DE4039648A1 - Measurement value processing system for biological object - mathematically evaluates and compares with given measurement value structures - Google Patents

Abstract

A measurement value processor (7) is coupled to the biological object (1) via multichannel measurement value detectors (2,3,4,5,6) e.g. various sensors. The processor is coupled to auxiliary devices (12,13,14). The processor has a first device (8) for entering the measurement values from at least one of the detectors, a second device (9) for comparing the measurement values with given values, a third device (10) for starting the auxiliary devices when a first predetermined value is attained and a fourth device (11) for stopping them when a second predetermined value is attained. The second device has given single or multi-channel measurement value structures. The third device operates when a first measurement structure has been attained. The fourth device operates when a second structure has been attained. A fifth device (20) between the first and second devices serves as an evaluator for calculating characteristic values from series of measurement values for reduction of data. USE/ADVANTAGE - Human or veterinary medicine. Operates in real time.

Description

The invention relates to a measured value processing device those mentioned in the preamble of patent claim 1 Art.

Such measured value processing devices, which have multi-channel, z. B. designed as sensors measured value acquisition devices associated with a biological object are general known. Recorded by the measured value acquisition devices Measured values are compared with specified values. If they match, the recorded values are saved, an alarm device is actuated or a control of the device under test performed. Modern due to high sampling rates Measured value acquisition devices and high number of measuring channels results a large amount of data that is difficult to process is.

The invention is based on the object of a simply constructed To create measured value processing device, which at reliable functioning a multi-channel data acquisition and processing in real time.

According to the invention, the recorded measured values are also specified Measured value structures compared. For data reduction the recorded measured values are evaluated mathematically. This results in particularly short measurement processing times.

Further developments of the invention are in the subclaims specified.

The invention is explained in more detail with reference to the drawing.

Show it:

Fig. 1 is a representation of a first embodiment of the invention,

Fig. 2 is a trigger module in a first embodiment;

Fig. 3 is a trigger module in a second embodiment,

Fig. 4 is a trigger module in a third embodiment;

Fig. 5 shows a measurement processing in a second embodiment,

Fig. 6 shows a measurement processing in a third embodiment;

Fig. A1, a data acquisition system with input and output relations,

Fig. A5, the operation of a trigger module,

Fig. A6 the schematic construction of the structure and pattern recognition using complex triggering.

As shown in FIG. 1, a biological object is connected to a measured value processing device 7 via several measuring channels k 1 to k 5 from measured value acquisition devices 2, 3, 4, 5, 6 . The measured value processing device 7 comprises a reading device 8 for reading in measured values of one or more channels k 1 ; k 2 ; k 3 ; k 4 ; k 5 of the measured value acquisition devices 2, 3, 4, 5, 6 . The measured values of the at least one channel k 1 ; k 2 ; k 3 ; k 4 ; k 5 are fed to an evaluation device 20 which is designed to calculate values of parameters from sequences of measured values.

The calculation of the parameters by the evaluation device 20 results in a considerable data reduction. The calculated parameters are fed to a comparison device 9 which compares the parameters of the measured values with predetermined measured value structures. Evaluation device 20 and comparison device 9 together form what is known as a trigger module, which will be explained in more detail below.

If the parameters of the measured values agree with the predetermined measured value structures, a starting device 10 is activated, which starts an additional device 12, 13, 14. The additional device can be designed as a control device 12 which is provided for controlling the biological object 1 . The additional device can, however, also be designed as a storage device 13 in order to store the read-in measured values from the activation of the starting device 10. Another possibility is that the additional device is designed as a warning device 14 for displaying an undesired functional state of the biological object 1 . Of course, the additional devices 12, 13, 14 explained can be connected to the measured value processing device 7 individually or in combination. The stop device 11 is activated when there is agreement with a second predetermined measured value structure. The stop device 11 is connected to the additional devices 12, 13, 14 analogously to the start device 10 .

FIGS. 2 to 4 show the possible structure of trigger modules TM. The trigger module TM in FIG. 2 has two evaluation devices 20 arranged in parallel, to which different measurement channels k 1 ; k 2 ; k 3 ; k 4 ; k 5 are assigned. The determined parameters of both evaluation devices 20 are fed to the comparison device 9 , which, if they match the predetermined measured value structures, activate the start device 10 or the stop device 11 .

The trigger module TM in FIG. 3 shows two evaluation devices 20 which are connected in series and which have different measuring channels k 1 ; k 2 ; k 3 , k 4 ; k 5 are assigned. The series-connected evaluation devices 20 are, as already explained above, connected to the comparison device 9.

The trigger module in FIG. 4 shows a combination of the trigger modules from FIGS. 2 and 3. Two evaluation devices 20 are each connected in series and a further two evaluation devices 20 are connected in parallel, the respective evaluation devices 20 having different measurement channels k 1 ; k 2 ; k 3 ; k 4 ; k 5 are assigned.

As will be explained in more detail below, the predetermined measured value structure can be a simple pattern of data, e.g. B. be a sequence of numbers, but can be one have a much more complicated structure.

The evaluation device 20 is constructed so that parameters such. B. peak values of measured value sequences, mean values of measured value sequences, effective values of measured value sequences, quantile values of measured value sequences.

The evaluation device 20 can also be used to form moving average value estimates, moving moment function estimates of the measured value sequence, to form moving moment function estimates, recursive estimates of the moment function, recursive estimates of the centered moment function of the measured value sequence, to form recursive estimates for values of the autocorrelation function, and recursive estimates of functions of accumulated differences in the measured value sequence, of recursive estimates of the quantile value interval limits of the measured value sequence, of recursive estimates of the mean value in the form of the quantile interval center of the measured value sequence, of adaptive mean values of the absolute amount of the measured value sequence, for the formation of adaptively formed mean values of a corrected measured value sequence, for the formation of recursive cross-correlation functions and for the formation of recursive estimates of functions of accumulated cross differences of the measured value sequences for determining the parameters be trained.

The evaluation devices 20 present in a plurality, such as, for. B. shown in FIGS. 2 to 4, can be designed to determine different parameters.

In Fig. 5 a further embodiment is shown in comparison between device 9 and starting device 10 and between the comparison device 9 and stop device 11 is a time adjustment device 30 is connected. The time influencing device 30 is designed in such a way that the activation of the start device 10 and / or the stop device 11 occurs temporally after agreement with the predetermined measured value structures or temporally before agreement with the predetermined measured value structures. Of course, one of the devices 10; 11 in time according to agreement and the other of the facilities 11; 10 can be activated before the predetermined measured value structures match.

In embodiments that are not shown, the time influencing device 30 is connected between two evaluation devices 20 , between two comparison devices 9 and / or between evaluation device 20 and comparison device 9.

As FIG. 6 shows, a setting device 40 for setting the sampling frequency is connected between the measured value acquisition devices 2, 3, 4, 5, 6 and the reading device 8 . The setting device 40 can be designed in such a way that the sampling frequency is determined by means of recorded mean value crossings of the measured value sequences or by means of recursively recorded mean value crossings of the measured value sequences.

The following is a presentation of the mathematical fundamentals for the subject matter of the invention, as well as the explanation of the Subject of the invention on the basis of specific exemplary embodiments.

A. Generalized triggering theory

In the following sections, terms such as trigger, Triggering, trigger criterion, adaptive triggering, complex triggering, Structure recognition by means of triggering, among other things. mathematically and compared to known technically implemented trigger processes to be generalized.

The term triggering is generalized with regard to:

• - type and structure of the triggering signal state,
• - the process adaptability by means of the stochastic method Approximation (adaptive triggering),
• - the complexity of the trigger connections,
• - the timing mode of triggering.

A. 1. Triggering in data acquisition systems

Triggering is generally understood to mean the start or cancellation of a process by a signal a (t) (pulse, edge change, etc.). The signal a (t) is also referred to as the trigger signal or trigger for short; it is applied to a trigger channel k _a . For measured value acquisition systems, this definition must be specified in more detail in relation to the (measuring) process to be started or stopped and the trigger signal a (t).

A measurement process should be based on a technical problem started or stopped when on the trigger channel sets a signal structure that can be determined from preliminary investigations. For this purpose, the signal a (t) on the trigger channel is monitored and in the case of a digital data acquisition system using mathematical methods rated. From the requirements of the technical problem this is followed by the definition of the signal state (more defined Signal state or trigger criterion) according to which the Trigger signal a (t) is monitored and after its occurrence the Triggering is triggered.

Let S be a measured value acquisition system, see also FIG. A1.
Let K = {k ⁱ } _{i = 1} ^L be the set of input channels of the system S and K _i : i ∈ {1,. . ., L}, be the i-th input channel of the system S;
L - maximum number of channels.

The signal applied to the channel k ⁱ , iL, i∈ N is ^{denoted by x i} (t) and

let the channel k ⁱ at times t₁,. . ., Recorded series of measurements consisting of N _i <∞, N _i ∈ N measured values. Furthermore, let K _a ⊂K be a subset of the input channels (K _a - set of trigger channels) with K _a = {k _a ¹,. . . k _a ^R } with RL, R∈ N.

Definition A1

Triggering is the start / stop of a (measuring) process on one or more input channels k _i ∈K at time t _a after the first occurrence of a defined signal state (e.g. pulse, edge change) at time t _e on at least one of the input channels k _a ⁱ ∈K _a .

The input channels k _a ⁱ are also referred to as trigger or release channels.

According to the type of initiation of the measuring process, one obtains finite series of measurements of the shape

with
t _a t₁ for the start of the (measuring) process or
t _a for stopping the (measuring) process.

With regard to the timing of the triggering, the Points in time

t _a - time of triggering the start / stop of a (measuring) process and
t _e - time of the first occurrence of a defined signal state on at least one of the triggering channels

differentiated.

The following applies:

t _a = τ _v + t _e with τ _v ∈ R
τ _v = t _a -t _e and τ _v - delay time of the triggering

Definition A2

1. pre-triggering: = a triggering according to definition 1, where the following applies: t _a <t _e ; (τ _v <0), ie the triggering of the start / stop of a measuring process at time t _a takes place before the occurrence of a defined signal state at time t _e on a triggering channel from K _a .
2. post-triggering: = a triggering according to definition 1, where the following applies: t _a <t _e ; (τ _v <0), ie the triggering of the start / stop of a measuring process at time t _a takes place after the occurrence of a defined signal state at time t _e on a triggering channel from K _a .
3. (com) triggering: = a triggering according to definition 1, where the following applies: t _a = t _e ; (τ _v = 0), ie the occurrence of a defined signal state on a triggering channel from K _a triggers the start / stop of a measuring process without a time delay.

Let ∈K _{a be} a trigger _{channel and a l} (t) the trigger signal applied to the channel, which is monitored for the occurrence of a defined signal state (pulse, edge change). If the course of the signal a _l (t) is followed over longer periods of time, this requires a corresponding indexing of t _a and t _{e of} the form:

t _e ⁱ - i-th point in time when a defined signal state occurs on the triggering channel,
t _a ⁱ - i-th point in time at which the start / stop of a (measuring) process is triggered. (A 1/1)

Previous considerations of the signal a _l (t) applied to the triggering channel assume its continuity. If the signal is only recorded in discrete form (e.g. in digitized form via the data acquisition system of a computer) or if it is only available at discrete points in time, one speaks of discrete triggering.

There is a transition to an equidistant time base of the form:

t = r Δt, with r∈ G and f _A - sampling frequency,

with which the values are recorded on the release channel,

A defined signal state can now only occur a discrete point in time

t _e = s · Δt, s∈ G

be recognized.

After the introduction and definition of the term triggering in data acquisition systems takes place according to the requirements Systems with changing operating conditions affect the construction process-adapted trigger method. This is done initially on the basis of deterministic trigger signals and is used in the Sections A 3 / A 4 generalized to stochastic trigger signals. In the following sections A 5 / A 6 will be Basics for process-adapted trigger methods for indication complex signal patterns and structures.

A. 2. Triggering based on deterministic signals

The focus of the following 3 sections should be the possibility of a mathematical description of the defined signal state (trigger criterion) and the determination of the point in time t _e when the trigger criterion occurs for the first time.

Let T and A be two arbitrary sets from R. Let σ _T and σ _A be sigma algebras over Borel sets from T and A. Let the trigger signal a (t) with a: t → a (t) be a (σ _T , σ _A ) measurable function with the domain T and values in A. Let the function h describing the type and structure of the trigger criterion be a function of values from A × A in a set H⊆ R and (σ _{A × A} , σ _H ) - measurable, where σ _{H is} a sigma algebra of subsets from H.

Definition A3

t _e ∈T is the time t _{e of} the first occurrence of the trigger criterion if:

t _e : = min {t: h [a (t), a (t *)] ∈H _e ; H _e ∈σ _H ; t, t * ∈T}.

The condition h (a (t), a (t *)) ∈H _e defines the trigger criterion according to which the trigger signal a (t) is monitored; it is also referred to as the trigger condition. H _e is the trigger set.

If the trigger criterion can only be characterized by several conditions, then t _{e is} defined by

now be a function vector of the form

= (h₁,..., h _Q ), Q∈ N and
H _e = H _e ¹ ×. . . × H _e ^Q ∈σ H _e ¹ ×,. . . × H _e ^Q ,
where the h _i {σ _A × σ _A ,} - are measurable functions of values from A × A in H ⁱ ⊆H⊆ R and is a system of subsets of H _i ,

The terms and definitions introduced should be based on a few Examples are illustrated.

1. Limit value triggering

The following applies:

h = I - identical figure
Q = 1, t = t *
H = A,

so is

h [a (t), a (t *)] = a (t) and
^d t _e ^g = min {t: a (t) ∈H _e , t∈T}.

If H _e = (a _g ^- , a _g ⁺) is a connected open interval from R with a _g ^- = -∞ or a _g ^- ∈ R fixed and a _g ⁺ = ∞ or a _g ⁺∈ R fixed and a _g ^- a _g ⁺, the following known trigger criteria are obtained.

2. Difference triggering

The following applies:

Q = 1, t * = t-σ, σ <0, fixed
H _e = [a _g ^- , ∞), a _g ^- 0.

h is defined by

h [a (t), a (t *)] = | a (t) -a (t *) |

and the time of the first occurrence of the trigger criterion by

^d t _e ^g = min {t: | a (t) -a (t-σ) | <a _g ^- , t∈T}.

The defined signal state is characterized here by a sudden change in the signal course, ie the point in time t _e is defined as the point in time at which the signal course a (t) changes within a defined time interval σ by more than the value a _g ^- <0.

Remarks

• 1. In addition to the pure difference or distance triggering, an evaluation of the edge direction low-high or high-low can also be used to characterize the defined signal state; this is also referred to as trend triggering. The function h is then defined without the amount: h (a (t), a (t *)): = a (t) -a (t *) t _{e is also} determined with differential triggering
  Low-high change through, and with for high-low changes
• 2. For discrete triggering with equidistant sampling frequency, the following applies:

The time of the first occurrence of the trigger criterion (Differential triggering) is obtained from:

3. Monotonic triggering

The following applies:

The trigger criterion here is based on the monotony behavior of the Triggers a (t) definable, the following applies:

A.3. Stochastic approaches for triggering measurement processes

Previous considerations and definitions of triggering require deterministic trigger signals. If process signals are used for triggering, this is no longer the case (see Section A.2.). The signal detected at the trigger _{signal k a} is therefore interpreted below as a stochastic process (X _i ) _{i ∈ T} , where T ia is interpreted as a set of points in time (time base).

X _T. = (X _i ) _{i ∈ T} , (T ≠ 0) be a family of random variables over a common probability space [Ω,, P] with values in a measurable space [E,].

X (t, ω) can be viewed as a mapping of T × Ω into E. For fixed t∈T as a function of ω, X (t, ω) (,) -measurable X (t, ω) = X _i (ω) for fixed t.

In the following, let _i ⊆ be the σ-algebra of the events that can occur in connection with the process X _i up to time t. _{Let i} be the σ-algebra generated by the quantities (X _{s , st), t} = σ (X _s , st).
( _i ) _{i ∈ T} is then an increasing family of σ-subalgebras of, of the form _s ⊆ _i ⊆, s, t∈T with st.

Definition A4:

Let [Ω,, P] be a probability space and let ( _i ) _{i ∈ R ⁺ be} an increasing family of σ-subalgebras of B ([0, t]) be a σ-algebra of Borel sets over [0, t]. A random process (X _i ) _{i ∈ R ⁺} defined on [Ω,, P] with values in [E, B ] is called measurable with respect to the family ( _i ), if for every t∈ R ⁺ the mapping (t, ω ) → X _i (ω) from (0, t) × Ω in [E,] is measurable with respect to the σ-algebra generated by B ([0, t]) × _i.

First of all, the concept of stopping time should be introduced.

Definition A5:

A mapping τ from a non-empty set Ω _t ⊆Ω with values in T is called a stop time with respect to ( _i ) _{i ∈ T} or short ( _i ) -stop time if t∈T the relation {τt} ∈ _i holds.

From now on, the set T will be considered as a set of discrete points in time with T = {t¹, t²,. . ., t ^N,. . .}, N∈ N.

Sentence A1

be a sequence of random quantities over [Ω,, P] with values in [E, B ], t _i ∈T⊆ R and T = {t¹, t²,. . ., t ^N,. . .}.

Let h be a measurable map of E₁x. . . xE _M → R : M∈ N , M <∞ and E _i ⊆Ei. Then is the picture

t _e (ω): t _e : Ω → T with

an ( _i ) stop time.

proof

There h a measurable representation of

E₁′x. . . xE _M → R holds

h ^-1 (S ′) ∈σ _{E₁x. . . xE} ,
S′∈ _R.

It remains to be shown:

{ω: t _e (ω) t} ∈ _t with _i = σ (X _s , st); t∈T.

This applies according to the following conditions

In the following, the point in time t _{e of} the first occurrence of the trigger criterion on X _{τ is to} be introduced as the stop time, the trigger criterion now being defined via stochastic properties and features of the process X _t .

An illustration h of serves to describe the trigger criterion

with values in a measurable space [H,], with

For fixed t₁, t₂,. . ., t _N as a function of ω

Definition A6

Let H _{e be} a subset of H and let h be a map of

in H. Then one denotes the condition

as a trigger criterion (trigger condition). H _e is the trigger set.

Definition A7

The mapping t _e with t _e : Ω → T is called the time of the first occurrence of the trigger criterion {h∈H _e } if:

Corollary A1

Let N = 1; [H,]: = [E,] and h (t, ω) = X (t, ω) a random process. _{Let i} be an increasing family of σ-subalgebras of.

Then is the picture

t _e : = inf (t: X _i (ω) ∈H _e , t∈T) an ( _i ) -stop time.

proof

It follows from Theorem A1 with M = 1 and h = I identical mapping ∎

Be it

a sequence of temporally consecutive, possibly dependent, random variables defined on [Ω,, P] with values in [E,]. Let EX _t = µ <∞.

The trigger criterion is described below via properties of Sampling functions of the sequence

introduced. The trigger point in time t _e is estimated using the sliding estimation of sample functions.

a) Determination of t _e using the moving average value estimate Corollary A2

Let h¹ be a map of

where applies

an ( _i ) stop time.

proof

Follows from Theorem A1 and measurability properties

that is, h can be measured again as a linear combination of measurable quantities X _{t (ω).}

b) Determination of t _e via sliding moment function estimation Corollary A3

Let ^m h ^{K be} a map of

where applies

a ( _t ) stop time.

proof

^{According to measurability properties, m} h ^K is a measurable function, the statement thus follows from Theorem A1 ∎

comment

For K = 2, the sliding estimate corresponds to the 2nd moment function an estimate of the rms value squared the sequence

c) Determination of t _e via sliding centered moment function estimation Corollary A4

Let ^Z h ^{K be} a map of

where applies

an ( _i ) stop time.

proof

Analogous to the proof of consequence A3 ∎

Remarks

• 1. For K = 2, the sliding estimate of the 2. centered moment function corresponds to the estimate of the spread of the sequence
• 2. If µ is unknown, µ becomes through the estimate replaced according to Corollary A2.

The main disadvantage of stochastic approaches to the definition of trigger times on the basis of sample functions lies in the inertia of the indication of the trigger criterion as well as in the limited possibilities of adaptability to special signal structures of the triggering channel. The implementation of this triggering by means of computer-aided data acquisition systems requires a large amount of memory and extremely high processing speeds. These disadvantages can be overcome using stochastic approximation methods for the construction of recursive trigger methods.
Of great practical interest are processes that can be implemented by means of computer-aided measured value acquisition systems in real time and with low storage space requirements.

A.4. Triggering based on the stochastic method Approximation

The trigger time is also determined here via the Estimation of characteristics of sample functions. To estimate the construction of recursive procedures of the form takes place:

where W is a measurable function defined on R × E with values in R. The following applies to the sequence γ _t:

The convergence of these procedures can be achieved by means of stochastic methods Approximation can be demonstrated. The Theory of stochastic approximation both the possibility to construct a large number of such recursive estimators, as well as basic ideas for an adaptive design of this Procedures.

Adaptivity is an adaptation of the estimation algorithms to changed structural conditions (unsteadiness) of the processes understood, possibly a waiver of convergence takes place in the classic sense. This approach will described in detail in section B.

On the basis of the general approaches to stochastic approximation, the construction of recursive estimation methods for parameters of sample functions takes place. In the following, results from Section B are used to define and determine the trigger time t _e . The methods are characterized by a high degree of process adaptability.

The choice of the sequence γ _n , via which the speed of adaptation of parameters to signal structures and patterns can be controlled, is of great importance. Section B.2 deals in detail with the mode of action and possible definitions of the sequence γ _n.

a) Estimation of t _e via the recursive estimation of moment functions Corollary A5

Let ^m h ^{K be} a map of

where ^m h ^K recursively according to

is calculated, an ( _i ) -stop time.

proof

h can be measured again as a function of measurable quantities and thus fulfills the conditions of sentence 1 ∎

Remarks

• 1. For K = 1, the mean value function of the sequence {X _t } is estimated.
• 2. For K = 2, the root mean square value of the sequence {X _t } is estimated.

b) Estimation of t _e based on the estimation of the centered moment function Corollary A6

Let ^z h ^{K be} a mapping of H × E into H, H _e ⊆H; and,

where ^z h ^K recursively according to

is calculated, a ( _t ) stop time.

proof

Analogous to the proof of consequence A5 ∎

Remarks

1. For K = 2, the scatter of the sequence {X _t } is estimated.

2. If µ is unknown, µ becomes through the estimate

replaced according to corollary A5.

c) Estimation of t _e based on the estimation of quantile values

May be

a sequence of identically distributed random variables with the distribution function F _X.

Definition A8

Z _α is called the α-quantile of the distribution F _X if the following applies

F _X (Z _α ) αF _X (Z _a +0) with 0 <α <1.

Section B.1 describes a recursive method for estimating of the α-quantile

described.

Based on this procedure, two are defined Tolerance range limits for the random variables

which is a percentage α · 100% of this Include random sizes.

The tolerance _{range limits Z i} ⁺ and Z _i ^- become recursive according to ⁺ = z₀⁺ start value, fixed, arbitrary.

and

certainly

Z _i ⁺ - is the upper threshold value at time t _i ;
Z _i ^- - is the lower threshold value at time t _i .

In the following, trigger times t _{e of} the first occurrence of a defined signal state are introduced, the definition of the trigger criterion taking place via stochastic parameters on the basis of the threshold values Z _t ⁺, Z _t ^- .

Strong fluctuations as a result of the random variables

are increased by increasing the width of the interval

z. B. displayed over a limit value a _g.

Statements about the monotony behavior of the sequence of random variables

you get z. B. by

Here, ^Q t _e ^W indicates the point in time at which the sequence of random variables

shows a monotonic trend (increasing monotonically) of length n _w .

The same applies to sections of length n _F , in which the sequence

falls monotonously:

with

Remarks

• 1. The threshold values Z _t ⁺ and Z _t ^- can be used to describe a large number of signal properties that can be used to define trigger criteria at the same time.
• 2. An optimal adaptation of parameters via quantile value estimates can be achieved by a suitable choice of γ _t and α.

d. Recursive estimation for values of correlation functions

Be there

two episodes of Random sizes with

= f (X, ν∈ R ⁺, f - measurable, it holds

τ = ν · Δt, Δt: = (t _{i + 1} -t _i ), t

and be a sequence of independent random variables.

By

a recursive estimate for the autocorrelation of the process X _{T is} given.

With ^τ t _e ^Aut , where

the trigger time is introduced at which the recursive point estimate of the autocorrelation function of the process X _t lies in the trigger set H _e.

Analogously, a trigger criterion can be defined via the cross-correlation of two processes X _T , Y _T (according to the above definition).

and thus the first point in time at which the recursive estimate for the Cross-correlation of the consequences

Assumes values in H _e .

The convergence of both recursive methods for estimating Values of the auto and cross correlation function for sequences of Zu case sizes according to the above conditions follow from Sentence A 2.

Sentence A 2

Be it

a sequence of two-dimensional too due vectors with

a sequence of independent random variables and

g and f are measurable maps.

By

be a sequence of random variables

Are defined

is the σ-algebra of the events, those up to time u in connection with the random process have occurred. Under the condition that

with t = 0, 1, 2,. . ., c = constant, the estimation function sequence {M _t } converges with

M _o : = m _o , any, start value,

M _{t + 1} : = M _{t + γ t} (Z _t ^K - M _t ), K ∈ N

with probability one against the expected value

EZ _t ^K = µ _K.

proof

The proof of the theorem follows from point B0. In the terminology there with ξ _t = Z _t , a (t) = γ _t , it was shown that a stationary, strongly mixing sequence of random variables Z _t (ω) over a probability space [Ω,, P], where for

is fulfilled and, under conditions a) and b) of the theorem, the estimation function sequence (M _t ) defined above converges _{to EZ t} ^K = µ _K with probability one. It remains to show that _u satisfies the condition (*). Let τ <t _{N + v} -t _N.

The following applies:

be one Sequence of independent random variables.

The following applies:

are independent random vectors with i ≠ j ≠ K ≠ l, where o. B. d. A. i <j <K <l apply.

Because of measurability properties, the following applies to the random Sizes

with f-measurable Function that f₁ (ω) and f₂ (ω) are independent.

If one sets t _i = t _N ; t _j = t _{N + v} ; t _K = t _{N + w} and t _l = t _{N + W + V} with t _{N + W} -t _N τ and w <v, then the random variables

independent. If one sets o. B. d. A. s = t _N , it follows because of the independence of

α (τ) = 0 and thus the proof of theorem ∎

Remarks

• 1. For g = I a recursive estimation method for the autocorrelation function of {X _t } is obtained.
• 2. Applies this gives a recursive estimation method for the cross-correlation between the structure is often given in practical applications, such as shock propagation via a wave or influences between neighboring Meßstel len biomedical signals.
• 3. A method similar to autocorrelation, which can be used in practical applications (especially for monitoring special frequency components), is obtained from

3.5. Construction of procedures for structure and pattern recognition by means of complex triggering

Let {k¹,. . ., k ^L } a subset of the set of release channels K _a and a¹,. . ., a ^L , the trigger signals detectable according to the release channel (deterministic, stochastic or mixed structures). To simplify the representation, a graphic symbol is cf. Fig. A 5 introduced to identify trigger modules.

The trigger ^{module TM i} denotes an algorithm A _i (computer program module) which determines the point in time t _e ^{i of} the first occurrence of a trigger criterion {h ⁱ (a) ∈ H _e ⁱ }.

Let a = (a 1,..., A ^L ) be the vector of the trigger channels and

a sample function of the measuring channel i with the time base {t _{N-M + 1,.} . ., t _N }. Let h be a measurable function with

h: E 11x. . .xE _M ¹x. . .xE₁ ^L x. . .xE _M ^L → R.

Then a multi-channel trigger criterion can be constructed by {h (a¹, ^{..., A L} ) ∈ H _e , H₂∈ R}.

Definition A 9

Triggering is called complex triggering if the point in time t _{a of} the initiation of the start / stop of a measuring process can be represented as a logically arithmetic expression of points in time t _{e of} the first occurrence of trigger criteria and time delay constants t _v .

A general illustration of complex triggering is given in FIG . Complex trigger processes are used to indicate signal patterns and structures that are defined according to a technical or medical problem. Applications of this method are described in the exemplary embodiment, using the following construction principle.

It is based on several procedural steps:

• 1. Preliminary investigations into the signal structure of the trigger signals
  • - reproducibility,
  • - variety and width,
  • - temporal regimes in patterns,
  • - Time synchronicity between different channels to determine delay constants.
• 2. Signal segmentation
  Investigation of the trigger signals with regard to mathematically be writable signal properties, determination of signal sections with the same properties and characterization of the time sequence of segments with different signal properties.
• 3. Selection, adaptation and optimization of the trigger method for recognizing the signal properties in the individual segments. Selection of the trigger criterion, selection of the sequence γ _n , setting of controlled variables (α, etc.).
• 4. Arithmetic and logical linkage, see FIG. A6 of the process-adapted trigger method according to the time regime of the patterns and structures to be recognized.

Remarks

• 1. For multi-channel examinations, either the independence of the trigger signals a¹,. . ., a ^L , or it is assumed in the model, so that no additional considerations are necessary with stochastic signal structures with regard to the corresponding probability spaces.
• 2. When generalizing the trigger terms, starting from the technically known state, offers the introduction of mathemati cic structure stop time the possibility of con to define structured trigger criteria in a uniform form and taking advantage of properties of stop times to one generalized model of triggering to arrive. Of great of practical importance is the fact that logical and arithmetic operations (multi-channel analysis) in general do not lead out of stop time models.

A. 6. Time dynamic triggering in data acquisition systems

Let a (t) be a trigger signal and let k _a ∈ K be the associated trigger channel. The consequence

characterize release time points regarding trigger criteria on a (t) according to (A. 1/1).

Definition A 10

Temporally dynamic triggering of measured value acquisition systems means the acquisition (measurement) of values x _i on one or more measuring channels

with the Zeitbasisa, where Ta is the sequence of the tripping time as the time base Points

serves. As a result of a Measured value acquisition triggered dynamically over time are measurement series the form

before.

To determine the times t _a ⁱ , methods according to A.1. to A.4. can be used.

The practical importance of procedures of the temporally dynamic Triggering lies in its contribution to solving problems gen like

• 1. Optimization of the ratio of sampling frequency to data volume,
• 2. Data reduction through event-related recording of measured values,
• 3. Construction of low memory monitoring methods space requirements.

Based on the optimization of sampling frequencies, the following Methodology and construction principles of dynamic time triggers procedures are represented.

Algorithms for selecting the sampling frequency f _A (or the time base T)

The question of the choice of the sampling frequency f _A in the sense of optimizing the relationship between sampling error and data volume is known and dealt with in the literature as the sampling theorem, which is required

For the power spectrum S _AA (f) of a (t), this assumes that the following applies:

S _AA (f) = 0; f <f _g . (A.6 / 1)

In most practical applications, a (t) is a stochasti cal signal with low-pass behavior and thus (A.6 / 1) only in a few given cases. Among other things, procedures are described in the literature for determining the sampling time and corresponding error estimates given. The basic idea behind the procedures specified there is the optimization of the sampling frequency by determining the time points of the mean value runs or the times of occurrence at least from signal extremes. Identifying this time You can score points using the following adaptive trigger methods carry out.

Determination of the times of the mean value runs

Let T ^{m be} an equidistant time base with

a stochastic process with the time base T ^m . From the practical problem statement it must be ensured that τ ^m can be chosen to be smaller than the smallest time span between two mean value passes. Because of this,

applied, where f _A ^{max represents} the maximum possible sampling frequency of the data acquisition system. Otherwise it is unsuitable for the problem. ^{The figure m} h ¹ from corollary A5 (adaptive averaging) is used to construct the method. For the sake of simplicity, one sets ^m h ¹ = h.

The following applies:

A mean value run is then carried out

characterizes,
one defines [a, b] *: = [min {a, b}, max {a, b}].

The mean time between two successive mean value passes in the interval I _k with I _k = [t₁ ^m , t _k ^m ] is then calculated

whereby

is applicable . By

a simple rule for calculating ϒ ^{k is} given. The times of the mean value runs

are because of the trigger criterion

according to conclusion A5 stop times with regard to _t .

The determination of a process-adapted time base

for the scanning of the process {X _t } is based, among other things, on the following parameter with

possible.

B. Application of methods of stochastic approximation for Construction of adaptive estimator sequences B.0. Stochastic approximation

In the theory of stochastic approximation it is assumed that an unknown function R (x) "measured" at any point x on the real axis E₁ can. Only certain characteristics of the function R (x), re their continuity, monotony, etc. are given and in particular let us know that the equation

R (x) = α, α ∈ E₁ (B./1)

has a unique solution x _o . One task is now to construct _{a sequence of consistent point estimates for x o} with the aid of the measurement data of R (x) (the measurement error is not negligible).

This task can be solved as follows:

Let Y _{t + 1} (X (t), ω) be the result of the measurement in X (t) at time t + 1.

In a simple situation, e.g. B.

Y _{t + 1} (X (t), ω) = R (X (t)) + G (t + 1, X (t), ξ (t + 1, ω)) (B./2)

where the ξ (t, ω) are a sequence of independent random variables that defined over a certain probability space [Ω,, P] are.

E G (t, x, ξ (t, ω)) = 0 for any x ∈ E₁, t = 1,2,. . . and G (t, x, y) an unknown function of the variables t, x, y is.

The measured values Y _t are now used as "correction variables" in a recurrently defined estimation function sequence for x _{o as} follows:

X (0) = x x ∈ E₁, t = 0, 1, 2,. . .

X (t + 1) -X (t) = a (t) Y _{t + 1} (X (t), ω) (B./3)

Here a (t) is a sequence of positive numbers that satisfy the conditions

enough.

Under suitable conditions at R (x), this process, defined in a fundamental work by Robbins and Monro in 1951, converges to x _o .

The results of Robbins and Monro were generalized in numerous subsequent works. Instead of the first shown convergence in the square mean, under weaker conditions also for the case that x and R (x) are vectors from E _n (n-dimensional Euclidean space) and modifications of the process (B./3), convergence with probability Proved one thing.

Sentence B1.

A random process X ^{m, x (m)} (t) with discrete time is defined according to the recurrent relationship
X (t + 1) = Φ (t + 1, X (t), ω) (B./5)

Φ (t, x, ω), t = 0, 1, 2,. . . x ∈ E₁, let be a lot more vectorial Quantities given over a probability space [Ω,, P] and meet the following conditions:

• A1. Let the function Φ (t, x, ω) with values from E _{l be l} x - measurable for every t = 0, 1, 2,. . . ( _l denotes the σ-algebra of the Borel sets)
• A2. Let there exist a family of σ-algebra _n of subsets of the set Ω such that _m ⊂ _n for m <n, and the family Φ (n, x, ω) is _n - measurable and independent of _n-1 .

Then the process X ^{m, x (m)} (t) with the initial condition X (m) (measurable with respect to the σ-algebra _m ) is Markovian. Its transition function is given by

P (u, x, u + 1, Γ) = P (Φ (u + 1, x, ω) ∈ Γ)

with Γ ∈ _l .

With the notation C _z ^o for the set of real-valued twice continuously differentiable functions whose second partial derivatives are bounded,

for the distance of the point x from the Quantity B,

U _ε (B) = {x ∈E _l : ρ (x, B) <ε} (epsilon neighborhood of B)

and U _{ε , 1 / e} (B): = (E _l / U _ε (B)) ∩ {x: | x | <1 / ε}

is applicable.

Proposition B2: Let X ^x (n) be a Markovian process according to the recursion formula

X (t + 1) -X (t) = a (t) [R (X (t)) + G (t + 1, X (t), ω)] (B./6)

with the initial condition X ^{x (o)} = x, and there exists a non-negative function V (x) ∈ C _z ^o that corresponds to the condition

and the inequalities

enough.

G (t, x, ω) + R (x) satisfy the conditions A1 and A2 of Theorem (B./1) and be it

E G (t, x, ω) ≡0. (B./9)

Also applies

and let B: = {x⁶: R (x) = 0}.

Then applies

In the following, some requirements are specified that are necessary for Retention of the convergence statements of the procedure (B./6) under Weakening of condition A2 and thus an application of Methods of stochastic approximation to problems enable stationary random processes.

Let there exist a growing family of σ-algebras _e ^t ⊂ (0 st ∞) such that one of the two groups of conditions is satisfied.

B1.) For every x and t let G (x, t, ω) be represented in the form

The families _e ^t are strongly mixed, ie

B2.) For every x and t let the random quantities _t ^t - be measurable and let the family _e ^t be absolutely regular, ie

where for quantities of the form

We have α (τ) β (τ).

Theorem B3: The following conditions are met:

Equation (B./1) has a unique solution r.
There is a symmetrical pos. Matrix D and a λ <0 such that for all x ∈ E ⁿ

It ex. σG / σx and

or if G can be factored

with γ = 2 + m and m even
For the mixing coefficient β (t) in B1 or B2, let β (t) C (ln t) ^{-z (m + z) (1 + h) / m} for h <0, and β (t) C t ^-h .

Then the process (B./6) converges with probability One against r.

B.1. Strongly more consistent with the construction Estimation function sequences

The computer-aided realization of basic tasks of the mathematical statistics, e.g. B. the construction more consistent Estimator sequences, is usually supported by some additional accompanying practical demands that are not out of the usual Quality criteria for estimates are derived. These are questions the computing speed, the storage space efficiency, the continuous evaluability of a sequence of estimation functions each subsequent point in time as well as questions about a rapid adjustment of the Algorithms for changes in the estimation conditions (Structural breaks) and the robustness against injuries in made conditions. Defined recurrent Estimation function sequences provide a solution to such problems important basis.

In some cases, sequences of estimators can easily be brought into the desired form. A classic example is the expected value estimation M _n , which is based on the realization of a sequence of independent, identically distributed random variables (ξ _n ) _{n = 0,1,2. . .} based. If we denote by x _i the realization of the random variable ξ _i , then is

strongly consistent sequence of estimators for E ξ _i . M _n can now be represented recurrently via trivial conversions:

M _o = x _o (B./17)

Computationally implemented in this form, the development of the sequence can be directly included in evaluations. In addition, the sequence of {x _i } _{needs i = 0,1,2,. . .} not to be saved. Only the past estimated value M _n , the “current measured value” x _{n + 1} and the “point in time” n + 1 are included in the calculation of the new estimated value M _{n + 1} . A disadvantage of this procedure is that the most computationally complex operation for this task, division, must be carried out at all times.

The apparatus of ideas of the stochastic approximation now initiates both construction methods (also for the recursive Representability not as simple estimation algorithms) as well Generalizations that help solve the above practical Contribute to requirements. The estimation function sequences get im generally a form of the species

S _o = s _o (start value)

S _{n + 1} = S _n -a _n K (S _n x _{n + 1} ) (B./18)

where {a _n } _{n = 0,1,2,. . .} a sequence of numbers and K a correction variable for the estimate S _n , which depends only on S _n and the current realization _{value x n + 1.}

A few examples are given below.

In these, let {ξ _t } _{t = 0,1,2,. . .} Sequence of independent, identically distributed random variables over a probability space [Ω,, P] with the distribution function F.

Sentence B4. Be it

M _o = m _o = const. (Start value)

M _{t + 1} = m _t -a _t (M _t -ξ _{t + 1} ^k ) (B./19)

with {a _t } _{t = 0,1,2,. . .} Sequence of real numbers that satisfy the conditions (B./4). Furthermore, E ξ _t ^zk <∞ applies.

Then the sequence of estimators {M _t } _{t = 0,1,2,. . .} defined according to (B./19) with probability one against µ _k : = E ξ₁ ^k .

Proof: Let us denote by _n : = σ (ξ _o ,..., Ξ _n ) that of ξ _o,. . ., ξ _n generated partial σ-algebra of, then the process defined recurrently with (B./19) has that in theorem B2. Required shape (in particular, conditions A1 and A2 of sentence B1 are met).

It will now

Y _{t + 1} (x, ω) = R (x) + G (t, x, ξ _{t + 1} (ω)): = (µ _x -x) + (ξ _{t + 1} ^k -µ _k ) = ξ _{t + 1} ^k -x (B./20)

set.

Then it remains to show, including theorem B2:

• EG (x, ω) ≡ 0, which is obviously trivially fulfilled. (B./21)
• With ν (x): = x² and B: = {µ _k }

So for known E ξ₁ =: µ with

S _o ² = s _o ²

S _{t + 1} ² = S _t ² - a _t (S _t ² - (ξ _{t + 1} -µ) ²) (B./24)

also a highly consistent sequence of estimators for Var ξ₁ =: σ² If µ is unknown, µ in (3.3.24) becomes through its estimate according to (B./19) replaced with k = 1.

Sentence B5. Let the density f (x) of the random variables ξ exist and f be _{continuous in x α} (α-quantile of F _ξ ), f (x _α ) <0.

Be further

X (0) = x (start value) (B./25)

X (t + 1) = X (t) + a _t Y _{t + 1} (X (t), ξ _{t + 1} (ω)) t = 0, 1, 2,. . .

and {a _t } _{t = 0,1,2,. . .} Sequence of real numbers that satisfy the conditions (B./4).

Then the sequence {X (t)} _{t = 0,1,2,. . .} with probability one against x _α .

Proof: Let R (x) = α-F _ξ (x)

Using theorem B2. remains to show:

EG (t, x, ξ _t ) = E (Y (x, ξ _t ) - R (x)) = (α-1) F _ξ (x) -R (x) + α (1-F _ξ (x )) = -F _ξ (x) + α-α + F _ξ (x) ≡ 0

because of f (x _α ) <0 and the continuity of f.

R² (x) + EG² (t, x, ξ _t ) = EY² (x, ξ _t (ω)) = (α-1) ²F (x) -α² (1-F (x)) 2

One possible method for estimating probability densities is based on the representation of a square integrable Density function f (x) as a series of orthogonal functions

f (x) = Σ R _j ϕ _j (x) (B./27)

with {ϕ _j } _{j = 0,1,2,. . .} orthogonal function system. The problem then is to estimate the Fourier coefficients

R _j = ∫ f (x) ϕ _j (x) dx j = 1,2,. . . (B./28)

These are unbiased estimates

As an estimate of the density can then

to get voted.

The sequence of estimators _{g (n)} (x) is strictly consistent if a mathematical sample {X _n } _{n = 1,2,. . .} underlying. The consistency can also be proven for the case that {X _n } _{n = 1,2,. . .} is a stationary, strictly mixing sequence of bounded random quantities.

The shape of the

R _j = Eϕ _j (X)

not only motivates the estimators (B./29) again, but also suggests the idea of recursively estimating _{the parameters R j.}

Sentence B6. Let f (x) be square integrable density function and {X _n } _{n = 1,2,. . .} independent, identical random quantities distributed according to f (x).

Be further

R _j ^⁰ = R (initial value, fixed but arbitrary)

R _j ^{n + 1} = R _j ⁿ -a _n (R _j ⁿ -ϕ _j (X _n )) j = 1,. . ., g _n (B./31)

where {a _n } is a sequence of numbers that satisfy the conditions

Σa _n = ∞, Σ a _n ² <∞

enough.

Then the sequence {(R ₀ ⁿ ,..., R _g ⁿ )} converges to (R ₀ ,..., R _g .

As in Section B.0. already indicated, can be the above specified construction methods for recurrent highly consistent Estimation function sequences also under certain dependency relationships realizing the observation values.

Let {ξ _t (ω)} _{t = 0,1,2,. . .} a stationary, strongly mixing sequence of random variables over a probability space [Ω,, P]. In particular, let the sequence of partial σ-algebras of _u : = σ (ξ ₀ ,..., Ξ _u ) satisfy the condition (B./12).

As a recurrent sequence of estimators,

M ₀ = m ₀ (start value, arbitrary, but fixed)

M _{t + 1} = M _t + a _t (ξ _t ^k -M _t ) k ∈N (B./32)

set. Then applies Sentence B7.

Under the condition that

converges the estimation function sequence {M _t } defined according to (B./32) with probability one to E ξ _t ^k = µ _k .

Proof: It will

Y _t (x, ω): = ξ _t ^k (ω) -x = - (x-µ _k ) + (ξ _t ^k -µ _k )
=: R (x) + G (t, ξ _t ^k (ω) (B./35)

set.

Then there remains, taking into account theorem B.3. to show:
-) EG (x, ω) ≡ 0 is trivially fulfilled.
-) R (x) · (x-µ _k ) -λ (x-µ _k ) (x-µ _k ) holds for example with λ = 1.
-) G (x, t, ω) is, as can be seen from (B./35), trivially factorizable in U (x, t) = 1 and V (t, ω) and E | V (t, ω) | ^γ <C with γ = 4 and because of (B./33).

Comment: With sentence B7. the consistency statements of recursive ones also remain Receive estimation function sequences that are based on the one hand Construction (B./32) and on the other hand on Observation values that correspond to a strongly mixing stationary process (see for example recursive density estimation).

Similar convergence could be found for other dependency structures say not yet to be derived. Also the continuity be conditions on the density function for consistent quantile value estimates so far haven't been significantly weakened, though in the use of this parameter even with discrete distributions no adverse effects were observed. The solution to this Problems are reserved for further research.

B.2. Practical design of the correction factor - Convergence acceleration

The design of the number sequence (a _t } in the recursion rule (B./18) is of great practical interest

is suggested by the conditions (B./4) and also in Method of stochastic approximation used. In particular however, if the start value is unfavorable, the corresponding one The sequence of estimators can often no longer be used in practice because the correction quantity quickly becomes very small and the convergence very small is slow. This fact even led to the practical Overall usability of algorithms of this kind is questioned became.

In the literature, with evidence of the preservation of the strong consi stence of the estimation function sequences the following design is proposed.

Be it

and

In terms of content, this choice of the sequence {a _t } means that the factor in front of the correction term is only reduced when the "correction direction" changes in the iteration sequence (observed over three values).

This approach has the advantage, as with numerous practical ones Applications could be proven that even with "far from Convergence value removed "start value the iteration sequence itself relatively fast (usually less than 20 steps) in one practically already satisfactory environment of the convergence value found.

These observations have been made through simulation studies for a lot number of distributions confirmed again and the different Behaviors of the usual (B./36) and the iteration sequences illustrated with (B./37) with respect to the acceleration of convergence. The realization of recursive algorithms using Sequences of numbers according to (B./35) but also according to (B./37) on microcomputer systems in real time and as part of z. B. one Signal analysis involves the necessary arithmetic operation "Division" further problems in itself. One would be cheap Limitation in the divisor to powers of 2, because then on every computer in the form of fast register rotation commands is feasible.

These conditions and the requirement (B./4) meet the consequences of shape

t = 1.2,. . ., k natural and p integer.

For example, for p = -1, 0, 1 these are the following:

The condition

is trivially fulfilled.

Next is

For p = 0 applies

and thus the 2 ^k -th member agrees with the 2 ^k -th member of the sequence

match.

In the computational implementation, it should of course be noted that the possible numbers of rotations are limited _{by the total number of bits in the representation of the variables a t.}

It should also be noted that sequence constructions of the form (B./38) can be combined with those according to (B./37).

B.3. Adaptive (inconsistent) statistics

The sequence {a _t } _{t = 1,2,. . .} and the arbitrary starting value of the iteration enable an adaptive design of estimation function sequences. As described in Section B./2, this sequence has a significant influence, at least in a practical sense, on the speed of convergence.

In numerous applications, it must now also be be assumed that z. B. signals to be analyzed only can be modeled piece by piece by stationary random sequences. In this class of tasks fall as well as those in the literature problems of robustness of statistics under the Condition of possible structural breaks, as well as the Localization of so-called "change points".

In this regard, the question arises as to how the estimation function sequences described in B./1 behave under conditions of structural breaks and trends. Simulation results show that structural breaks z. B. in the form of an abrupt change in the mean value of the time series under consideration then have less of an influence on the convergence behavior of the recursive estimation algorithms if the correction element is given by the sequence {a _t } _{t = 1,2,. . .} is still little compressed (i.e. when the structural break occurs at a small value of t).

This fact suggests the idea of using the design of this sequence for the automatic adaptation of the estimation function sequence to the value to be estimated (also under the condition of a possible change in the form of structural breaks and trends). As a rule, however, the degree of adaptivity and the convergence behavior are mutually influencing factors. If a _t = c = const. selected, a very quick adjustment can be achieved, but a convergence in the mathematical sense is no longer present because of the violation of the conditions (B./4). Using the example of the moment estimation, however, it can be made clear that this case is of great practical interest. It can be shown that recursive statistics designed in this way correspond to the frequently used method of exponential smoothing.

Sentence B8

For a _t = c = const., 0 <c <1, the estimation function sequence (B./19) (oBdA for k = 1) can be represented in the following form:

If µ: = E ξ _i and σ²: = Var ξ _{i are} denoted, the following also applies:

Proof: Let M ₀ start value and

M _{t + 1} = M _{t3072 00070 552 001000280000000200012000285913296100040 0002004039648 00004 32953UB <-c (M t -ξ t + 1 ). Then M₁ and M₂ can easily be brought into the form (p.3.39): M₁ = M₀-c (M 0 -ξ₁) = (ic) M 0 + c · ξ₁ M₂ = M₁-c (M₁-ξ₂) = ( 1-c) M₁ + c · ξ₂ = (1-c) ² M 0 + c (1-c) ξ₁ + c · ξ 2 For any t, (B./39) is shown by complete induction. Be it Then follows what was to be shown. With slight transformations one gets the further statements of the sentence: Note: If M 0 = ξ 0 (first measured value) is selected, the following applies The invention is to be explained on the basis of exemplary embodiments. 1) Epilepsy monitoring For diagnosis, therapy and therapy progress control of epileptic seizure disorders, the use of an automatic detection and further processing of epileptic grapho elements of the electroencephalogram (EEG) is necessary. With the aid of measuring electrodes of the EEG (electroencephalogram - electrical brain activity) on the human scalp 1, one or more channels are recorded and fed to a measured value recording device 2, 3, 4, 5, 6 via appropriate amplifier technology. The measured value acquisition device 2, 3, 4, 5, 6 consists either of a single or multi-channel analog / digital converter, with the help of which the measured values (EEG) are digitized in a computer-compatible manner, or of a portable recording device (Holter technology telemetry), the ensures long-term recording because it can be attached to the patient. The registration by means of a recording device can be done digitally, then an analog / digital conversion is carried out before the storage (e.g. magnetic cassette). The digitized measured values are transferred to a measured value processing device 7 , in which the evaluation takes place. By means of an additional device 14 , the result of the evaluation is either communicated immediately to the patient or the doctor or is recorded or stored for a later evaluation. In the exemplary embodiment, the measured value processing device consists of a computer unit. Here, the processing of the measured values begins with the measured values being transferred to the memory location of the computer by a reading device 8. This read-in device 8 is formed by computer assemblies in such a way that the data either from the analog / digital converter or from the recording medium (e.g. digital magnetic cassette) reach the main memory via the computer bus. You can then start processing the measured values. This is done with the aim of recognizing those signal patterns (grapho elements) in the signal curves that have a characteristic that deviates from the normal EEG and that characterize epileptic activity. These can be so-called spikes, spike-wave complexes, sharp waves and epileptic seizure activity in various forms. In order to be able to recognize such signal characteristics, a device for comparing values in the form of a comparison device 9 is required. This facility ensures that the course of EEG values is constantly compared with given values. These values can be stored in terms of value as a priori knowledge and called up by the comparison device 9 or determined by the comparison device 9. It is to be assumed here that the comparison device 9 is to be represented by calculation algorithms which are implemented in the program for the computer implementation. This algorithm (algorithms) ensures the comparison of a value-related undershoot or overshoot between EEG values and specified values. According to the visual classification of epileptic grapho elements, these differ from the so-called background activity of the EEG (background activity of the EEG characterizes the normal EEG signal characteristics as assessed by the doctor) by distinguishable changes in amplitude, frequency, pattern expression (spike wave) and duration of Grapho elements. Thus, medical experience (specified values) can be used for comparison, ie taken into account by the algorithm. Another possibility is the calculation of comparison values (predetermined values) from the EEG values themselves. This is done, for example, by comparing past EEG values with current values. An epileptic grapho element can then be identified and delimited with regard to its duration in such a way that the current EEG values exceed the previous EEG values multiplied by a factor and mark the beginning of the grapho element and the fall below its end. The previous EEG values would then characterize the characteristics of the background activity prior to the appearance of the grapho element. The degree of deviation of the epileptic grapho element to be specified from the background activity is introduced into the comparison algorithm by the multiplication factor (multiplication of evaluation results of previous EEG values). Both a starting device 10 and a stopping device 11 are controlled by the comparison device. These in turn control an additional device 14 which is switched on (activated) by the starter device 10 and switched off (deactivated) by the stop device. When an epileptic grapho element is recognized, it is removed from the sequence of EEG values and stored in a separate memory area (additional device 14 or memory 12 ). The starting device 10 is realized by an algorithm which, after the detection of the beginning of the epileptic grapho element, activates the memory area and writes the memory with the sequence of EEG values coming after the starting point of the grapho element. When the end of the grapho element is recognized, writing to the memory is ended. This is done by the algorithm which the stop device 11 implements. The additional device 14 consists of a unit which makes the stored grapho elements with the necessary additional information available to the doctor. Such a unit can be implemented by computer modules in that the stored grapho elements are displayed in a suitable form on the screen. Necessary additional information is the time of the beginning and the end of the grapho element (temporal context) as well as the identification for the registration channel (topographical context). The additional information is saved by the start or stop device 10, 11 . The additional device 14 is completed by a classifier. The stored grapho elements can thus be classified and displayed with regard to their signal properties. An evaluation device 20 is connected between the reading in of the measured values by the reading-in device 8 and the comparison with predetermined values in the comparison device 9. This serves for data reduction and processes the measured values (EEG measured values) into EEG values. Calculations of the EEG measured values can be used, which can characterize the different signal properties of background activity and epileptic grapho elements. The following calculations are available for this: - Peak values of the measured values (a) - Averaging (b) - RMS value (c) - Quantile values (d) - Moving mean value estimation (e) - Moving moment function estimation (f) - Moving centered moment function estimation (g) - recursive estimation of the moment function (h) - recursive estimation of the centered moment function (i) - recursive estimation for values of the autocorrelation function (j) - recursive estimation of functions of accumulated differences (k) - recursive estimation of quantile value interval limits (m) - recursive estimation of the mean value in Form of the quantile value interval middle (n) - Formation of adaptive mean values of the absolute amount (o) - Cross correlation (p) - Adaptively determined number of zero crossings of the measured value sequence corrected by the adaptively formed mean value (q) - Recursive estimation of the mean value of the absolute values of the quantile value interval exceedances (r) The calculations mentioned are both parallel and al s also applied one after the other (in a row). This means that evaluation devices 20 are operated in parallel as well as in series. The criteria for distinguishing between the background activity and the epileptic grapho elements as well as between the grapho elements themselves are hereby met. For spike detection, the recursive estimation of the mean value of the absolute values of the quantile value interval exceeded (r) is used as the first calculation step (amplitude criterion). This is done for all channels (parallel connection of evaluation devices 20 ). In parallel with this, the adaptively determined number of zero crossings of the measured value sequence (q) (frequency criterion) corrected by the adaptively formed mean value is determined. Then a cross-correlation calculation (p) is applied to the parameter curves (of all or selected channels), which result from the recursive estimation of the mean value of the absolute values of the quantile value interval exceedances (r). The selection is made on the basis of previous detection results and is supplied by the comparison device 9 as a selection criterion. The results of the calculations are fed to the comparison device 9. A spike is recognized when there is both a value-based exceeding of the recursive estimate of the mean value of the absolute values of the quantile value interval exceeding and a value-based exceeding of the cross-correlation values (inter-channel relations). In order to increase the efficiency of the detection, the value range of the adaptively determined number of zero crossings of the measured value sequence corrected by the adaptively formed mean value (frequency criterion) is checked. The values must lie within a certain range of values. Epileptic convulsive activity is detected by performing a recursive estimation of functions of accumulated differences (k). An estimate is made for an average frequency content of the EEG measured values. In parallel, the recursive estimation of quantile value interval limits is calculated as an amplitude criterion. Both parameter curves are fed to the comparison device 9 for each registration channel. Epileptic convulsive activity is recognized when the result of the comparison is that predetermined values of the comparison device 9 have been exceeded. The efficiency of both detection methods can be increased by the fact that the evaluating device 20, the arithmetic operation of the measured values or values calculated already in the upstream evaluation units 20 allows. If this link takes place in the form of the so-called 4-NN (Nearest Neighbors) interpolation algorithm, the detection can be carried out in the topographical context. As an example, the spike detection will be explained in a topographical context. The detection conditions already defined above serve as a basis for this. It is also checked how the topographical distribution of the mean value of the adaptively determined number of zero crossings of the measured value sequence (q) corrected by the adaptively formed mean value behaves in the detected spike interval in relation to the temporal environment of the spike (background activity). This reflects the fact that the topographical distribution of this parameter is stable from spike event to spike event and unstable in intervals of background activity. For this additional test, the mean value of the adaptively determined number of zero crossings of the measured value sequence (q) corrected by the adaptively formed mean value is determined in the interval already detected. This mean value is determined for all channels and processed into an image matrix using the 4-NN interpolation algorithm. This image matrix can be quantified by parameters (for example the structure of the image due to the variance of the image matrix and the mean value of the image matrix) and compared with predetermined values in the comparison device 9. The direct comparison of image matrices (current and specified) is also possible. In the case of generalized spike or convulsive activity, the aforementioned image parameters form a possibility of increasing the reliability of the detection. The flexible, context-related structure of the evaluation device 20 (parallel processing, processing in series) results in the necessity of a correspondingly adapted structure of the comparison device 9 . A parallel connection of comparison devices 9 is necessary when several measuring channels are processed. The degree of parallelization then increases again with the number of calculation functions used per channel in the evaluation device 20. Comparison devices 9 arranged in a row are necessary for hierarchically structured detection methods. The spike detection should be used again for explanation. The method explained so far for this purpose was based on the assumption that certain parameter profiles were calculated in the evaluation device 20 and a detection carried out by comparison with predetermined values in the comparison device 9 . Additional detection efficiency is achieved by including topographical parameters (calculated from the image matrix) of the image matrix calculated from the mean value of the adaptively determined number of zero crossings of the measured value sequence corrected by the adaptively formed mean value. The comparison of the topographical parameters takes place after the comparison of the parameter curves that are used for detection. Both comparison devices 9 are accordingly connected in series. For the detection and classification of epileptic grapho elements it is necessary to work in a defined range of sampling frequencies. For all calculations (evaluation device 20 ) and the detection method applied to it (comparison device 9 ), it is important to have more than 20 sampling points available per half period as a rule. Since the epileptic grapho elements have basic frequencies that vary both among themselves and in relation to individuals, an optimal adjustment of the sampling frequency is necessary. For this purpose, an adjusting device 40 is connected between the measured value acquisition unit 2, 3, 4, 5, 6 and the reading device 8. In this, the sampling frequency of the measured value acquisition device is controlled on the basis of calculations of the instantaneous frequency or the mean frequency of the EEG measured values. For this purpose, calculation functions for determining the adaptively determined number of zero crossings of the measured value sequence corrected by the adaptively formed mean value are implemented in the setting device 40. If the epilepsy monitoring is to be used especially for the determination of spike activity, the sensitivity of the detection algorithms of the comparison device 9 is sufficient. The additional device 14 can then be designed as a counter which evaluates the frequency of spike activity related to the time unit. If a critical frequency value is exceeded, an alarm function is triggered that informs the patient of the associated increased risk of seizures. This means that life-threatening accident situations (handling boiling water, climbing stairs, etc.) can be avoided before and during seizures, as the patient can prepare for a possible seizure. A prerequisite is a high efficiency of the spike detection, since a high number of false positive detections leads to false alarms and thus to additional alarming of the patient. 2) Control for biofeedback and functional electrical stimulation as well as prosthesis control by means of surface electromyogram The single or multi-channel registration of the surface electromyogram (EMG - electrical muscle activity) can be used for the three application areas of biofeedback, functional electrical stimulation (FES) and prosthesis control with appropriate context-related processing. The EMG is registered and amplified by means of surface electrodes. The amplified EMG measured values are fed to a measured value acquisition device 2, 3, 4, 5, 6 , then evaluated in the measured value processing device 7 and values resulting from the processing are transferred to the additional device 12 (hereinafter referred to as the control device). The control device 12 generates signals that a) control devices for biofeedback that enable the patient to visually or acoustically control body functions that deviate from the normal function and to be treated (e.g. muscle tension), b) a stimulus device for functional electrical stimulation control, which attempts to correct and approximate the body functions deviating from normal function and to be treated by electrical stimulation of normal function, c) generate control signals for the prosthesis control, which induce the prosthesis to make directed movements through conscious and targeted activation of muscles that reproduce natural movements Measured value acquisition consists of a single or multi-channel analog / digital converter that digitizes the EMG measured values. A read-in device 8 enables the digitized EMG measurement values to be stored. The read-in device 8 consists of computer modules which transport the EMG measured values from the analog / digital converter via the computer bus into the main memory of the computer. In order to be able to recognize defined signal characteristics, a device for comparing values in the form of a comparison device 9 is required. This device ensures that there is a constant comparison of the course of EMG values with given values. These values can be stored in terms of value as a priori knowledge and called up by the comparison device 9 or determined by the comparison device 9. It is to be assumed here that the comparison device 9 is to be represented by one or more calculation algorithms which are implemented in the program for the computer implementation. The algorithms guarantee the comparison of a value-related falling below or exceeding between EEG values and given values. Furthermore, the comparison device 9 must ensure that the correspondence of the value range of the current EMG values with a predetermined value range is determined, the departure from a predetermined, unnecessarily coherent value range is signaled, the change in the current EMG values in a predetermined time interval by more than one predetermined constant recognized and states of the sequence of the current EMG values such as monotonically increasing or monotonously decreasing or monotonously increasing or monotonously decreasing or constant or an inverse state of the above-mentioned states can be determined. The described states of EMG values can represent the deviation from the normal state (a priori specification or learned from the algorithm), which the comparison device 9 determines and controls a start device 10 and a stop device 11 with the result. These in turn control a control device 12 which is switched on (activated) by the start device 10 and switched off (deactivated) by the stop device. A few concrete examples should clarify this. For biofeedback, it is essential to determine whether an EMG activity has exceeded or fallen below. In this case, the control device 12 consists of a device that enables the patient, visually or acoustically, to monitor body functions that deviate from the normal function and are to be treated (e.g. muscle tension). The signal effect is set by the patient or doctor in such a way that either the exceeding or undershooting is communicated to the patient or the remaining in the specified range of values is signaled. The choice is made dependent on the desired therapy goal or on the patient's individual criteria. The starting device 10 then starts the control device 12 , which generates the signaling, and the stop device 11 switches off the signaling by the control device 12 . In this case, the control device 12 can consist of an acoustic device or a light signal device. In the case of functional electrical stimulation and prosthesis control, dynamic properties of the EMG values such as the change in the current EMG values in a predetermined time interval by more than a predetermined constant and the dynamics of the sequence of the current EMG values such as monotonically increasing or monotonically decreasing or monotonically strictly increasing or monotonously strictly decreasing or constant or an inverse state of the states indicated above is evaluated. In accordance with the evaluation result, the control device 12 is switched on via the start device 10 and switched off via the stop device 11 . In this case, the control device 12 represents the actuator or part of the actuator of a closed control loop. The controller consists of the comparison device 9 with a target / actual value comparison (comparison between predetermined and current EMG values). In this case, the control device 12 can be switched off (stop device 11 ) if extremely unphysiological EMG values (exceeding or falling below) are detected. This can be caused by measurement artifacts (detachment of the electrodes, cable breakage, etc.). The control intervention by the control device 12 would lead to excessive stimulation currents in the functional electrical stimulation and undesired deflections in the prosthesis control. If the EMG values remain in a specified range of values, a control effect is effective that can be modified with regard to its control dynamics. This is done by connecting or disconnecting correction elements of defined dynamics, which are known from control technology. The start or stop device 10, 11 is also used for this purpose. Which correction elements are switched on or off depends on whether the change in the current EMG values in a predetermined time interval by more than a predetermined constant or states of the sequence of current EMG values such as monotonically increasing or monotonically decreasing or strictly monotonously increasing or monotonously decreasing or constant or an inverse state of the above states were determined. With biofeedback, this regulation with dynamic correction is carried out by the person (patient) himself. An evaluation device 20 is connected between the reading in of the measured values by the reading-in device 8 and the comparison with predetermined values in the comparison device 9. This serves for data reduction and processes the measured values (EMG measured values) into EMG values. Calculations of the EMG measured values can be used, which characterize the different signal properties. The following calculations are available for this: - Peak values of the measured values (a) - Averaging (b) - RMS value (c) - Quantile values (d) - Moving mean value estimation (e) - Moving moment function estimation (f) - Moving centered moment function estimation (g) - recursive estimation of the moment function (h) - recursive estimation of the centered moment function (i) - recursive estimation for values of the autocorrelation function (j) - recursive estimation of functions of accumulated differences (k) - recursive estimation of quantile value interval limits (m) - recursive estimation of the mean value in Form of the quantile value interval middle (n) - Formation of adaptive mean values of the absolute amount (o) - Cross correlation (p) - Adaptively determined number of zero crossings of the measured value sequence corrected by the adaptively formed mean value (q) - Recursive estimation of the mean value of the absolute values of the quantile value interval exceedances (r) The calculations mentioned are both parallel and al s also applied one after the other (in a row). This means that evaluation devices 20 are operated in parallel as well as in series. The following calculations are used as EMG values: - the peak values (a) - the averaging (b) - the effective value calculation (c) - the quantile value determination (d) from the value sequences that quantify the EMG activity (EMG measured values). These are - sliding moment function estimation (f) - sliding centered moment function estimation (g) - recursive estimation of the moment function (h) - recursive estimation of the centered moment function (i) - recursive estimation for values of the autocorrelation function (j) - recursive estimation of quantile value interval limits (m ) - Formation of adaptive mean values of the absolute amount (o) - Cross-correlation (p). To achieve smoothing effects, the - sliding mean value estimation (e) - recursive estimation of the mean value in the form of the quantile value interval middle (n) can be applied to all value sequences. The efficiency of the evaluation of Oberflächenmyogrammen can be increased in that the evaluating device 20, the arithmetic operation of the measured values or values calculated already in the upstream evaluation units 20 allows. If this link is made in the form of the so-called 4-NN (Nearest Neighbors) interpolation algorithm, the evaluation can be carried out in the topographical context. Electrodes for registering the surface electromyogram are attached to a muscle area and enable the electrical muscle activity to be evaluated over a large area. The EMG measured values are fed to the evaluation device 20 via amplifiers, the measured value acquisition device 2, 3, 4, 5, 6 and the read-in device 8. The parameters that are equivalent to performance (physical performance) are calculated here. These are alternatively - rms value (c), - quantile values (d), - sliding moment function estimation (f), - sliding centered moment function estimation (g), - recursive estimation of the moment function (h), - recursive estimation of the centered moment function (i), - recursive estimates for values of the autocorrelation function (j) and recursive estimation of quantile value interval limits (m). These parameters (EMG values) are made equivalent to power density by averaging (b) in defined time intervals. This mean value is determined for all channels and processed into an image matrix using the 4-NN interpolation algorithm. This image matrix can be quantified by parameters (e.g. the structure of the image by the effective value (c) of the image matrix and the mean value (b) of the image matrix) and compared with predetermined values in the comparison device 9. The direct comparison of image matrices (current and specified) is also possible. This means that topographic activity patterns can also be included for biofeedback application, functional electrical stimulation and prosthesis control. The flexible, context-related structure of the evaluation device 20 (parallel processing, processing in series) results in the necessity of a correspondingly adapted structure of the comparison device 9 . A parallel connection of comparison devices is necessary when several measuring channels are processed. The degree of parallelization then increases again with the number of calculation functions used per channel in the evaluation device 20. To the functioning of biofeedback, Functional Electrical Stimulation it is necessary that between the comparison device 9 and starting device 10 and / or between the comparison device 9 and stop device 11 (cases A) and / or between judging means 20 and a further evaluation device 20 and / or between the comparing means 9, and a further comparison device 9 and / or a time delay device 30 is connected between the evaluation device 20 and the comparison device 9 (cases B). This corrects time differences that result from - the different computing times for the individual algorithms and - z. B. Delay differences that result from the different window lengths in sliding calculations (case B). Furthermore, it is necessary to use time delay devices for dynamic correction of the control in functional electrical stimulation and prosthesis control (case A). Since the registered surface electromyograms have frequency ranges that vary with one another as well as with individual persons, an optimal adjustment of the sampling frequency is necessary. For this purpose, an adjusting device 40 is connected between the measured value acquisition device 2, 3, 4, 5, 6 and the reading device 8. In this, the sampling frequency of the measured value acquisition device 2, 3, 4, 5, 6 is controlled on the basis of calculations of the instantaneous frequency or the mean frequency of the EEG measured values. For this purpose, calculation functions for determining the recursive estimation of functions of accumulated differences (k) or the adaptively determined number of zero crossings of the measured value sequence (q) corrected by the adaptively formed mean value are implemented in the setting device 40.}

Claims (56)

1. A measured value processing device ( 7 ) which is connected to a biological object ( 1 ) via multi-channel measured value acquisition devices ( 2, 3, 4, 5, 6 ) and is coupled to an additional device (12, 13, 14 ),
wherein the measured value processing device ( 7 ) comprises:
a first device ( 8 ) for single-channel or multi-channel reading of measured values from at least one of the measured value acquisition devices ( 2, 3, 4, 5, 6 ) into the measured value processing device ( 7 ),
a second device ( 9 ) for comparing the measured values with predetermined values,
a third device ( 10 ) for starting the additional device ( 12, 13, 14 ) when a predetermined first value is reached,
a fourth device ( 11 ) for stopping the additional device ( 12, 13, 14 ) when a predetermined second value is reached,
characterized,
that the second device ( 9 ) is arranged for comparing the measured values with predetermined single-channel or multi-channel measured value structures,
that the third device ( 10 ) is arranged for starting the additional device ( 12, 13, 14 ) when a predetermined first measured value structure is reached,
that the fourth device ( 11 ) is arranged for stopping the additional device ( 12, 13, 14 ) when a predetermined second measured value structure is reached, and
that between the first ( 8 ) and second device ( 9 ) a fifth device ( 20 ) designed as an evaluation device for calculating values of parameters from measured value sequences for the purpose of data reduction is arranged. 2. measured value processing device according to claim 1, characterized in that the comparison device ( 9 ) is arranged so that a pattern of measured values or values of the at least one channel (k 1 , k 2 , k 3 , k 4 , k 5 ) with a predetermined Pattern is compared, if they match, the start device ( 10 ) or the stop device ( 11 ) is activated. 3. Measured value processing device according to claim 1, characterized in that the comparison device (9 ) is arranged so that the start device ( 10 ) or the stop device ( 11 ) is activated when a predetermined measured value or value is exceeded. 4. Measured value processing device according to claim 1, characterized in that the comparison device (9 ) is arranged so that the start device ( 10 ) or the stop device ( 11 ) is activated when a predetermined measured value or value is not reached. 5. Measured value processing device according to claim 1, characterized in that the comparison device (9 ) is arranged so that the start device ( 10 ) or the stop device ( 11 ) is activated when the measured values or values leave a predetermined, unnecessarily coherent range of values. 6. Measured value processing device according to claim 1, characterized in that the comparison device (9 ) is arranged so that the start device ( 10 ) or the stop device ( 11 ) is activated when the measured values or values in a predetermined time interval by more than a predetermined one Change constant. 7. Measured value processing device according to claim 1, characterized in that the comparison device (9 ) is arranged so that the start device ( 10 ) or the stop device ( 11 ) is activated when the measured value sequence or the sequence of values monotonically increasing or monotonically decreasing or monotonously strictly increasing or monotonously decreasing or strictly decreasing or constant or an inverse state of the states indicated above. 8. measured value processing device according to claim 1 to 7, characterized in that the comparison device (9 ) the measured value sequence or the sequence of values of one of the channels (k 1 , k 2 , k 3 , k 4 , k 5 ) are supplied and that the comparison device ( 9 ) a second comparison device (9 ) is connected in series, to which the sequence of measured values or the sequence of values of at least one further channel (k 2 , k 3 , k 4 , k 5 ; k 1 ) are fed. 9. Measured value processing device according to claim 1 to 7, characterized in that the comparison device (9 ) is supplied with the sequence of measured values or the sequence of values of one of the channels (k 1 , k 2 , k 3 , k 4 , k 5 ) and that the comparison device ( 9 ) a second comparison device (9 ) is connected in parallel to which the sequence of measured values or the sequence of values of at least one further channel (k 2 , k 3 , k 4 , k 5 ; k 1 ) are fed. 10. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) is designed to determine peak values of the measured value sequence or the sequence of values as one of the parameters. 11. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) is designed to form mean values of the measured value sequence or the sequence of values as one of the parameters. 12. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) is designed to form effective values of the measured value sequence or the sequence of values as one of the parameters. 13. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) is designed to form quantile values of the measured value sequence or the sequence of values as one of the parameters. 14. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) is designed to form sliding mean estimates of the measured value sequence or the sequence of values as one of the parameters. 15. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) is designed to form sliding torque function estimates of the measured value sequence or the sequence of values as one of the parameters. 16. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) is designed to form sliding, centered moment function estimates of the measured value sequence or the sequence of values as one of the parameters. 17. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) is designed to form recursive estimates of the moment function of the measured value sequence or the sequence of values as one of the parameters. 18. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) is designed to form recursive estimates of the centered moment function of the measured value sequence or the sequence of values as one of the parameters. 19. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device ( 20 ) is designed to form recursive estimates for values of the autocorrelation function of the measured values as one of the parameters. 20. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) is designed to form recursive estimates of functions of accumulated differences in the measured value sequence or the sequence of values as one of the parameters. 21. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) is designed to form recursive estimates of quantile value interval limits of the measured value sequence or the sequence of values as one of the parameters. 22. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device ( 20 ) is designed to form recursive estimates of the mean value in the form of the quantile value interval center of the measured value sequence or the sequence of values as one of the parameters. 23. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device ( 20 ) is designed to form recursive estimates of the mean value of the absolute values of the quantile value interval exceedances of the measured value sequence or the sequence of values as one of the parameters. 24. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) is designed to form adaptive mean values of the absolute amount of the measurement sequence or the sequence of values as one of the parameters. 25. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device ( 20 ) is designed to form adaptively determined numbers of zero crossings of the measured value sequence corrected by the adaptively formed mean value or the sequence of values as one of the parameters. 26. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) for forming recursive cross-correlation functions on the basis of the measured value sequences or the sequences of values of two channels (k 1 , k 2 , k 3 , k 4 , k 5 ) is designed as one of the parameters. 27. Measured value processing device according to one of claims 1 to 9, characterized in that the evaluation device (20 ) is designed to form recursive estimates of functions of accumulated cross differences of the measured value sequences or the sequences of values from two channels as one of the parameters. 28, measurement processing according to any one of claims 1 to 27, characterized in that the evaluation device (20) the measured values of at least one channel (k 1, k 2, k 3, k 4, k 5) is supplied and that the evaluation means (20) a second evaluation device (20 ) is connected in series, to which the measured values of at least one further channel (k 2 , k 3 , k 4 , k 5 ; k 1 ) are fed. 29, measurement processing according to any one of claims 1 to 27, characterized in that the evaluation means (20) the measured values of at least one channel (k 1, k 2, k 3, k 4, k 5) is supplied and that the evaluation means (20) a second evaluation device (20 ) is connected in parallel, to which the measured values of at least one further channel (k 2 , k3 , k 4 , k 5 ; k 1 ) are fed. 30, measurement processing according to any one of claims 1 to 27, characterized in that the evaluation means (20) the measured values of multiple channels to the channels (k 1, k 2, k 3, k 4, k 5) are supplied and that the evaluation device (20 ) is designed for arithmetic linking of the measured values of the channels (k 2 , k 3 , k 4 , k 5 ; k 1 ). 31. Measured value processing device according to one of claims 2 to 30, characterized in that several evaluation devices ( 20 ) and / or comparison devices (9 ) are connected in series and / or in parallel with one another. 32. Measurement processing device according to one of claims 1 to 31, characterized in that between the comparison device (9 ) and the start device ( 10 ) and / or between the comparison device (9 ) and the stop device ( 11 ) and / or between the evaluation device (20 ) and another Evaluation device ( 20 ) and / or a time delay device (30 ) is connected between the comparison device (9 ) and a further comparison device ( 9 ) and / or between the comparison device (9 ) and the evaluation device (20). 33. Measured value processing device according to claim 32, characterized in that the time delay device (30 ) is arranged so that the start device ( 10 ) or the stop device ( 11 ) is activated in time after the measured value structure corresponds to the predetermined measured value structures. 34. Measured value processing device according to claim 32, characterized in that the time delay device (30 ) is arranged so that the start device ( 10 ) or the stop device ( 11 ) is activated before the measured value structure agrees with the predetermined measured value structures. 35. Measured value processing device according to one of claims 1 to 34, characterized in that the reading device ( 8 ) is preceded by an adjusting device (40 ) for setting the sampling frequency of at least one measured value acquisition device ( 2, 3, 4, 5, 6 ). 36. Measured value processing device according to claim 35, characterized in that the setting device (40 ) is designed for setting the sampling frequency by means of recorded mean value crossings of the measured value sequences. 37. Measured value processing device according to claim 35, characterized in that the setting device ( 40 ) is arranged such that the sampling frequency is set by means of recursively recorded mean value crossings of the measured value sequence. 38. Measured value processing device according to one of claims 1 to 37, characterized in that the evaluation device ( 20 ) is designed such that the recursively formed parameters are designed as adaptively formed parameters. 39. Measured value processing device according to one of claims 1 to 38, characterized in that the additional device is a control device ( 12 ) for controlling the biological object ( 1 ). 40. Measured value processing device according to one of claims 1 to 39, characterized in that the additional device is a memory device ( 13 ) for storing the measured values. 41. Measured value processing device according to one of claims 1 to 40, characterized in that the additional device is a warning device ( 14 ) for displaying an undesired state of the biological object ( 1 ). 42. Measurement processing device according to one of claims 1 to 41, characterized in that the multi-channel measurement recording device ( 2, 3, 4, 5, 6 ) is designed to record an electroencephalogram. 43. Measured value processing device according to one of claims 1 to 41, characterized in that the multi-channel measured value acquisition device ( 2, 3, 4, 5, 6 ) is designed to acquire an electromyogram. 44. Measured value processing device according to one of claims 1 to 41, characterized in that the multi-channel measured value acquisition device ( 2, 3, 4, 5, 6 ) is designed to detect evoked potentials. 45. Measured value processing device according to one of claims 1 to 41, characterized in that the multi-channel measured value acquisition device ( 2, 3, 4, 5, 6 ) is designed to acquire cortical direct voltage potentials. 46. Measured value processing device according to one of claims 1 to 45, characterized in that the multi-channel measured value acquisition device ( 2, 3, 4, 5, 6 ) is designed for the additional acquisition of polygraphic data. 47. Measured value processing device according to one of claims 1 to 46, characterized in that the multi-channel measured value acquisition device ( 2, 3, 4, 5, 6 ) is designed for the additional acquisition of mechanographic data. 48. measured value processing device according to claim 46, characterized in that the polygraphic data the Measurements of breathing, heart rate, electrocardiogram, Respiratory rate, blood pressure, electrooculogram and temperature contains. 49. measured value processing device according to claim 47, characterized in that the mechanographic data the Measurement of force, distance, speed, acceleration, Includes compression, tension, torque and torsion variables. 50. Measured value processing device according to one of claims 1 to 39, characterized in that the control device (12 ) contains an acoustic and / or visual signal effect for biofeedback. 51. Measured value processing device according to one of claims 1 to 39, characterized in that the control device ( 12 ) contains electrical stimulation for functional electrical stimulation. 52. Measured value processing device according to one of claims 1 to 39, characterized in that the control device (12 ) contains a control and regulation of prostheses. 53. A measured value processing device (7 ) connected to a biological object ( 1 ) via multi-channel measured value acquisition devices ( 2, 3, 4, 5, 6 ), which is coupled to an additional device ( 12, 13, 14 ), characterized in that,
that the measured value processing device ( 7 ) comprises a reading device ( 8 ) for multi-channel reading in of measured values from at least one of the measured value acquisition devices ( 2, 3, 4, 5, 6 ), and
that with the reading device ( 8 ) an adjusting device (40 ) for setting the sampling frequency of at least one measured value acquisition device ( 2, 3, 4, 5, 6 ) is connected upstream. 54. Measured value processing device according to claim 53, characterized in that the setting device (40 ) is designed for setting the sampling frequency by means of recorded mean value crossings of the measured value sequences. 55. Measured value processing device according to claim 53, characterized in that the setting device ( 40 ) is arranged such that the sampling frequency is set by means of recursively recorded mean value crossings of the measured value sequence. 56. With a biological object ( 1 ) via several channels (k 1 , k 2 , k 3 , k 4 , k 5 ) of measured value acquisition devices ( 2, 3, 4, 5, 6 ) connected measured value processing device ( 7 ), which with a Additional device ( 12, 13, 14 ) is coupled, characterized in that the measured value processing device ( 7 ) comprises:
a first device ( 8 ) for single-channel or multi-channel reading of measured values from at least one of the measured value acquisition devices ( 2, 3, 4, 5, 6 ) into the measured value processing device ( 7 ),
a second device ( 9 ) for comparing the measured values with predetermined values, a further channel (k 2 , k 3 , k 4 , k 5 , k 6 ) being read in when a first predetermined measured value structure is reached.