SU1008888A1 - Adaptive filter - Google Patents

Abstract

1, the ADAPTIVE FILTER containing a multi-channel model, whose information inputs are the input of an adaptive filter, and the control inputs are connected to the outputs of the weighting function of the weight function, characterized in that, in order to improve the noise immunity of receiving repetitive signals of unknown shape, it has been entered two adders and a threshold element, while the outputs of the multichannel model are connected to the inputs of the first adder and to the information inputs of the weighting function of the weighing function, the control input of which is connected via a threshold element with the output of the first adder, and vnhody - a second adder inputs. 2. The filter according to Claim 1, which means that the shaper of the weighting function contains N channels, each of which consists of a series-connected key, a storage unit and an amplitude modulator, and the information inputs of the keys are informational. the shaper inputs of the weight function, the control inputs of the keys are combined and are the control input of the shaper of the weight function, the second inputs of the amplitude modulators are connected to the outputs of the common generator, and the outputs are the outputs of the shaper of the weight function.

Description

00 CX) 00 X The invention relates to radio engineering and can be used in devices for receiving and processing signals of an unknown fortiy; An adaptive multichannel filter is known that contains a multichannel model and shaper of a weight function, and the information input of the multichannel model is the filter input, the control input of the multichannel model is connected to the output of the weight function shaper, and the output of the multichannel model is the output of the multichannel filter on a model that performs parallel filtering of N signals, where N is the number of channels of Model 1. The FORM of the frequency response of each of the channels of a multichannel filter can be changed without changing the structure of the device by replacing the weighting function associated with the impulse response of the filter with the ratio V (x) T-g (where VCx) is the weighting function; g (y-) g (t) -pulse characteristic of the synthesized filter; . Td is the signal processing time of the model torus; V is the recording speed of the weight function on the non-inertial photoresistive layer of the modulator. Filters of this type have constant parameters. They allow for optimal filtering of the signal if its parameters are precisely known at the receiving point. If the form of the received signal is unknown, for optimal filtering it is necessary to adapt the filter parameters to the parameters of the received signal of an unknown form, that is, a known filter does not allow for the reception of signals of unknown forms with sufficient noise immunity. The purpose of the invention is to increase the interference immunity of receiving repetitive signals of unknown shape. The goal is achieved by the fact that the filter containing a multi-channel model, whose information inputs are the input of the adaptive filter, and the control inputs are connected to the outputs of the weight function generator, two adders and a threshold element are entered, while the outputs of the multi-channel modeldator are connected to the inputs the first adder, and to the information inputs of the driver of the weighting function, the control input of which is connected via a threshold element with the output of the first adder, and the outputs - c. inputs of the second adder. Moreover, the weight function generator contains N channels, each of which consists of a series-connected key, a storage unit and an amplitude modulator, and the information inputs of the keys are information inputs of the weight function generator, the control inputs of the keys are combined and are the control function of the weight function The second inputs of the amplitude modulators are connected to the outputs of the master oscillator, and the outputs are the outputs of the weight function generator. FIG. 1 depicts a structural electrical adaptive filter circuit; in fig. 2-structured electrical circuit for shaping weight function. The adaptive filter (Fig. 1) contains a multichannel model 1, a shaper 2 weight functions, an adder 3, a threshold element 4, an adder 5, a low-pass filter 6. Shaper 2 weight functions (Fig, 2) contains the master oscillator 7, the amplitude module. tori 8, storing blocks 9 and keys 10. Adaptive filter works as follows. The input signal Uj (t), which is an additive mixture of the unknown repeat signal IF (1) and the fluctuation noise (noise) Un (t) U (t) Uc (t) + U (t) is fed to the combined information input mn ( 5 channel model 1. The number of used channels of model 1 is determined by the expression N B, where B is the base of the received signal Ug (t), Ffng is the top frequency of the signal spectrum Uc (t), T is the signal duration and (t) . The control inputs of all N channels of the model 1 from the corresponding outputs of the imaging device 2 are subjected to the voltage of the weighting functions. The weighting functions of the shaper 2 are produced as follows. From the N outputs of the oscillator 7 of the shaper 2 to the second inputs of the corresponding amplitude modulators 8 of the shaper 2, short pulses of constant amlita (S-pulses) are received, duration cg, selected from the condition where Xa - the size of one element of the conductive layer - modeler 1 along the direction of recording the weight function. The interval of following 5 pulses at each of the outputs of the master generator 2 must not be longer than the duration of the signal Uc; (t1, and 5-imp pulses on the i-output of the master oscillator 7 f) are relative to 6 pulses on the (1-1) th course of the oscillator 7 on the integral shaft, defined by the Kotelnikov theorem. To the input of each of the amplitude modulators 8 of the forcing 2 with co The corresponding storage unit 9 receives the voltage IJ- (t) (where), which modulates the amplitude of 6 pulses, applied to the amplitude modulators 8 s of the master oscillator. Modulated in amplitude 5, the pulses from the outputs of the amplitude modulators 8 of the former 2 are fed to the corresponding control inputs. E1Y of the multichannel model 1 .. when the adaptive filter turns on the output of the model 1 and the voltage is practically zero obviously below the initial threshold voltage of the threshold element 4. At the same time, the control voltage and the output of the threshold element 4 are missing keys 10 of the driver 2, the control inputs of which are interconnected h connected to the output of the threshold element 4 are closed, the voltage Olji (t) on memorized on eSlocks 9 of former 2 are absent. At the same time, the non-stimulated 5th pulses are controlled by the control echojs of all N channels of model 1 j. Through the interval TQ at the output of each of the channels of model 1, the output voltage L ,, is obtained (), tor defined by the expression:, (H, 0, () 04 where the current reference point of the input signal in the i-th channel torus, which runs through the values from O to T with the velocity of the interchange of the weight function over the weight layer of the i-th channel of model 1, and to, -, where K is the proportionality coefficient depending on the type of modulator 1. and its design functional characteristics, Tp is the processing time of the input signal U, (t) in modulator 1. Thus, o6pai3bM, at the moment on during all N channels, the modulator and 1 is the voltage Ugj, x (T), which represents samples of the instantaneous values of the input signal taken at intervals At. All N samples of the input signal from the outputs of model 1 are fed to the adder 3, the output voltage of which is determined by the correlation Ultrasound vS to. Voltage Uj is applied to Horn element 4, where it compares Chp (t) Unopo uUnop (t) with a variable voltage with a change in the threshold voltage, where the component of the threshold voltage is generated from the signal applied to the input the threshold element 4. At the time of the inclusion of the adaptive filter the horn voltage of pores () is equal to the initial value of PORv- during the whole time of reception of a repetitive signal Ug (t) voltage. Uj, generated as the sum of samples of the input signal U (t), is applied to the threshold element 4. This voltage varies over time both due to the change in the level of the input signal UgCt), which is a mixture of useful signal and noise, and the account of the adaptive approximation of the filter parameters on the modulator 1 to the optimal ones. If the voltage at the su1 output of the viator 3 does not exceed (t), then the voltage U at the output of the threshold element 4 is absent, its threshold voltage does not change, the keys 10 in the driver 2 are closed, the voltage on the storage unit and there are no, unmodulated S-pulses are fed to the control inputs of the Kanshov model 1 and the adder 5. From adder 5, the unmodulated successor of the HFCTb 5 impulses shifted from one another relative to each other by g is fed to. filter 6 .. At the same time, at the output of filter 6, a constant voltage is separated, which is the envelope of the unmodulated S-impulse. SOB. . in the case when the voltage at the output of the adder 3 exceeds the threshold voltage U |, op (t) of the threshold element 4, the output of the voltage Ujj unlocking the keys 10 of the driver 2 at the output, In addition, in the very threshold element 4 automatic adjustment of the threshold voltage ni. When the keys 1C are unlocked, l 2 will be transferred to the storage blocks 9 of the generator 2, the readings of the input signal J (t) from the output of the model 1 are transferred, modulating (in amplitude modulators 8) S-pulses from the master oscillator 7 From the output of amplitude modulators B: shaper 2 amplitude-modulated $ -pulses

served on the board; The inputs of the multichannel model 1 to the adder 5, the output of which is modular, the bath in amplitude, a sequence of 5 pulses, a low-pass filter, which distinguishes the voltage proportional to the envelope of this sequence.

If the input signal Ug (t) contains only the noise Uf (t), then the response times of the threshold element 4 will be completely random and the values of the input signal samples carried by c. the outputs of the multichannel modulator 1 through the keys 10 into the storage units 9 of the imaging unit 2 will also be random, as a result of the voltage. Uaj (t) on storage units 9 will randomly fluctuate, remaining on average close to zero. At the same time, from the driver 2 to the control inputs of the model 1. and the adder 5 of the readout signal receives 5 pulses, the amplitude of which fluctuates relative to the mean value in very small limits. At the output of the adaptive filter, the noise voltage Uewxnft is applied.

In that case, when the input signal is the sum of the noise voltages and f .t) and the useful signal U (., (T), the probability of triggering of the threshold element 4 increases due to the increase in the energy of the Input signal U ;, (t), consequently, the total voltage of the input signal Ug (t) samples to the threshold element 4, the higher the signal-to-noise ratio at the input of the adaptive filter, the more likely the triggering of the threshold element 4 at the instants of the appearance of the useful signal Uc ( t). But even with small signal-to-noise ratios, the triggering of the threshold element 4 will occur more often if there is a useful signal at the input of the adaptive filter than when it is not present. After several repetitions of the useful signal that caused the triggering of the threshold element 4, the values of the stresses (t) on the storage units 5 of the imager 2 approximate m samples of the useful signal .. This is due to the averaging of the voltage surges Uq (t) caused by noise, with the accumulation of samples of the input signal on the storage blocks 9. The voltages UgiCt modulate the 5-pulses applied to the amplitude modes l tori 8 oscillator 7. This value stresses the weighting functions supplied from generator 2, approach the optimal, which correspond to the instantaneous values of Samples Leznov Ueftl signal. By approximating the weight functions being formed to the optimal ones, the frequency of correct (when a useful signal appears) triggering of threshold element 4 will increase, as the response of each channel of model 1 to the useful signal will increase, which will lead to an increase in voltage. Uj (t) at the output of the adder 3. The adaptation process is accelerated. For single temporarily decreasing the random operation of the threshold element 4 under the influence of noise, at intervals, when the useful signal Ucff) is absent at the input, the response threshold of the prog element 4 increases automatically with increasing voltage Usft) at the output of the adder 3.

Thus, when, at the input voltage of the variable filter U | (t) there is no repeating signal U (. (T), the random element 4 triggers, the average values of the voltages on the memory blocks 9 of the generator 2 remain zero and threshold) in threshold element 4 saves the value close to 1} n

When the repetitive signal UcCt) appears, the percentage of random triggering of the threshold element 4 gradually decreases. The voltage on the storage units 9 increasingly approaches the optimal values, modulated by these voltages 6-pulses applied to the adder 5, the output samples signal, tends to the value of the samples and the useful signal, the threshold voltage of the threshold element 4 increases, and the voltage at the output of the filter 6 more and more accurately repeats the useful signal Uc (t), since it is an envelope STI useful signal samples U (t), taken received through interval T dt (V, which allows accurately reconstruct the shape of the signal.

Thus, the proposed device, by adapting to the parameters of a repeating signal of unknown shape, taken against the background noise, and automatically forming an optimal weighting function, makes it possible to increase the noise immunity of the reception of this signal.

The proposed device provides not only an increase in the noise immunity of receiving any repeated signals of unknown shape by their optimal filtering (if the duration and the upper frequency of the signal spectrum are known), but also the optimal detection of the regular signal, because when the levels of the modulated 5 pulses are approached.

To the optimum, the voltage at the output of the adder 3 reaches a maximum corresponding to the optimal detector.

The proposed device can be used in detection systems, radio astronomy, radiolocation and radio direction finding. The time the adaptive filter enters the optimal filtering mode depends on the signal-to-noise ratio i at the filter input and the repetition frequency of the input signal.

Claims (2)

1. ADAPTIVE FILTER containing a multi-channel simulator, the information inputs of which are the input of the adaptive filter, and the control inputs are connected to the outputs of the weight function shaper, characterized in that, in order to increase the noise immunity of the known adder of repeated informal signals, two and a threshold are introduced into it element, while the outputs of the multichannel simulator are connected to the inputs of the first adder and to the information inputs of the forwarder of the weight function, the control input of which is connected through thresholds of the output element of the first adder and the outputs - to the inputs of the second adder. 2. The filter according to claim 1, with the fact that, the shaper of the weight function contains N channels, each of which consists of a series-connected key, a storage unit and an amplitude modulator, moreover, the information inputs of the keys are information inputs of the shaper of the weight function, § control inputs of keys are combined | and are the control input of the generator of the weight function, the second to the inputs of the amplitude modulators of the connection. They are to the outputs of the master oscillator, I and the outputs are the outputs of the generator of the weight function. δ □ O ·. □ o □ o eo