Rainfall Prediction Using Rule-based Fuzzy Inference System

Dossier / Travail , 2023 , 7 Pages , Note: 1.0

Informatique - Intelligence artificielle

My main goal in this paper is to lessen the likelihood of flooding and a drought. My main area of interest is the use of temperature, atmospheric and variance to forecast rainfall. Applying rule-based reasoning and fuzzy logic, I attempted to predict when it will rain. Mamdani implication is used to prepare the fuzzy rule foundation. Fuzzy tool box and MATLAB Simulink are the programs used for this. The predicted results are derived using the triangle membership function. The input variables for our model are humidity, atmospheric pressure, temperature, and clouds. All but clouds have three membership functions. There are three membership functions for the output variable. The software used for all implementations is MATLAB 7.9.

Rainfall Prediction Using Rule-based Fuzzy Inference System

Abdur Rahman

International Islamic University Chittagong

Abstract:

My main goal in this paper is to lessen the likelihood of flooding and a drought. My main area of interest is the use of temperature, atmospheric and variance to forecast rainfall. Applying rule­based reasoning and fuzzy logic, I attempted to predict when it will rain. Mamdani implication is used to prepare the fuzzy rule foundation. Fuzzy tool box and MATLAB Simulink are the programs used forthis. The predicted results are derived using the triangle membership function. The input variables for our model are humidity, atmospheric pressure, temperature, and clouds. All but clouds have three membership functions. There are three membership functions for the output variable. The software used for all implementations is MATLAB 7.9. All the images are the author's own work and generated by MATLAB 7.9.

Keyword: Fuzzy logic, Fuzzy interface system, Mamdani, Rainfall, Rules.

Problem Definition:

Water is necessary for both life and the entirety of human activity. The maintenance of human health, as well as the development of the economy and society, are totally reliant upon quick access to sufficient water resources. Nowadays, rainfall forecasting is an essential and vital procedure since rain and flooding cause hundreds of deaths and displaced people every year. Different models can be used to forecast rainfall events. However, weather forecasting is one of the most crucial and difficult operational tasks. Meteorological services around the world perform these duties. It involves many different specialist disciplines of knowledge and is a challenging process. The challenge lies in the fact that all decisions in the field of meteorology must be made in conditions of uncertainty. As a result of its capacity to handle ambiguity and imprecise requirements, intuition artificial intelligence has since been studied in weather forecasting. Fuzzy logic is ideal for a wide range of applications because it can accommodate erroneous and inconsistent real-world data. Classical logic is expanded by fuzzy logic. There's no need for inputs. Because it has a lower forecast error than other mathematical models, fuzzy logic takes precedence over other mathematical models or methods, when weather parameters like humidity, temperature, etc. reasonably fluctuate. The fundamental ideas of a fuzzy logic system are a fuzzy set, an IF-Then fuzzy rule base, linguistic variables, and possibility distribution. The If-Then rule base is used to decode the mathematical input data relationships.

Details about the problem:

The problem in this paper has been simplified by using four variables. The four inputs are:

1. Humidity
2. Atmospheric Pressure
3. Temperature
4. Clouds

The basic solution to the issue is depicted in Figure 1. For the sake of simplification, the fuzzy controller only requires two inputs. It then processes the data and outputs the likelihood of rain. It is up to the various sensors and equipment how to obtain these four inputs. This research is unconcerned with how the sensors function. My presumption is that we have these inputs on hand. The following is a brief introduction to the four points that have been made. Determined by humidity gadget for a hygrometer. Rain is more likely to occur with greater humidity levels. Rain probability decrease with decreasing humidity. On the other hand, a barometer is used to calculate atmospheric pressure. The likelihood of rain increases as the reading decreases. The likelihood of rain decreases as the reading increases. Thermometers measure temperature; they show that the likelihood of rain decreases with decreasing temperature and increases with increasing temperature. Four categories of clouds are recognized. They are Cumulonimbus, Altocumulus, Stratocumulus, and Cirrus. Altocumulus clouds, Stratocumulus clouds, Cumulonimbus clouds, and Cirrus clouds all produce light rain, drizzle, and showers, respectively. Therefore, we can get the data we need for our fuzzy controller using a fairly simple sensor system and devices.

Abbildung in dieser Leseprobe nicht enthalten

Figure 1: Basic block diagram of the Fuzzy Inference Model

Details about the set applied:

Prior to discussing the fuzzy controller's specifics, it is necessary to establish the range of potential values for the input and output variables. These membership functions are used to convert measurement values from the actual world to fuzzy values so that operations can be performed on them (in fuzzy set theory terminology). The labels for the input and output variables, as well as the membership functions connected to them, are displayed in Figure 2. Humidity, Atmospheric Pressure, Temperature, and Clouds values for the input variables are standardized and span the sensor and device domains and range from 0 to 1. The rules that are kept in the database are what the fuzzy controller refers to when making a decision. In a set of rules, these are kept. The rules are just if-then statements, which are straightforward and simple to comprehend because they are just regular English sentences. Rules utilized in this work are developed from common sense and information from a normal context. Here, the following sets of rules are applied to generate the output:

1. If (Humidity is Low) and (Atomospheric_Pressure is High) and (Temperature is Cold) and (Clouds is Cirrus) then (Rainfall is No_Rain)
2. If (Humidity is Low) and (Atomospheric_Pressure is Medium) and (Temperature is Cold) and (Clouds is Cirrus) then (Rainfall is No_Rain)
3. If (Humidity is Low) and (Atomospheric_Pressure is Medium) and (Temperature is Moderate) and (Clouds is Cirrus) then (Rainfall is No_Rain)
4. If (Humidity is Medium) and (Atomospheric_Pressure is High) and (Temperature is Cold) and (Clouds is Cirrus) then (Rainfall is No_Rain)
5. If (Humidity is Medium) and (Atomospheric_Pressure is Medium) and (Temperature is Moderate) and (Clouds is Altocumulus) then (Rainfall is Drizzle)
6. If (Humidity is Medium) and (Atomospheric_Pressure is High) and (Temperature is Moderate) and (Clouds is Altocumulus) then (Rainfall is Drizzle)
7. If (Humidity is Medium) and (Atomospheric_Pressure is High) and (Temperature is Hot) and (Clouds is Altocumulus) then (Rainfall is Drizzle)
8. If (Humidity is Medium) and (Atomospheric_Pressure is High) and (Temperature is Hot) and (Clouds is Strarocumulus) then (Rainfall is Drizzle)
9. If (Humidity is High) and (Atomospheric_Pressure is Medium) and (Temperature is Hot) and (Clouds is Strarocumulus) then (Rainfall is Drizzle)
10. If (Humidity is High) and (Atomospheric_Pressure is Medium) and (Temperature is Moderate) and (Clouds is Strarocumulus) then (Rainfall is Drizzle)
11. If (Humidity is High) and (Atomospheric_Pressure is High) and (Temperature is Moderate) and (Clouds is Strarocumulus) then (Rainfall is Drizzle)
12. If (Humidity is High) and (Atomospheric_Pressure is High) and (Temperature is Hot) and (Clouds is Cumulonimbus) then (Rainfall is Shower)
13. If (Humidity is High) and (Atomospheric_Pressure is High) and (Temperature is Moderate) and (Clouds is Cumulonimbus) then (Rainfall is Shower)
14. If (Humidity is High) and (Atomospheric_Pressure is Medium) and (Temperature is Moderate) and (Clouds is Cumulonimbus) then (Rainfall is Shower)
15. If (Humidity is Medium) and (Atomospheric_Pressure is Medium) and (Temperature is Moderate) and (Clouds is Cumulonimbus) then (Rainfall is Shower)

Abbildung in dieser Leseprobe nicht enthalten

Because the rules are imprecisely specified, their values are also fuzzier than they should be. After being obtained from the sensors and devices, the four input parameters are fuzzified in accordance with the membership functions of the corresponding variables. These are used in addition to the membership function curve to reach a conclusion (using some criteria). As a final result, the precise value of the rainfall is discovered.

Abbildung in dieser Leseprobe nicht enthalten

Figure 3: Labels and membership functions for output variable Rainfall

Result and Discussion:

The sensors collect the input values, which are fuzzyfied using the stated model. The output fuzzy function is then constructed using basic if-else rules and other simple fuzzy set operations, and using the criteria, the output value for Rainfall is obtained. The input-output relations' response surface, as determined by FIU, is shown in Figure 4. Fuzzy Interface Unit is referred to as FIU. The application interface FIDE encodes controller information in this fundamental unit.

Abbildung in dieser Leseprobe nicht enthalten

Figure 4: Input/Output response surfaces.

The outcomes (shown in the plot above) demonstrate how the machine will behave under various circumstances. For instance, if we assume that Humidity and Clouds are both 1, the Rainfall that the model predicts is also 1. This is quite appropriate and persuasive.

Summary:

On the basis of a fuzzy inference system, I attempted to create intelligent models for rainfall event prediction. I have demonstrated that, while employing such a method, it is desirable to combine the practical knowledge of forecasters with theoretical research as well as the effectiveness and precision of computer systems through an algorithm-based process. Instead of the complexity associated with other forecasting techniques, I defined the necessary membership functions for all parameters in this algorithm and implemented effective procedures for gathering the membership functions. The findings indicated a high degree of agreement with the real data, and the methodology was simple to drive and apply.

Future Work:

The models presented can be further improved by increasing the set of input parameters and adjusting the set of rules to get multiple weather phenomena forecast, e.g., fog and/or thunderstorms. Because the set of rules is primarily responsible for the model's accuracy, it is highly dependent on the experience of the person putting the rules in place as well as the length of the training data set; this reflects the limited ability of fuzzy inference systems to learn. In the future, a hybrid intelligent approach that combines the fuzzy inference system with neural networks may provide the ability to learn while reducing the need for experts.

[...]

• Université: University of Chittagong
• Note: 1.0
• Auteur: Abdur Rahman (Auteur)
• Année de publication: 2023
• Pages: 7
• N° de catalogue: V1321941
• ISBN (PDF): 9783346813862
• Langue: anglais
• mots-clé: Fuzzy logic, Fuzzy interface system, Mamdani, Rainfall, Rules