Planning Assessment
Spatial Suitability Scoring For Rail Transit Network Extension by Developing ‘Rail Transit Fuzzy Logic Model’ (RFLM) in Matlab Environment
2024 Volume 12 Issue 1 Pages 222-241
Details
2024 Volume 12 Issue 1 Pages 222-241
This research has focused on the fuzzy logic model for evaluating the necessity of expanding the rail transit network to achieve sustainable transport development. It is essential to distinguish input parameters that can be obtained and updated frequently in a cost-effective way. A framework that can be employed to systemize the planning process by using the ‘Rail Transit Fuzzy Logic Model’. It can be characterized as a spatial suitability oriented fuzzy approach by using five inputs (slope, geology, population, land use and stream), 1 output (Suitability Score) and 108 If-Then rules to assist in combining these data sources in the MATLAB environment. This study consists of three main steps: (i) development of a fuzzy inference system (FIS), (ii) design of the graphical user interface (GUIDE toolbox), and (iii) suitability score database production for the study area. Fuzzy rule-based systems have been applied with parameters represented by fuzzy diagrams that are integrated into an aggregate form to obtain a spatial suitability score. This analysis allows us to assign suitability scores for any specific type of study area. The study area is composed of 208 rows and 216 columns that are uploaded and modelled in MATLAB R2021b as a 216-by-208 matrix. This method commonly has not been employed and processed for a rail network that covers a large geographical area spanning several regions in a city. The proposed methodology offers systematic qualitative data available for decision-makers to be used in the evaluation of rail transportation investments.
The most visionary and impactful rail transit networks are conceived not only to improve mobility and transport resources, but also to transform urban development patterns and improve the accessibility, sustainability, social and economic development and liveability of the city.
Urban rail transit becomes the indicative of significant consequence for supporting liveability and economic vitality, improving land use structure and promoting urban sustainable development. Most metropolitan countries have a large percentage of road congestion and air pollution as a consequence of population and building density. Transit authorities are searching for solutions by extending urban rail transit systems (Xie, 2012).
The development of a new inner city rail line represents one of the largest investments in a city. In developing countries public resources are limited. For instance, the construction of a new urban rail line might use resources that would be spent on other infrastructure projects. Urban rail investments, especially metro projects are much more costly. Railway projects should be planned carefully with a long term, integrated, cost effective and multi-actor vision for urban development. According to (B. Sharma, S. Sharma, et al., 2010), the linguistic preference of the public should be taken into consideration, as this could assure the participation of a larger population. Tools for interactive learning and awareness creation that can aid public in understanding the elaborated and intricate planning processes can help with their involvement. Public participation should be organized concurrently with the plan formation, as opposed to consulting citizens after the plan is prepared. Rail transport serves as a competitive transport mode in comparison with automobile services such as travel time and travel cost. Population density, traffic blackbone zones, land structure, building density, the concentration of education and health services, alternative transport options have complementary impact levels on travel patterns.
This study aims to develop an overarching framework to effectively obtain the suitability scores for different input criteria occupying large geographical areas to evaluate station and corridor-based railway transport planning. ‘Rail Transit Fuzzy Logic Model’ developed to systemize planning process using spatial suitability oriented fuzzy approach. The objectives of this research are to: (1) develop a fuzzy logic model to evaluate rail transit network suitability for expanding the entire network for densely built environments in Istanbul Metropolitan area; (2) to derive the input parameters for suitability analysis of a study area which is represented by matrix sized with 216 rows and 208 columns (3) to extend rail transit suitability scores by software implementation of the model in MATLAB environment. This research methodology has not been applied to rail transport planning issues in previous academic studies. The novel approach proposed in this study will allow transport planners, urban developers, regulators, operators and experts to interactively and independently address all different kind of goals, criteria seamlessly adaptable to all kinds of transportation system evaluations.
Teodorovic (1999) studied fuzzy logic systems for transportation engineering. Fuzzy logic yields considerable results for solving various traffic and transportation engineering problems. The study proposed using a combination of objective and subjective knowledge together. Additionally, the study emphasizes that fuzzy logic models gave significant results compared with the logit modelling for solving route choice problems.
From perceptive of decision support systems, Ocalir-Akunal (2016) examines decision support systems in the transport planning area by selecting case studies from the literature. Avineri, Prashker, et al. (2000) developed a fuzzy expert system to facilitate transportation investment decision making. The fuzzy rule base in the model is introduced as a suitable and flexible tool for characterizing transportation policy decisions.
Bailey (2005) distinguishes approximate reasoning methods to three main types: Fuzzy MCE, Fuzzy inference systems and pairwise comparison methods. Fuzzy MCE is an extension of the multi-attribute decision-making process. Fuzzy inference systems are the most commonly used form of approximate reasoning consisting of a set of linguistic rules that build the inference engine to evaluate the problem statement with adding rules. For instance, if-then statement can be identified as ‘if the slope is moderately suitable and geology is unsuitable and population is low and land use is unsuitable and stream is low then suitability is poor’. The pairwise comparison method is based on fuzzy binary preference aggregation.
Şen (2014) introduces a new engineering aspect as ‘logic engineering’ which indicates that scientific conclusions derived through antecedent assumptions that logically coherence to the study being thought about and discussed. These harmonious relations of parts are verbal and linguistic statements to each other, and they can include vagueness and inexactness in the initial philosophical thinking. Soft computation approaches are made possible by fuzzy logic, which captures quantitative and empirical knowledge and allows for simultaneous simulation of various processes and non-linear relationships.
Urban rail transit system serves as a basis supporting economic and social issues, for instance changing the status of urban development, precipitating the growth, easing the travel of transport users and increasing the quality of living, capturing a liveable environment, stimulating urban vitality, moving to a higher urbanization stage, and affording integrated urban mobility (Bao, 2018). Different perspective of planning strategy was performed by (Liu, Pai, et al., 2018) for green transit-oriented development using a multi-objective planning model which objectives are determined as minimum allocated scale, maximum allocated capacity, budget limit, identification of transit station area etc. The pre-modelling factors are the definition of the area that can or cannot be developed and public transportation system planning in response to make the model be applied in developable urban area with given transportation network.
Istanbul’s urban transport system accommodates 7.321.516 passengers (average weekday) in May 2022 (Metro, Bus, Marmaray, Sea transport, Metrobus, Light railway system). The percentages of trips by metro rail system 57 % , the percentage of trips by light rail system 24 % and the percentage of Marmaray 19 %. These statistics are taken from (Istanbul Metropolitan Municipality, 2022) as shown in Figure 1. Istanbul’s total railway network length reached 321.40 kilometers in the existing situation (Istanbul Metropolitan Municipality, 2023). There are 119.70 kilometers new railway lines are planning to be under operation in 2025.
The number of public transport passengers has declined significantly in Istanbul according to ‘COVID-19’ outbreak. Private car use and ownership, motorcycle, and scooter usage increased in Istanbul. Vulnerable groups such as elderly people and children should avail of public transport taking by new policies, service improvements after the pandemic. Walking, cycling and micro-mobility integration and accessibility to rail transport services should be developed.
In this article, applications of fuzzy logic modelling to systemize the planning process of urban rail transit expansion are reviewed and assessed based on a fuzzy rule-based system. The development of a fuzzy inference system, utilization of fuzzy logic designer and graphical user interface, acquisition of suitability data (suitability scores) carried out respectively.
Zhao and Deng (2011) developed a decision support system to identify alternatives for urban rail transit projects in China that are both effective and functional. The decision-making methodology for selecting urban rail transport projects considers traveller attraction, environmental protection, project viability and operation. By screening urban rail transit projects, multilevel fuzzy comprehensive evaluation is employed to outline the facts of primary sources. According to Karlson, Karlsson, et al. (2016), railway corridor design should be based on ecological and geological factors of the proposed location to improve the sustainability performance of transportation infrastructure development.
Şal and Çubuk (2021) developed a fuzzy multi-criteria assessment model for feasibility studies of transportation projects. Transportation related evaluation criteria are Transportation network, Trip characteristics, Trip rate, Trip Length, Trip production and generation, Trip distribution, Transport demand forecasts, Trip demand forecast modelling, Traffic Analysis, Travel Desire Lines, Target year projections and generation of alternatives.
Travel demand data for the existing year and projected years is one of the most fundamental constituents to get information about travel patterns, transport dynamics of the city. A demand forecasting model was developed within the context of the Istanbul Transport Master Plan Study. The study involves these phases: (1) production and attraction of trips within the scope of traffic analysis regions, (2) traffic volumes of trips according to their purposes, (3) the separation of trips by types on the basis of traffic analysis regions, and (4) the assignment to the road network and the public transport network according to the transportation types. While the city's inner sections have resembling trip distribution to one another, the city will grow east and west in 2023 and relations with the peripheries will improve. Kocakuşak (2021) analyzed urban transport master plans in Turkish cities from the perspective of climate change. The study emphasized that modern transportation planning focuses on solutions that aim to change travel behavior, make better use of existing infrastructure, manage travel demand, improve safety and health, and reduce environmental impacts rather than providing mechanical and demand-meeting solutions to problems. Yılmaz (2020) investigated the use of ‘Analytic Hierarchy Process’ and ‘Spatial Interpolation Methods’ as a travel demand forecasting approach. The study draws attention to the need for alternative travel demand forecasting methods. It is difficult to conduct these surveys and build models according to work time, work force and high budgets. There is no contemporary and official ‘Master Plan Study’ for Istanbul metropolitan region that is occupied and shared with governmental and educational institutions. For this reason, daily district based and zone based trip generation and attraction values are not taken as an input parameter to the Rail Transit Fuzzy Logic model and are not included in the content of this study.
To represent the multi-directional nature and size of urban regions in thematic maps, Gopal, Tang, et al., (2016) suggested a fuzzy sets notion of urban categorization. Using various combinations of input values in the database, fuzzy sets can bring the specific characterizations of urban class into agreement among components. They defined the five inputs (1) Land use, (2) Population, (3) NDVI- Normalized Vegetation Index, (4) VMT-Vehicle Miles Travelled, (5) CVMT-Commercial VMT to define an ‘Urbanness’ output. The ‘Fuzzy Urban Index’ is the output index.
The study area is composed of 208 rows and 216 columns that are uploaded and modelled in MATLAB R2021b as a 216-by-208 matrix. The five input layers were stored into the raster format with a pixel size of 100 × 100 m in the GIS environment. Every cell has a 100 m grid resolution in raster format. The total study area occupies an area of approximately 227 km2 as shown in Figure 2.
The Fuzzy Logic Toolbox in MATLAB software was used to develop a fuzzy logic system (Rail Fuzzy Logic Model-RFLM) consisting of five input variables and one output variable. The inputs are specified by five variables: Slope, Geology, Population, Land use and Stream. Geographic information system (GIS) is used for generating database sets. Datasets were created corresponding to each of the five input variables. These input variables can be changed to research objectives and applicable to different kind of transportation project evaluation and prioritization, resource planning, scenario planning, transit development, urban form, functions and urban development. Data production is explained detailly in the following section 2.3.
Rule box contains 108 rules and the output variable is Suitability Score as shown in Figure 3.
Fuzzy inference system is widely used computer aided system which is called as fuzzy-rule-based system, fuzzy expert system, fuzzy logic designer as well (Karimov, 2010).
The membership function (MF) exemplifies the degree of membership ranging between 0 to 1. If the membership values are zero, there is no membership. When the value converges to 1, the membership is coming closer to the full membership stage. In this study, triangular and trapezoidal membership function types are used.
In Fuzzy set theory, fuzzy operators serve the purpose of combining fuzzy sets by ‘AND’, ‘OR’ operators. A system of fuzzy IF-THEN rules is applied to characterize interpretative provisions in linguistic terms as a knowledge-based composition.
For instance, The fuzzy rule set can be generalized for n rules by this formation:
Rule 1 : if input 1 is A1, …..input n is An , then output 1 is B1
Rule 2 : if input 1 is A2, …..input n is An , then output 1 is B2
…..
Rule n : if input n is An, …..input n is An , then output 1 is Bm
In this study, Mamdani's method is adopted for the inference engine, which was proposed by (Mamdani and Assilian, 1975). Graphical representation of Mamdani inference method is shown in Figure 4. Behret, Uçal, et al. (2011) explained the mechanism of Mamdani inference method as follows: (1) If there is more than one input in the rule, fuzzy set operations should be applied to achieve a single membership value; (2) then implication method (min) is applied to reach each rule’s conclusion; (3) the outputs obtained for each rule are combined into a single fuzzy set, using a fuzzy aggregation operator (max).
Fuzzy Logic is mixed symbolic/numeric approach for rule-based reasoning. It is contingent to construct a fuzzy model with various decision criterions by predicting boundary parameters and membership function plots. This study seeks to formalize qualitative, symbolic, textual statements to a quantitative, numerical values by taking advantage of fuzzy sets.
Fuzzification of input and output variables, selection of fuzzy membership functionsFuzzification is a process in the fuzzy inference system when the values of input variables to system are considered as crisp type (Nandi, 2013). The crisp value transforms into a degree of membership value by using a membership function. In this study, the FIS Editor (Fuzzy Inference System) in the Fuzzy Logic Toolbox and GUI (graphical user interface) within the framework of MATLAB R2021b (The MathWorks, Inc.) software was used for modelling and simulation purposes. Several combinations of membership functions, boundary parameters, fuzzy diagrams were retrained with different levels to investigate the best fit fuzzy-logic model structure. Fuzzy Logic Designer Variables and Parameters identified for this study are shown in Table 1.
| Variables | Membership Function Name | Boundary Parameters | Type | ||||
|---|---|---|---|---|---|---|---|
| A | B | C | D | ||||
| Input | Slope | Low(Very Suitable) | 0 | 0 | 2 | 5 | Trapmf |
| Medium(Moderately_Suitable) | 2 | 5 | 7 | 10 | Trapmf | ||
| High(Unsuitable) | 7 | 10 | 40 | Trimf | |||
| Geology | Unsuitable | 0 | 0 | 2 | Trimf | ||
| Suitable | 4 | 8 | 10 | 10 | Trapmf | ||
| Population | Low | 0 | 0 | 10000 | 30000 | Trapmf | |
| Medium | 20000 | 45000 | 70000 | Trimf | |||
| High | 60000 | 98400 | 98400 | Trimf | |||
| Land Use | Unsuitable | 0 | 0 | 6 | Trimf | ||
| Suitable | 6 | 10 | 10 | Trimf | |||
| Stream | Low | 1 | 2 | 3 | Trimf | ||
| Medium | 2 | 4 | 6 | Trimf | |||
| High | 5 | 6 | 7 | Trimf | |||
| Output | Suitability | Poor | 0 | 0 | 10 | 40 | Trapmf |
| Good | 25 | 50 | 75 | Trimf | |||
| Excellent | 60 | 90 | 100 | 100 | Trapmf | ||
The input variable Slope has three attributes: VS – very suitable, MS – moderately suitable, U– unsuitable. A membership function of type Π (trapmf) was used. The input variable Geology has two attributes: U – unsuitable, S – suitable. A membership function of type I (trimf) was used. The input variable Population has three attributes: L-Low, M-Medium, H-High, The input variable Land use has two attributes: U – unsuitable, S – suitable. The input variable Stream has three attributes: L-Low, M-Medium, H-High. The output variable S(Suitability) has with three attributes: P – Poor, G – Good, E – Excellent. A membership function of type Π (trapmf) was used as shown in Figure 5.
Fuzzy if-then rules implementation consists of a three-stage process: (Tavana, Arteaga, et al., 2017) Fuzzification of inputs, Applying fuzzy operators in rules with multiple inputs and applying the inference method to deduce and determine output values.
The fuzzy inference system's fundamental structure components are the ‘rule base’, ‘database’, and ‘reasoning process’. A rule base comprises a set of fuzzy rules, a database describes the membership functions used in the fuzzy rules, and a reasoning mechanism applies the inference technique to the rules and facts to provide a reasonable output or conclusion (Castillo and Melin, 2001).
Dutu, Mauris, et al. (2018) analyzed the problem of learning fuzzy rule-bases from the standpoint of finding a complementary balance between the system's accuracy, the time it takes to learn the rules, and the interpretability of the rule-bases that are acquired. As a result, they introduce the Precise and Fast Fuzzy Modelling technique, which is a comprehensive design procedure for learning and optimizing the system rule-base.
Şen (2003) stated that in fuzzy logic system, a set of propositions are required for obtaining crisp values by performing fuzzy inference. This set of propositions is named as fuzzy rule-base. For instance, obtaining information in the laboratory environment, making observations in the field work, measuring different variables and collecting the results as a database are stages as a matter of principles in an experimental study, similarly the fuzzy rule base is the systematic set of decisions that is used by the model builder to solve the problem based on his knowledge and experience about the situation. In order to make fuzzy logic propositions, each variable must be divided into sub-fuzzy sets, and the split subsets must cover the entire change space of the superset in which they are located. In this study, 3 membership functions for input A, 2 membership functions for input B, 3 membership functions for input C, 2 membership functions for input D, 3 membership functions for input E and 3 membership functions for input F are defined for a fuzzy system consisting 5 input (A, B, C, D, E) and 1 output (F). One of the membership functions of the output set is referenced for each different probability of each input set and the rule base is created by 108 (3*2*3*2*3) as shown in Table 2.
Setting fuzzy rules is crucial in the design of a fuzzy system. Rules are generated from rule matrix and the relationship between the input and the output variables of the fuzzy model is determined by 108 fuzzy rules as shown in Table 3.
| Slope | Geology | Population | Land use | Stream | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Slope | Low | Low | Medium | High | Suitable | Unsuitable | Low | Medium | High | Unsuitable | Suitable | Low | Medium | High |
| Medium | Low | Medium | High | Suitable | Unsuitable | Low | Medium | High | Unsuitable | Suitable | Low | Medium | High | |
| High | Low | Medium | High | Suitable | Unsuitable | Low | Medium | High | Unsuitable | Suitable | Low | Medium | High | |
| Geology | Suitable | Low | Medium | High | Suitable | Unsuitable | Low | Medium | High | Unsuitable | Suitable | Low | Medium | High |
| Unsuitable | Low | Medium | High | Suitable | Unsuitable | Low | Medium | High | Unsuitable | Suitable | Low | Medium | High | |
| Population | Low | Low | Medium | High | Suitable | Unsuitable | Low | Medium | High | Unsuitable | Suitable | Low | Medium | High |
| Medium | Low | Medium | High | Suitable | Unsuitable | Low | Medium | High | Unsuitable | Suitable | Low | Medium | High | |
| High | Low | Medium | High | Suitable | Unsuitable | Low | Medium | High | Unsuitable | Suitable | Low | Medium | High | |
| Land use | Unsuitable | Low | Medium | High | Suitable | Unsuitable | Low | Medium | High | Unsuitable | Suitable | Low | Medium | High |
| Suitable | Low | Medium | High | Suitable | Unsuitable | Low | Medium | High | Unsuitable | Suitable | Low | Medium | High | |
| Stream | Low | Low | Medium | High | Suitable | Unsuitable | Low | Medium | High | Unsuitable | Suitable | Low | Medium | High |
| Medium | Low | Medium | High | Suitable | Unsuitable | Low | Medium | High | Unsuitable | Suitable | Low | Medium | High | |
| High | Low | Medium | High | Suitable | Unsuitable | Low | Medium | High | Unsuitable | Suitable | Low | Medium | High | |
Note: In the Slope, Low=Very Suitable, Medium=Moderately suitable, High=Unsuitable
| Rule no. | Rule decription |
|---|---|
| 1 | If Slope is Low, Geology is Suitable, Population is Low, Land use is Suitable and Stream is Low Then Suitability is Good. |
| 2 | If Slope is Low, Geology is Suitable, Population is Low, Land use is Suitable and Stream is Medium Then Suitability is Good. |
| 3 | If Slope is Low, Geology is Suitable, Population is Low, Land use is Suitable and Stream is High Then Suitability is Good. |
| 4 | If Slope is Low, Geology is Suitable, Population is Low, Land use is Unsuitable and Stream is Low Then Suitability is Poor. |
| .…… | .…… |
| 108 | If Slope is High, Geology is Unsuitable, Population is High, Land use is Unsuitable and Stream is High Then Suitability is Poor. |
Results of MATLAB Implementation in Figure 8, the result of experiment 30 is shown in rule viewer. First column of rule viewer represents Slope, second column represents Geology, third column represents Population, fourth column represents Land use and fifth column represents Stream. These five columns represent input values. The sixth column is for output, which represents the suitability score. In Figure 7, the value of Slope is 0.926, Geology is 1, Population is 59800, Land use is 1.56 and Stream is 6.53. The output of this rule’s result is 16.9.
Similar to fuzzification, defuzzification is the conversion of a fuzzy amount to a crisp quantity. In a fuzzy inference system, defuzzification is used to convert the output's fuzzy value (Nandi, 2013). This study applies ‘centroid’ defuzzification method (And Method = min, Or Method = max, Implication Method = min, Aggregation Method = max, Defuzzification=centroid).
The Centroid of Area (the expected values of probability distribution) the most widely adopted in defuzzification (Nandi, 2013):
ÝCOA= ∫ µ(y)ydy / ∫ µ(y)dy (1)
where ÝCOA is the crisp value; µ(y) is the membership function corresponding to the value Ý.
The overall structure of the model proposed in this study is based on fuzzy logic system as shown in Figure 8.
Sample set of input data was entered into the fuzzy inference system where rule firing occurs for 108 rules and output results are taken which is called ‘Suitability Score’. Examples are given for three Suitability categories (Excellent, Good and Poor). Parallel visualizing diagram allows Fuzzy inference system variables (input-output) to be viewed in two dimensions as shown in Figure 9.
Surface Viewer is used to show the input changes and the output value in three-dimensional mesh graphics as shown in Figure 10. Surface viewer shows the output surface for two pairs of input data (Slope-Population, Slope-Land Use, Slope-Geology, Stream-Slope).
The GUIDE toolbox is used to create the graphical user interface in MATLAB. This toolbox contains a number of graphical user interface components that can be used to create the desired GUI.
MATLAB GUIs can be built by using GUIDE (GUI Development Environment) with a figure that can be set with components by the help of a graphic layout editor. GUIDE generates a code file that contains callbacks for the GUI and its many components. Both the figure (as a FIG-file) and the code file are saved by GUIDE. The second way to build MATLAB GUI is creating code files that generate GUIs as functions or scripts.
Rail Transit Suitability GUI Interface developed for this work to provide non-expert users to obtain the output score by giving the input values manually and then after clicking the ‘Proses’ tab, output value is automatically come into the screen as shown in Figure 11. The users can easily use this interface separately which was designed previously by the Fuzzy Logic Editor in this study. The software implementation of the RFLM model can be available for users, researchers as an ‘Open Source’ to the academic community.
In this study, this methodology is applied to a geographic dataset using a matrix file. This matrix file is composed of 208 rows and 216 columns. There are 5 input raster datasets to represent the input variables. (Slope, Geology, Population, Land use and Stream) Digital topographical maps are used to create Digital Elevation Model and derive layers such as slope and stream delineation. Geological data are taken from (Istanbul Metropolitan Municipality, 2021a) and created geological raster map. Population data are taken from (Turkish Statistical Institute, 2021) for the study area. Digital topographical maps and Land use data is taken from (Istanbul Metropolitan Municipality, 2021b). All parameter maps were converted and stored in a raster format with 100 m grid resolution. ArcMap 10.8 software was used for the generation of all raster files. These raster files are converted to ASCII text files by using Arc GIS conversion tools. Converted text files are copied into an Excel file for preparing input datasets which can be seen in Figure 12. Then, these input files are imported into the MATLAB environment to implement the ‘Suitability. FIS’ (Fuzzy Inference System).
Each input matrix file is imported as a numeric matrix in MATLAB and loaded in the workspace. Five input matrices are saved as jeo_matrix (208*216), slope_matrix (208*216), pop_matrix (208*216), land_matrix (208*216), stream_matrix (208*216). Every column is saved separately in the workspace. (slope_1 to slope_216, jeo_1 to jeo_216, pop_1 to pop 216, land_1 to land_216, stream_1 to stream_216).
Matlab codes are written in the Command window and the model is run one by one for every column file separately which is shown below.
a=readfis(‘Suitability’); inputs=([slope_1 jeo_1 pop_1 land_1 stream_1 ]);
Model_1=evalfis(a, [slope_1 jeo_1 pop_1 land_1 stream_1]);
Model_2=evalfis(a, [slope_2 jeo_2 pop_2 land_2 stream_2]);
…..
Model_216=evalfis(a, [slope_216 jeo_216 pop_216 land_216 stream_216]);
The output columns are aggregated into a single output excel file, which can be seen in Figure 13.
The suitability scores obtained for five different input criteria presented in three-dimensional view occupying an area of approximately 227 km2 as shown in Figure 14. The study region is uploaded and modelled in MATLAB R2021b as a 216-by-208 matrix. Riad, Billib, et al. (2011) implemented the fuzzy inference method to randomly selected 35 points in a way to cover all the priority zones in that map. These points were chosen from various places in each priority zone on the suitability map to be manipulated by fuzzy logic in MATLAB to discover the optimal locations for artificial groundwater recharge. In this study, fuzzy model run for 44928 cells.
For checking the output values of Model file (208*216), a group of datasets are selected and presented in Table 4 as a 10*10 matrix and all calculations are repeated for this section for crosschecking. All the results are the same as the total output matrix file which was done previously. Defuzzified output value of 50 represents the default values that cells with no rules fired or out of range values. The score of poor suitability has a range from 0 to 40, good suitability has a range from 25 to 75, excellent suitability has a range from 60 to 100.
|
Matris Satır/ Sütun No |
145 | 146 | 147 | 148 | 149 | 150 | 151 | 152 | 153 | 154 |
|---|---|---|---|---|---|---|---|---|---|---|
| 21 | 64 | 81 | 81 | 85,2 | 81 | 81 | 84,3 | 81 | 50 | 50 |
| 22 | 81 | 81 | 81 | 81 | 63,3 | 50 | 81 | 81 | 50 | 50 |
| 23 | 81 | 63,4 | 63,4 | 65 | 81 | 85,2 | 81 | 81 | 50 | 50 |
| 24 | 81 | 81 | 63,4 | 81 | 81 | 81 | 81 | 81 | 81 | 81 |
| 25 | 81 | 81 | 85,2 | 84,3 | 81 | 81 | 81 | 81 | 81 | 81 |
| 26 | 81 | 83,7 | 81 | 81 | 81 | 85,2 | 85,2 | 85,2 | 81 | 81 |
| 27 | 50 | 85,2 | 81 | 81 | 81 | 85,2 | 85,2 | 81 | 81 | 81 |
| 28 | 85,2 | 85,2 | 81 | 81 | 81 | 81 | 81 | 81 | 81 | 81 |
| 29 | 81 | 81 | 81 | 81 | 81 | 81 | 81 | 81 | 81 | 85,2 |
| 30 | 84,6 | 81 | 81 | 81 | 81 | 81 | 81 | 81 | 81 | 81 |
The model output excel file is saved as a text file by adding these variables on top of the text document (ncols 216, nrows 208, xllcorner 381790,7065, yllcorner 4537104,0617, cellsize 100, NODATA_value 0) and opened Model_208_216.txt document in Arc GIS 10.8. Text document’s layer properties defined from the symbology tab to draw raster format map assigning a color to each suitability value as shown in Figure 15.
The flow chart of the methodology was given in Section 2.1.3. In order to make overall modelling process, Rail Fuzzy Logic Model (RFLM) is conducted by using The Fuzzy Logic Toolbox in MATLAB software to develop a fuzzy logic system consisting of five input variables and one output variable. To our knowledge, no prior studies have examined aggregating input decision criteria to one suitability output score value by the help of the fuzzy rule based inference system modelling approach in this research field. In the literature of this field, there is limited studies to build a framework that can be employed to systemize the planning process by a fuzzy approach. Akgun, Sezer, et al. (2012) developed a program by using Mamdani fuzzy inference system for the assessment of landslide susceptibility, topographical, geological, and environmental factors were used as input parameters. (Akil, Yudono, et al., 2023) identifies various supportive and inhibiting parameters and subsequently develops them into rules for spatial analysis. These parameters were obtained from interviews with traders, community members, and government officials. Fuzzy inference system can combine numerical and categorical data and produce numeric data for decision-making processes. According to Akgun, Sezer, et al. (2012), it is not possible to run a model including a very large geographic dataset using existing MATLAB software. In this study, we built a fuzzy logic model which can be applicable for large geographical area that covers an area of approximately 227 km2.
The applicability of this model output findings are tested on Mahmutbey-Bahçeşehir-Esenyurt Railway Line Financial and Economic Feasibility Study’s rail transit assignment results (Istanbul Metropolitan Municipality, 2017). In case of this railway project ‘passenger trips’ for each railway station, the peak hour cross-section traffic values of 2023 were compared to model results. Future research should further develop and confirm these findings by alternative feasibility studies.
In case of this railway project ‘passenger trips’ for each railway station, the peak hour cross-section traffic values of 2023 were compared to model results, suitability scores as shown in Figure 17. Stations between Tahtakale and Esenyurt Central stations are examined which coincides with the working boundary. For instance, it is observed that the Bahçeşehir station’s suitability score 64 on the scale of 100. The number of passenger entries and exits are totally 27369 in one hour morning peak period. Additionally, Esenyurt station’s suitability score 50 on the scale of 100. The number of passenger entries and exits are totally 21284 in one hour morning peak period. The outcome of the ‘Suitability Final Score Map’ values can be used as an assistive analysis for feasibility studies.
Istanbul Transport Master Plan (ITMP) which was approved in 2011 by Transport Coordination Center (UKOME) contains urban rail transit plans for short-mid and long terms. Bakırköy-Beylikdüzü Light Metro Project was planned to construct in mid-term (2013-2015) with a length of 25 km and T-2 project code. According to Istanbul’s Transport Annual Report 2016, The Yenikapı − Sefaköy metro line was planned as an extension of M2 Metro line to the west. By 2023 the line will be extended towards Beylikdüzü in the west. In existing situation, Bakırköy-İncirli-Sefaköy Metro Line, Sefaköy-TÜYAP Metro Line and Beylikdüzü-Sabiha Gökçen Airport high speed metro line projects are continuing at Project-stage and these lines are planning to put into operation in the end of the year 2029. These lines will serve densely populated areas with high demand of public transport system. The study area selected with respect to sustainable development and extension for rail transport network and fuzzy model findings promotes the railway line should be extended towards Beylikdüzü in the West region of Istanbul Metropolitan Area.
This study introduces a ‘Rail Transit Fuzzy Logic Model (RFLM)’ for the evaluation of the necessity to expand the rail transit network to achieve sustainable transport development. The model is flexible, easily modified and can be applied to various kinds of decision variables. The model design is based on the concepts of fuzzy logic theory.
The study’s framework provides a sufficient basis to afford the application of ‘Fuzzy Logic Designer’ by concerning the interactions of different input parameters. This study’s framework provide a potential mechanism that can be applied by regulators, operators, transport planners, experts with different goals and requirements to all kinds of urban transport modes.
A fuzzy inference system is developed by using MatlabTM software. Five input variables (Slope, Geology, Population, Land use, Stream) and one output variable (Suitability Score), membership function plots are defined for each of these variables in Fuzzy Logic Designer. A Mamdani-type fuzzy logic algorithm mechanism was utilized. Identifying the Fuzzy Inference variables is inherently linked to the designer of the system and data availability for any subject under consideration.
The graphical user interface is designed in MATLAB by using the GUIDE toolbox to contrive giving input parameters and processing the results by users manually while fuzzy inference system is working from background in the program environment. Software implementation of the RFLM model is developed MATLAB R2021b and source codes are available to share by user open-source format.
The study has an innovative methodology to assist experts to run a fuzzy inference model including a very large geographic dataset using MATLAB software. The study area is composed of 216 rows and 208 columns matrix size. The five input layers were stored into the raster format with a pixel size of 100 × 100 m in the GIS environment. Every cell has a 100 m grid resolution in raster format and the study area occupies an area of approximately 227 km2.
The selection of input and output variables need elaboration in order to indicate different aspects of rail transport planning. This study’s framework can be applicable to alternative evaluation criteria like environmental and ecological or social impacts of the study area. This methodology assists public transport planners to avoid the ambiguity during evaluation process and aggregate ‘difficult to measure’ attributes as well with the aid of using fuzzy logic. When the input values increase, the implementation of the methodology becomes more difficult to analyze and decreases the ability to standardisation and interpretation.
The overall objective of this study is to understand how the fuzzy logic theory can be used in spatial suitability evaluation by development of fuzzy inference system, combining inputs to find the consequent suitability class, designation of graphical user interface and output data production for rail transit system planning, especially for redevelopment processes. A comprehensive review of previous studies, this methodology is not applicated to a similar transport problem in this extent. Consequently, it was envisaged that this study would offer substantial contribution for the upper and medium scale, short-mid and long term urban rail transport plans and sustainable urban mobility plans. The topic of fuzzy logic systems and tools is a growing research area and offers practicable and extensive methods. Further research can be done by identifying and evaluating additional and goal-driven assessment criteria, fuzzy logic designer variables and parameters according to new goals and requirements in future researches. Additionally, it is recommended to improve the applicability in specific application areas such as road, rail, air, freight transport options.
In conclusion, this study reveals that fuzzy logic modelling is an useful approach that can be processed in a compositional and interpretable mechanism by quantifying different evaluation criteria to establish a replicable, adaptable multi-disciplinary model for Urban Rail Transport Development.
Conceptualization, B.Ç.; methodology, B.Ç.; software, B.Ç.; investigation, B.Ç.; resources, B.Ç.; data curation, B.Ç.; writing—original draft preparation, B.Ç.; writing—review and editing, B.Ç. Authors have read and agreed to the published version of the manuscript.
The authors declare that they have no conflicts of interest regarding the publication of the paper.
Software Implementation of the RFLM model is available in ÇALIŞKAN, BERNA (2022), ÇALIŞKAN, BERNA (2022), “Fuzzy Logic Designer_Berna”, Mendeley Data, V1, doi: 10.17632/kv7ff6n8h6.1
The open-source Rail Transit System Suitability Fuzzy Logic Designer and Matlab GUI, source Matlab codes and the application is developed by corresponding author Berna Çalışkan, the interface was launch in December 2021 using Matlab 2021b. All design stages are developed in MATLAB's graphical user interface (GUI). Source codes are written in script files using MATLAB language.
This research did not receive any grant from funding agencies in the public, commercial, or governmental sectors.
I would like to express my great appreciation to International Review for Spatial Planning and Sustainable Development (IRSPSD) for the valuable and constructive comments and suggestions for improving this research paper.
No funding was received to assist with the preparation of this manuscript.
