基于听觉掩蔽效应的数字助听器算法研究及DSP的实现
|
摘要
耳聋已成为一个世界性的问题,解决耳聋问题的最直接手段是给耳聋患者戴上助听器,提高人耳对语音感知。数字助听器最大的优点是能够植入各种语音处理算法,改变语音的分布,提高人耳感知语音的舒适度和分辨率。数字助听器的诞生给耳聋患者带来了希望。
本文重点研究了数字助听器的核心算法。将人耳的听觉掩蔽效应应用到数字助听器的核心算法中去,直接从人耳的听觉生理上出发,改善人耳对声音的听觉舒适度,并且把此算法在DSP上实现。
本文首先分析了数字助听器三个核心算法:语音增强,宽动态压缩和移频压缩。传统的数字助听器算法忽略了人耳的生理特性,语音听觉舒适度没提高,针对算法的缺点,提出用基于听觉掩蔽效应的语音增强算法取代原语音增强算法。这样利用人耳的生理特性降低噪声,提高语音感知的舒适度。
针对听觉掩蔽效应的语音增强算法,本文首先对噪声的估算运用了实时噪声估计,提高了噪声估计的准确性,降低了语音失真。还对原有算法中的听觉掩蔽阈值计算不准确,从而引起了谱减系数计算误差,导致语音失真过大的问题,提出了一种改进的计算方法。在原有的谱减系数计算公式中加入了一个修改参数θ来降低语音失真。实验仿真结果表明,本算法不但提高了语音信噪比,还改善了语音音质。
其次就传统的宽动态压缩算法,在实现频谱增益时忽略了语音共振峰,频谱增益后引起了共振峰的非线性变化,改变了共振峰的频域分布,针对此问题提出了一种基于共振峰估计的宽动态压缩算法,减少了共振峰的非线性失真,改善了听觉舒适度。
最后对数字助听器的三个核心算法进行了实验验证。利用合众达的SEED—DEC5416开发板,把数字数听器的三个核心算法植入其中,在此开发板上实现了三种语音处理功能。在算法的实现过程中,对其中出现的问题进行了分析研究。实践表明,该模块满足了语音的实时性并且达到了数字助听器语音处理的效果。
Deafness has become a global problem, the best solution for deafness means to wear hearing aids, voice through hearing aids for reduce noise, improve human ear on the voice level of comfort. Traditional analog hearing aids, only to strengthen the human ear to the voice of perception, this simple amplification can help the patients of conduction hearing loss. But the accounting for the majority of deaf patients with SNHL in patients with virtually no benefits, it will increase the ear to feel much bad. Solving the nerve deafness in patients with hearing loss, the most critical issue is to change the distribution of audio, let patients with limited hearing to feel the voice, and the core processors of digital hearing aids can add voice algorithm, and change the distribution of voice, the birth of digital hearing aids to deaf patients is a good news.
This paper focuses on the digital hearing aids of auditory masking. We use ear to the physiological characteristics-auditory masking effect of the digital hearing aids to the core algorithm, directly from the ear to the hearing of the physical to have an improvement of the human ear to the voice of comfort.
This paper firstly introduces the three core algorithm for digital hearing aids, on their strengths and weaknesses. We use the speech enhancement algorithms of masking effect to the digital hearing aids. This has the advantage of directly from the people to the character- ristics of the ears, ears of the people voice perception that the maximum comfort.
Auditory masking effect for voice enhancement algorithms, the paper of the original algorithm in the auditory masking threshold value is not accurate, resulting spectrum by coefficient calculation error; voice distortion caused too great a problem, an improved method of calculation. In the original spectrum by joining the formula for calculating the coefficient of revised parameters, the algorithm can reduce the voice distortion. Simulation results show that the algorithm can improve the voice signal to noise ratio also improved voice quality.
Then, the traditional wide dynamic compression algorithm, in the realization of the gain spectrum ignoring the voice formant, after the spectrum gain resonance from the peak of the non-linear changes, change the resonance frequency of the peak.For this issue, we bring forward a new WDRC algorithm which is based on the estimated formant, reduce the resonance peak of nonlinear distortion and improved the auditory comfort.
Finally, the three core algorithm of the digital hearing aids were verified .The using of the SEED-DEC5416 development board has heard several figures for the algorithm to make a prototype of the digital hearing aids. Practice shows that the module to meet the real-time voice and reach of digital hearing aids voice processing results.
本文重点研究了数字助听器的核心算法。将人耳的听觉掩蔽效应应用到数字助听器的核心算法中去,直接从人耳的听觉生理上出发,改善人耳对声音的听觉舒适度,并且把此算法在DSP上实现。
本文首先分析了数字助听器三个核心算法:语音增强,宽动态压缩和移频压缩。传统的数字助听器算法忽略了人耳的生理特性,语音听觉舒适度没提高,针对算法的缺点,提出用基于听觉掩蔽效应的语音增强算法取代原语音增强算法。这样利用人耳的生理特性降低噪声,提高语音感知的舒适度。
针对听觉掩蔽效应的语音增强算法,本文首先对噪声的估算运用了实时噪声估计,提高了噪声估计的准确性,降低了语音失真。还对原有算法中的听觉掩蔽阈值计算不准确,从而引起了谱减系数计算误差,导致语音失真过大的问题,提出了一种改进的计算方法。在原有的谱减系数计算公式中加入了一个修改参数θ来降低语音失真。实验仿真结果表明,本算法不但提高了语音信噪比,还改善了语音音质。
其次就传统的宽动态压缩算法,在实现频谱增益时忽略了语音共振峰,频谱增益后引起了共振峰的非线性变化,改变了共振峰的频域分布,针对此问题提出了一种基于共振峰估计的宽动态压缩算法,减少了共振峰的非线性失真,改善了听觉舒适度。
最后对数字助听器的三个核心算法进行了实验验证。利用合众达的SEED—DEC5416开发板,把数字数听器的三个核心算法植入其中,在此开发板上实现了三种语音处理功能。在算法的实现过程中,对其中出现的问题进行了分析研究。实践表明,该模块满足了语音的实时性并且达到了数字助听器语音处理的效果。
Deafness has become a global problem, the best solution for deafness means to wear hearing aids, voice through hearing aids for reduce noise, improve human ear on the voice level of comfort. Traditional analog hearing aids, only to strengthen the human ear to the voice of perception, this simple amplification can help the patients of conduction hearing loss. But the accounting for the majority of deaf patients with SNHL in patients with virtually no benefits, it will increase the ear to feel much bad. Solving the nerve deafness in patients with hearing loss, the most critical issue is to change the distribution of audio, let patients with limited hearing to feel the voice, and the core processors of digital hearing aids can add voice algorithm, and change the distribution of voice, the birth of digital hearing aids to deaf patients is a good news.
This paper focuses on the digital hearing aids of auditory masking. We use ear to the physiological characteristics-auditory masking effect of the digital hearing aids to the core algorithm, directly from the ear to the hearing of the physical to have an improvement of the human ear to the voice of comfort.
This paper firstly introduces the three core algorithm for digital hearing aids, on their strengths and weaknesses. We use the speech enhancement algorithms of masking effect to the digital hearing aids. This has the advantage of directly from the people to the character- ristics of the ears, ears of the people voice perception that the maximum comfort.
Auditory masking effect for voice enhancement algorithms, the paper of the original algorithm in the auditory masking threshold value is not accurate, resulting spectrum by coefficient calculation error; voice distortion caused too great a problem, an improved method of calculation. In the original spectrum by joining the formula for calculating the coefficient of revised parameters, the algorithm can reduce the voice distortion. Simulation results show that the algorithm can improve the voice signal to noise ratio also improved voice quality.
Then, the traditional wide dynamic compression algorithm, in the realization of the gain spectrum ignoring the voice formant, after the spectrum gain resonance from the peak of the non-linear changes, change the resonance frequency of the peak.For this issue, we bring forward a new WDRC algorithm which is based on the estimated formant, reduce the resonance peak of nonlinear distortion and improved the auditory comfort.
Finally, the three core algorithm of the digital hearing aids were verified .The using of the SEED-DEC5416 development board has heard several figures for the algorithm to make a prototype of the digital hearing aids. Practice shows that the module to meet the real-time voice and reach of digital hearing aids voice processing results.
引文
1.韩纪庆,张磊,郑铁然.语音信号处理[M].北京:清华大学出版社,2004.25-32
2.张雄伟,陈亮,杨吉斌.现代语音处理技术及应用[M].北京:机械工业出版社,2003. 9-11,250-252
3. Berouti M,Schwartz R,Makhoul J.Enhancement of speech corrupted by acoustic noise[C]. IEEE International Conference on Acouotics,Speech,and Signal Processing,1979,(4): 208-211
4. Lockwood P,Boudy J.Experiments with a nonlinear spectral subtractor (NSS), hidden markov models and projection, for robust speech recognition in cars [J]. Speech Commu- nication,1992,11(2):215-228
5. Nathalie V.Signal channel speech enhancement based on masking properties of human auditory system[J].IEEE Transactions on Speech an Audio Processing,1999,7(2):126-137
6. Lu C.Reduction of musical residual noise for speech enhancement using masking properties and optimal smoothing[J].Pattern Recognition Letter,2007,28(11):1300-1306
7. Zhong L,Rafik.G,Richard M.Noise estimation using speech/non-speech frame decision and subbed spectral tracking[J], Speech Communication,2008, 49(7): 542-557
8. Yoshifumi C,Kotaro M,Katsumori H,Hidetoshi N,etc.Real-time processing using the frequency domain binaural model[J].Applied Acousties,2007, 68(8):923-938
9. Kotta M, Preeti R.Speech enhancement in nonstationary noise environments using noise properties [J]. Speech Communication,2006, 48(1):96-109
10.陶智,赵鹤鸣,龚呈卉.基于听觉掩蔽效应和Bark子波变换的语音增强[J].声学学报, 2005,30(4):367-372
11.陶智,赵鹤鸣,吴迪.基于临界频带子波变换的语音增强[J].微电子学与计算,2004, 21(12):50-52
12.徐望,王炳锡,丁琦.一种基于信号子空间和听觉掩蔽效应的语音增强方法[J].信号处理,2005,20(2):112-116
13.王莉,胡剑波,徐盛.基于听觉掩蔽效应得语音增强算法的研究[J].电声技术,2006, 4(7):39-43
14.刘海斌,吴镇杨,赵力,等.非平稳环境下基于人耳听觉掩蔽特性的语音增强[J].信号处理, 2003,19(4):303-307
15.谢志文,尹俊勋.同时掩蔽效应的实验研究[J].声学技术,2006,25(5):446-450
16.同鸣,郝重阳,刘晓军.基于听觉掩蔽特性的扩频信息隐藏方法[J].计算机工程,2006, 32(10):137-139
17.清源科技.TMS320C54x DSP硬件开发教程[M].北京:机械工业出版社, 2003.195-202
18.清源科技.TMS320C54x DSP应用程序设计教程[M].北京:机械工业出版社,2003.2-5
19. Texas Instruments Incorporated著梁晓雯等编译.TMS320VC54系列DSP的CPU与外设[M].北京:清华大学出版社,2006.5-10
20. Sen M.Kuo BobH.Lee著.卢伯英译.实时数字信号处理[M].北京:中国铁道出版社,2004. 340-343
21.赵明忠,顾斌,王钧铭,等.DSP应用技术[M].西安:西安电子科技大学出版社,2004. 198-200
22.汪安民,程昱.DSP应用开发实用子程序[M].北京:人民邮电出版社,2005.252-285
23.贾彬彬.数字助听器产品的市场现状和技术展望[EB/OL].http://www.mednet.cn /News/NewsText/News_557.htm,2005-5-9
24.吴星华.基于DSP的数字助听器开拓[J/OL].http://www.icembed.com/info-15355.htm, 2007-2-10
25.库奇.了解数字助听器[EB/OL].http://www.kuqiqi.com/jtjy/tsjy/tzjy/211933553.html, 2007-10-16
26.肖宪波,王宁宇,胡广书.数字助听器中若干主要算法的发展和现状[J].生物医学工程学,2004,21(4):694-698
27.何振亚.自适应信号处理[M].北京:科学出版社,2002.239-241
28.陈广飞,应俊.数字助听器宽动态压缩算法研究[J].北京生物医学工程,2006, 25(5):454-456
29.应俊.基于DSP数字助听器关键技术的研究[D]:[硕士学位论文].北京:中国人民解放军军医进修学院,2006
30.梁勇.移频助听技术与移频助听器[J].中国听力语言康复科学杂志,2005,3(2):25-27
31.赵力.语音信号处理[M].北京:机械工业出版社,2003.77-80,272-273
32.马大猷.现代声学理论基础[M].北京:科学出版社,2004.50-51
33. Rainer M.Noise power spectral density estimation based on optimal smoothing and minimum statistics[J].IEEE Transactions on Speech and Audio Processing,2001,9(5):504-512
34. Kathryn H,Arehart J,Stephen G,etc.Evaluation of an auditory masking threshold noise suppression algorithm in normal-hearing and hearing-impaired listeners[J].Speech Communication,2003, 40(4):575-592
35.蔡汉添,袁波涛.一种基于听觉掩蔽模型的语音增强算法[J].通信学报,2002,23(8): 93-98
36. Ted P,Andreas S.Perceptual coding of digital audio[J].Proceedings of the IEEE,2000,88 (4):451-512
37. Lu C,Wang H.Speech enhancement using perceptually constrained gain factors in critical- band wave-packet transform[J].Electronics Letters,2004,40(6):394-396
38. Marka B,Elim S.Auditory masking of anuran advertisement calls by road traffic noise[J]. Animal Behaviour,2007,74(6):1765-1776
39.王晓亚.倒谱在语音基音和共振峰提取中的应用[J].无线电工程,2004,34(1):57-61
40.郁伯康,郁梅.LPC方法提取语音信号共振峰的分析[J].电声技术,2000,3(1):3-8
41.何峰,陈晓清,李国锁等.一种新的语音信号共振峰提取的算法[J].信号处理,2007,23(4): 618-621
42. Audiologic I, Boulder C.The continuous frequency dynamic range compressor[C]. IEEE ASSP Workshop on Applications of Signal Processing to Audio and Acoustics.New Paltz, 2003,(2)19-22
43. Teiero-Caladol J,Rutledqe J,Nelson P.Preserving spectral contrast in amplitude compres- sion for hearing aids[C]. Engineering in Medicine and Biology Society in Proceedings of the 23th Annual International Conference of the IEEE , 2001,(2):1453-1456
44. Teiero J,Bernal S,Hidaldo J.A digital hearing aid that compensates loudness for sensorineural hearing impaiements[C].Internation Conference on Acoustics, Speech, and Signal Processing,1995,(5): 2991-2994
45. Shinichi S,Katsuhiko G,Makoto T.Frequency compression hearing aid for severe-to- profound hearing impairments[J].Auris Nasus Larynx,2006,27(4):327-334
46. Brian C.Speech processing for the hearing-impaired:successes,failures,and implications for speech mechanisms[J].Speech Communication,2003,41(2):81-91
47. Jean-Baptiste M,Liesbeth R,Jan W.Comparison of adaptive noise reduction slgorithms in dual microphone hearing aids[J].Speech Communication,2007,48(8):957-970
48. Harold K,David B.Hearing aids:A review of what’s new[J].Otolaryngology-Head and Neck Surgery,2006,134(6):1043-1050
49. Jong W,Steven M,Ward R,ect.Improving performance in noise for hearing aids and coch- lear implants using coherent modulation filtering[J].Hearing Research ,2008,239(2):1-11
50.应俊,陈广飞,柴学宁.基于DSP的全数字助听器设计和实现[J].医疗卫生装备, 2006,27(7):8-9
51. Texas Instruments Incorporated.TMS320VC5416 Fixed-point Digital Signal Processor [DB/OL]. http://focus.ti.com/lit/an/spra095d/spra095d.pdf,2004
52. Texas Instruments Incorporated. TLV320AIC23B Stereo Audio Codec [DB/OL]. http://focus.ti.com/lit/ds/sgls240a/sgls240a.pdf, 2004
53. Texas Instruments Incorporated. TMS320C54X Code Composer Studio Tutorial [DB/OL]. http://focus.ti.com/lit/an/spra327d/spra327d.pdf,2004
54.王剑峰,马艳萍,刘云.McBSP在语音信号处理中的应用与实现[J].微计算机应用, 2006,27(3):349- 351
55.蔡斌,郭英,李虹伟,等.应用TMS320C54X系列DSP实现语音增强算法[J].仪器仪表学报, 2003,24(4):440-442
56.杨银堂,马峰,刘毅.DSP应用开发从实践到提高[M].北京:中国电力出版社,2007.50-56
2.张雄伟,陈亮,杨吉斌.现代语音处理技术及应用[M].北京:机械工业出版社,2003. 9-11,250-252
3. Berouti M,Schwartz R,Makhoul J.Enhancement of speech corrupted by acoustic noise[C]. IEEE International Conference on Acouotics,Speech,and Signal Processing,1979,(4): 208-211
4. Lockwood P,Boudy J.Experiments with a nonlinear spectral subtractor (NSS), hidden markov models and projection, for robust speech recognition in cars [J]. Speech Commu- nication,1992,11(2):215-228
5. Nathalie V.Signal channel speech enhancement based on masking properties of human auditory system[J].IEEE Transactions on Speech an Audio Processing,1999,7(2):126-137
6. Lu C.Reduction of musical residual noise for speech enhancement using masking properties and optimal smoothing[J].Pattern Recognition Letter,2007,28(11):1300-1306
7. Zhong L,Rafik.G,Richard M.Noise estimation using speech/non-speech frame decision and subbed spectral tracking[J], Speech Communication,2008, 49(7): 542-557
8. Yoshifumi C,Kotaro M,Katsumori H,Hidetoshi N,etc.Real-time processing using the frequency domain binaural model[J].Applied Acousties,2007, 68(8):923-938
9. Kotta M, Preeti R.Speech enhancement in nonstationary noise environments using noise properties [J]. Speech Communication,2006, 48(1):96-109
10.陶智,赵鹤鸣,龚呈卉.基于听觉掩蔽效应和Bark子波变换的语音增强[J].声学学报, 2005,30(4):367-372
11.陶智,赵鹤鸣,吴迪.基于临界频带子波变换的语音增强[J].微电子学与计算,2004, 21(12):50-52
12.徐望,王炳锡,丁琦.一种基于信号子空间和听觉掩蔽效应的语音增强方法[J].信号处理,2005,20(2):112-116
13.王莉,胡剑波,徐盛.基于听觉掩蔽效应得语音增强算法的研究[J].电声技术,2006, 4(7):39-43
14.刘海斌,吴镇杨,赵力,等.非平稳环境下基于人耳听觉掩蔽特性的语音增强[J].信号处理, 2003,19(4):303-307
15.谢志文,尹俊勋.同时掩蔽效应的实验研究[J].声学技术,2006,25(5):446-450
16.同鸣,郝重阳,刘晓军.基于听觉掩蔽特性的扩频信息隐藏方法[J].计算机工程,2006, 32(10):137-139
17.清源科技.TMS320C54x DSP硬件开发教程[M].北京:机械工业出版社, 2003.195-202
18.清源科技.TMS320C54x DSP应用程序设计教程[M].北京:机械工业出版社,2003.2-5
19. Texas Instruments Incorporated著梁晓雯等编译.TMS320VC54系列DSP的CPU与外设[M].北京:清华大学出版社,2006.5-10
20. Sen M.Kuo BobH.Lee著.卢伯英译.实时数字信号处理[M].北京:中国铁道出版社,2004. 340-343
21.赵明忠,顾斌,王钧铭,等.DSP应用技术[M].西安:西安电子科技大学出版社,2004. 198-200
22.汪安民,程昱.DSP应用开发实用子程序[M].北京:人民邮电出版社,2005.252-285
23.贾彬彬.数字助听器产品的市场现状和技术展望[EB/OL].http://www.mednet.cn /News/NewsText/News_557.htm,2005-5-9
24.吴星华.基于DSP的数字助听器开拓[J/OL].http://www.icembed.com/info-15355.htm, 2007-2-10
25.库奇.了解数字助听器[EB/OL].http://www.kuqiqi.com/jtjy/tsjy/tzjy/211933553.html, 2007-10-16
26.肖宪波,王宁宇,胡广书.数字助听器中若干主要算法的发展和现状[J].生物医学工程学,2004,21(4):694-698
27.何振亚.自适应信号处理[M].北京:科学出版社,2002.239-241
28.陈广飞,应俊.数字助听器宽动态压缩算法研究[J].北京生物医学工程,2006, 25(5):454-456
29.应俊.基于DSP数字助听器关键技术的研究[D]:[硕士学位论文].北京:中国人民解放军军医进修学院,2006
30.梁勇.移频助听技术与移频助听器[J].中国听力语言康复科学杂志,2005,3(2):25-27
31.赵力.语音信号处理[M].北京:机械工业出版社,2003.77-80,272-273
32.马大猷.现代声学理论基础[M].北京:科学出版社,2004.50-51
33. Rainer M.Noise power spectral density estimation based on optimal smoothing and minimum statistics[J].IEEE Transactions on Speech and Audio Processing,2001,9(5):504-512
34. Kathryn H,Arehart J,Stephen G,etc.Evaluation of an auditory masking threshold noise suppression algorithm in normal-hearing and hearing-impaired listeners[J].Speech Communication,2003, 40(4):575-592
35.蔡汉添,袁波涛.一种基于听觉掩蔽模型的语音增强算法[J].通信学报,2002,23(8): 93-98
36. Ted P,Andreas S.Perceptual coding of digital audio[J].Proceedings of the IEEE,2000,88 (4):451-512
37. Lu C,Wang H.Speech enhancement using perceptually constrained gain factors in critical- band wave-packet transform[J].Electronics Letters,2004,40(6):394-396
38. Marka B,Elim S.Auditory masking of anuran advertisement calls by road traffic noise[J]. Animal Behaviour,2007,74(6):1765-1776
39.王晓亚.倒谱在语音基音和共振峰提取中的应用[J].无线电工程,2004,34(1):57-61
40.郁伯康,郁梅.LPC方法提取语音信号共振峰的分析[J].电声技术,2000,3(1):3-8
41.何峰,陈晓清,李国锁等.一种新的语音信号共振峰提取的算法[J].信号处理,2007,23(4): 618-621
42. Audiologic I, Boulder C.The continuous frequency dynamic range compressor[C]. IEEE ASSP Workshop on Applications of Signal Processing to Audio and Acoustics.New Paltz, 2003,(2)19-22
43. Teiero-Caladol J,Rutledqe J,Nelson P.Preserving spectral contrast in amplitude compres- sion for hearing aids[C]. Engineering in Medicine and Biology Society in Proceedings of the 23th Annual International Conference of the IEEE , 2001,(2):1453-1456
44. Teiero J,Bernal S,Hidaldo J.A digital hearing aid that compensates loudness for sensorineural hearing impaiements[C].Internation Conference on Acoustics, Speech, and Signal Processing,1995,(5): 2991-2994
45. Shinichi S,Katsuhiko G,Makoto T.Frequency compression hearing aid for severe-to- profound hearing impairments[J].Auris Nasus Larynx,2006,27(4):327-334
46. Brian C.Speech processing for the hearing-impaired:successes,failures,and implications for speech mechanisms[J].Speech Communication,2003,41(2):81-91
47. Jean-Baptiste M,Liesbeth R,Jan W.Comparison of adaptive noise reduction slgorithms in dual microphone hearing aids[J].Speech Communication,2007,48(8):957-970
48. Harold K,David B.Hearing aids:A review of what’s new[J].Otolaryngology-Head and Neck Surgery,2006,134(6):1043-1050
49. Jong W,Steven M,Ward R,ect.Improving performance in noise for hearing aids and coch- lear implants using coherent modulation filtering[J].Hearing Research ,2008,239(2):1-11
50.应俊,陈广飞,柴学宁.基于DSP的全数字助听器设计和实现[J].医疗卫生装备, 2006,27(7):8-9
51. Texas Instruments Incorporated.TMS320VC5416 Fixed-point Digital Signal Processor [DB/OL]. http://focus.ti.com/lit/an/spra095d/spra095d.pdf,2004
52. Texas Instruments Incorporated. TLV320AIC23B Stereo Audio Codec [DB/OL]. http://focus.ti.com/lit/ds/sgls240a/sgls240a.pdf, 2004
53. Texas Instruments Incorporated. TMS320C54X Code Composer Studio Tutorial [DB/OL]. http://focus.ti.com/lit/an/spra327d/spra327d.pdf,2004
54.王剑峰,马艳萍,刘云.McBSP在语音信号处理中的应用与实现[J].微计算机应用, 2006,27(3):349- 351
55.蔡斌,郭英,李虹伟,等.应用TMS320C54X系列DSP实现语音增强算法[J].仪器仪表学报, 2003,24(4):440-442
56.杨银堂,马峰,刘毅.DSP应用开发从实践到提高[M].北京:中国电力出版社,2007.50-56