Neuro-fuzzy or hybrid systems, including fuzzy logic, neural networks, genetic algorithms and expert systems, are an effective tool for solving a wide range of problems in the real world.
Each intellectual method has its own individual characteristics (for example, the ability to learn, the ability to explain solutions), which make it suitable only for solving specific specific problems.
For example, while neural networks are successful in pattern recognition, they are ineffective in explaining how to reach their solutions.
Fuzzy logic systems, which are associated with inaccurate information, are verbally applied in explaining their decisions, but cannot automatically supplement the system of rules that are necessary for making these decisions.
These limitations have prompted the creation of intelligent hybrid systems, where two or more methods are combined in order to overcome the limitations of each method alone.
Hybrid systems play important role when solving problems in various applied fields. In many complex areas, there are problems associated with individual components, each of which may require its own processing methods.
Let there be two separate subtasks in a complex application area, for example, a signal processing task and a solution output task, then the neural network and the expert system will be used, respectively, for these separate tasks.
Intelligent hybrid systems have been successfully applied in many areas such as management, engineering, trade, credit, medical diagnostics and cognitive modeling. In addition, the application range of these systems is continuously growing.
While fuzzy logic provides a mechanism for logical inference from cognitive uncertainty, computational neural networks have notable advantages such as learning, adaptation, fault tolerance, parallelism, and generalization.
In order for the system to be able to handle cognitive uncertainties the way humans do, it is necessary to apply the concept of fuzzy logic in neural networks. Such hybrid systems are called fuzzy neural networks or fuzzy neural networks.
Neural networks are used to set up belonging functions fuzzy systems, which are used as decision-making systems.
Fuzzy logic can describe scientific knowledge directly using the rules of linguistic labels, but the process of designing and customizing the membership functions that define these labels usually takes a lot of time.
Neural network learning methods automate this process, significantly reducing the development time and cost of obtaining these functions.
Theoretically, neural networks and fuzzy logic systems are equivalent, since they are mutually transformable, however, in practice, each of them has its own advantages and disadvantages.
In neural networks, knowledge is automatically acquired by applying a backtracking inference algorithm, but the learning process is relatively slow and the analysis of the trained network is difficult ("black box").
It is impossible to extract structured knowledge (rules) from a trained neural network, as well as collect specific information about the problem in order to simplify the learning procedure.
Fuzzy systems are of great use because their behavior can be described using fuzzy logic rules, and thus can be controlled by adjusting these rules. It should be noted that the acquisition of knowledge is a rather complicated process, while the area of change of each input parameter must be divided into several intervals; the use of fuzzy logic systems is limited to areas where expert knowledge is acceptable and the set of input parameters is small enough.
To solve the problem of acquiring knowledge, neural networks are complemented by the property of automatically obtaining fuzzy logic rules from numerical data.
The computational process is the use of the following fuzzy neural networks. The process begins with the development of a "fuzzy neuron", which is based on the recognition of biological neural morphologies according to a learning mechanism. In this case, the following three stages of the computational process of a fuzzy neural network can be distinguished:
development of fuzzy neural models based on biological neurons;
synoptic connection models that introduce uncertainty into neural networks;
development of learning algorithms (method of regulation of synoptic weight coefficients).
On fig. P1.1 and P1.2 two possible models of fuzzy neural systems are presented.
The resulting linguistic statement is converted by the fuzzy logic interface block into the input vector of a multilevel neural network. The neural network can be trained to generate the necessary output commands or decisions
A multilevel neural network launches an interface fuzzy logic mechanism.
The main processed elements of a neural network are called artificial neurons, or simply neurons. Signal from neural inputs xj considered unidirectional, direction indicated by arrow, same for neural output
Rice. P1.2. The second model of fuzzy neural system
A simple neural network is shown in fig. P1.3. All signals and weights are given by real numbers.
The input neurons do not change the input signal, so the output and input parameters are the same.
When interacting with the weighting factor w t for signal x, we get the result p = wi xi, i = 1, …, n. The elements of the input information pi are added and as a result give the input value for the neuron:
The neuron applies its transfer function, which can be a sigmoid function of the form:
To calculate an output value:
This simple neural network that performs multiplication, addition and calculates the sigmoid function, let's call standard neural network.
Hybrid neural network is a neural network with fuzzy signals and weights, and fuzzy transfer functions. However: (1) one can combine Xj And w h using other continuous operations; (2) add the p1 components using other continuous functions; (3) the transfer function can be any other continuous function.
The processing element of a hybrid neural network is called fuzzyneuron.
It should be noted that all input, output parameters and weights of the hybrid neural network are real numbers from the interval .
Rice. P.4. Transfer function of a hybrid neural network
P1.2. fuzzy neurons
Definition 1 - fuzzy neuron I. The signals x, and w, are combined by the maximum operator and give:
The elements of the input information p, are combined using the minimum operator and as a result give the output information of the neuron:
Definition 2 - fuzzy neuron OR. Signal x, and weight w, are combined by the minimum operator:
The elements of the input information p, are combined using the maximum operator and as a result give the output information of the neuron:
Signal X, and weight w are combined by the multiplication operator:
Fuzzy neurons AND and OR perform standard logical operations on set values. The role of joins is to distinguish the specific levels of influence that individual inputs can have on the result of their join.
It is known that standard networks are universal approximators, i.e. they can approximate any continuous function on a compact set with any accuracy. A task with such a result is; non-constructive and does not provide information on how to build a given network.
Hybrid neural networks are used to implement IF-THEN fuzzy logic rules in a constructive way.
Although hybrid neural networks are not able to directly use the standard backtracking inference algorithm, they can be trained by steepest descent methods to recognize the parameters of membership functions, which are linguistic terms in rules.
History Block 09 -1 Unique deposits "White Tiger" and "Dragon" . JV "Vietsovpetro" (Viet. Sovpetro) - joint venture Russian company OJSC Zarubezhneft and the Vietnamese company Petro. Vienam, created in 1981. Batkho (Vietnamese Bạch Hổ, Russian White Tiger) is a large offshore oil field in Vietnam, located 120 km southeast of the port city of Vung Tau, on the shelf of the South China Sea. 2
Characteristics of the deposit 1) tectonic disturbance; 2) hydrocarbon deposit of the sedimentary cover; 3) the MBT well is located within the Kyulong depression, its length is 450–500 km, and its width is 75–110 km. Most of the wells drilled for foundation are high-rate. The maximum recovered thickness of the basement reaches 1700 m, the thickness of the sedimentary cover exceeds 4300 m. 3
Characteristics of the field Within the shelf of southern Vietnam, igneous and fractured basement reservoirs of the Mesozoic age are widely developed. In 1988, during a repeated test of the MSP-1-1 well at the White Tiger field in the Cuu Long depression, an oil fountain was obtained for the first time from a depth of 3150 m. The discovery of a unique deposit in the fractured granitoids of the Mesozoic basement has intensified exploration work on the formation of a magmatic basement on the shelf of Vietnam and the region as a whole. 4
More than 120 exploratory wells, production and injection wells have been drilled at the field. On the Central arch, a greater number of wells were drilled to a depth of 4500-4760 m. On the Northern arch - 4457 m. The deepest well BT-905 was drilled to a depth of 5014 m. In 1988, the first million oil was extracted. 2005 - 150 million tons of oil. 2008 - 170 million tons of oil. By the end of 2009, cumulative production amounted to 183 million tons. 2012 - 200 million tons of crude oil - "White Tiger" and "Dragon" fields. In 2012, Vietsovpetro produced 6,110 thousand tons, including White Tiger - 4,398 thousand tons, Dragon - 1,504 thousand tons.
Properties of oil Oil from the Vietnamese fields Bach Ho, Rong, in terms of their rheological properties, have general characteristics: high viscosity and high wax content. The pumping and transport of such oils indicate that in oil pipelines laid under water, intense heat exchange between the flow of pumped oil and environment leads to a sharp change in the thermohydrodynamic regime in the flow along the pipeline. The drop in oil temperature along the way causes a change in its rheological properties and is accompanied by phase transitions as a result of the saturation of the flow with heavy hydrocarbons, as well as the formation of near-wall oil deposits on the inner surface of the pipeline. These factors, under certain technological conditions, turn out to be the cause of a gradual spontaneous decrease in the throughput of the pipeline, which, first of all, increases the energy consumption for pumping, therefore, increases the cost pipeline transport. The oil produced in the country's fields has a low sulfur content of 0.035–0.14% (in Brent it is 0.2-1%, and in Urals 1.2-1.3%). 6
Field development At the White Tiger and Dragon fields, the following were built: 13 fixed offshore platforms 22 block conductors 2 technological platforms - maximum capacity: 38 thousand tons per day for oil, 46 thousand tons per day for gas-liquid mixture. 3 compressor stations with a capacity of 9.8 million cubic meters per day. one system collection of low pressure gas ensures the normal functioning of the entire technological process for the collection and transportation of gas to the shore, the preparation of gas lift gas and its use for the mechanized method of oil production at the fields of the JV Vietsovpetro, and also allows you to utilize up to 97% of the produced gas. One of the best onshore bases in Southeast Asia has been created at JV Vietsovpetro for the construction and offshore installation of process and satellite platforms for well drilling and oil and gas production. JV "Vietsovpetro" has four jack-up drilling rigs, more than 20 fleet units, including crane-mounting, fire-fighting, diving and transport-towing vessels, four 7 unmoored loading rigs.
Pipeline from the Dragon field At the end of 1994, a pipeline from the RP-1 production platform of the Rong field to the Central technology platform TsTP-2 of the Bach Ho field, laid along the bottom of the Vietnamese shelf, 33 km long for pumping highly paraffinic oil with a pour point of 250 C. To improve the rheological properties of this oil, Sepaflux ES-3266 pour point depressant produced by BASF is used. At the same time, it was possible not only to significantly reduce the pour point, which ensures reliable pumping of crude oil through an underwater non-thermal insulated pipeline, but also to reduce the plastic viscosity of oil by more than 7 times. 9
Field development Non-mooring loading unit "Vietsovpetro-01" - crude oil storage tanker Full load - 139 thousand tons of oil 9 anchors 10 -15 inclined shafts Divergence to the sides by more than 2 km 10
Oil refining in Vietnam The only operating oil refinery in the country is the Dung Kuat Refinery. Currently, the construction of a refinery in the north of the country is starting and construction is planned in the south. The Dung Kuat refinery was built in three years (from November 2005 to January 2009) and launched in February 2009. The Ngi Son refinery was planned to be built in the north of the country, its capacity, according to the basic design, is 10 million tons per year. Commissioning was scheduled for 2013-2014. The Long Son refinery will be located in the south of the country, its design capacity is also 10 million tons per year. The project is at an early stage of development, partners and investors have not been identified. Commissioning is scheduled for 2016–2020. eleven
Block 09 -3/12 is located in Yuzhno. Konshon oil and gas basin, 150 km southeast of Vung Tau and 20 km east of the White Tiger field. The prospects for oil and gas potential are associated with Oligocene-Miocene deposits and rocks of the crystalline basement. It is planned to process and interpret earlier seismic surveys, assess oil and gas content promising structures block and preparation for drilling the first exploration well Due to the fact that the "Sea Turtle" field is located in the overlap zone of block 09 -3 with the "Southern Dragon" field of block 09 -1, it was decided to combine the two fields into a Joint Operating Zone. In 2010, stable industrial oil production began at the combined South Dragon - Sea Turtle field, which in 2013 reached 12 one million tons.
Block 04 -3 is located 280 km southeast of Vung Tau. The Tien Yng - Mang Cau deposit was discovered within the block. The oil and gas prospects of the block are associated with the Oligocene and Lower Miocene deposits on the Bo Cau, Hoang Hak and Kim Loan structures prepared for drilling. In 2013, drilling of an exploration well was started at the Bo Cau structure. Block 04 -1 is located in the north of the South Kon Son basin, 250 km southeast of Vung Tau. In 2012, a prospecting well ST-2 X was drilled on the Sean-Tien-B structure. Taking into account the results of drilling, special processing and interpretation of seismic data is carried out to identify and prepare promising objects for drilling. 13
Block 42 is located in the Phu Quoc oil and gas basin within the Gulf of Thailand, 400-450 km west of Vung Tau. The prospects for oil and gas potential are associated with the Paleozoic-Mesozoic complex. An oil contract was signed on the terms of the PSA. Under preparation "Agreement on joint activities» between JV Vietsovpetro and PVEP (a subsidiary of KNG Petrovietnam) Block 12/11 is located within Yuzhno. Konshon oil and gas basin, 350 km southeast of Vung Tau. Oil and gas prospects are associated with Oligocene and Lower Miocene deposits within the identified Thien Nga, Chim Cong, Chim Ung, Hong Hac/Hoang Yen and Quyt structures. It is planned to carry out 3D seismic surveys in the block in 2013 to start prospecting and exploration drilling. 14
15
Features of the lithological composition and reservoir properties of horizonsVII + VIIIlower Oligocene age oil field White Tiger (Vietnam)
Bui Khak Hung
National Research Tomsk Polytechnic University, Tomsk
Scientific Supervisor Associate Professor
The White Tiger field is a unique field in Vietnam in terms of oil reserves. It is located on the shelf of southern Vietnam, 120 km southeast of the coastline. The geological section of the deposit is represented by pre-Cenozoic crystalline rocks of the basement and Cenozoic terrigenous rocks of the sedimentary cover, in which sandy-silty and clayey rocks of the Oligocene, Neogene and Quaternary age are distinguished. The basal Lower Oligocene deposits, which wedged out on the slopes of basement blocks occupying a high hypsometric position, are distinguished by the greatest variability in thickness and composition. Among the Lower Oligocene deposits, horizons VII + VIII are the most oil-saturated and belong to oil deposits. industrial value. Therefore, the study of the features of the lithological composition and reservoir properties of horizons VII + VIII is of great importance.
Using the Surfer program, a structural map was constructed for the top of the VII+VIII horizons of the Lower Oligocene and modeled it in 2D (Fig. 1A).
(A) (B)
upper - well / lower - mark (m) upper - well / lower - thickness (m)
Rice. 1.Structural map (A) and isopach map (B) of horizons VII+VIII of the lower
Oligocene of the White Tiger deposit
Figure 1A shows that the drawing of the structural maps of the northern section (horizons VII+VIII of the Lower Oligocene) of the White Tiger deposit changes greatly. In well 1013, the lowest mark -4161m along the top and -4225m along the bottom was discovered, that is, a depression zone is noted in the east direction. And the highest mark is -3336 m along the top and -3381 m along the base in the northwest in well 4, in the area of which the dome of the structure is clearly distinguished. The amplitude of the dome is 470 meters along the contouring isohypse - 3850m. For a visual representation of the distribution of capacities, an isopach map was constructed. (Figure 1B)
Figure 1B shows a northeast strike of disjunctive faults. It can be seen that the most maximum power reaches 94 m in well 10 and is represented by sandstones of continental genesis. And the minimum thickness is 22m and 17m in wells 64 and 83, in the western part of the site.
Formation of the thickness of deposits is possible in two directions of sedimentation conditions. The reduction in the thickness of deposits in the arch and its increase on the limbs of uplifts is due to the erosion of this hill and the filling of depressions with destruction products.
An increase in the thickness of deposits on the slopes of paleo-elevations indicates the accumulation of sediments in the shallow water zone during wave activity.
According to the methodology developed and well logging data, maps of lithological composition and sand content were built (Fig. 2).
https://pandia.ru/text/79/171/images/image004_29.gif" alt="E:\Nam 5\diploma\lithological map of 7+8 Lower Oligocene horizons.jpg" width="258" height="337"> !}
( A) (B)
upper - well upper - well
lower - net-to-gross ratio (%) lower - αPS value
on the right - coefficient of plasticity (%) on the right - thickness (m)
Rice. 2. Map of net-to-gross and clasticity coefficients (A) and map of lithological composition (B) of horizonsVII + VIIILower Oligocene (0-0.2: clays and silty-clayey rocks; 0.2-0.4: siltstone and clayey-silty rocks; 0.4-0.6: mixed sandy-silty-clayey rocks; 0.6 -0.8: Fine-grained sandstone; 0.8-1: Coarse-medium-grained non-clay sandstone)
Figure 2A shows the distribution of type A reservoirs (PS value in the range 1-0.8) in the zone of wells 83, 64, 4, 14, 602, 1014, 1003. The distribution zone of type B reservoirs (PS value in the range 0.6- 0.4) in wells 10, 1013. The distribution zone of type B reservoirs (0.6-0.8) in wells 114, 116, 907. The distribution zone of non-reservoirs is identified in the east, northeast (well 9), in the south ( wells 1106, 12).
In Fig. 2B, we see that the zone of high distribution of sand bodies is located in the area of wells 14; 116 and 1014 with an average thickness of 23 m. The maximum value of the net-to-gross ratio is in well 1014 and corresponds to 70.2%. The maximum value of the coefficient of plasticity is also observed in the well 1.3%. The decrease in the net-to-gross ratio on the crest and its increase on the slopes and at the foot of the uplifts is due to the activity of flows that erode the upland and form alluvial fans of the erosion products.
Along the line of wells 16-9, a geological profile of the VII+VIII horizons of the Lower Oligocene was constructed (Fig. 3).
Rice. 3. Geological profileVII + VIIILower Oligocene horizons at the White Tiger oil field (Vietnam) along the line of wells 10 - 14 - 145 - 116 - 9
Horizons VII+VIII represent an anticlinal fold complicated by faults. On the profile we can see the change in the thickness of the horizons by wells. In well 10, the deposit thickness reaches 94 m. And in well 14, the deposit thickness decreases to 33 m. A fault is noted between wells 14 and 145. And between wells 116 and 9, 2 disturbances were identified, differing in a significant width of the rock crushing zone. The lithological composition of the deposits is heterogeneous. In well 10, we see an alternation of clay and sandy-silty rocks. Clay thickness is 40 m. Clay deposits wedge out and completely disappear in well 14. In well 14, only sandy-alveuritic rocks with a thickness of 33 m are observed. Clay deposits are observed in wells 145, 116 and the thickness of clays in well 9 increases. horizon as a layer. The thickness is insignificant in comparison with the thickness of sandstones and is 6-7 m. In well 9, the thickness of the clay layer increases by 2 times. On the profile, we mark the zones of the highest reservoir properties in wells 14, 145, 116, in which the porosity coefficient varies from 12% to 14% and the oil saturation coefficient is 0.6-0.66 units. Of all the studied wells, the highest oil flow rate was obtained in the well m3/day. With such low porosity values (practically non-reservoir), high oil production rates can be explained by the proximity of the zones of two tectonic faults.
Thus, a complex type of reservoir of porous-fractured horizons VII+VIII was revealed in the northern block of the White Tiger field. In wells drilled close to the zones of tectonic faults, high oil flow rates were obtained. In wells that have only a porous type of reservoir and are far from zones of disjunctive disturbances, much lower oil flow rates are obtained.
Bibliography:
1. P, G, et al. Geology and oil and gas potential of the basement of the Sunda Shelf. M., Oil and gas, 1988, 285s.
2. Yezhov interpretation of geophysical data; Tomsk Polytechnic University. - 3rd ed. - Tomsk: TPU Publishing House, 200p.
3. Pospelov foundation: geological and geophysical methods for studying the reservoir potential and oil and gas potential - Moscow 2005.
