solve a fuzzy logic problem

I have to use fuzzy to find the likelihood of a fire in a car based on the following conditions, The judgment set are as follows: u1—（0.5，0.3，0.1，0.1）Likelihood of a fire occurring while charging a car u2—（0.3，0.2，0.2，0.3）The evaluation set for a car crash fire u3—（ 0 ，0.1，0.4，0.5）The judgment set for a car flooding fire u4—（0.4，0.4，0.1，0.1）The critical set affected and ignited

which means – considering u1 - Likelihood of a fire while charging a car, 50% said it's big, 30% said it was larger, 10% said it was average, 10% said it's small same for u2,u3 and u4

The evaluation set of these four factors is composed of the evaluation matrix：

The fuzzy comprehensive decision-making model of fire risk of new energy vehicles is as follows:

Due to the 0.5＋0.3＋0.2＋0.2＝1.2，In order to find the percentage, the normalization process is performed, namely：

It can be seen from the formula that the comprehensive decision of the four factors affecting the new energy vehicle fire is as follows: 41% people think such accidents are very dangerous, 25% people think relatively dangerous, 17% people think the danger is average, and 17% people think there is no danger.

The above is my question to solve, I have opened fuzzy interface and created 4 inputs Then my doubts are What type of membership function MF type is suitable for this problem, shall i use gaussianmf ? Also have doubt in how to give inputs based on these conditions - like How many MFs (membership functions should i give) What value range shoud i give in params How to set the rules? What is their weight value I should assign? How many outputs should i give, is it one or four I have referred a video on Getting Started with Fuzzy Logic Toolbox, but i'm still not clear to how to solve the problem. I would be of great help if someone could really assist me in solving this?

Answer 1

[Sam Chak]

Hi @Elysi Cochin

Take this slowly. First, specify what those 4 inputs to the fuzzy system are, and the outputs of the fuzzy system. (I don't know what symbols you have chosen, so I just use for inputs and for outputs.)

Before determining the type of MF to be used, and the number of MFs, try constructing the structure of the Fuzzy If-Then Rules from your expert opinions:

IF is 𝒜 AND is ℬ AND is 𝒞 AND is 𝒟, THEN is 𝒫 AND is 𝒬 AND is ℛ AND is 𝒮

where

are the input fuzzy sets and are the output fuzzy sets.

[Elysi Cochin]

Sir, is the rule you created based on the problem I gave?

But sir, dont i need the type and number of MF before i create the rules?

Thats how i saw in the video i mentioned in the question?

Still not clear, thats why so much doubt.

[Sam Chak]

Hi @Elysi Cochin

You have described about the likelihood of a fire in a car, but I cannot relate them to a fuzzy system because your information are insufficient. Therefore, I have written a general guideline so that you can start with something.

Come back to your question in the comment, I'm actually referring to the generic structure of the Fuzzy If-Then Rules. They are not specific rules because I don't know what your input and output fuzzy sets are. Only you, the expert of the fire protection system, know. In fact, you can also decide whether to use the 'AND' operator, or the 'OR' operator, or a mixed of BOTH.

There is nothing wrong with the design procedure you have seen in the video. After all, the example in the video is built by an Expert who knows the Fuzzy If-Then Rules of the system very well. Naturally, having the IF-THEN Rule-base structure in mind, the Expert knows the desired:

• number of inputs and outputs
• type of membership function
• number of membership functions

to be designed for the fuzzy system as shown in the video. To be able to design something like the Expert in the video, you must also have something in mind at first, and therefore, you are advised to perform the requested tasks, or else you can't move forward. The structure of the Fuzzy If-Then Rules can be viewed as the overall picture of the design.

I guess the outputs are the percentage of people in a population, who think such accidents are

1. very dangerous,
2. relatively dangerous,
3. a little dangerous, and
4. not dangerous at all.

However, I don't understand. If you can use the evaluation matrix effectively, why don't you continue using that method?

[Elysi Cochin]

Sir can you tell me, from the datas i have given, what else is needed to design a fuzzy system, because i'm also in the process of collecting data. If you could help me, i could collect it correctly

[Sam Chak]

I don't know what data to be collected because you don't provide the info that I'm seeking in my TWO (2) comments above. Perhaps, the comments were so long-winded that you overlooked some of the very important points that are necessary in the fuzzy system design. It's okay. Try finding more info from the papers on the fuzzy modeling approaches on ScienceDirect, IEEE Xplore, SpringerLink, etc.