روش ترکیبی هوشمند برای کنترل کیفیت صنعتی ترکیب شبکه های عصبی، منطق فازی و تئوری فراکتال

The application of type-2 fuzzy logic to the problem of automated quality control in sound speaker manufacturing is presented in this paper. Traditional quality control has been done by manually checking the quality of sound after production. This manual checking of the speakers is time consuming and occasionally was the cause of error in quality evaluation. For this reason, by applying type-2 fuzzy logic, an intelligent system for automated quality control in sound speaker manufacturing is developed. The intelligent system has a type-2 fuzzy rule base containing the knowledge of human experts in quality control. The parameters of the fuzzy system are tuned by applying neural networks using, as training data, a real time series of measured sounds produced by good sound speakers. The fractal dimension is used as a measure of the complexity of the sound signal.

In this paper, we present the use of an intelligent hybrid approach, combining type-2 fuzzy logic and neural networks, to the problem of quality control in the manufacturing of sound speakers. The quality control of the speakers was done manually by checking the quality of sound achieved after production [4]. A human expert evaluated the quality of sound of the speakers to decide if production quality was achieved. Of course, this manual inspection of the speakers was time consuming and occasionally resulted in errors in quality evaluation [8]. For this reason, it was necessary to consider automating the quality control of the sound speakers using intelligent techniques and fractal theory. The problem of measuring the quality of the sound speakers can be outlined as follows: (1)First, we need to extract the real sound signal of the speaker during the testing period after production. (2)Second, we need to compare the real sound signal to the desired sound signal of the speaker, and measure the difference with some appropriate metric. (3)Third, we need to decide on the quality of the speaker based on the difference found in step 2. If the difference is small enough then the speaker can be considered of good quality, otherwise it is of bad quality. The first part of the problem was solved by using a multimedia kit that enables us to extract the sound signal as a file, which basically contains 108,000 points over a period of time of 3 s (this is the time required for testing). We can consider that the sound signal is expressed as a time series [3], which captures the basic characteristics of the speaker. The second part of the problem was addressed by using a neuro-fuzzy approach to train a fuzzy model with the data coming from the good quality speakers [9]. We used a neural network [6] to obtain a Sugeno fuzzy system [22] with the time series of the ideal speakers. In this case, a neural network [5], [17] and [21] is used to adapt the parameters of the fuzzy system with real data of the problem. With this fuzzy model, the time series of other speakers can be used as checking data to evaluate the total error between the real speaker and the desired one. The third part of the problem was solved by using another set of type-2 fuzzy rules [13], which basically are fuzzy expert rules to decide on the quality of the speakers based on the total checking error obtained in the previous step. Of course, in this case we needed to define type-2 membership functions for the error and quality of the product, and the Mamdani reasoning approach was used. We also use as input variable of the fuzzy system the fractal dimension of the sound signal. The fractal dimension [9] is a measure of the geometrical complexity of an object (in this case, the time series). We tested our fuzzy-fractal approach for automated quality control during production with real sound speakers with excellent results.

We described in this paper the application of fuzzy logic to the problem of automating the quality control of sound speakers during manufacturing in a real plant. Type-2 fuzzy logic was used for the fuzzy rules of quality evaluation because human experts have uncertainty in membership function specification. We have implemented an intelligent system for quality control in MATLAB language using the new approach. We also use the fractal dimension as a measure of geometrical complexity of the sound signals. The intelligent system performs rather well considering the complexity of the problem. The intelligent system has been tested in a real manufacturing plant with very good results.