P3 static constraint
1.1.1 daily observable time
The daily observable time is defined as the sun below 18 degrees under the horizon.
Due to the influence of the global turn, the time every day is 3 minutes 56 seconds, and the result is that the starting red trisss of the observed belt are about 1 degree. The range in the deficitway can observe almost no greater than 100 degrees (-10 ° ~ 90 °).
At the time of minced, it can be calculated, so it can be accurate before observation, when observations can be started, when to end the observation.
1.1.2 month
The moon affects the observable time and observable star. Every month, eight to twenty-two is the big month of the month, and to avoid the time of rising the moon to observe. The rest is the small moon night, and the launch of the launch of the moon is also possible to observe the relatively bright goals. It can accurately calculate the situation of a certain day.
1.1.3 Observed Observing Tianshi
Due to the design characteristics of Lamost, telescope can only observe the Tianzhu Near Zhongtian. That is to say, the center position of the telescope designated by the observation plan is constrained by the observation start time. Lamost needs to track the observation of the west to the east, this small area is also asymmetrical in the middle of China, which is generally point to the center of the telescope within 1.5 hours to China.
1.1.4 stars and other constraints
Lamost design observed star, etc.
-20.5m
To achieve a consistent signal-to-noise ratio, the time of the star exposure such as different stars is different. Therefore, in the same observation, it is necessary to select the basic consistent brightness of the brightness to observe.
1.1.5 Star
The current star scheme tends to place four star CCDs in the position of the four fixed focal panels on the focal plane, and position and track the limestar imaging. The degree of constraints developed to the planned strategy are mainly reflected in the requirements of Lamost to the brightness of the star, and if the star is required to be bright, the corresponding number of stars is relatively small, and the SSS is over-coverage. The impact is affected.
1.1.6 observation priority
The observation priority can be divided into two: one is in the goal of the patrol observation library, some goals should take precedence over additional goals, when formulating a plan; otherwise, it is possible to temporarily designate a batch of target priority observations. . Observation priority affects the observable target.
1.1.7 focal plane shape constraint
Due to the outside of the focal plane, there is still a position of 4 fixed focal panels and the central position of the S-H test device, and there is a position of 143 fiber units. Therefore, the observation star table will leave a hole in the corresponding position and need to be added in the later observation plan.
1.1.8 Fiber Positioning device anti-collision constraint
In order to make the entire focal plane without observation dead zones, each fiber-optic head is in diameter
33mm
Circle, greater than two fiber units
25.6mm
. Therefore, the optical fiber head of the adjacent two units has an overlap region, and when the target chooses in this overlap region, a mechanical collision may occur. Therefore, when selecting the observed star in a fiber unit, it is necessary to ensure a safe distance with the stars of adjacent cells.
1.1.9 Observation Target Status
The long-term plan is to integrate the observation goals in the approvalful subject, add observation target library, and give observation priority to the targets in various topics. Observing plans need to be produced from the observation target library available from the long-term plan. Long-term plan may change, mainly possible to have a new topic will join the observation target library.
The generation of an observational plan mainly dispenses unbursed targets from the observation target library, so the observational plan first relies on the state of the mid-star target of the observation plan library, the state of the target is observed in the initial time, with Observing, completely successful observation goals are in the "observation success" state, and have been submitted in an observation plan, and there is no observation that the observations have been observed.
When the final observation plan is generated, the target library is the target library, and the observation state is "unworthy".
P4 dynamic constraint
1.1.10 Meteorological Condition 1) The clouds of clouds may not observe, and the cloud is not observed under the conditions of cloud.
2) Wind will lead to atmospheric disturbance, affect imaging quality. The dark target cannot be observed when the wind speed is relatively large.
3) Rain, snow causes observation that cannot be carried out.
4) The atmosphere is considering the quality of the image, which in turn affects the observed star.
1.1.11 rack status
When the telescope rack is abnormal, it may result in the normal pointing to some days, and the SSS must eliminate this part of the Tianzhu when developing a observation plan, and the current observable range of the telescope is defined in the non-defense area.
1.1.12 focal plane status
The focus surface is mainly the state of the fiber head in the fiber unit on the fiber unit in the focal plane. If some optical fiber positioning is abnormal, the failed fiber head must be ignored when performing the target matching to increase the matching rate of other fibers.
1.1.13 Daylight Observation
Lamost is observed to observe background light to remove the effect, the actual operation is to reserve a certain number of fibers to perform weather observations in the targetless area. The fiber optic observation of the sky is preferably distributed uniform on the focal plane, and there are two ways to allocate in the optical fiber of the sky.
1) Optics that do not assign targets are used to measure daylight.
2) If 1) Insufficient optical fibers in the middle, then a number of uniformly removed from the matched target in the focal plane is used to measure the weather.
1.1.14 people constraint
Specifies a specific elest constraint condition by manual designation.
P5 constraint condition
From the analysis of observation conditions, it can be seen that various observations have affected the development of observations from several aspects:
1) Optional Nik
2) Optional goals in the sky
3) Available fiber unit
4) Observation strategy
P6 observation strategy
The efficiency mentioned here is mainly measured by the following indicators.
P9 static elevation algorithm
At present, the existing star algorithm is mainly based on the completeness of overlay. Its main idea is to develop all observations before the start of the tour, and these observations are asked to meet the requirements of overlay. Then picked a plan to observe from the actual observation. Observing plans no longer change in the patrol process. Since such a star algorithm is generated a fixed constant observation plan, it is called a static elevation algorithm.
P11 Lamost features
Therefore, it is not possible to use the SDSS to cover the way in honeycomb.
Since each fiber has a fixed area of the fixed area, it can be placed on the focal plane in the focal plane as the SDSS, and the problem of fiber utilization is required.
Observation efficiency problem, the center of the static elest star telescope is also fixed, and the observation efficiency will decrease rapidly as observations.
The static elevation algorithm cannot solve the accident in observation, and cannot solve some of the objectives in the observation plan fail to successfully observe.
It can be seen that the observation conditions of Lamost are demanding relative to SDSS. Therefore, it is not possible to prepare all the observation plan in advance. When the observation plan is developed, it must be flexiblely developed according to the current observation conditions, referred to as dynamic elest stars.
P12 dynamic elest strategy
Observation conditions tend to limit the observable objects in the observable days and the sky, how to choose in one part of the local area, so that the results of multiple observations meet the overall coverage and optimization of the overall coverage and optical fiber utilization The requirement is a key issue. In order to solve this problem, we first introduce the concept of optimization problems:
Priority strategy is a relatively intuitive problem solving method for solving optimization problems. Its features are currently the best, that is, it expects to produce a global most Exquisite. It adopts the "step-by-step construction" method, making a current optimal decision in every stage, and once the decision makes it, it will no longer be changed. The basis of making decisions is called "priority guidelines."
It should be noted that the priority policy does not always produce the optimal solution, but for solving dynamic elest problems, priority policies can generate a comparison optimized solution. Because first Lamost is completed each time the observation plan, the spatial distribution of the target is not only changed, and the observable area has also changed. The correlation between the two decision stages is relatively weak. It is more suitable for applying "Optimization Strategy." P13 "Maximum Density, Uniform Distribution" Guidelines
How is the key to applying priority strategies to determine what is the best phase, followed by determining how decisions can make the phases optimal, that is, find priority guidelines. From the previous analysis, the optimality of the observation of the tour is: uniform sampling and fiber utilization. So how do I determine priority guidelines? For Lamost, an obvious priority is to select the Tile that is not observing the density density, and the spatial distribution of the target in the TILE is selected and assigned to the fiber is uniform, referred to as "maximum density, uniform distribution" guidelines. Let's explain this guideline:
1) Priority observation of the TILE that is not observing the target density. Obviously, the Tile target with large density is relatively large, the fiber matching rate is high, and it is easy to obtain a relatively good observation efficiency.
2) Try to make the remaining unbaudible target in the space evenly distributed in the space. Such efficiency is the highest in the long history. Since the fiber unit of Lamost is uniformly distributed over the focal plane, only one fiber can be included in each fiber unit. If the non-observed target distribution is uneven, it will lead to a portion of the fiber "full", and the optional Star is too many, and the other part of the fiber "hunger", there is no observation star. This will reduce the observation efficiency of Lamost.
"Maximum density" and "uniform distribution" have intrinsic consistency. The maximum density will result in uniform target spatial distribution in the pursuit of maximum observation efficiency.
P15 telescope
1) First exclude fixing conditions (those that can be obtained before running: time, moon phase, weather, preliminary screening. Looking for the largest Tile of Tile, Tile of Tile as a telescope in the screening observable in the screening. Point to the position.
2) The maximum implementation of density can use discretization methods to discrete the observable tiane, and then move the center point of the telescope to obtain the center point position of the maximum density. However, this way is more than a long calculation time, we can use two level points to algorithms. That is, Mr. Many TILE, each TILE records its central location, overridden target number, and the number of targets that are not observed. When running, you will choose to observe the number of unbained targets. It should be pointed out that the way TILE is flexible, the more the generated Tile, the less fine it is. And you can add a letter to the Tile to add a letter as needed.
P16 fiber match
After the telescope points to the determination, how can the remaining unbained target space are even uniform? As you can see, we will ultimately lead to uniform distribution of surveyed star density distribution throughout the day. Then we only need to distribute even every observable star in the sky, then the remaining stars will be uniform. Of course, it takes a process here. Here we can use the "most recent distance method". Because the fiber unit is uniformly distributed on the focal plane, it only needs to allow each fiber unit to select its coverage within the center of the fiber unit.
Since the Lamost's fiber unit is arranged in a honeycomb, a target may occur simultaneously within two fiber units. In order to avoid this situation, the fiber distribution conflict triggered, we used "from the inside out" allocation: first of all the fiber units in the focus surface, the target collection is removed from the matching target collection. Then, then the target match of the fiber unit near the center, sequentially complete the matching of all fiber units in turn.
P17 simulation calculation
To verify the validity of the dynamic elegant algorithm, we have chosen a test day for simulation observation. The results of the simulation are shown in Figure 2-2. The circle in the figure represents an observation, and some round bottoms are more impressed by many observations at the same location. It can be seen that the dynamic elevation algorithm will increase more observational efficiency in more observation in the lower left corner of the observation target. The statistical results of the simulation test are shown in Table 2-2. The overlay showed that the dynamic starry scheme can reach a theoretical coverage of 98.5% (SDSS 92%) and the actual coverage of 95.8, and the average observation efficiency is as high as 91%. Fully meet the requirements of the Lamost program. It should be pointed out that in small range simulations, the test day we selected is square, and the edge coverage efficiency in the sky is lowered, and when over-day, the coverage will be higher than the above analog value. The overlay effect is shown in Figure 2-3.
P19 SSS system design
Depending on the observation process, the working time of Lamost can be divided into four different time periods:
Table Lamost Working Period
The non-observation period Lamost is in the preparation of the preparation time of the downtime. Lamost is turned on, pre-preparatory program preparatory observation procedures per observed before performing observations, performing observation
The SSS must complete the generation and submission of the observation plan before each observer preparation period ends. The decision time is constrained, so it is necessary to complete a part of the decision. In fact, according to the previous analysis, the SSS decision factor can be divided into two parts: Part of the decision-making factor known before observation, called "static conditions", static conditions do not change over time, or it has transformation, but The law of the change is known; the other part is the factor that cannot be pre-known is called dynamic conditions. The dynamic conditions will change over time, such as "weather", its change is irregular, only the specific situation can be obtained when it starts observing.
P20 SSS system design
SSS is primarily responsible for providing a patrol plan, which is generally a decision system. For a decision system. The most important thing is the rules of decision making. What kind of treatment is needed, causing what output. In general, these rules are reflected in the process of processing.
P21 automatic decision support
For a decision support system, its most important factor is "model library", which should have analysis processing capabilities for the external conditional set. But this processing ability is not evacuation. It is based on the accumulation of conventional treatment. With the accumulation of knowledge, the system can handle the problems that the problem can be handled by means of a conventional experience when encountering similar situations.
As mentioned earlier, SSS has uncertainty in decision rules, with the largest problem that the SSS observation mode has its particularity, and the conditions for decision make are very complicated. So there is no good rules to guide us how to develop decisions. For a given conditional set, what strategy should be taken to develop a plan without a ready-made solution.
Therefore, it is necessary to introduce a mechanism for decision support in the SSS system. Its main task is to collect data in observations and find regularities. With each observation conditions, observation plans, observations as input, the output is the effect of observations, and summarizes the law, and uses this law as the basis for the future observation plan. It is a "self-feedback" process.
P22 observation time discretization
According to the observation process, each observation night observation 5-8 times. Each observation time is almost uncertain. That is to say that Lamost can initiate a observation at any time. Due to time limitations, SSS needs to use two-stage decision-making methods: the first phase develops telescope center position; the second phase is mainly determined to match the fiber. Due to time and space limitations, SSS cannot develop plans at any point in time. The number of plans is endless, and SSS can only prepare a plan on discrete time point, as shown in Figure 3-3. Although this is limited to the observation time of Lamost. However, only the discrete points are small enough, and this limitation is negligible. The plan to be developed in the first stage is a batch of plans at the time of time, and each fixed point in time has a batch of plans, and each batch includes an optional plan at this moment. Any time OCS can Request to the SSS, but SSS will return the nextmost time point plan.
P30 system operation development environment
As a long-term running system, SSS requires a stable operating system as a support. At present, we mainly use free Linux as a platform for system development and testing. Linux is likely to be the ultimate run platform for the system.
SSS may require a cross-platform support. In order to avoid cross-platform migration issues, we use the Java platform as the main development and operation platform of SSS. The Java platform not only has a cross-platform feature, but also provides a large number of mature tools on graphical interface and system development, which can shorten the development time and ensure the quality of development of the system.
In a distributed heterogeneous environment, the SSS needs to interact with software systems developed in different languages and other operations, so we use CORBA technology as a distributed platform. CORBA provides distributed component features, which makes the system good flexibility. If the Java's running efficiency in the future system does not meet the requirements, the key computing part of the SSS system is relatively high. High language such as C / C development, these critical calculation portions can be integrated with CORBA seamlessly.
In actual development, we selected the Java of CORBA implementation Jacorb. Tests show that JACORB can achieve TAO integration in the component interaction, naming services, etc. and C C .
SSS needs to manage millions of observed objectives. The fault tolerance and security of data are particularly important, so we use large relational database Oracle to serve as a support platform for relational databases.
