordinate the research necessary, develop the software tools (expert systems) and integrate all available information for quality gradingTRE, GRE, COMPOLAB: incorporate material parameters from the multisource data extracted along the harvesting chainGRAPHITECH: integration with the classification rules for commercial assortments, linkage with the database of market prices for woody commoditiesMHG: propagate information about material characteristics along the value chain (tracking) and record/forward this information through the cloud database BOKU: validation of the grading system

Task 4.6: Implementation of the grading system

Objectives

The goals of this task are:• to develop reliable models for predicting the grade (quality class) of the harvested log/biomass.• to provide objective/automatic tools enabling optimization of the resources (proper log for proper use)• to contribute for the harmonization of the current grading practice and classification rules

• provide more (value) wood from less trees


Deliverables

D.4.01 Existing grading rules for log/biomassReport: This deliverable will contain a report on existing log/biomass grading criteria and criteria gap analyses

Delivery Date: October 2014 (M.10) DONE

D.4.12 Implementation and calibration of prediction models for log/biomass quality classes and report on the validation procedurePrototype: This deliverable will contain a report on the validation procedure, and results of the quality class prediction models, and integration in the SLOPE cloud data base

Delivery Date: June 2016 (M.30)

T4.6: Implementation of the log/biomass grading system


CNR:







CNR, BOK:




D04.01

CNR

D04.12

CNR



the resources planned: 8.0 M/Mthe resources utilized:PROBLEMS: Delay related to other tasks within WP4SOLUTIONS: intensify efforts, implement ready theoretical solutions developed up-to-data

31.06.2016

draft: October 2014

accepted: July 2015


The concept (logic)

3D quality index (WP 4.1)

NIR quality index (WP 4.2)

HI quality index (WP 4.3)

SW quality index (WP 4.4)

CP quality index (WP 4.5)

Data from harvester

Other available info

Quality class

Threshold values and variability models of

properties will be defined for the

different end-uses (i.e. wood processing industries, bioenergy

production).

(WP5)


implementation#1: Quality index concept

Each index can be between:0 – bad, not suitable, low, , …

and1 – good, proper, perfect, appreciated, , …

Computed for: Suitability modeled separately for different destination fields:

resonance wood, structural timber, pulp/paper, chemical conversion…

Presence of various defects, such as: Rotten wood, knottiness, compression wood, eccentric pith…

Compatibility with standard quality classes

For each task of WP4 series of quality indexes will be computed as default


implementation#2: Quality index computation

Set of experimental sampleswith characteristics representingpoor quality QI = “0”

Set of experimental sampleswith characteristics representingsuperb quality QI = “1”

PLS models for prediction

validation of models

implementation of modelsfor routine data processing

never ending tuning process


implementation#3: summary of QI + weights

weight for each quality aspect

rangeconstruct.

woodbiomass

/fuel pulp plywood class A class DT4.2 moisture 0 - 1 0,2 1

density 0 - 1 1 1 1 1 1carbohydrate content 0 - 1 1lignin content 0 - 1 1 1calorific value 0 - 1 1rotten wood progress 0 - 1 -100 1 1 1early/late wood ratio 0 - 1 0,2 1width of sapwood 0 - 1 0,1pith eccentricity 0 - 1 0,5 0,8 1width of bark 0 - 1 0,2 1 1 1presence of reaction wood 0 or 1 1 1 1 1presence of resin 0 or 1 0,2 1 1presence of rot 0 or 1 -100 0,7 1presence of bark 0 or 1 -0,5 0,2 1 1presence of contamination –soil 0 or 1 -0,1 -0,1presence of contamination – oil 0 or 1 1

T4.3 ovalness 0 - 1 1 2 1ratio of knot area 0 - 1 0,2 1knot count 0 - 1 0,2 1

T4.4 velocity 0 - 1 1 0,8 1homogenity velocity 0 - 1 1 1 1density 0 - 1 1 0,8 1elasticity 0 - 1 1 0,3 1suitability for pales 1

T4.5 knotines 0 - 1 0,5 0,6 1knots size 0 - 1 2 0,6 1knot spatial distribution 0 - 1 1 1 1log density 0 - 1 1 1 1 1easy for processing 0 - 1 1 1 1 1


implementation#4: maths behind

For each log:

∑∑ ⋅

=i

iimarket w

QIwQ

where:Qmarket – log quality for specific use/marketwi – weight of quality indexQIi – quality index assessed by sensor

)( ii wtresholdQI >∀

where:treshold(wi) – minumum value of QIi

AND/OR*

* - depending on application


implementation#4: quality map

Map of knots

Map of quality

QIT4.4

QIT4.1

QIT4.2

QIT4.3

QIT4.5


The concept (diagram)Measure 3D shape of

several trees

Measure NIR spectra of tree X in forest

Extract 3D shape of tree X

Compute 3D quality in-dexes for log X.1 … X.n

Measure NIR spectra of tree X on processor

Measure NIR spectra of tree X on the pale

Compute NIR quality in-dex for tree X

Compute NIR quality in-dexes for log X.1 … X.n


Data base for harvest data

Data base for Forest In-formation System

Determine quality grade for log X.1 … X.n

T4.1

T4.2

Measure hyperspectral image of tree X in forest

Measure cross section image of log X.1 … X.n


Compute HI quality index for tree X

Compute HI quality in-dexes for log X.1 … X.n


T4.3

Measure stress waves on tree X in forest

Measure stress waves of tree X on processor

Measure stress waves of log X.1 …X.n on the pale

Compute SW quality in-dex for tree X

Compute SW quality in-dexes for log X.1 … X.n


T4.4

Measure delimbing force on log X.1 … X.n

Measure cross-cutting force on log X.1 … X.n

Compute CF quality in-dexes for tree X

Compute CF quality in-dexes for log X.1 … X.n

T4.5


The concept (diagram)#1

Measure 3D shape of several trees

Measure NIR spectra of tree X in forest

Extract 3D shape of tree X


Measure NIR spectra of tree X on processor






Determine quality for log X.1 …

T4.1

T4.2

Measure hyperspectral image of tree X in forest

Measure cross section image of log X.1 … X.n





T4.3



Measure stress waves on tree X in forest

Measure stress waves of tree X on processor

Measure stress waves of log X.1 …X.n on the pale

Compute SW quality in-dex for tree X



T4.4

Measure delimbing force on log X.1 … X.n

Measure cross-cutting force on log X.1 … X.n

Compute CF quality in-dexes for tree X

Compute CF quality in-dexes for log X.1 … X.n

T4.5








Data base for Forest In-formation System

Determine quality grade for log X.1 … X.n



Compute HI quality in


data flow & in-field hardware

NI CompactRio master

Database

NI CompactRio client

FRID

wei

ght

fuel

???

Data storageCP N

IR HI SW

cam

era

kine

ct


Challenges

What sensors set is optimal (provide usable/reliable information)?

How to merge various types of indexes/properties?

Can the novel system be accepted by “conservative” forest (and wood transformation) industry?

How the SLOPE quality grading will be related to established classes?

the final answer possible only after demonstrations

Thank you very much

Education

3rd Technical Meeting - WP4