The invention relates to a method that allows analysis of machine sounds and estimation of malfunction.

More specifically, the invention relates to a method that allows the machine operators or customers to receive the sounds of the machines online or offline through a mobile application on their mobile devices, to analyze the mentioned sounds with artificial intelligence algorithms and to report the fault information to the operator or the customer through the mobile application.

Prior Art

With the development of technology, machines are developing day by day and their usage areas are increasing. The machines, which are located in many areas, are taken to the service periodically and thus, the necessary maintenance and controls are performed and the malfunctions are eliminated. The maintenance performed extends the lifetime of the machines as well as eliminating malfunctions and functioning properly.

Ensuring that the machines functioning properly is very important in terms of labor, time and cost. Particularly, when malfunction occurs in the heavy equipment, the repair processes prevent the operation of the machinery, causing the work to be impossible and the work plan to be troubled. In this regard, the rapid detection of faults occurring in the heavy equipment and learning from which component the problem is caused will lead to shorter repair time by directing qualified personnel and spare parts to the faulty machine.

Recently, vibration and frequency analysis of the sound has been used in order to determine which part of the machine the fault is in. The operating sounds of the machines give information about their current status. In the current methods, it can often be determined whether there is a problem by regularly checking the vibration values of the machine sounds with measuring devices. However, current solutions do not allow instantaneous learning of the defective part through an online or offline platform when a problem occurs with the machines. Due to insufficient data related to the malfunction, technicians cannot be assigned to the works to provide optimal efficiency and the estimating of malfunction process is prolonged due to misdiagnosis, missing equipment, spare parts. This situation made it necessary to develop a method to meet the need of technicians and spare parts according to the nature of the fault by analyzing the sounds of the machine and making the fault determination.

The document US6772633 mentions a method for facilitating diagnosis of the operation of devices or machines based on acoustics. In this regard; it is ensured that obtaining a sound while a device is operating, processing the sound, acquiring one or more acoustics-based fault-signatures associated with the device, the fault-signatures being stored in a database, analyzing the processed sound based upon the one or more acquired fault-signatures and based upon such analysis, determining whether the device has a fault condition evidenced by the sound that it emitted. Therefore, there is no solution for receiving the sounds of the machines online or offline through a mobile application on the mobile devices of operators or customers, and analyzing the said sounds with artificial intelligence algorithms and reporting the fault information to the operator or customer through the mobile application. There is also no method for the direct detection of the defective part.

The document CN102809493A mentions a method an abnormal sound diagnosis device. A device has been developed to identify abnormal sounds in machines. Fault noise can be detected by time- frequency analysis of machine sound vibrations coming from a data acquisition unit. However, there is no solution to meet the need for technicians and spare parts according to the nature of the fault by analyzing the sounds of the machine by means of the application running online or offline only in a mobile device.

The document US982451 1 mentions methods and systems for diagnosing vehicle systems using vehicle sounds, vibrations, or both. A system comprises one or more sensors and a processor. The sensors measure sound and vibrations from the vehicle. The processor identifies the vehicle system or device based on the measured sound and vibration. Therefore, there is no solution for receiving the sounds of the machines online or offline through a mobile application on the mobile devices of operators or customers, and analyzing the said sounds with artificial intelligence algorithms and reporting the fault information to the operator or customer through the mobile application. In addition, no estimating of malfunction is mentioned.

Consequently, the need to listen to the sounds of the machines online or offline through a mobile application on the mobile devices of operators or customers, and to analyze the said sounds using the fault classification algorithms, and to inform the operator or customer through the mobile application required the emergence of the solution according to the present invention. Objectives and Short Description of the Invention

The aim of the present invention is to provide a solution to meet the need for technicians and spare parts according to the nature of the fault, by analyzing the machine noises and performing estimation of malfunction.

Another aim of the invention is to provide a solution for playing the sounds of the machines online or offline through a mobile application on the mobile devices and analyzing the said sounds with the fault classification algorithms and reporting the fault information to the relevant people through the mobile application or an internet platform.

Another aim of the invention is assignment of qualified personnel to repair operations by means of fast and practical detection of defective parts of machines on an online or offline platform.

One another aim of the invention is to prevent the prolonged estimating of malfunction process due to missing equipment and spare parts incompatibility by means of fast and practical detection of defective parts of machines on an online or offline platform.

The other aim of the invention to provide a method that enables the reduction of logistics and personnel expenses by means of the determination of the personnel and spare parts requirements according to the nature of the fault by means of the estimating of malfunction realized online or offline.

In order to achieve the above aims, the present invention is a malfunction estimation system by analyzing machine sounds, and it comprises at least one mobile device which comprises a mobile application that allows the identification of malfunction sounds and sending them to the audio database, at least one data processing device which includes an internet platform that allows sharing of fault information with the relevant units,

an ERP system that allows the storage and processing of various data,

a data warehouse that allows data exchange between different environments, an audio database that stores data about fault sounds,

a server that allows noise analysis of fault sounds,

a learning component and estimation component for mentioned server,

an offline estimation component for a mobile device that allows offline malfunction estimation. The invention is a method that enables the analysis of machine noises to detect faults and comprises the following process steps: recording the sound of the machine, which is considered to be a malfunction, for a certain period of time via online or offline application on the mobile device and sending it to sound database,

sending the audio data recorded in the sound database to the server via a data warehouse, to determine the component of the malfunction by performing noise analysis of the data from the audio database on the server,

following the determination of which component the fault is in, recording the fault detection information from the server via the data warehouse by sending it to the ERP system via the data warehouse,

sending the diagnostic information recorded in the ERP system to the mobile device and data processing device of the relevant person through the data warehouse

At the method of the invention also comprises continuous improvement of the failure estimation results performed by using the sound data converted into numerical values, by the mentioned server exchanging data together with the audio database via the data warehouse and instant data exchange.

Description of the Figures

In Figure 1 , system components of the method subject to the invention and interaction between them are shown.

Reference Numbers

10. Mobile Device

20. ERP System

21. Data warehouse

30. Audio Database

40. Server

41. Learning Component

42. Estimation Component

43. Offline Estimation Component

50. Data processing device Detailed Description of the Invention

The invention relates to a method that allows the need for personnel and spare parts to be met according to the nature of the fault by means of noise analysis by machine noise analysis.

Machine sounds are recorded online or offline via a mobile application and a platform that works on data processing devices (50), such as machine operators, technicians or customers' mobile devices (10). Mentioned online sounds and audio data from an audio database (30) are analyzed on the server (40) with artificial intelligence algorithms and the fault information is reported to the mobile devices (10) or data processing device (50) of the mentioned personnel or customers through the mobile application or internet platform.

Offline sounds can be evaluated by the algorithm written on the mobile application for mobile devices (10). An offline estimation component (43) for the mobile device is used while estimation of malfunction. The audio data received in the offline estimation component (43) are processed instantly. The mentioned offline estimation component (43) does not have a self-learning ability. The offline estimation component (43) is updated periodically in the form of a week or a month by using the data generated at the learning component (41). It is ensured that the defective part is detected instantly with only one mobile device (10) without the need for additional equipment by means of mentioned method. In this regard, a fast and practical method is presented.

System components in which the present innovative method is applied and interaction between them are shown in Figure 1. The mentioned system mainly has a mobile device (10) containing a mobile application that enables the identification of malfunction sounds and sending them to the audio database and a data processing device (50) that includes an internet platform, an ERP system (20) that keeps various data, a data warehouse (21) that allows data exchange between different environments, an audio database (30) that stores data associated with fault sounds, a server (40) that enables the analysis of noise related to fault sounds, learning component (41) and estimation component (42) associated with the said server (40), an offline estimation component (43) associated with the mobile device (10) that allows offline failure estimation.

In the present invention, the sound of the machine, which is considered to be a malfunction, is recorded for a certain period of time via an online or offline application on the mobile device (10) and sent to the audio database (30). The audio data recorded in the audio database (30) is sent to the server (40) via a data warehouse (21). The server (40) constantly interacts with the audio database (30) and exchanges data via the data warehouse (21). The data accumulated in the audio database (30) is continuously improved with the diagnostic analysis that processes in the server (40). Data is processed daily within the learning component (41) of the server (40), and data is processed instantly within the estimation component (42). Otherwise, by means of a data flow from the learning component (41) to the estimation component (42) at certain times, continuous improvement is made in the estimating of malfunction system. Hereby, noise analysis of the audio data obtained from the audio database (30) is provided. Artificial neural network algorithms are used in the analysis processes performed on the server (40). Variant subtraction techniques are used in noise analysis. By using methods such as Spectrum, Audio Spectrum, Audio Spectrum Envelope and MFCC, sound data are taken and numerical values are converted into numerical values. By means of sound filtering methods, it is ensured that machine sounds are purified from external sounds. After determining which component the fault is in, the fault detection information received from the server (40) via data warehouse (21) is sent to the ERP system (20) and recorded. The diagnostic information recorded in the ERP system (20) is sent to the mobile device (10) of the customer or technical personnel via the data warehouse (21) and to the data processing device (50) in which an internet platform is running. Thus, users can quickly and practically learn what component the malfunction is in.

The present invention, unlike all other methods, enables the detection of the defective part quickly without the need for additional equipment when a problem occurs with the machines. Practical estimating of malfunction via a mobile device (10) only installed on the application provides time and economic gain. When the right personnel are sent to the right job, the time spent by the works is pending is minimized by reducing the planning errors. With the determination of the fault, depending on the nature of the fault, the need for technical personnel and spare parts is determined, and logistics and personnel expenses are reduced.