CN-115828065-B - Electromagnetic shielding intelligent mower boundary inside and outside recognition method based on DTW algorithm

Abstract

The invention discloses an electromagnetic shielding intelligent mower boundary internal and external identification method based on a DTW algorithm. The method comprises the steps of firstly recording the condition of no background noise, establishing normal and effective signal library members, carrying out internal and external classification and marking, periodically collecting induction current signals by an induction probe of the intelligent mower, and extracting time domain characteristic values of the induction current signals. Secondly, background noise suppression is carried out on the induced current signals, approximate effective signal waveform fragments are intercepted, and sorting and merging are carried out. And finally, calculating the matching value of the approximate effective signal waveform fragment and the normal effective signal library based on a DTW algorithm, and searching an optimal reference signal by combining the time domain characteristic value and the matching value of each fragment so as to determine the judgment of the internal and external states of the boundary of the intelligent mower. The invention has strong anti-interference capability and wide application scene, and can improve the accuracy of the inside and outside judgment of the system boundary by adding the effective signal template under the specific environment into the normal effective signal library.

Inventors

• CHEN KEMING
• CHEN LIYU
• LAI SHENGJI
• QI LINGLING

Assignees

• 杭州电子科技大学

Dates

Publication Date

20260508

Application Date

20221206

Claims (3)

1. 1. The method for identifying the inside and outside of the boundary of the electromagnetic shielding intelligent mower based on the DTW algorithm is characterized by comprising the following steps: A. Normal effective signal library establishment: A-1, recording waveform data of effective signals inside and outside a boundary and waveform data of signal superposition generated by adjacent systems under the condition of no background noise, and establishing normal effective signal library members; a-2, carrying out internal and external classification and marking on each effective signal in A-1; B. the method comprises the steps of collecting an original induction current signal f (n) and preprocessing: B-1, periodically sampling an induced current signal by the intelligent mower to obtain f (n); B-2, f (n) related characteristic values are obtained; C. f (n) background noise suppression and approximation valid waveform set f_n (n) is established: c-1, performing autocorrelation filtering on f (n), and inhibiting background noise to obtain ff (n); c-2, cutting ff (n) to obtain approximate effective waveform fragments; c-3, combining all approximate valid waveform segments into a set f_n (n); D. Based on the normal and effective signal library, matching is carried out, and the judgment result inside and outside the boundary is output: d-1, carrying out matching value calculation on members in the DTW algorithm set f_n (n) and a normal effective signal library in sequence; d-2, reserving f_n (n) members with higher similarity to form a new set f_n' (n); D-3, calculating a priority value of members in the set f_n' (n) based on the peak-to-peak value and the matching value; d-4, determining the internal and external states of the boundary by taking the maximum member of the priority value as a final reference signal; in the step A-1, storing effective signal waveform data in a normal state to a normal effective signal library, wherein the normal state refers to an effective signal waveform segment with larger signal intensity acquired by an inductance probe in an environment without background noise or weak background noise; the waveforms also include such cases as waveform stacking; in the step A-2, the basis of the internal and external classification is the phase of an effective signal, wherein a positive pulse part in the effective signal temporally precedes a negative pulse part in the effective signal to represent the boundary line, and a signal with the opposite phase represents the outside of the boundary line; in the step B-2, calculating and storing related characteristic values of f (n), wherein the related characteristic values comprise a maximum value, a minimum value and a mean value; In step C-1, if the original signal is denoted as x (t), τ represents the time domain offset, and the signal autocorrelation function R (τ) is denoted as: If the original signal is a periodic signal, the period is T, and the autocorrelation function is still a periodic signal with the same frequency, so that only one period T is shifted: whereas for digital signals its autocorrelation function evolves as: Wherein N is the number of digital signal period sequences; in step C-2, the mean value is calculated And (3) with For a sequence average value of 0 or more in the median value of the signal ff (N), N + represents the sequence length, For a sequence average value of less than 0, N - represents the sequence length: Selecting out more than Or is smaller than The length of each continuous sequence is greater than a threshold G, and if no continuous sequence meeting the conditions exists, the final result is judged to be an invalid signal; In the step C-3, the continuous sequences screened in the step C-2 are combined, if the time domain interval of the two continuous sequences is smaller than T gap , the two sequences are combined, wherein T gap is the maximum time interval for judging that the two continuous sequences are unified and integral, and the combined sequences form a set f_n (n); in step D-1, the DTW algorithm calculates the similarity between two time series by extending and shortening the time series, so that one signal sequence of the normal effective signal library is Q, and one signal sequence of f_n (n) is C: Q=q 1 ,q 2 ,q 3 ,……q n C=c 1 ,c 2 ,c 3 ,……c m An n x m matrix grid is constructed, the matrix elements (i, j) represent the distance d (q i ,c j ) between the two points q i and c j , the euclidean distance d (q i ,c j )＝(q i -c j ) 2 , each matrix element (i, j) represents the alignment of the points q i and c j , the shortest path is defined as the canonical path, and the sequence is used to represent: W=w 1 ,w 2 ,w 3 ,……w K max(n,m)≤K≤n+m-1 The boundary of the rounding path is determined and has continuity and monotonicity, and the whole is expressed as: the cumulative distance is the similarity between two sequences, and the current cumulative distance γ (i, j) represents the current lattice point distance d (q i ,c j ), that is, the sum of the Euclidean distances of the points q i and c j and the cumulative distance of the smallest neighboring element reaching the point: γ(i,j)=d(q i ,c j )+min{γ(i-1,j-1),γ(i-1,j),γ(i,j-1)} calculating a matching value gamma of each member in f_n (n) and a member in a normal effective signal library through a DWT algorithm; in the step D-2, determining the normal effective signal library member with the highest similarity of each member in f_n (n), marking, and recording the corresponding matching value and the type inside and outside the boundary of the f_n (n) member; setting up the highest matching threshold omega, discarding members with gamma being more than or equal to omega, and combining members meeting the conditions into a new set f_n' (n); In step D-3, each member in the set f_n' (n) calculates a priority value based on the signal peak-peak value and the matching value, and sorts the signals according to the priority value, wherein the weight of the peak-peak value is ρ 1 , and the weight of the matching value is ρ 2 , and the steps are as follows: ρ 1 +ρ 2 =1 In total, h signal members are included in f_n' (n), the peak-to-peak value of the ith member is p_p i , and the calculation formula of the priority value beta i is as follows: in step D-4, based on the determination of the type inside and outside the member boundary with the largest priority value in step D-3 as the final result, if the set f_n' (n) is empty, it is determined as an invalid signal.
2. 2. The method for identifying the inside and outside of the boundary of the electromagnetic shielding type intelligent mower based on the DTW algorithm of claim 1, wherein in the step B-1, the sampling period T a is more than twice the induction current period T c .
3. 3. The method for identifying the inside and outside of the boundary of the electromagnetic shielding intelligent mower based on the DTW algorithm of claim 1, wherein in the step C-2, the threshold G is changed according to the effective signal length.

Description

Electromagnetic shielding intelligent mower boundary inside and outside recognition method based on DTW algorithm Technical Field The invention belongs to the field of data analysis, and particularly relates to an electromagnetic shielding type intelligent mower boundary internal and external identification method based on a DTW algorithm. Background With the development of social economy, the urban greening level is continuously improved, the areas of public lawns and private gardens are continuously enlarged, and the demands of people on maintenance equipment of the lawns are also increasing, wherein the intelligent mower equipment is in an important role. At present, the main current mode of the intelligent mower for cutting grass is electromagnetic shielding type, namely, a closed-connection cable is paved at the periphery of the area, a periodic pulse current signal is introduced into the cable, a variable magnetic field can be generated within a certain range, a corresponding induction current is generated by an inductance probe on the intelligent mower, and the equipment extracts an effective waveform section in the induction current signal and analyzes the characteristics of the effective waveform section to realize judgment of current position information. The method is a key ring for extracting effective signal waveforms and analyzing features in induced current signals, and fundamentally determines the internal and external recognition accuracy of the boundary of the intelligent mower. At present, a simple time domain analysis is often adopted for an effective induction current signal waveform extraction and analysis method, the time domain characteristics of the acquired digital signals are paired, an effective signal waveform interval is obtained, and the internal and external states are analyzed. The method is only used for the environment without interference or with weak interference, and the recognition accuracy inside and outside the boundary is poor under the environment with strong interference, so that the reliability and stability of the intelligent mower are reduced. The interference mainly comprises background environment noise and intersystem mutual interference, namely effective waveform superposition, the induced current signal has periodicity, the environment interference noise is mostly white noise, the induced current signal can be processed through an autocorrelation filtering technology, the signal is subjected to feature matching calculation through a dynamic time warping algorithm, meanwhile, effective signal waveform superimposed signal data can be added in a matching source, the influence of intersystem interference is eliminated, and the accurate identification of the boundary internal and external states of the current position of the intelligent mower is finally realized. Disclosure of Invention Aiming at the problems, the invention provides an electromagnetic shielding type intelligent mower boundary internal and external identification method based on a DTW algorithm. The DTW algorithm is a dynamic time warping algorithm, and the similarity degree between the two sequences is calculated based on a dynamic programming thought. According to the invention, the induction current data acquired by the inductance probe is analyzed, background noise is suppressed by utilizing an autocorrelation filtering technology, possible effective signal waveform fragments are cut out, and then a DTW algorithm is utilized to carry out matching calculation, so that an accurate boundary internal and external recognition result is obtained. The invention combines the autocorrelation filtering technology and the DTW matching algorithm to realize the accurate judgment of the internal and external states of the boundary of the electromagnetic shielding intelligent mower. The invention can detect and autonomously analyze the induced current signal in real time, has strong anti-interference capability and has the capability of correctly identifying and judging under the condition that effective signal waveforms are overlapped. To achieve the above object, the present invention comprises the steps of: A. Normal effective signal library establishment: And A-1, recording waveform data of effective signals inside and outside the boundary and waveform data of signal superposition generated by adjacent systems under the condition of no background noise, and establishing normal effective signal library members. A-2, carrying out internal and external classification and marking on each effective signal in A-1. B. the method comprises the steps of collecting an original induction current signal f (n) and preprocessing: B-1, periodically sampling an induced current signal by the intelligent mower to obtain f (n). And B-2, f (n) related characteristic values are obtained. C. f (n) background noise suppression and approximation valid waveform set f_n (n) is established: and C-1, performing autocorrelation f