阅读说明：本技术 Ai装置、激光雷达装置以及风力发电厂控制系统 (AI device, laser radar apparatus and wind power plant's control system ) 是由 小竹论季 大塚浩志 于 2017-04-26 设计创作，主要内容包括：在现有的风力发电厂控制系统的结构中,很难取得高空间分辨率的信息,存在不能为了使机器学习充实化而得到充分学习所需要的信息的课题。本发明的AI(Artificial Intelligence：人工智能)装置具有：学习装置,其对风向量进行机器学习,预测风车的发电量,并且对发电量的预测值与测定值进行比较,在预测值与测定值之差为设定的阈值以上的情况下,选择测定风向量的激光雷达装置,导出测定参数；以及控制装置,其向激光雷达装置发送学习装置导出的测定参数。(In the structure of existing wind power plant's control system, it is difficult to obtain the information of high spatial resolution, the project of information required for sufficiently learning cannot be obtained to make machine learning substantialization by existing.AI (Artificial Intelligence: artificial intelligence) device of the invention includes learning device, it carries out machine learning to wind direction amount, predict the generated energy of windmill, and the predicted value of generated energy is compared with measured value, in the case where more than the threshold value that the difference of predicted value and measured value is setting, the laser radar apparatus for the amount of determining the wind direction is selected, location parameter is exported；And control device, location parameter derived from learning device is sent to laser radar apparatus.)

1. a kind of AI (Artificial Intelligence: artificial intelligence) device, the AI device include

Learning device carries out machine learning to wind direction amount, predicts the generated energy of windmill, and to the predicted value of the generated energy It is compared with measured value, in the case where more than the threshold value that the difference of predicted value and measured value is setting, selection measures the wind The laser radar apparatus of vector exports location parameter；And

Control device sends location parameter derived from the learning device to the laser radar apparatus.

2. AI device according to claim 1, which is characterized in that

The learning device investigation, which whether there is, to be within the distance in setting range away from the windmill and obtained from last time The wind direction amount at the time of from have passed through the region of setting time or more, there are the region, selection measurement The laser radar apparatus in the region, exports the location parameter.

3. AI device according to claim 1, which is characterized in that

The learning device investigation, which whether there is, to be within the distance in setting range away from the windmill and obtained from last time The wind direction amount at the time of from have passed through the region of setting time or more, in the case where the region is not present, selection is surveyed The laser radar apparatus of the wind direction amount in the front of the fixed windmill, exports the default parameters of the laser radar apparatus As the location parameter.

4. AI device according to claim 1, which is characterized in that

In the case where the difference of the predicted value of the generated energy and measured value is smaller than the threshold value, the turbulent flow of the wind direction amount is calculated Degree calculates the distance and bearing angle of non-observation area in the case where the turbulence level is bigger than the 2nd threshold value, selects described in measurement The laser radar apparatus of non-observation area, calculates the location parameter.

5. AI device according to claim 2, which is characterized in that

The learning device is excellent according to the pulsewidth of the laser radar apparatus, beam diameter, focal length and non-coherent accumulation number Choosing sequence change values, the SNR (Signal to Noise Ratio: signal-to-noise ratio) in the case where calculating after changing, export the arteries and veins Wide, the described beam diameter, the focal length and the non-coherent accumulation number are as the location parameter.

6. a kind of laser radar apparatus, which is included

Optical generator exports laser；

Optical modulator is modulated the laser of optical generator output；

Optical system, the laser after exporting the light modulator modulates receive object in the hair as light is sent Reflected light after sending the output destination of light to reflect is as reception light；

Optical receiver carries out heterodyne detection to the reception light that the optical system receives, extracts and receive signal；

Range intervals divider divides the reception signal according to the range intervals of setting；

Fast fourier transform processor carries out Fourier to the reception signal after range intervals divider segmentation Transformation calculates the spectrum of the reception signal of each range intervals；

Accumulator accumulates the spectrum according to each range intervals after range intervals divider segmentation；

Direction of visual lines course-and-speed computer calculates Doppler frequency shift ingredient according to the spectrum after accumulator accumulation, according to The Doppler frequency shift ingredient calculates direction of visual lines air speed value；

Wind direction amount arithmetic unit calculates wind direction amount using multiple direction of visual lines air speed values；

System parameter controller, according to the location parameter that is received from AI device described in claim 1, described in setting The non-phase of the pulsewidth of optical modulator, the beam diameter of the optical system, the focal length of the optical system and the accumulator Dry cumulative number；And

Data communication section is sent to the AI device using the pulsewidth, the beam diameter, the focal length and described non- The wind direction amount that coherent accumulation number is found out.

7. a kind of wind power plant's control system, which is included

AI device described in claim 1；

Optical generator exports laser；

Optical modulator is modulated the laser of optical generator output；

Optical system, the laser after exporting the light modulator modulates receive object in the hair as light is sent Reflected light after sending the output destination of light to reflect is as reception light；

Optical receiver carries out heterodyne detection to the reception light that the optical system receives, extracts and receive signal；

Range intervals divider divides the reception signal according to the range intervals of setting；

Fast fourier transform processor carries out Fourier to the reception signal after range intervals divider segmentation Transformation calculates the spectrum of the reception signal of each range intervals；

Accumulator accumulates the spectrum according to each range intervals after range intervals divider segmentation；

Direction of visual lines course-and-speed computer calculates Doppler frequency shift ingredient according to the spectrum after accumulator accumulation, according to The Doppler frequency shift ingredient calculates direction of visual lines air speed value；

Wind direction amount arithmetic unit calculates wind direction amount using multiple direction of visual lines air speed values；

System parameter controller, according to the location parameter that is received from AI device described in claim 1, described in setting The non-phase of the pulsewidth of optical modulator, the beam diameter of the optical system, the focal length of the optical system and the accumulator Dry cumulative number；And

Data communication section is sent to the AI device using the pulsewidth, the beam diameter, the focal length and described non- The wind direction amount that coherent accumulation number is found out.

8. a kind of wind power plant's control system, which is included

AI device as claimed in claim 2；

Optical generator exports laser；

Optical modulator is modulated the laser of optical generator output；

Optical system, the laser after exporting the light modulator modulates receive object in the hair as light is sent Reflected light after sending the output destination of light to reflect is as reception light；

Optical receiver carries out heterodyne detection to the reception light that the optical system receives, extracts and receive signal；

Range intervals divider divides the reception signal according to the range intervals of setting；

Fast fourier transform processor carries out Fourier to the reception signal after range intervals divider segmentation Transformation calculates the spectrum of the reception signal of each range intervals；

Accumulator accumulates the spectrum according to each range intervals after range intervals divider segmentation；

Direction of visual lines course-and-speed computer calculates Doppler frequency shift ingredient according to the spectrum after accumulator accumulation, according to The Doppler frequency shift ingredient calculates direction of visual lines air speed value；

Wind direction amount arithmetic unit calculates wind direction amount using multiple direction of visual lines air speed values；

System parameter controller, according to the location parameter that is received from AI device as claimed in claim 2, described in setting The non-phase of the pulsewidth of optical modulator, the beam diameter of the optical system, the focal length of the optical system and the accumulator Dry cumulative number；And

Data communication section is sent to the AI device using the pulsewidth, the beam diameter, the focal length and described non- The wind direction amount that coherent accumulation number is found out.

9. a kind of wind power plant's control system, which is included

AI device as claimed in claim 3；

Optical generator exports laser；

Optical modulator is modulated the laser of optical generator output；

Optical system, the laser after exporting the light modulator modulates receive object in the hair as light is sent Reflected light after sending the output destination of light to reflect is as reception light；

Optical receiver carries out heterodyne detection to the reception light that the optical system receives, extracts and receive signal；

Range intervals divider divides the reception signal according to the range intervals of setting；

Fast fourier transform processor carries out Fourier to the reception signal after range intervals divider segmentation Transformation calculates the spectrum of the reception signal of each range intervals；

Accumulator accumulates the spectrum according to each range intervals after range intervals divider segmentation；

Direction of visual lines course-and-speed computer calculates Doppler frequency shift ingredient according to the spectrum after accumulator accumulation, according to The Doppler frequency shift ingredient calculates direction of visual lines air speed value；

Wind direction amount arithmetic unit calculates wind direction amount using multiple direction of visual lines air speed values；

System parameter controller, according to the location parameter that is received from AI device as claimed in claim 3, described in setting The non-phase of the pulsewidth of optical modulator, the beam diameter of the optical system, the focal length of the optical system and the accumulator Dry cumulative number；And

Data communication section is sent to the AI device using the pulsewidth, the beam diameter, the focal length and described non- The wind direction amount that coherent accumulation number is found out.

10. a kind of wind power plant's control system, which is included

AI device as claimed in claim 4；

Optical generator exports laser；

Optical modulator is modulated the laser of optical generator output；

Optical system, the laser after exporting the light modulator modulates receive object in the hair as light is sent Reflected light after sending the output destination of light to reflect is as reception light；

Optical receiver carries out heterodyne detection to the reception light that the optical system receives, extracts and receive signal；

Range intervals divider divides the reception signal according to the range intervals of setting；

Fast fourier transform processor carries out Fourier to the reception signal after range intervals divider segmentation Transformation calculates the spectrum of the reception signal of each range intervals；

Accumulator accumulates the spectrum according to each range intervals after range intervals divider segmentation；

Direction of visual lines course-and-speed computer calculates Doppler frequency shift ingredient according to the spectrum after accumulator accumulation, according to The Doppler frequency shift ingredient calculates direction of visual lines air speed value；

Wind direction amount arithmetic unit calculates wind direction amount using multiple direction of visual lines air speed values；

System parameter controller, according to the location parameter that is received from AI device as claimed in claim 4, described in setting The non-phase of the pulsewidth of optical modulator, the beam diameter of the optical system, the focal length of the optical system and the accumulator Dry cumulative number；And

Data communication section is sent to the AI device using the pulsewidth, the beam diameter, the focal length and described non- The wind direction amount that coherent accumulation number is found out.

Technical field

The present invention relates to AI device, laser radar apparatus and wind power plant's control systems.

Background technique

In the past, cup type airspeedometer/blade type by being equipped on windmill is used in deflection/pitching/direct torque of windmill The information that registering weather vane obtains.But the value for being set to the above equipment at windmill rear is the air speed value after being stirred by blade, therefore It is different from the real wind that flows into, it include error.In addition, in order to measure through wind speed/wind direction after blade, using being chased after always from change The form of dynamic wind, cannot predict, this will lead to the loss of generated energy.In this regard, wind direction/wind speed of far point can be measured by carrying Radar obtain the information of arrival wind, generated energy can be improved in advance carrying out the control of windmill.

Radar installations is fluctuated to space emitting electromagnetic wave or sound wave etc., the fluctuation after being reflected by target object is received, to it Signal is analyzed, to measure the distance and angle of radar installations to object.There is known following weather radars in radar: It can be with the particle (aerosol) of the small liquid or solid to swim in an atmosphere for object, according to the phase of the fluctuation after reflection Position rotation amount understands the movement speed i.e. speed of wind of aerosol.In weather radar, especially with light as electromagnetic wave Laser radar, the diffusion of the wave beam of radiation is minimum, can with higher angular resolution observed objects, be used as wind direction and wind velocity thunder It reaches.In the calculating of wind direction amount, usually using multidirectional direction of visual lines air speed value, pass through VAD (Velocity Azimuth Display: Velocity Azimuth is shown) method, vector operation etc. calculated.

It is improved except wind-power electricity generation amount except through using such radar installations to obtain the nearly following wind speed information, such as Shown in patent document below, mode is also existed as follows: based on the weather information for using past weather information and the same day Machine learning carries out the prediction of wind information, and wind-power electricity generation amount is improved using the prediction.

Summary of the invention

Subject to be solved by the invention

But even with the weather information obtained by satellite and the wind information obtained according to formula mast is often set, by It is low in spatial resolution and the precision of the prediction is made to be lower, it is also possible to deteriorate the improvement ability of generated energy.In addition, in order to logical The high efficiency that windmill generating amount is realized in machine learning is crossed, good quality and the information content for being necessarily required to information content are more.Cause This needs the measurement of high spatial resolution, such as there is the method for carrying out multiple mast construction to solve the above-mentioned problems, but It is to lead to the problem of cost to get higher.

On the other hand, the case where it is also contemplated that using the laser radar apparatus that can be measured over long distances.

Fig. 1 is the simulation drawing measured using the wind regime for the existing laser radar apparatus that can be measured over long distances.It is using In the case where such structure, the directive property of laser is high, and the deviation between more remote then its wave beam is bigger.In addition, there are windmill that In the case where the veil of sample, the air speed value at the veil back side cannot be obtained.In addition to this, even if being hided with reducing sample rate Keep away the mode of the veils such as blade, i.e., waiting blade is measured after passing through, since wind at every moment changes, not The prediction error of observation area becomes larger.Preferably it requires to carry out to high-resolution anemometry with higher sample rate, but It is that, as described above, masking causes to observe, there are following projects: due to needing the scanning for waiting next time and measuring Scanning spends time etc. when wider range, and wind is caused to change, measurement accuracy deterioration.

Fig. 2 is the simulation drawing of the wind regime measurement in the case where being set with existing laser radar apparatus on windmill.In this way, In the case where each windmill is equipped with laser radar, setting and observed pattern be it is fixed, therefore, laser project direction And Observable distance be it is fixed, generation be unable to observation area.

In conventional structure, be difficult obtain high spatial resolution information, exist cannot in order to make machine learning enrich Change the project of information required for obtaining sufficiently learning.

Means for solving the problems

AI (Artificial Intelligence: artificial intelligence) device of the invention includes learning device, to wind Vector carries out machine learning, predicts the generated energy of windmill, and be compared with measured value to the predicted value of generated energy, is predicting In the case that the difference of value and measured value is the threshold value of setting or more, the laser radar apparatus for the amount of determining the wind direction, export measurement are selected Parameter；And control device, location parameter derived from learning device is sent to laser radar apparatus.

Invention effect

According to the present invention, carry out following control: being observed in the larger Shi Yigao sample rate of the disorder of wind, disorder compared with Hour expands observation area and improves the precision of machine learning so as to increase sample and provisioning information for study.

Detailed description of the invention

Fig. 1 is the simulation drawing measured using the wind regime for the existing laser radar apparatus that can be measured over long distances.

Fig. 2 is the simulation drawing of the wind regime measurement in the case where being set with existing laser radar apparatus on windmill.

Fig. 3 is the structure chart for showing a configuration example of wind power plant's system for embodiments of the present invention 1.

Fig. 4 is the structure chart for showing a configuration example of laser radar apparatus for embodiments of the present invention 1.

Fig. 5 is the structure chart for showing a configuration example of signal processor 1010 for embodiments of the present invention 1.

Fig. 6 is the structure chart for showing a configuration example of data integration device 2 for embodiments of the present invention 1.

Fig. 7 is the structure chart for showing a configuration example of AI device 3 for embodiments of the present invention 1.

Fig. 8 is the figure for showing the mapping of the wind direction and wind velocity observation situation in the wind power plant of embodiments of the present invention 1.

Fig. 9 is the feelings that the control of non-observation area is observed using the laser radar apparatus of embodiments of the present invention 1 Simulation drawing under condition.

Figure 10 is the sequence for showing the judgement of the non-observation area of embodiments of the present invention 1 and measuring the decision sequence in region Column figure.

Specific embodiment