COSMIC RADIATION CONCENTRATING FACILITY EQUIPPED WITH A REFLECTIVE OPTICAL SURFACE CONTROL SYSTEM

Abstract

The invention relates to a cosmic radiation concentrating installation from a celestial object, equipped with - a concentrating optical surface capable of reflecting the incident cosmic radiation towards a target surface o'x'y', and capable of exhibiting local surface and pointing and orientation errors, - a control system for the reflecting optical surface, - means for acquiring images of the optical surface from different viewpoints m'mn (x'mn, y'mn) located in the target surface, m varying from 1 to m, n varying from 1 to n, so as to obtain mxn images of the optical surface illuminated by the cosmic radiation, with m viewpoints along x' and n viewpoints along y', with m>1, n>1, mN≥30, - and a processing unit for the m.N acquired images adapted to: o calculate the slopes δδ(p) / δx and δδ(p) / δy for each point p(x,y) of the reflective optical surface according to: (formula i) l(m'mn, p) being the luminance at a point of the image corresponding to the point p(x,y) of the reflective optical surface (2) observed from the viewpoint m'mn, ε0 the apparent angular radius of the celestial object, gx and gy predetermined coefficients, and o determine from these slopes δδ(p) / δx and δδ(p) / δy, a local surface error δp(x,y) at the point p(x,y) of the reflective optical surface (2).

Description

[0001] The field of the invention is that of the control of the reflective optical surfaces of a cosmic radiation concentrating installation, used for example in a solar energy concentrating installation intended for the production of helio-thermal electricity.

[0002] The conversion of concentrated solar power into domestic or industrial electricity is one of the most promising avenues for renewable energy production in the 21st century. Future solar power plants will typically consist of a heat receiver 1 installed at the top of a tower several hundred meters high, and hundreds or thousands of heliostats placed at ground level, each with a reflective surface 2 for incident rays 3, tracking the sun and concentrating its radiation in a fixed direction towards the receiver 1, as shown. figure 1aHeliostats are orientable, while the receiver is fixed. Other types of cosmic ray concentrators include a single concentrator directed towards the radiation source (shown). figure 1b but which concentrates the radiation towards a receiver 1 in a direction varying with the orientation of the concentrator (concentrating reflector surface 2-receiver 1: orientable), a double-reflection solar furnace comprising a field of orientable flat heliostats, each comprising a flat surface 2b reflecting incident rays 3 and a fixed concentrating surface 2a which concentrates the radiation towards a fixed receiver 1 shown figure 1c , like the one in Odeillo, France, whose dimensions are shown in the figure.

[0003] Among the many remaining technological challenges are those of the time and effort devoted to the adjustment and control of heliostats or more generally of reflective optical surfaces before the operational commissioning of the installation, and of their regular monitoring in operation.

[0004] Each concentrating reflective surface, also called a mirror, is generally segmented into several facets or segments, but not necessarily. It is parabolic, or even spherical or flat, as indicated in the previous examples.

[0005] The main optomechanical defects characteristic of a cosmic radiation concentrating reflective surface 2 (or 2a), segmented into reflective facets 21, are shown on the figure 2 ; in this figure we can see two facets O i and O j. Optomechanical defects are essentially divided into three categories: Local surface errors δl represent the deviation from the ideal parabolic or spherical shape of the reflecting facet. They can have numerous origins: manufacturing quality, deformations under mechanical stress, environmental effects, etc. They are generally highly variable from one facet to another. Orientation errors δr between facets; positioning errors along the X, Y, and Z axes have a negligible influence on the concentration factor. Finally, the overall pointing error δp of the heliostat or concentrator manifests as the addition of a mean plane inclined relative to all the facets.

[0006] In the case of a concentrating surface not segmented into facets, the orientation errors δr of the facets relative to each other do not have a reason to exist; the optomechanical defects then arise from the local surface errors δl and the global pointing error δp.

[0007] Once the optomechanical defects have been identified, the adjustment process involves correcting the shape of the mirrors, typically using mechanical actuators located at the rear. These actuators allow, depending on the case, modification of their orientation (platform), their average curvature, or higher-order defects. However, experience gained from existing solar installations (for example, the THEMIS power plant in Targasonne, France, or the 1000 kW solar furnace in Odeillo) suggests that these error measurement and adjustment operations could take several months or even years when applying current techniques to an industrial-scale installation of 10 megawatts or more. Furthermore, these operations sometimes require occupying the focal point of the installation, thus reducing the boiler's operating time.

[0008] These optomechanical defects are measured either by carrying out laboratory measurements, but then the final shape of the facets in operation is not precisely known, or by carrying out field measurements.

[0009] Most current techniques, such as optimizing the flux collected by a detector located at the focal point of the installation, or remotely observing a target in the focal plane, disrupt the operation of the solar installation: these methods effectively obstruct the access of solar rays to the boiler (the heat receiver). They are based on measuring irradiance in the target plane of the receiver (flux densities), from which overall information about the quality of the reflective surfaces is derived, but do not allow for the determination of local surface errors δl.

[0010] Another solution involves using a deflectometry technique (observing a grid or fringe pattern through reflective surfaces), but this does not work for installations such as those described in relation to the figures 1b And 1c and applies to a configuration very different from that of final use, for example in the laboratory, or for a pair of non-conjugate points. It also requires complex, or even impossible, image processing.

[0011] We also know the backsight method, which consists of placing a detector in the middle of the target plane from which we directly observe the images of the radiation source, or luminance distributions on the reflective surface, as described in the publication by F. Hénault and C. Royère: "Concentration of solar radiation: analysis and evaluation of impulse responses and alignment errors of reflective facets," J. Optics 1989, vol. 20, no. 5, pp. 225-240. However, this method does not allow for a quantitative measurement of local surface errors δl and provides only a rough measurement of the orientation errors δr between the facets.

[0012] US documents 2013 / 202215 and US 5,982,481 describe systems and methods for aligning individual facets of a concentrator, using a plurality of artificial radiation sources.

[0013] US document 7,667,833 describes a method for aligning a linear receiver parabolic trough solar concentrator, based on the TOP (“Theoretical Overlay Photographic”) method.

[0014] Consequently, there remains to this day a need for an installation whose reflective surfaces can be controlled without disrupting the operation of the installation, and this in a reduced time.

[0015] The principle of the invention consists of observing, from several observation points located in the target surface of an operating cosmic ray concentrating installation, the distributions of visible luminance on the surface of the reflective surfaces, and deducing quantitative information on their local surface errors, pointing and possibly orientation errors.

[0016] More specifically, the invention relates to a cosmic radiation concentrating installation from a celestial object, equipped of a concentrating optical surface capable of reflecting incident cosmic radiation towards a target surface O'X'Y' and capable of exhibiting local surface errors and pointing and orientation errors of a control system of the reflecting optical surface.

[0017] It is primarily characterized by the fact that the control system includes: means for acquiring images of the optical surface from different viewpoints M' mn (x' mn , y' mn ) located in the target surface, m varying from 1 to M, n varying from 1 to N, so as to obtain M x N images of the optical surface illuminated by cosmic radiation, with M viewpoints along X' and N viewpoints along Y', with M>1, N>1, MN≥30, and a processing unit for the acquired MN images adapted to calculate the slopes δΔ(P) / δx and δΔ(P) / δy for each point P(x,y) of the reflecting optical surface according to: ∂ Δ P ∂ x = g X ε 0 ∑ m = 1 M ∑ n = 1 N sign x ′ mn L M ′ mn , P ∑ m = 1 M ∑ n = 1 N L M ′ mn , P ∂ Δ P ∂ y = g Y ε 0 ∑ m = 1 M ∑ n = 1 N sign y ′ mn L M ′ mn , P ∑ m = 1 M ∑ n = 1 N L M ′ mn , P L(M' mn , P) being the luminance at a point of the image corresponding to the point P(x,y) of the reflective optical surface observed from the viewpoint M' mn , ε 0 the apparent angular radius of celestial object, gx and gy predetermined coefficients, and ∘ determine from these slopes δΔ(P) / δx and δΔ(P) / δy, a local surface error ΔP(x,y) at the point P(x,y) of the reflective optical surface.

[0018] The main advantages of the invention are as follows: It allows for the measurement of local adjustment, pointing, and surface errors of reflective surfaces in situ and enables regular checks to be carried out without disrupting the operation of the installation, such as the solar power generation process. It allows for the control and adjustment of all reflective surfaces of the installation in the shortest possible time. It requires only a limited number of observation points (M'mn).

[0019] The means of acquiring images from different viewpoints may include several image acquisition devices respectively located in different fixed or mobile positions in the target surface.

[0020] The target surface can be flat and is then designated as the target plane.

[0021] According to one feature of the invention, the reflective optical surface to be controlled is orientable. In this case, the means for acquiring images from different viewpoints comprise at least one image acquisition device located in the target surface and means for modifying the orientation of the reflective optical surface to be controlled.

[0022] The reflective optical surface is typically segmented into facets and the processing unit is adapted to further determine orientation errors of the facets relative to each other and a setting error of each facet.

[0023] The reflective optical surface is, for example, mounted in a heliostat.

[0024] The reflective optical surface is parabolic, or even spherical or flat.

[0025] The cosmic radiation concentrator installation may include several reflective optical surfaces.

[0026] The radiation is, for example, solar or lunar.

[0027] The invention also relates to a tower power plant or an individual concentrator or a double-reflection solar furnace or a Cherenkov telescope comprising a cosmic radiation concentrating installation as described.

[0028] Other features and advantages of the invention will become apparent from the following detailed description, given by way of non-limiting example and with reference to the accompanying drawings in which: THE Figures 1 The already described schematically represent different types of solar radiation concentrating installations, a tower power plant illuminated by a field of focusing heliostats ( fig 1a ), an individual concentrator directed towards the sun ( fig 1b ), a double-reflection solar furnace with a field of flat-plate heliostats and a fixed concentrator ( fig 1c ), there figure 2The already described illustration highlights the main errors characteristic of a cosmic radiation concentrator installation with a faceted, segmented reflective surface, the figures 3 schematically represent the coordinate systems and scientific notations used, in the case of an overview including the radiation source, the concentrating surface and the target plane ( fig 3a ) and at the level of the concentrating installation with the concentrating surface, the target plane in which different viewpoints of the concentrating surface are located (four in the example in the figure), and the plane of the images of the concentrating surface obtained from these viewpoints ( fig 3b ), THE figures 4 schematically represent different examples of embodiments of the invention in the case of a segmented reflective surface (8 segments or facets in the figure) with a camera moving in the target plane ( fig 4a), a fixed camera scanned by the adjustable reflective surface ( fig 4b ), several fixed cameras scanned by the adjustable reflective surface ( fig 4c ), several mobile cameras scanned by the adjustable reflective surface ( fig 4d ), THE Figures 5 illustrate the physical interpretation of the calculation of optomechanical defects according to the invention from a theoretical Gaussian sun ( fig 5a ), and a uniform sun seen in practice ( fig 5b ), there figure 6 is a flowchart showing the steps of an example of data processing acquired from different viewpoints according to the invention, the figures 7 illustrate the results obtained for an installation according to the invention as described figure 1b a segmented, steerable reflective surface associated with a camera, and represented schematically ( fig 7a ), with a sequence of images ( fig 7b) obtained by the camera, the actual, measured errors and the corresponding measurement errors presented in false colors ( fig 7c ) and in three dimensions ( fig 7d ), THE figures 8 illustrate the results obtained for an installation according to the invention as described figure 1a , with a segmented, steerable reflective surface associated with a camera, and represented schematically figures 4 , with a sequence of images ( fig 8a ) obtained by the camera, the actual, measured errors and the corresponding measurement errors presented in false colors ( fig 8b ) and in three dimensions ( fig 8c ).

[0029] From one figure to another, the same elements are identified by the same references.

[0030] In the following description, a solar energy concentrator will be used primarily as an example of an installation. However, the invention also applies to a lunar energy concentrator and to Cherenkov telescope arrays used for high-energy astrophysics, whose specifications are similar to those required for a solar concentrator. A target plane will also be used as an example of a target surface.

[0031] The main coordinate systems and notations used subsequently are indicated on the figures 3 Three main indicators are used: The Oαβγ frame of angular coordinates in the sky, where the Oγ axis is directed along the vector Stowards the center of the sun (or towards any other celestial object being targeted), and where neighboring directions are designated by their angular coordinates (α, β). In practice, α and β are considered to be first-order quantities with respect to 1. The OXYZ frame of reference is associated with the moving or fixed concentrating surface, where O is the center of the reflecting surface, and the OZ axis is directed along the vector N is normal to the surface at O, and OX and OY are axes perpendicular to OZ. The points P on the reflecting surface are defined by their Cartesian coordinates (x, y, z). The frame of reference O'X'Y'Z' is associated with a target plane O'X'Y' (where the receiver is located). The point O' is defined as the intersection of the solar ray reflected on the reflecting surface at O ​​and directed along the vector R , with the target plane; F being the focal length of the concentrating surface, we have OO'=F. The O'Z' axis is normal to the target plane, and is not necessarily parallel to the vectorR (as in the case of heliostats in a solar power plant, for example). Similarly, the OZ and O'Z' axes are generally not coincident, except in the case of concentrators directly pointed towards the sun, an example of which is shown on the figure 5 The points M' of the target plane are defined by their Cartesian coordinates (x', y').

[0032] Note that the vectors S, N And R are linked together by Descartes' first law of reflection.

[0033] As we can see figure 3b Some points M' are denoted M'mn, where the indices m and n denote a limited number of different "viewpoints" located in the target plane, with 1 ≤ m ≤ M along the O'X' axis and 1 ≤ n ≤ N along the O'Y' axis. The coordinates of point M'mn in the target plane are denoted (x'mn, y'mn). In the example of the figure 3b , we have four observation points: M = N = 2.

[0034] The general principle of the invention consists of observing, from different viewpoints M'mn, the luminance distributions L(M'mn, P) reflected by all the points P of the reflecting surface that one wishes to monitor and then adjust regularly. Observation devices, such as commercial CCD cameras possibly equipped with zoom, or even simple webcams, are placed at these observation points M'mn, aimed at the reflecting surfaces to be monitored. M'mn designates the position of the observation device in the target plane. These observation points are, of course, located within the image spot of the sun (or more generally, of the celestial object) in the target plane; among these observation points, one may find the point of concentration of the concentrating surface, but not necessarily. By extension, the corresponding observation device is also designated by M'mn.Subsequently, a camera is used as an example of an observation device.

[0035] Several observation strategies are possible, differing essentially in the number and type of movements applied to the reflective surfaces and / or the cameras themselves: mobile camera in the target plane, fixed camera swept by the mobile reflective surface, system composed of several fixed cameras, "mixed" observation strategy. ❖ Mobile camera in the target plane.

[0036] According to a first embodiment of the invention shown figure 4aImages representing the luminance distributions L(M'mn, P) for all points P are accumulated by moving a single camera (image acquisition means 4 = a single camera) over different observation points M'mn located in the target plane, using mechanisms associated with this camera that allow scanning all viewpoints M'mn. However, this solution has drawbacks in terms of mechanical complexity and the time required to move the camera in the O'X'Y' plane, during which the orientation of the heliostat, and therefore of the reflecting surface 2, may have been modified to track the Sun's path. Furthermore, in the case of a mobile or fixed parabolic concentrator, such as for a Cherenkov telescope, it disrupts the receiver's operation. ❖ Fixed camera scanned by the energy concentrator system.

[0037] An alternative to the previous solution involving only one camera is illustrated. figure 4b This method takes advantage of the fact that for small changes in the orientation of the heliostat, and therefore of the reflecting surface 2 in its movement to track the sun, the observed luminance distributions 5 are identical to those obtained by moving the camera to a fictitious observation point M'mn located in the target plane. This results in a decisive advantage over the previous solution, because flat or focusing heliostats are by design capable of tracking the sun's path across the sky with a refresh rate on the order of a few Hz, thus allowing a complete sequence of L(M'mn, P) 5 image acquisitions to be performed much more quickly.

[0038] It is also possible to take advantage of the mobility of the open-loop controlled heliostats (and therefore the reflecting surfaces 2) to direct them towards a target plane O'X'Y' independent of the receiver. In the case of a tower power plant, the target plane can be located a few meters below the focal volume of the installation, allowing the heliostats to be controlled or adjusted to be directed there individually in turn without disrupting the operation of the heat receiver. Similarly, in the case of a flat heliostat illuminating the fixed concentrator of a double-reflection solar furnace, the target points can be located in directions different from that of the concentrator's axis and situated at sufficiently distant distances in the landscape.The strategy is also applicable to fixed or sun-facing solar concentrators, provided that the observation points M'mn are located within the focal volume but outside the receiver, which is itself located at the center of this focal volume. This restriction does not apply to Cherenkov radiation collecting telescopes, as the center of the image plane is traditionally left free for installing their calibration systems. Finally, it should be noted that in all cases, it is possible to take advantage of the natural movement of the observed object (typically the Sun or the full Moon) to eliminate one of the two rotations required to track it across the sky. Furthermore, this option allows for the automation of the image acquisition sequence by synchronizing it (electronically or digitally) with the movements of the heliostats: thus, image acquisition time can be significantly reduced.

[0039] This observation strategy is well-suited to a fixed solar concentrator on a solar furnace, or to concentrators directly tracking the sun. In both cases (as with Cherenkov telescopes), it is preferable to perform the checks on the reflective surfaces at night, for example, during a full moon, without disrupting the daytime operation of these systems. Cherenkov telescopes, being too sensitive to stray light, cannot perform scientific observations during full moon nights; therefore, the checks can be carried out at night, with a simplified observation system and without disrupting the telescope's normal operation.

[0040] In the case of the fixed concentrator of a solar furnace, the angular scanning is carried out in practice by means of the flat heliostats which illuminate the concentrator (this assumes that their settings have themselves been optimized beforehand). ❖ System composed of several fixed cameras

[0041] A radical alternative to the two previous cases is illustrated on the figure 4c It consists of placing in the target plane O'X'Y' as many fixed cameras as there are viewpoints M' mn required (image acquisition means 4 = MN fixed cameras), thus allowing the acquisition of the different luminance distributions L(M' mn ,P) 5 simultaneously. This observation strategy has the following advantages and disadvantages.

[0042] Benefits : Simultaneous image acquisition 5 allows the movements of the sun or moon in the sky to be frozen, thus improving overall measurement accuracy. It also avoids the need for complex digital processing to correct the L(M' mn ,P) 5 images based on their acquisition time if the measurement time required for the strategies described above is too long (typically exceeding 5 minutes).

[0043] Disadvantages: The major drawback is undoubtedly the total cost of the necessary equipment, resulting from the significant increase in the number of cameras required. Furthermore, to achieve simultaneous data acquisition, it becomes necessary to ensure the temporal synchronization of the data acquired by the different cameras, thus complicating the computer control of the measurement system. Finally, the camera aiming axes must be adjusted in orientation to converge on the monitored reflective surfaces, potentially introducing additional mechanical complexity.

[0044] As with the previous strategy, this one is particularly well-suited to the heliostats of single-reflection tower power plants. It appears more difficult to implement on solar concentrators directly pointed at the sun or on double-reflection solar furnaces (shown respectively figures 1b And1c ), due to the number of cameras required to be positioned near the focal volume. ❖ "Mixed" observation strategy

[0045] In order to benefit from the advantages offered by the two previous implementation methods (using a limited number of cameras without sacrificing the speed of the complete acquisition sequence), it seems natural to combine the two strategies, as illustrated for example on the figure 4dHere, a line of fixed cameras is arranged along the horizontal axis O'X' at observation points that we will denote M'm1 (image acquisition means = 4 cameras in a line). The additional viewpoints M'mn (with n ≠ 1) are obtained by means of the altitude rotation of the heliostats. Alternatively, the cameras can be aligned vertically and scanned by azimuth rotation. More generally, a discrete distribution of observation cameras in the target plane O'X'Y' can be combined with an altitude- and azimuth-optimized heliostat scanning sequence.

[0046] The mixed observation strategy appears perfectly applicable to the focusing heliostats of a tower power plant or the concentrating surface of a solar furnace, and to the flat-plate heliostats of a double-reflection solar furnace. In the case of a solar furnace, images are formed of the concentrating surface to control for defects in this surface, and images are formed of the flat-plate heliostats to control for their defects.

[0047] Finally, it should be noted that the constraint of not disrupting the operation of the solar installation can be met through two different strategies: 1) Use of a target plane distinct from that of the receiver, and 2) Measurements carried out at night by aiming at the Full Moon. The full moon is indeed also an excellent observable source for the adjustment and control of planar or focusing heliostats controlled in open loop: observing the full moon, whose apparent diameter is close to that of the sun, makes it possible to carry out all the operations of characterizing optomechanical defects during the night, thus allowing the optimization of mirrors and heliostats to be carried out even more quickly. The observed sources can be either the sun during the day or the full moon during the night.

[0048] Regardless of how the invention is implemented, it offers the following advantages: a limited number of commercial CCD cameras, or even simple webcams. Measurement of reflective surfaces in situ, making the best use of their natural ability to track the movements of the sun or moon.

[0049] The next step in the process of controlling and adjusting reflective surfaces consists mainly of innovative digital processing of the acquired data, which is the subject of the following section.

[0050] The general principle of image processing can be intuitively explained from the figures 3b And 5 Let us first assume that the apparent surface of the sun is a disk with a Gaussian profile characterized by the angular luminance law L(α,β): L α β = L 0 exp α 2 + β 2 / 2 ε 0 2 , where ε₀ is the apparent angular radius of the sun (also shown Figures 1), and L0 the luminance at its center. Now consider an observation device M'mn whose position is located by the coordinates (x'mn, y'mn) in the target plane O'X'Y', and from which all points P of the reflecting surface are simultaneously sighted. If the optomechanical defects of the mirror at point P with coordinates (x, y), including local surface, orientation, and pointing errors, are represented by the function Δ(P) = Δ(x, y), then the rays from the sun reflected at point P and collected by the observation device M'mn have angular coordinates in the frame Oαβγ of the figure 3a : α ≈ − x ′ mn F + 2 ∂ Δ P ∂ x β ≈ − y ′ mn F + 2 ∂ Δ P ∂ y where F is the focal length of the concentrating reflector surface, and where the pair (α,β) and the angles of incidence on the mirror can be considered small within the framework of a first-order approximation (This is always true for the angles α and β; when it is not true for the angles of incidence on the heliostats of a tower power plant, this restriction does not affect performance). From relations (1-2), we can then determine the expression for the solar radiances L(M' mn ,P) observed from points M' mn and deduce the relations that link them to the partial derivatives of the optomechanical defects of the mirror Δ(P). L(M' mn ,P) is the radiance at a point of the image corresponding to the point P(x,y) of the optical surface observed from the viewpoint M' mn .

[0051] In the particular case of four viewpoints M'11, M'12, M'21 and M'22 located at the corners of a rectangle in the target plane of length 2δx' along the X' axis and 2δy' along the Y' axis as shown on the figure 3b The coordinates of the four viewpoints M'11, M'12, M'21, and M'22 in the O'X'Y' plane are therefore respectively (-δx', -δy'), (-δx', δy'), (δx', -δy'), and (δx', δy'). Substituting these coordinates into relations (2) and then inserting these relations into the solar luminance law (1), we obtain a system of four equations with two unknowns δΔ(P) / δx and δΔ(P) / δy: L M ′ 11 , P = L 0 exp − 1 2 ε 0 2 2 ∂ Δ P ∂ x + δ x ′ F 2 + 2 ∂ Δ P ∂ x + δ y ′ F 2 , L M ′ 12 , P = L 0 exp − 1 2 ε 0 2 2 ∂ Δ P ∂ x + δ x ′ F 2 + 2 ∂ Δ P ∂ x + δ y ′ F 2 , L M ′ 21 , P = L 0 exp − 1 2 ε 0 2 2 ∂ Δ P ∂ x + δ x ′ F 2 + 2 ∂ Δ P ∂ x + δ y ′ F 2 , L M ′ 22 , P = L 0 exp − 1 2 ε 0 2 2 ∂ Δ P ∂ x + δ x ′ F 2 + 2 ∂ Δ P ∂ x + δ y ′ F 2 .

[0052] By taking the natural logarithms to the left and right of the = sign in each relation, we obtain a linear system of four equations with two unknowns, which we solve analytically using the least-squares method. We thus find relations (3): ∂ Δ P ∂ x = − F ε 0 2 4 δx ′ log L M ′ 22 , P L M ′ 21 , P L M ′ 12 , P L M ′ 11 , P ∂ Δ P ∂ y = − F ε 0 2 4 δy ′ log L M ′ 22 , P L M ′ 12 , P L M ′ 21 , P L M ′ 11 , P ,

[0053] From a physical point of view, the principle of the method consists of using the luminance profile of the solar disk (previously assumed to be Gaussian), decentered with respect to the theoretical observation directions defined by the points M'mn due to the effect of optomechanical defects Δ(P). These defects can then be calculated from the luminance differences measured at the points M'mn, as illustrated. figure 5a . This figure shows the measurements taken at two observation points M' 11 and M' 12 on a well-aligned mirror (=black points) and the measurements taken on a misaligned mirror (=white points).

[0054] But in reality, the law of solar luminance is rarely Gaussian, although this does occur under certain cloud cover. In the case of a non-Gaussian but monotonically decreasing profile from the center to the edge, we obtain the more general mathematical relationships ∂ Δ P ∂ x = g X ε 0 L M ′ 22 , P − L M ′ 21 , P + L M ′ 12 , P − L M ′ 11 , P L M ′ 22 , P + L M ′ 21 , P + L M ′ 12 , P + L M ′ 11 , P ∂ Δ P ∂ y = g Y ε 0 L M ′ 22 , P − L M ′ 12 , P + L M ′ 21 , P − L M ′ 11 , P L M ′ 22 , P + L M ′ 21 , P + L M ′ 12 , P + L M ′ 11 , P , where gX and gY are gain coefficients along the two axes X and Y. The numerical values ​​of these coefficients obviously depend on those of the parameters ε0, δx' and δy', but also on the luminance law of the observed object at the time of the measurements. This is why, in practice, the values ​​of gX and gY are determined experimentally.

[0055] Finally, in the most common case where the sun's profile is uniform as illustrated figure 5b The previous equations do not give satisfactory results. Indeed, the slopes of the sun's profile, which are directly related to the partial derivatives of the slopes of the reflective surface δΔ(P) / δx and δΔ(P) / δy, are too steep.

[0056] The solution according to the invention is based on increasing the number of observation points. This amounts to extending relations (4) to the case of M x N different observation points of known position (with M x N > 4), distributed in the target plane; these M x N observation points are advantageously located on a regular grid centered on point O'. These new relations can then be written as: ∂ Δ P ∂ x = g X ε 0 ∑ m = 1 M ∑ n = 1 N sign x ′ mn L M ′ mn , P ∑ m = 1 M ∑ n = 1 N L M ′ mn , P ∂ Δ P ∂ y = g Y ε 0 ∑ m = 1 M ∑ n = 1 N sign y ′ mn L M ′ mn , P ∑ m = 1 M ∑ n = 1 N L M ′ mn , P , where the analytic function sign(u) is equal to u / |u|. The physical interpretation of these latter relations is illustrated on the figure 5bThanks to the accumulation of different viewpoints M'mn, they allow us to transform the real sun, which most often appears as a disk of uniform luminance, into a "fictitious sun" with gentler slopes. This latter can be understood as the product of a convolution of the uniform solar luminance law with a distribution representing the positions of the observation points M'mn in the target plane. The luminance curve of the fictitious sun thus generated then exhibits less steep slopes, consistent with equations (5).

[0057] The overall logic of data processing is presented in the flowchart of the figure 6 It can be broken down into six main steps. A) Reorganization of the acquired image sequence. This step is optional and depends on the different observation strategies presented. The simplest case is obviously that of a single camera moving in the focal plane, where the coordinates x' mn and y' mn are directly returned by the camera's movement mechanisms, and where equations (5) can be applied as is. In all other cases, it is necessary to reconstruct the fictitious coordinates of the virtual observation points M' mn from characteristic data of the observation sequence, such as the direction of the sun (itself a function of the date and time of measurement), the position and movement of the heliostats, and the location of the cameras in the target plane. B) Summation of the images and their partial derivatives. This step consists of calculating the double sums of the numerators and denominators in equations (5).Depending on the geometric outline of the observed surfaces, the observation area of ​​points M'mn can also be restricted to a circular region with a radius close to F. ε0. C) Smoothing of the sums calculated during the previous step. This is an optional procedure: depending on the total number of images acquired from points M'mn, their weighted sums calculated during the previous step may appear "grainy," with breaks in slope corresponding to the contours of the image of the solar disk moving across the surface of the mirrors. It is then possible to smooth these images using a filter (Gaussian or other) in order to "soften" the profile of the fictitious sun. figure 5bThis option proves particularly useful when the total number of points M'mn is small (typically less than 10x10). D) Calculation of the slopes ∂Δ(P) / ∂X and ∂Δ(P) / ∂Y of the reflecting surface. Here, the analytical relationships defined by equations (5) are directly applied to measure the partial derivatives ∂Δ(P) / ∂X and ∂Δ(P) / ∂Y of the optomechanical defects. This requires identifying, experimentally for example, the optimal values ​​of the gain coefficients gX and gY appearing in relationships (5). E) Determination of the optomechanical defects Δ(P) from their slopes. It is, of course, possible to determine the optomechanical defects of the mirrors Δ(P) from their partial derivatives determined in the previous step. This is in fact a classic problem in adaptive optics which was - among others - discussed by Southwell in the publication: "Wave-front estimation from wave-from slope measurements", J. Opt. Soc. Am. Vol 70, p 998-1006.Here we used the "zonal reconstruction type A" described in the original article. F) and G) Calculation of surface errors δl, orientation errors δr of the facets relative to each other, or global pointing errors δp of the mirror. These calculations can be carried out in two different ways, as shown in the flowchart: first, it is possible (step G) to estimate the slopes of the different facets of the segmented mirror from the surface defects Δ(P) measured during the previous step, for example, by means of a Zernike polynomial decomposition allowing the calculation of angular errors and focusing errors.

[0058] It is also possible (step F) to directly calculate the orientation errors δr by averaging the values ​​of the partial derivatives ∂Δ(P) / ∂X and ∂Δ(P) / ∂Y measured on the surface of the individual facets during step D. In all cases, the pointing errors δp of the reflecting surface can be estimated as the arithmetic means of the errors affecting each individual facet. The main difference between the two procedures, both of which were tested in the numerical simulations presented in the following section, lies essentially in their actual computation time and the nature of the information sought (alignment errors, defocusing, higher-order defects, etc.).

[0059] The entire process of measuring local surface errors δl, adjusting δr and pointing δp of the reflective surfaces of the heliostats can be fully automated.

[0060] Having determined the errors, these are converted into instructions for adjusting the reflecting surfaces, which can be executed manually or automatically if the concentrating surface is equipped with controlled actuators.

[0061] To validate these methods for optimizing reflective surfaces, several numerical simulations of the performance of solar energy concentrating systems were carried out. Two main types of concentrators were studied: first, a segmented parabolic concentrator directly controlled by the sun, and then a focusing heliostat located in the field of a solar power tower.

[0062] Case of a parabolic concentrator: Let us first consider the case of a parabolic concentrator directly locked to the sun (this case is also applicable to Cherenkov radiation-collecting telescopes) shown figure 7a (which corresponds to the case of the figure 1bfor which the angle of incidence of cosmic rays between the vectors S And N is almost zero: we have S → = N → = R → ) or illuminated by one or more flat heliostats (as in the case of a solar furnace, figure 1c These are generally concentrator systems with very high numerical apertures, whose typical parameters are given in the second column of Table 1 below. This table also indicates the main characteristics of the measurement device (area to be scanned and number of measurement points in the target plane), as well as those of the optomechanical defects to be measured. In the numerical simulations, these defects are modeled by random samples whose standard deviations are indicated in the last three rows of the table. The random errors δr of facet alignment around the X and Y axes of the concentrator ("tip-tilt") are introduced by normal distributions with a standard deviation of 2 mrad.

[0063] Finally, the local surface errors δl of the facets are simulated using the deviations from their radii of curvature R and their aspherization coefficients ε given in the table (taking, for this parabolic surface, the theoretical values ​​of the radius of curvature and the aspherization coefficient to be R = 20 m and ε = -1 respectively), which are chosen as the averages of the Gaussian error laws. This allows the introduction of shape errors up to the fourth order.

[0064] We have δp=0. Table1 : NUMERICAL SIMULATION PARAMETERS Parabolic concentrator Focusing heliostat Dimensions of the concentrating reflective surface Diameter 5 m 7 x 6 m² Focal length of the concentrating reflector surface 10m 100 m Numerical aperture of the concentrating reflective surface F / 2 F / 11 Number of facets 18 4 x 2 Dimensions of the facets Diameter 1 m 1.4 x 2.8 m² Area swept in the target plane 0.1 x 0.1 m² 0.7 x 0.7 m 2< Number of observation points 24 x 24 24 x 24 Standard deviation of the orientation errors δr of the facets 2 mrad 0.5 mrad Standard deviation of radius of curvature errors 0.5 m - Standard deviation of errors on the aspherization coefficient 0.5 -

[0065] Taking into account all the preceding parameters, the figure 7breproduces an acquisition sequence of 32 L(M'mn,P) images viewed from the target plane, extracted from the complete sequence of 24 x 24 images simulated by ray tracing software. Although all numerical calculations are performed here with a uniform solar luminance law, the figure 7b This diagram presents the images that would be observed when aiming at the full moon, both to aid understanding and for aesthetic purposes. A possible sequence of temporal scanning is indicated by arrows in the figure.

[0066] The results obtained for calculating the slopes along the X and Y axes and for determining the optomechanical defects Δ(P) are presented in the upper half of Table 2 below. They are quantified there in terms of Peak-Valley (PTV) errors and RMS standard deviations characterizing the actual optomechanical defects (first column), those actually measured by the system (second column), and their absolute and relative differences (third and fourth columns). Views of these different defect maps are also shown in false color and in three dimensions in the figures 7c And 7dThe results obtained for slope measurement appear particularly promising, as the PTV and RMS errors are less than 15% and 6% respectively. However, it is reasonable to expect that after a further realignment sequence, followed by concentrator checks using the same procedure, these figures can be significantly improved. The measurement error maps displayed on the upper right panels show... figures 7c And 7d that the local surface errors δl and the orientation errors δr have disappeared.

[0067] It was also noted that axial positioning errors on each of the facets of the reflective surface ("pistons") do not disrupt the calculation of slopes. Table 2 : REELS MEASURES MEASUREMENT ERRORS RELATIVE ERRORS (%) PTV RMS PTV RMS PTV RMS PTV RMS Parabolic concentrator 6.43 1.09 5.40 0.75 5.97 1.24 92.9 113.3 Δ(P)(mm) 6.20 1.10 6.81 1.10 0.84 0.12 13.5 10.7 Slopes along X (mrad) 3.93 1.05 4.23 1.04 0.58 0.06 14.8 5.7 Slopes along Y (mrad) Focusing heliostat 10.52 1.77 6.92 1.03 5.27 0.75 50.1 42.6 Δ(P)(mm) 7.28 1.66 4.40 0.86 0.40 0.04 5.5 2.5 Slopes along X (mrad) 5.88 1.31 4.40 1.01 0.35 0.05 6.0 3.9 Slopes along Y (mrad)

[0068] In conclusion, the measurement errors of the optomechanical defects Δ(P) of a large numerical aperture solar energy concentrator surface, as simulated here, appear to be very satisfactory and compatible with the performance sought on this type of installation.

[0069] Case of a focusing heliostat: Consider the reflective surface of a focusing heliostat located within the field of a solar power plant tower. This case is more general but also more complex than that of concentrators directly (or indirectly, in the case of a solar furnace illuminated by its flat-plate heliostats) pointed towards the sun, because the heliostat must then perform a dual function: on the one hand, track the sun in its diurnal movement across the sky, and on the other hand, concentrate the rays onto the receiver (the boiler) located at the top of the tower. The main consequence is that the vectors S And Nidentifying the directions of the sun and the normal to the reflective surface of the heliostat (see the figure 3a ) are constantly changing direction: therefore, we no longer have S → = N → = R → . The optimal shape of the heliostat's reflective surface is then empirically defined as a parabola with a radius of curvature equal to twice its distance from the boiler. However, this type of reflective surface exhibits two intrinsic geometric aberrations, known as coma and astigmatism, whose amplitudes increase sharply with the dimensions of the heliostat and the value of the angle of incidence between the vectors. S And N Consequently, relations (2) can now be written as: α ≈ − x ′ mn F + 2 ∂ Δ P ∂ x + ∂ Δ Coma P ∂ x + ∂ Δ Astig P ∂ x β ≈ − y ′ mn F + 2 ∂ Δ P ∂ y + ∂ Δ Coma P ∂ y + ∂ Δ Astig P ∂ y , where the additional parameters Δ Coma (P) and Δ Astig (P) denote the "natural" coma and astigmatism aberrations of the focusing heliostat. It is therefore necessary to eliminate these aberrations during data processing. In practice, this involves subtracting them from the slopes and optomechanical defects calculated during steps D and E of the data processing procedure described in relation to the flowchart of the figure 6 Calculation of wavefront errors Δ Coma (P) + Δ Astig (P) and their reference slopes, that is, assuming the reflective surface of the heliostat is perfectly spherical and free from orientation errors δr and surface errors δl, can be carried out using ray tracing software already used to simulate L(M') images min ,P) of the figures 7b And 8a . Reference maps obviously depend on the general parameters of the configuration studied: position of the sun in the sky, location of the heliostat in the field of the tower power plant, coordinates of the target, focal length of the reflective surface of the heliostat, etc.

[0070] The main results obtained are presented in the same way as for the previous example. The geometric parameters of the heliostat's reflective surface are given in the third column of Table 1. Only the orientation errors of the facets of the heliostat's reflective surface, with a standard deviation of 0.5 mrad, are shown here. In order to demonstrate the full potential of the method, a rather unfavorable geometric configuration was used, where the angles of solar incidence on the heliostat's reflective surface are on the order of 25 degrees, thus generating strong natural aberrations of coma and astigmatism. A sequence of L(M'mn,P) images observed on the surface of the heliostat, aimed at the full moon, is presented on the figure 8a .

[0071] The measurement errors of the optomechanical defects Δ(P) and their slopes along the X and Y axes in terms of Peak-Valley and RMS errors are presented in the lower half of Table 2. False-color and three-dimensional views of the calculated error maps are reproduced in the figures 8b And 8c The general appearance of these figures is similar to that of the previous section, except for their first lines showing the natural aberrations of coma and astigmatism of the ideal reflective surface of the heliostat.

[0072] Finally, the measurement errors δr of the altitude and azimuth misalignments of the eight facets of the focusing heliostat's reflecting surface, performed after two iterations of the control and adjustment sequence, are shown in Table 3. These figures and the graphical representations allow us to draw the following conclusions: - The Peak-Valley and RMS measurement errors of the slopes of the focusing heliostat's surface are even better than those of the parabolic concentrator, being less than 6% and 4%, respectively. This is probably due in part to the heliostat's small numerical aperture (here, F / 11). - Furthermore, the relative errors in the angular misalignments of the heliostat's reflecting surface facets are on the order of 2% Peak-Valley and 1.5% RMS, an excellent performance that can likely be improved in further iterations.However, this does not seem truly necessary, as alignment residuals of less than 0.01 mrad are largely negligible compared to the apparent angular diameter of the sun (approximately 10 mrad). Table 3 : ALIGNMENT ERRORS δr (mrad) MEASUREMENT ERRORS (mrad) RELATIVE ERRORS (%) Azimuth Altitude Azimuth Altitude Azimuth Altitude 0.31 0.42 0.00 -0.01 0.20 2.37 -0.216 -0.233 0.002 0.002 1.0 1.0 0.233 0.225 -0.012 -0.002 5.4 0.7 -0.199 0.313 0.009 -0.010 4.6 3.3 0.153 0.204 -0.002 -0.002 1.5 0.8 0.020 -0.046 0.000 0.001 1.5 1.9 -0.135 0.266 0.005 -0.014 3.8 5.2 RMS errors 0.006 0.006 2.0 1.5

[0073] The numerical simulations presented here show that the measurement accuracies achieved are largely compatible with the expected performance of these different types of solar energy concentrator systems.

[0074] Compared to existing techniques, the solution according to the invention, which is extremely adaptable, has the essential advantage of being able to be implemented on an operating installation, without disrupting helioelectric production in the case of a solar installation.

[0075] Furthermore, it significantly reduces the commissioning time of the installation and is applicable to the main types of existing point-source solar energy concentrators (tower plants, concentrators directly pointed towards the sun, double-reflecting solar furnaces, etc.). This allows, for example, the detection of heliostats that are becoming detached so that they can be returned to their resting position, for example, for aviation safety reasons.

[0076] It allows for the control and adjustment of all the reflecting surfaces in the installation in the shortest possible time. By automating the adjustment process, human intervention is reduced to a simple periodic readjustment of the reflecting surfaces based on predetermined numerical values. In the case of a tower power plant, an ambitious but achievable goal is to be able to adjust around one hundred heliostats per day. This minimum time can itself be reduced to real-time control and adjustment when the mirrors themselves are equipped with actuators that can be remotely controlled from the processing unit responsible for the operation.

Claims

1. An Installation for concentrating cosmic radiation originating from a celestial object with a non-zero apparent angular radius, comprising - a concentrating optical surface (2) able to reflect incident cosmic radiation (3) toward a target surface O'X'Y' so as to form an image of said celestial object here, and liable to contain local surface errors and aiming and orientation errors, - a system for inspecting the reflective optical surface, characterised in that the inspecting system comprises: - means (4) for acquiring images (5) of the optical surface from various viewpoints M'mn (x'mn, y'mn) that are located inside the image mark of the celestial object in the target surface, m varying from 1 to M, and n varying from 1 to N, so as to obtain M x N images of the optical surface illuminated by the cosmic radiation, with M viewpoints along X' and N viewpoints along Y', where M>1, N>1 and M.N≥30, - and a unit for processing the M.N acquired images, which unit is suitable for: ∘ calculating the slopes δΔ(P) / δx and δΔ(P) / δy for each point P(x,y) of the reflective optical surface, where: ∂ Δ P ∂ x = g X ε 0 ∑ m = 1 M ∑ n = 1 N sign x ′ mn L M ′ mn , P ∑ m = 1 M ∑ n = 1 N L M ′ mn , P ∂ Δ P ∂ y = g Y ε 0 ∑ m = 1 M ∑ n = 1 N sign y ′ mn L M ′ mn , P ∑ m = 1 M ∑ n = 1 N L M ′ mn , P L(M'mn, P) being the luminance at a point of the image corresponding to the point P(x,y) of the reflective optical surface (2) observed from the viewpoint M'mn, ε0 the apparent angular radius of the celestial object and gx and gy pre-set coefficients, and ∘ determining from these slopes δΔ(P) / δx and δΔ(P) / δy, a local surface error ΔP(x,y) at the point P(x,y) of the reflective optical surface (2).

2. The installation for concentrating cosmic radiation according to the preceding claim, characterised in that the means (4) for acquiring images from various viewpoints include a plurality of devices for acquiring images, which devices are respectively located at various fixed or movable positions on the target surface.

3. The installation for concentrating cosmic radiation according to either of the preceding claims, characterised in that the reflective optical surface (2) to be inspected is orientable.

4. The installation for concentrating cosmic radiation according to Claim 1 taken in combination with Claim 3, characterised in that the means (4) for acquiring images from various viewpoints include at least one image-acquiring device located on the target surface and means for modifying the orientation of the reflective optical surface to be inspected.

5. The installation for concentrating cosmic radiation according to any of the preceding claims, characterised in that the reflective optical surface (2) is segmented into facets (21) and in that the processing unit is suitable for furthermore determining errors in the orientation of the facets with respect to one another and an adjustment error of each facet.

6. The installation for concentrating cosmic radiation according to any of the preceding claims, characterised in that the reflective optical surface (2) is mounted in a heliostat.

7. The installation for concentrating cosmic radiation according to any of the preceding claims, characterised in that the reflective optical surface (2) is parabolic.

8. The installation for concentrating cosmic radiation according to any of the preceding claims, characterised in that it includes a plurality of reflective optical surfaces (2).

9. The installation for concentrating cosmic radiation according to any of the preceding claims, characterised in that the incident cosmic radiation (3) is solar or lunar.

10. The installation for concentrating cosmic radiation according to any of the preceding claims, characterised in that the target surface is planar.

11. A central-tower power plant or individual concentrator or double-reflection solar furnace or Cherenkov telescope including an installation for concentrating cosmic radiation according to any of the preceding claims.