WO2006057566A1

WO2006057566A1 - Tree stem or log appraising apparatus

Info

Publication number: WO2006057566A1
Application number: PCT/NZ2005/000304
Authority: WO
Inventors: Jeffrey Robert Parker; Gregory John Searles
Original assignee: Fibre-Gen Instruments Limited
Priority date: 2004-11-24
Filing date: 2005-11-17
Publication date: 2006-06-01
Also published as: NZ536818A; AU2005203427A1; AU2008207613A1; AU2008207613B2

Abstract

The present invention relates to methods and apparatus for or useful in determining the velocity V, V2, or some function of V of an impact induced longitudinal vibration response in a stem, log or board. This involves providing a conveyor to serially advance stems, logs or boards in a conveying direction so as to present an end of each stem, log or board to an impact zone where the end is impacted by an impact head mounted on a pivotable support arm. A sensor is provided to detect the spectrum resulting from the impact and a processor responsive to the sensor determines velocity V, V2 or some function of V.

Description

TREE STEM OR LOG APPRAISING APPARATUS

FIELD OF INVENTION

The present invention relates to apparatus and methods useful for determining the speed of travel within tree stems or logs of the speed of sound or non audible like travel within the tree stems or logs resultant from an impact or striking of the log or otherwise inducing such travel in the stem or log.

BACKGROUND

In our New Zealand Patent Specification 333434 we disclose a log cutting procedure reliant upon an assessment of tree stems for longer logs as a consequence of a manual impacting of an end of a tree stem or log and detecting by a sensor the spectrum resulting from the impact. Our New Zealand Patent Specification 515734 shows manual apparatus separating a sensor head (with a compliant mounting of a piezoaccelerometer) from its powered processor unit.

The present invention recognises that there is an advantage in being able to handle without careful operator intervention the tree stems passing to a merchandiser for breakdown and/or logs being fed into a saw mill and it is to automated systems having a requisite accuracy and performance capability that various aspects of the present invention are directed whether in respect of the apparatus itself, retro-fitted apparatus (i.e. using the existing conveyor), or as an overall procedure. This and other objects will become apparent in the following description of the invention.

BRIEF DESCRIPTION OF THE INVENTION

In a first aspect the present invention provides an apparatus for or useful in determining the velocity V, V² or some function of V, of an impact induced longitudinal vibration response in a stem, log or board of length L comprising: a conveyor to serially advance stems, logs or boards in a conveying direction so as to serially present an end of each stem, log or board to an impact zone, an impact head; an impact arm or other means ("arm") carrying the impact head; a pivoted, articulated, or other mounted support of the arm whereby the arm and the impact head can cycle from a cycle commencement condition; driving means capable under signal actuation of assisting and/or allowing:

(i) the commencement of an impact causing swing from the cycle commencement condition of the arm and its distally carried impact head, and

(ii) the retrieval of the arm and its distally carried impact head to the cycle commencement condition; an encoder to ensure the end to be impacted by the impact head of each stem or log is, in turn, impacted by the impact head during a said cycle; a sensor to detect the spectrum or part thereof of the longitudinal vibration response caused by the impact of the impact head with each stem, log or board; and a processor responsive to the sensor for determining the velocity V, V² or some function of V of the impact induced longitudinal vibration response; wherein the apparatus is operable to obtain V, V² or some function of V for each stem, log or board on the conveyor.

Preferably the sensor is mounted on a pivoted, articulated or other sensor arm. Alternatively the sensor is mounted on the impact arm. Desirably, said apparatus is capable reproducibly of being caused and/or allowed to move to the or an impact zone to encounter an end of each stem, log or board to provide the impact required.

Preferably the swing is a downward swing.

Preferably the encoder registers the presence of the stem, log or board of a predetermined point along the conveying direction and releases the impact arm at a predetermined time thereafter.

Preferably the encoder registers the presence of the stem, log or board at a predetermined point along the conveying direction and releases the sensor arm and/or impact arm at a predetermined time thereafter. Desirably the stem, log or board lies longitudinally or substantially longitudinally of the conveying direction and the advance of the stems, logs or boards is continuous.

Desirably the sensor is moveable between a first position proximate to the or an impact zone and a second position remote from the or an impact zone; and the sensor is configured to be repeatably moved between said first position immediately prior to or at the time of the impact and said second position after the impact.

Preferably the driving means is a pneumatic system. Preferably said driving means assists the commencement of a downward swing only of the arm and by the time the impact head is to encounter a stem, log or board end the arm is swinging with its arm without pneumatic acceleration. The pneumatic means may comprise a double acting pneumatic piston which is for the purpose of (i) caused and/or allowed to move towards one limit of its movement without pneumatic assistance other than any that might be required to overcome resistance to travel.

Optionally (but necessary to derive the requisite data) the apparatus includes a sensor to detect speed by time of passage of the stem, log or board or to detect the spectrum or part thereof.

Preferably the detected spectrum is of frequencies in the range from 150 to 2000 Hz.

Preferably said impact head is vibrationally isolated and/or dampened from said arm (at least for frequencies of consequence). Preferably in addition said arm may be dampened from the frame or other structure from which it is pivoted, articulated or otherwise mounted.

Desirably the apparatus includes an encoder to regulate the swing down and/or uplift of the arm to cause the impact head to encounter an end of each and every said stem, log or board.

Described is a striking apparatus for use in the assessment of tree stems, logs or boards, said apparatus comprising: an impact head; an arm or other means ("arm") carrying the impact head; a pivoted, articulated, or other mounted support of the arm whereby it can be allowed and/or caused to swing down under and/or up against the influence of gravity to distally move the impact head therewith; and pneumatic means capable under signal actuation of assisting and/or allowing (i) the commencement of an impact causing swing from a cycle commencement condition of the arm and its distally carried impact head, and

(ii) the retrieval of the arm and its distally carried impact head to its cycle commencement condition. In yet another aspect the present invention provides an apparatus to determine the velocity V, or some function thereof, of an impact induced travel or impact induced longitudinal vibrational response in a stem, log or board of length L, said apparatus comprising: a conveyor to advance the stems, logs or boards longitudinal of their longitudinal axes serially to and past an impact zone, striking apparatus having: an impact head; an arm or other means ("arm") carrying the impact head; a pivoted, articulated, or other mounted support of the arm whereby it can be allowed and/or caused to swing down under and/or up against the influence of gravity to distally move the impact head therewith; and pneumatic means capable under signal actuation of assisting and/or allowing

(i) the commencement of a downward swing from a cycle commencement condition of the arm and its distally carried impact head, and (ii) the retrieval of the arm and its distally carried impact head to its cycle commencement condition. a sensor mounted on a pivoted, articulated or other sensor arm to detect the spectrum or part thereof of the longitudinal response caused by the impact of the impact head with each stem, log or board; and a processor responsive to the sensor for determining the velocity V or some function thereof.

Preferably the striking apparatus is caused to swing down and/or uplift so as to encounter each and every (preferably advance) end of a said stem, log or board.

In another aspect the present invention provides a method of treating tree stems passing to a merchandiser for breakdown and/or logs being fed into a saw mill which comprises, on a conveyor, endwise serially feeding the stems and/or logs serially to an impact zone where a drop impact head, on a swing or the like arm without operator intervention, is caused, in turn, and substantially reproducibly, to encounter the leading end of the or each stem and/or log, and in turn, without operator intervention, sensing for each stem or log the outcome of the impact and providing from that outcome an indicator or instruction output reliant on velocity V, V² or some function of V, V being a measure or estimate of the longitudinal velocity of the impact caused longitudinal response within each stem or log.

In a further aspect the present invention provides a method of merchandiser breakdown or saw milling which relies on treating tree stems and/or logs respectively according to a method or an apparatus of the present invention. The invention also consists in the various outcomes.

In yet another aspect the present invention provides an apparatus or a method substantially as herein described with reference to any Example thereof and with or without reference to the accompanying drawings. In a further aspect the present invention provides an apparatus for or useful in determining the velocity V, or some function thereof, of an impact induced travel in a stem or log of length L, (preferably in association with or for association with a conveyor to advance the stems and/or logs longitudinal of their longitudinal axes to and past an impact zone), said apparatus being an impact device capable reproducibly of being caused and/or allowed to move to the or an impact zone to encounter an end of each stem and/or log to provide the impact required, said device having an impact head, an arm or other means ("arm") carrying the impact head, a pivoted, articulated, or other mounted support of the arm whereby it can be allowed and/or caused to swing down under and/or up against the influence of gravity to distally move the impact head therewith, and pneumatic means capable under signal actuation of (i) assisting the commencement of a downward swing from a cycle commencement condition of the arm and its distally carried impact head, and (ii) retrieving the arm and its distally carried impact head to its cycle commencement condition.

DEFINITIONS As used herein, the following terms have the folio wing specified meanings:

As used herein "(s)" following a noun means the plural and/or singular forms of the noun.

The term "comprising" means "consisting at least in part of, that is to say when interpreting independent paragraphs including that term, the features prefaced by that term in each paragraph will need to be present buy other features can also be present.

As used herein the term "adapted to derive the velocity V (or some function thereof)" refers to any measure which can, if desired, in conjunction with the length L of the board and/or the density p (irrespective of whether or not the machine has provision for measuring length L or density p) can be used (for example as V² or MOE (modulus of elasticity)) as an indicator of relative stiffness. The velocity V can instead be simply an assessment of acoustic or stress wave travel velocity substantially longitudinally of the board irrespective of the downstream purpose to which it might be put. The velocity V can be derived by time of travel means or by Fast Fourier Transformation ("FFT") as discussed below of a sensor detection of the spectrum of the longitudinal response or part thereof (e.g. that retrieved and/or comb filtered.

Fast Fourier Transform is a recognised means of calculating the frequencies present in an amplitude vs time resonant signal. From practical experiments the minimum and maximum possible velocities of longitudinal vibration for particular species have been found. With the length of the log known, the maximum and minimum fundamental frequencies can be derived. A search is done within these bounds, and any high energy frequencies are tested, by developing a comb of multiples of these frequencies to find if they have higher harmonics.

For example if the peak to be tested has a frequency of 300 Hz, a comb consisting of 300, 600, 900, Hz is constructed, and the energy measured within that comb by adding the power at the comb frequencies. For short logs, a deviation of a few percent is allowed, i.e. energy is considered to be part of the comb if it falls within a predetermined band about the expected centre, to take account of the effects described earlier which are encountered in practice.

This is repeated for the other frequencies found within the initial range. The preferred result is the peak whose comb accounts for the greatest quantity of spectrum power.

Preferred forms of the present invention will now be described with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

Figures IA through IH shows a sequence of operation when viewed in plan showing the relationship of the log direction and logs #1 and #2 in respect of the striking apparatus. The sequence of operation and the mode of operation is substantially as depicted in such drawings, the full textual content of which is here included by way of reference,

Figures 2 through 5 show in more detail the apparatus of the present invention, Figure 6 illustrates a mechanism to deploy and retrieve the impact head,

Figure 7 illustrates the impact head of the present invention approaching and striking a log end, Figure 8 depicts one embodiment of the present invention showing the impact head and sensor mounted separate arms,

Figure 9 illustrates the embodiment depicted in Figure 8 in which the separately mounted impact head and sensor are deployed and located in the path of an approaching log end, Figure 10 is an illustration of the operation of a comb filter on a power spectrum,

Figure 11 is a histogram showing the signal to noise ration for different microphone set ups, and

Figure 12 illustrates how whole stem velocity information, combined with a knowledge of typical velocity profiles along a stem, can predict velocities within logs subsequently cut from the stem.

DETAILED DESCRIPTION OF THE INVENTION

The present invention without a need for human intervention once in operation provides an endwise feed through of logs or stems with the advance ends being detected to initiate immediately, after a time or in conjunction with conveyor speed/time or conveyor travel, the movement of a retractable striker which is to encounter that advance end in a substantially reproducible manner.

Desirably the microphone is mounted in a manner allowing it to move to a position proximate to the impact zone before the impact head strikes the log and move out of position once the spectrum has been detected thereby avoiding the advancing log. This embodiment is particularly advantageous when measuring the spectrum of logs having greatly varying diameters. A static microphone would have to be mounted at least the radius of the largest log away from the impact zone to allow for clearance of the log. At such a distance and in a noisy environment the signal to noise ratio of the spectrum detected is reduced and the accuracy of the measurement can be affected. By using a microphone which moves into a position proximate the impact zone and then moves out of the way of the advancing log an improved signal to noise ratio can be obtained as illustrated by Figure 11. One technique for providing such a moveable microphone is to mount the microphone on a swinging sensor arm as illustrated by Figures 8 and 9. When the trailing end of a log has cleared the swing path of the sensor arm, the arm swings down bringing the microphone into the path of an advancing log. Then, once the leading edge of a log to be tested reaches a predetermined point, the impact arm carrying the impact head swings down and strikes the log. Once the resultant signal has been detected both the impact arm and sensor arm move out of the path of the oncoming log. This sequence of steps may then be repeated. An alternative embodiment locates the impact head and the microphone on the same arm, although such an arrangement has the disadvantage of vibrations in the arm caused by the impact affecting the detection of the spectrum. Therefore it such a configuration it is necessary to provide dampening means to protect the microphone from vibrations in the arm.

A typical layout is as shown in the plan view as shown in Figure 2 where the direction of log travel is as depicted, there being a log 3, a chain track 4, a microphone location 5, a through beam sensor 6 and the hammer 7 at rest (i.e. in a retracted condition).

The log 3 travels along the conveyor, breaks the laser beam of laser 6, and after a certain delay, the hammer is driven down by a pneumatic ram from its rest position for a small period of time and is then allowed to fall freely until it strikes the end face of the log in its bottom dead centre position. The pneumatic ram is used to initiate the fall to ensure a consistent time for the hammer head to go from rest to bottom dead centre position. It is switched to gravity feed to ensure that the hit is not too hard and possibly ruining the sound signal by crushing the fibres of the log, sending a large shock (and therefore noise) through the entire structure and prolong the life of the unit. There is very little drag on the hammer due to the pneumatic system. During the downward stroke of the hammer, the ram initially starts the movement by activating one side of the piston. Once this has been done, both sides of the piston are left open to the atmosphere so that there is no pressure or vacuum that may form on any side of the piston and restrict its movement. Pressure is then applied on the opposite side of the piston to lift the hammer back up.

The reason for the rubber coupling in the middle of the hammer handle (as hereinafter described) is to remove the possibility of oscillations or noise in the frequency range we are looking at (150 - 2000Hz). The coupling will reduce any oscillations in the hammer handle to well below this.

It is important (or at least preferable) to hit in the bottom dead centre position as that is when the head of the hammer preferably is flush with the end of the log, ensuring efficient excitation of the log. The length is determined by counting the number of encoder pulses that the laser was blocked for, then multiplying this number by the number of mm/pulse.

Figure 3 shows a preferred hammer inlet in accordance with the present invention with reference to the following reference numerals.

6 - hammer head 7 - lead in plate to ensure that the end of the log does not catch in the 90 degree grove between the head and the handle

8 - hydraulic hose fittings

9 - flexible rubber hose is used to couple the hammer head to the rest of the hammer handle 10 - the hammer handle is hollowed out to concentrate the weight of the hammer at the head end

11 - two grease nipples are attached to the handle to allow grease to be pumped in at high pressure to increase the hammer life.

We have ensured that when hitting the log, no other noise is produced in either the hammer handle ("arm") or the supporting structure. Normal sledge hammer handles were either too rigid to withstand the high shock loading or weak enough to vibrate at frequencies within our target frequency range. Therefore we came up with a completely new hammer design to overcome all these problems as in Figure 3.

Figures 4, 5 and 6 shows plan end and side views in which 12 is the log and 13 the microphone location. Figure 5 shows as the area 14 the smallest log (for example 10 inches in diameter) and the largest log (15) (for example 32 inches in diameter) that the apparatus is adapted to cater for.

The apparatus need not be, but optionally can be, of a kind with processing algorithmic (e.g. FFT) and/or comb filtering approaches as disclosed in New Zealand Patent No. 515734.

A dominating consideration of the processor is the high rate of decay of the signal coming from the wood. Measurements of the attenuation of acoustic signals in wet wood show that the signal can fall by 6OdB in 0.1s, in an approximately exponential fashion. The process of Fourier analysis in this application can be thought of as a simple way of averaging the echo times of many reflections, since the fundamental frequency f₀ found by Fourier analysis is the inverse of the echo time T. The reception of many echoes leads to an accurate average. It is for this reason that resonance-type instruments produce more consistent answers than single transit stress-wave timers. However the echo time in a long stem is typically 10ms. To detect 20 echoes necessitates detecting signal for 200ms, and clearly by this time the amplitude will be very low if the attenuation is 60dB/100ms. To obtain useful signals for a duration of 0.1 to 0.4s₅ the gain of the analogue amplifier is made to increase at a constant exponential rate, for example 20 to 6OdB, over the course of the event to partially offset the natural attenuation. Amplifier offset voltages must be carefully controlled with such a strategy to prevent dc contamination of the final spectrum. In conjunction with this, high resolution A/D converters, typically 14 bit, are used so that useful resolution can still be obtained where the signal has fallen into the microvolt range (but is still above the noise background). If the initial acoustic signal is converted to a 3 V amplitude signal, the level 100ms after this might be 3mV, which would give some resolution on a 14 bit converter set to 3 V scale, since the least significant bit is 0.19mvolt. However, signals beyond the 100ms time frame would quickly fail to be digitized.

The provision of time-dependent gain is vital to extend the period over which signals can be usefully digitized. 20 dB of gain over the 100 ms described above would raise the signal at that time to 30mv, enabling the time of useful digitization to be considerably extended.

The processor records a number of digitized values over an ensuing time. Typically, 2048 readings will be taken over 400 ms, following which the analogue amplifier and A/D converter are turned off. The data are then Fourier transformed following appropriate windowing and filtering. The particular data record described combination will yield a maximum frequency of 2.5kHz with a resolution of 2.5Hz, which suits forest applications, but could be changed to suit other needs.

The power spectrum is then analysed by the processor using algorithms discussed in the next section to extract a fundamental resonance fo. This can consist of a single value for velocity, (assuming a prior log length has been entered into the unit), using the formula

V = 2 f₀ L where L is the length, or the value can be converted to a speed class.

It is well known that exciting a beam or log of wood into longitudinal oscillation produces a disturbance which can be Fourier analysed into a series which is harmonic, and in which the speed of sound in the wood is given by

V = 2Lf₀

V is the speed of longitudinal compressional motions along the member, and since the lateral boundaries are stress free, is given by the well known relation

V² = E/p where E is Young's modulus, and p the material density.

In samples of regular cross section, particularly where these are slender, higher resonances are closely harmonically related to the fundamental. Extraction of the modulus using the two equations above is simple since the fundamental is easily identified. The number of harmonics detected depends on the frequency characteristics of the exciting impulse. Wet wood is soft. Typically a hammer is arrested in a time of the order of a millisecond and the spectra cannot be expected to contain harmonics greatly in excess of the inverse of this time, i.e. greatly above 1 kHz. In practice, there is a variety of circumstances where this picture requires modification to extract reliable values of the modulus. In practice, where a sample is wet few harmonics will be detected in the log compared with the sawn wood. A decision on which frequency should be identified as the fundamental may be less clear for the log. We have found that this can be exacerbated by the presence of unwanted noise spikes in the spectrum, or unwanted resonances arising from less than optimum hammer blows. Situations of poor spectra have been found to be inevitable in some physical locations, for example when obtaining spectra from the logs of cross-cut stems, when the log faces are relatively inaccessible. In development work, it is possible to repeatedly take a spectrum until by chance it is "clean". In a production tool, a high success rate in analysis must be available, and a built-in indication of the confidence in the answer is desirable. However, some short logs, measured in isolation on bearers, still show a small but measurable departure from a harmonic series, usually with the higher harmonics at frequencies below what would be expected. This would lead to a difference of two in the predicted value of stiffness. All the foregoing situations must be allowed for in the analysis software.

Finite Element modelling of the eigenmodes of the logs has been carried out to gain an understanding of the factors involved in departures from harmonic series.

The results show that for a cylindrical log, the lowest resonance frequencies are closely harmonic. This remains true when the anisotropic elasticity of wood is included. The frequency of the fundamental mode is only slightly affected by the value chosen for Poisson's Ratio, which is fortunate since this parameter is ill-defined in wood. Further, no evidence was found that radial structure in logs, approximated by an inner core of low stiffness surrounded by a stiffer outer cylinder produced other than some average spectrum of the two; i.e. such internal structure is not responsible for anharmonic effects. At a frequency when the wavelength across the log approaches the wood diameter, the longitudinal frequencies become lower than expected i.e. a harmonic pull-down of the kind described earlier is seen. Due to the fact that the sound speed across the log is of the order of one tenth the longitudinal speed, this condition may be reached at what may be surprisingly low harmonic numbers in "fat" logs. Model results showed that ill-defined body resonances prevailed at higher frequencies. In other words, the spectra of short fat logs might be expected to show a small lowering of higher harmonics compared to the fundamental, but few harmonics will be seen. This roughly accords with our observational experience. The theory shows that for non-tapering logs the best indication of stiffness comes from the fundamental.

The situation for stems is different because of their taper. Taper is the only parameter found which causes the resonances following the fundamental to be sharply lowered in frequency. However, the modelling shows that it is the low harmonics which are raised above the value expected from the wood modulus, while the high harmonics still indicate stem stiffness. As with non-tapered logs, when the transverse wavelength of a resonance frequency approaches the stem diameter, the harmonic frequency tends to fall lower than expected. Because for stems, the frequency at which this is predicted to occur is high, the effect is unlikely to be seen and indeed we have not observed it.

Tapered-log modelling shows that it is the taper per wavelength which is important. The imbalance or asymmetry occurring in the oscillating mass and spring forces about each node in the log is the underlying cause of frequency shift. Thus the fundamental mode, where the stem is half a wavelength long, can be strongly affected. The taper per wavelength in the N harmonic is only 1/N of that in the fundamental. The higher harmonics are much less affected by the taper and yield the correct stiffness. Modelling shows, and our experience confirms, that to a reasonable approximation, if the fundamental resonance frequency is raised by a factor ke^" over its value expected on the basis of the stem length and stiffness, the N harmonic will be raised by a factor ke^" over its harmonic value. Resonances therefore fairly quickly reach their harmonic values. Other expressions which express the deviation of the overtones from a harmonic series can be derived. Assuming f^ is the frequency of the N^th member of the actual resonance series, and f₀ is the "true" fundamental, or lowest member of the series, from which the velocity and stiffness can be found, the lowest member fi coincides with fO if the log is slender and non- tapered, but there may be no resonance energy seen at f₀, for example with stem spectra.

This background of observation and modelling results provides the basis of the algorithms used to analyse spectra. While a velocity can be judged by an operator from a screen display of spectra, an automatic system needs to allow for noise peaks, non harmonic effects, and perhaps most confusing to an automatic process, missing spectral peaks which confuse the identification of a series.

The algorithm must reject occasional noise peaks in the spectrum, which means that as many as possible of the resonant peaks should be identified, since random noise spikes will not occur in harmonic ratios. It must allow for the fact that frequencies may be non-harmonic to a small extent in short logs and greatly so in stems and it should not require all members of a series to be present.

The identification system first considers only spectral signals above a threshold, for example those within 20% of the power of the largest spectral peak. It may be advantageous to smooth data in the frequency domain before doing this if signals are noisy to limit the number of peaks to be considered.

Given the length of a log and a likely range of sound speed, the possible range of frequencies for a fundamental is calculated and spectral peaks sought within that range. The search is done within velocity windows whose ranges are less than 2:1. Within such a window, the range of possible fundamental frequencies cannot overlap the consequent second harmonic range, and so allows fundamental and second harmonic to be distinguished. If no successful identification is ultimately made within this window, subsequent searches are made within modified velocity windows. This is generally not required. Most green P. radiata logs have velocities between 2.5 and 4km/s which fulfills the velocity criterion. For each potential candidate for a fundamental resonance, a filter comb is constructed. For example, if the peak to be tested had a frequency of 300 Hz, a comb consisting of 300, 600, 900, Hz is constructed, and the energy measured within that comb by adding the power at the comb frequencies. For short logs, a deviation of a few percent is allowed, i.e. energy is considered to be part of the comb if it falls within a predetermined band about the expected centre, to take account of the effects described earlier which are encountered in practice. The velocity, and modulus, is then calculated from the second harmonic by assuming that this is the frequency 2f₀.

This procedure is repeated for all peaks which are candidates for the fundamental within its allowed frequency range. The preferred identification is that spectral peak whose comb accounts for the greatest quantity of spectrum power. A numerical confidence measure which follows from this procedure is the ratio of the power accounted for in the peaks within the comb to the sum of power in other peaks plus the background noise level.

In the search to identify harmonic members, no power is considered in peaks which fall at frequencies which would lead to impossibly low velocities. The reason for this is that such peaks can be generated by moving the accelerometer head during the course of recording data. Nevertheless, their inclusion in the confidence measure gives operator warning that such an event might have happened.

It will be occasionally found, particularly with short "fat" logs, that only one resonance is seen. In that case, provided it produces a plausible velocity, it must be assumed to be the fundamental. The procedure is modified for stems where taper is important resulting in a grossly non-harmonic series. A range of fundamental frequencies is sought as before, but the comb generated is considerably modified. Because the procedure is more complex and suits the presence of many harmonics, it is only applied to logs above a preset length, for example 12m. If fo is as before the "true" fundamental from which the speed in the tapered log can be found and the modulus calculated, the exponential deviation from a harmonic series described earlier can be expressed as (fN - NfO)/fN = ke^'N

Here fN is the frequency of the N harmonic, and k is a constant between 0 and 1, which must be determined. Other expressions are possible. One such alternative expression has the form WNf₀ - 1 = (constant)/N²

When for example using the former expression, having identified one peak as a possible fundamental (i.e. N=I), for a given value of k, a value of f₀ is defined, and a comb of frequencies can then be generated at which the other harmonics should fall. The power falling within the comb is summed as before, and the procedure repeated with different values of k to find the optimum match for that presumed fundamental mode.

The comb filter process is illustrated in Figure 10, with reference to the analysis of a stem, using the exponential expression above to analyse the spectrum sketched in Figure 10 panel (c). This spectrum shows a noise floor, from which a genuine resonance sequence occurs near the frequencies f_l5 f₂, f₄ and fs, but the member f ₃ is missing and there is a noise pulse or unwanted resonance mode between ft and f₂. To test whether ft is indeed the first member of series, value f₀ is chosen, which defines the harmonic series nf₀ in Figure 10(a) and the value of k. This frequency f₀ is that which, together with the stem length, defines the true wave propagation speed sought. A series of displaced frequencies ft, f₂... which are the centre frequencies of the comb filter can now be generated from the exponential expression. Passbands for the filter are created by opening narrow windows about these centre frequencies, thus defining the comb filter shown in Figure 10(b). The power spectrum in Figure 10(c), minus a threshold representing the noise floor, is multiplied by the filter to yield the output of Figure 10(d), which is the spectral power falling in the windows of the comb. The sum of the power coming through the filter is a measure of how well the original harmonic series describes the actual spectrum. A range of values of f₀ and k are tested to find the combination which produces the best fit. Note that the noise spike between ft and f₂ is ignored, and the absence of the third overtone is merely regrettable, not catastrophic, in generating a fit.

This procedure will sometimes yield two values of k which generate equal summed powers. A second measure is therefore taken at each value of k to express how closely the comb is fitted. This is the sum of the deviations of each peak from its comb centre frequency. The choice is made on the basis of the most power and the best comb fit. The next candidate resonance for the fundamental is then tested, and classed as a better identification or not on the basis of both the resonance power accounted for, and the closeness of fit to the comb. With a fast processor, computation time is acceptably short.

In effect, a transformation is being done to best fit the given resonances to a harmonic set, and does not require all member of a series to be present. It could begin by generating a comb by assuming that a particular peak was the N^th harmonic and generating a comb from that. In fact, the algorithm does this, testing each peak in turn to be a particular harmonic of an assumed series, and finding the goodness-of-fit for each combination. This is useful since some stem signatures have an ill-defined fundamental frequency. The complexity of these procedures is frequently not needed because many resonance spectra have an obvious interpretation. Their need is in the general case, when a reliable answer is needed in a high percentage of cases from less than perfect data, and the data itself must be used to indicate to unskilled operators whether or not the answer is reliable.

Further, stem average velocities can be advantageously used to more intelligently break stems into logs. We have found that the velocity varies along stems in a broadly similar way and can be predicted.

It can be represented by a sum of a cubic expression involving the position along the stem and a constant term. With reference Figure 12, a constant term in the cubic can been adjusted by calculation so that the transit time derived by integrating the speeds from the cubic expression along an actual stem equals the time found from the averaged velocity V along the stem. For example, A x³ + B x² + C x + D where x is the distance along the stem and A, B, C and D are constants.

The curve drawn is the resulting prediction of speed along that stem. Also shown in Figure 3 as the stepped line are speeds subsequently measured in the sequence of logs made from that stem. Clearly in this example, a combination of reference information and stem- average measurement has enabled a considerable improvement to be made in velocity or stiffness estimation along the stem prior to making cuts.

A better non cubic predictive model might be of the form ^v/_Vτ = aL^b _M(l-LM)^c + d where a, b, c and d are constants, V is velocity of the log VT is velocity of the tree stem from which the log is to be derived, and L_M is the mid-point relative length of the log.

Data such as that in Figure 12 gives confidence in the comb filter technique. Generally, the transit time deduced for a stem based on a determination of f₀ agrees within about 1 to 2% with the sum of the transit times in each log cut from that stem, each of which has been analysed by a comb filter. This agreement is very satisfactory. A speed based on the lowest resonant frequency in a stem, and simple interpretations of the log spectra, could not approach such accuracy.

Stiffness measurement is a parameter which has had recent prominence, both in regard to log and timber stiffness and the implications it has for the basic constituent fibres of the wood. Measurement of stiffness using so-called stress wave timers, that is to say electronic instruments which detect the time of flight of a sonic impulse along or across a piece of wood have been in use for many years. While it is generally accepted that they measure a quantity indicative of mechanical stiffness, for forest use, they tend to be of marginal accuracy, and relatively insensitive (due their inherent broadband nature) and therefore difficult or impossible to apply to long logs. Their fatal flaw is that they require double ended operation, i.e. detectors need to be placed at each end of the log under test. Logistically, this is unacceptable in forest use.

Where in the foregoing description reference has been made to elements or integers having known equivalents, then such equivalents are included as if they were individually set forth.

Although the invention has been described by way of example and with reference to particular embodiments, it is to be understood that modifications and/or improvements may be made without departing from the scope or spirit of the invention.

Claims

WHAT WE CLAIM IS:

1. Apparatus for or useful in determining the velocity V, V² or some function of V, of an impact induced longitudinal vibration response in a stem, log or board of length L comprising: a conveyor to serially advance stems, logs or boards in a conveying direction so as to serially present an end of each stem, log or board to an impact zone, an impact head; an impact arm or other means ("arm") carrying the impact head; a pivoted, articulated, or other mounted support of the arm whereby the arm and the impact head can cycle from a cycle commencement condition; driving means capable under signal actuation of assisting and/or allowing:

(i) the commencement of an impact causing swing from the cycle commencement condition of the arm and its distally carried impact head, and (ii) the retrieval of the arm and its distally carried impact head to the cycle commencement condition; an encoder to ensure the end to be impacted by the impact head of each stem or log is, in turn, impacted by the impact head during a said cycle; a sensor to detect the spectrum or part thereof of the longitudinal vibration response caused by the impact of the impact head with each stem, log or board; and a processor responsive to the sensor for determining the velocity V₅ V² or some function of V of the impact induced longitudinal vibration response; wherein the apparatus is operable to obtain V, V² or some function of V for each stem, log or board on the conveyor.

2. An apparatus a claimed in claim 1 wherein the sensor is mounted on a pivoted, articulated or other sensor arm.

3. An apparatus as claimed in claims 1 or 2 wherein the sensor is mounted on the impact arm.

4. An apparatus as claimed in any one of the preceding claims wherein said apparatus is capable reproducibly of being caused and/or allowed to move to the or an impact zone to encounter an end of each stem, log or board to provide the impact required.

5. An apparatus as claimed in any one of the preceding claims wherein the swing is a downward swing.

6. An apparatus as claimed in any one of the preceding claims wherein the stem, log or board lies longitudinally or substantially longitudinally of the conveying direction.

7. An apparatus as claimed in any one of the preceding claims wherein the driving means is a pneumatic system

8. An apparatus as claimed in any one of the preceding claims wherein the advance of the stems, logs or boards is continuous.

9. An apparatus as claimed in any one of the preceding claims wherein the encoder registers the presence of the stem, log or board of a predetermined pointe along the converying direction and releases the impact arm at a predetermined time thereafter

10. An apparatus as claimed in claim 2 wherein the encoder registers the presence of the stem, log or board at a predetermined point along the conveying direction and releases the sensor arm and/or impact arm at a predetermined time thereafter.

11. An apparatus a claimed in claim 4 wherein the sensor is moveable between a first position proximate to or an impact zone and a second position remote from the or an impact zone.

12. An apparatus as claimed in claim 10 wherein the sensor is configured to be repeatably moved between said first position immediately prior to or at the time of the impact and said second position after the impact.

13. An apparatus as claimed in any of the preceding claims wherein said driving means assists the commencement of a downward swing only of the arm and by the time the impact head is to encounter a stem, log or board end the arm is swinging with its arm without pneumatic acceleration.

14. An apparatus as claimed in claim 12 wherein the driving means comprises a double acting pneumatic piston which is for the purpose of (i) caused and/or allowed to move towards one limit of its movement without pneumatic assistance other than any that might be required to overcome resistance to travel.

15. An apparatus as claimed in any of the preceding claims including a sensor to detect speed by time of passage of the stem, log or board or to detect the spectrum or part thereof.

16. An apparatus as claimed in any of the preceding claims wherein the detected spectrum is of frequencies in the range from 150 to 2000 Hz.

17. An apparatus as claimed in any of the preceding claims wherein said impact head is vibrationally isolated and/or dampened from said arm.

18. An apparatus as claimed in any of the preceding claims wherein said arm is dampened from the frame or other structure from which it is pivoted, articulated or otherwise mounted.

19. An apparatus as claimed in any of the preceding claims including an encoder to regulate the swing down and/or uplift of the arm to cause the impact head to encounter an end of each and every said stem, log or board.

20. Apparatus to determine the velocity V, or some function thereof, of an impact induced travel or impact induced longitudinal vibrational response in a stem, log or board of length L, said apparatus comprising: a conveyor to advance the stems, logs or boards longitudinal of their longitudinal axes serially to and past an impact zone, striking apparatus having: an impact head; an arm or other means ("arm") carrying the impact head; a pivoted, articulated, or other mounted support of the arm whereby it can be allowed and/or caused to swing down under and/or up against the influence of gravity to distally move the impact head therewith; and pneumatic means capable under signal actuation of assisting and/or allowing (i) the commencement of a downward swing from a cycle commencement condition of the arm and its distally carried impact head, and

(ii) the retrieval of the arm and its distally carried impact head to its cycle commencement condition. a sensor mounted on a pivoted, articulated or other sensor arm to detect the spectrum or part thereof of the longitudinal response caused by the impact of the impact head with each stem, log or board; and a processor responsive to the sensor for determining the velocity V or some function thereof.

21. An apparatus as claimed in claim 19 including an encoder whereby the striking apparatus is caused to swing down and/or uplift so as to encounter each and every (preferably advance) end of a said stem, log or board.

22. A method of treating tree stems passing to a merchandiser for breakdown and/or logs being fed into a saw mill which comprises, on a conveyor, endwise serially feeding the stems and/or logs serially to an impact zone where a drop impact head, on a swing or the like arm without operator intervention, is caused, in turn, and substantially reproducibly, to encounter the leading end of the or each stem and/or log, and in turn, without operator intervention, sensing for each stem or log the outcome of the impact and providing from that outcome an indicator or instruction output reliant on velocity V, V or some function of V, V being a measure or estimate of the longitudinal velocity of the impact caused longitudinal response within each stem or log.

23. A method as claimed in claim 21 wherein the longitudinal vibration response is of frequencies in the range from 150 to 2000 Hz.

24. A method of merchandiser breakdown or saw milling which relies on treating tree stems and/or logs respectively according to a method of claims 21 or 22 or using an apparatus of claim 1 to 21.

25. An apparatus as claimed in claim 1 or 19, substantially as herein described with reference to any Example thereof and with or without reference to the accompanying drawings.

26. A method as claimed in claim 21, substantially as herein described with reference to any Example thereof and with or without reference to the accompanying drawings.