設(shè)計網(wǎng)站名稱,seo任務(wù)優(yōu)化網(wǎng)站排名,好享管家安卓下載,wordpress3.0問題#xff1a;窗口切片是將連續(xù)信號或長序列分割成多個較短、可能重疊的片段的過程。這個重疊是怎么個重疊方式#xff0c;是數(shù)值上的疊加嗎? 一、概念解釋#xff1a; 窗口重疊指的是相鄰窗口在時間軸上共享一部分相同的信號樣本#xff0c;而不是對這些樣本的值進(jìn)行數(shù)…問題窗口切片是將連續(xù)信號或長序列分割成多個較短、可能重疊的片段的過程。這個重疊是怎么個重疊方式是數(shù)值上的疊加嗎?一、概念解釋窗口重疊指的是相鄰窗口在時間軸上共享一部分相同的信號樣本而不是對這些樣本的值進(jìn)行數(shù)學(xué)疊加。舉例說明假設(shè)窗口長度L100L 100L100個樣本跳躍步長H50H 50H50個樣本重疊長度OL?H50O L - H 50OL?H50個樣本那么窗口1覆蓋樣本 0-99窗口2覆蓋樣本 50-149窗口3覆蓋樣本 100-199注意窗口1和窗口2共享了樣本50-99但這50個樣本在提取時被分別放入兩個不同的窗口在每個窗口內(nèi)獨(dú)立處理不是把這些樣本的值加起來2. 可視化展示3.可視化代碼importnumpyasnpimportmatplotlib.pyplotasplt# 創(chuàng)建簡單的示例信號tnp.linspace(0,1,200)signalnp.sin(2*np.pi*5*t)0.5*np.sin(2*np.pi*10*t)# 窗口參數(shù)L50# 窗口長度H25# 跳躍步長50%重疊overlapL-H# 重疊長度# 計算窗口數(shù)量K(len(signal)-L)//H1# 創(chuàng)建圖形fig,axesplt.subplots(2,2,figsize(12,10))plt.rcParams[font.sans-serif][SimHei,Arial Unicode MS,DejaVu Sans]plt.rcParams[axes.unicode_minus]False# 1. 信號上的窗口位置展示ax1axes[0,0]ax1.plot(t,signal,b-,linewidth2,alpha0.7,label原始信號)# 用不同顏色標(biāo)記前3個窗口colors[red,green,blue]forkinrange(3):start_idxk*H end_idxstart_idxL# 繪制窗口覆蓋區(qū)域ax1.axvspan(t[start_idx],t[end_idx],alpha0.3,colorcolors[k],labelf窗口{k1}ifk0elseNone)# 標(biāo)記窗口邊界ax1.axvline(xt[start_idx],colorcolors[k],linestyle--,alpha0.7)ax1.axvline(xt[end_idx],colorcolors[k],linestyle--,alpha0.7)# 標(biāo)注重疊區(qū)域ifk0:prev_end_idx(k-1)*HL overlap_startstart_idx overlap_endmin(end_idx,prev_end_idx)# 用更深的顏色標(biāo)注重疊區(qū)域ax1.axvspan(t[overlap_start],t[overlap_end],alpha0.5,colorpurple,label重疊區(qū)域ifk1elseNone)ax1.set_xlabel(時間 (秒))ax1.set_ylabel(幅度)ax1.set_title(f1. 窗口在信號上的位置 窗口長度{L}, 跳躍{H}, 重疊{overlap}個樣本)ax1.legend(locupper right)ax1.grid(True,alpha0.3)# 2. 提取的窗口數(shù)據(jù)不重疊顯示ax2axes[0,1]# 提取前3個窗口的數(shù)據(jù)windows_data[]forkinrange(3):start_idxk*H end_idxstart_idxL window_signalsignal[start_idx:end_idx]windows_data.append(window_signal)# 在y軸方向偏移顯示offsetk*2.5time_windowt[start_idx:end_idx]ax2.plot(time_window,window_signaloffset,colorcolors[k],linewidth2,labelf窗口{k1})# 填充區(qū)域ax2.fill_between(time_window,offset,window_signaloffset,alpha0.2,colorcolors[k])ax2.set_xlabel(時間 (秒))ax2.set_ylabel(幅度偏移顯示)ax2.set_title(2. 提取的窗口數(shù)據(jù)獨(dú)立顯示)ax2.legend(locupper right)ax2.grid(True,alpha0.3)# 3. 詳細(xì)展示重疊區(qū)域ax3axes[1,0]# 重點(diǎn)關(guān)注窗口1和窗口2的重疊部分win1_idx0win2_idx1win1_startwin1_idx*H win1_endwin1_startL win2_startwin2_idx*H win2_endwin2_startL# 提取兩個窗口的數(shù)據(jù)win1_signalsignal[win1_start:win1_end]win2_signalsignal[win2_start:win2_end]win1_timet[win1_start:win1_end]win2_timet[win2_start:win2_end]# 繪制重疊區(qū)域?qū)Ρ萶verlap_start_idxwin2_start# 重疊開始o(jì)verlap_end_idxwin1_end# 重疊結(jié)束overlap_lengthoverlap_end_idx-overlap_start_idx# 窗口1的數(shù)據(jù)紅色ax3.plot(win1_time,win1_signal,r-,linewidth2,label窗口1)ax3.fill_between(win1_time,0,win1_signal,alpha0.2,colorred)# 窗口2的數(shù)據(jù)綠色在重疊區(qū)域高亮ax3.plot(win2_time,win2_signal,g-,linewidth2,label窗口2)ax3.fill_between(win2_time,0,win2_signal,alpha0.2,colorgreen)# 特別標(biāo)記重疊區(qū)域overlap_timet[overlap_start_idx:overlap_end_idx]overlap_signal_win1win1_signal[-overlap_length:]# 窗口1的后半部分overlap_signal_win2win2_signal[:overlap_length]# 窗口2的前半部分# 繪制重疊區(qū)域的點(diǎn)對點(diǎn)比較foriinrange(0,overlap_length,5):# 每隔5個點(diǎn)畫一條線x_posoverlap_time[i]y1overlap_signal_win1[i]y2overlap_signal_win2[i]ax3.plot([x_pos,x_pos],[y1,y2],k--,alpha0.3,linewidth0.5)# 標(biāo)記重疊區(qū)域邊界ax3.axvline(xt[overlap_start_idx],colorblack,linestyle--,linewidth1.5,alpha0.7)ax3.axvline(xt[overlap_end_idx],colorblack,linestyle--,linewidth1.5,alpha0.7)# 填充重疊區(qū)域背景ax3.axvspan(t[overlap_start_idx],t[overlap_end_idx],alpha0.1,colorpurple)ax3.set_xlabel(時間 (秒))ax3.set_ylabel(幅度)ax3.set_title(f3. 重疊區(qū)域詳細(xì)對比 {overlap_length}個樣本重疊 ({overlap_length/L*100:.0f}%))ax3.legend(locupper right)ax3.grid(True,alpha0.3)ax3.set_xlim([t[win1_start],t[win2_end]])# 4. 修復(fù)的矩陣視角展示樣本如何被多個窗口使用ax4axes[1,1]ax4.axis(off)# 關(guān)閉坐標(biāo)軸用于顯示示意圖# 創(chuàng)建示意圖顯示樣本索引如何被多個窗口覆蓋sample_indiceslist(range(200))# 200個樣本# 修復(fù)正確初始化window_coveragewindow_coverage{}# 初始化所有樣本索引foridxinrange(200):window_coverage[idx][]# 計算覆蓋關(guān)系forkinrange(K):start_idxk*H end_idxstart_idxLforidxinrange(start_idx,min(end_idx,200)):ifidx200:window_coverage[idx].append(k)# 繪制覆蓋關(guān)系圖y_offset0sample_display_rangerange(60,140)# 顯示一部分樣本foridxinsample_display_range:windows_coveringwindow_coverage[idx]# 繪制樣本點(diǎn)ax4.plot([idx,idx],[y_offset-0.2,y_offset0.2],k-,linewidth1)ax4.text(idx,y_offset-0.3,str(idx),hacenter,fontsize8)# 繪制覆蓋該樣本的窗口forwin_idxinwindows_covering:# 計算窗口位置win_startwin_idx*H win_endwin_startL# 繪制窗口覆蓋條ifwin_startidxwin_end:colorcolors[win_idx%len(colors)]# 調(diào)整y位置避免重疊y_posy_offset0.5(win_idx%3)*0.3ax4.plot([win_start,win_end],[y_pos,y_pos],colorcolor,linewidth3,alpha0.7)# 標(biāo)記連接點(diǎn)ax4.plot([idx,idx],[y_offset0.2,y_pos],k:,alpha0.3,linewidth0.5)# 標(biāo)記窗口編號ifidxwin_start5:# 在窗口開始位置附近標(biāo)記ax4.text(win_start2,y_pos0.05,f窗口{win_idx1},fontsize7,colorcolor)# 添加說明文本explanationax4.text(0.05,0.7,explanation,transformax4.transAxes,fontsize9,bboxdict(boxstyleround,facecolorlightyellow,alpha0.9))ax4.set_xlim([55,145])ax4.set_ylim([-1,3])ax4.set_title(4. 樣本被多個窗口覆蓋示意圖)plt.tight_layout()plt.show()