Method and a device for identifying material types of spatial objects

Abstract

The present invention relates to a method for identifying material types of spatial objects characterized in that the method comprising obtaining an acoustic signal from each identified object by deforming the objects mechanically, recording said acoustic signal and comparing it to an acoustic model being obtained on the basis of analysis of reference objects of multiple material types. The present invention also relates to a device for identifying material types of spatial objects, comprising a deformation chamber (K), a mechanical deformation system (F), at least one electro-acoustic transducer (1), an acoustic signal recording assembly (2) and a data processing unit (3) with installed acoustic model being obtained on the basis of analysis of reference objects of multiple material types.

Claims

1. A method for identifying material types of spatial objects, in particular plastic objects, the method comprising obtaining an acoustic signal from each identified object by deforming the objects mechanically, recording said acoustic signal and comparing it to an acoustic model being obtained on the basis of analysis of reference objects of multiple material types, wherein the method is enabling to thoroughly differentiate between various types of plastics preferably using statistical classification of acoustic signals, wherein the objects are set in motion, wherein the deforming step is carried out in such a manner that the objects are crushed, and wherein the deforming step is carried out in such a manner that the objects are set in motion with a velocity of at least 5 mm/s in relation to a solid barrier (PR), which is located in the trajectory of the moving objects.

2. The method of claim 1, wherein the objects are set in motion by gas for best results by compressed air.

3. The method of claim 1, wherein the objects are set in motion by gravity.

4. The method of claim 1, wherein the objects are set in motion by a stream of liquid.

5. The method of claim 1, wherein the deformation step is carried out in such a manner that the objects are hit by a solid body moving at a velocity of at least 5 mm/s.

6. The method of claim 5, wherein the body is set in motion by gas, for best results by compressed air.

7. The method of claim 5, wherein the body is set in motion by gravity.

8. The method of claim 5, wherein the body is set in motion by a stream of liquid.

Description

(1) Exemplary embodiments of the present invention are illustrated in the following examples and in the accompanying drawings, in which

(2) FIG. 1 is a schematic view of the device for identifying material types of objects according to one embodiment of the disclosure,

(3) FIG. 2 is a schematic view of the device for identifying material types of objects according to the embodiment in which objects are crushed,

(4) FIG. 3 presents the example of the analysed acoustic signal with the whole registered signal shown in the upper part, and the chosen fragment of the signal shown below it,

(5) FIG. 4 presents a graphical illustration of the identification results for PET and HDPE plastics materials with the comparison to the acoustic model obtained on the basis of analysis of reference objects,

(6) FIG. 5 presents a graphical illustration of the identification results for polystyrene and PVC plastics with the comparison to the acoustic model obtained on the basis of analysis of reference objects,

(7) FIG. 6 presents a graphical illustration of identification results for aluminium cans as well as plastic packages and packaging waste with the comparison to the acoustic model obtained on the basis of analysis of reference objects,

(8) FIG. 7 is a schematic view of the device for identifying material types of objects according to the embodiment in which objects are collided with a stiff barrier,

(9) FIG. 8 is a schematic view of the device for identifying material types of objects according to the embodiment in which objects are driven into motion with the use of an actuator,

(10) FIG. 9 is a schematic view of the device for identifying material types of objects according to the embodiment in which objects are hit with a stream of crushed solid body,

(11) FIG. 10 presents the graphical illustration of the identification results for three sample groups of objects analysed by the device according to the embodiment in the FIG. 7 with the comparison to the acoustic model,

(12) The embodiments are discussed in the following examples provided to further illustrate the presented invention. The examples are not meant to limit in any manner the scope of the invention.

EXAMPLE 1

Identification of Packages and Packaging Waste from PET and HDPE

(13) a) Construction of the Device

(14) In an embodiment of the invention the device consists of a deformation chamber K, mechanical deformation system F, one electro-acoustic transducer 1 in a form of a microphone placed in the deformation chamber K, which is connected to the assembly 2 which registers acoustic signal, to which the data processing unit is connected, in a form of a computer 3 on which the acoustic signal model obtained from the analysis of model objects of multiple material types is installed. In this embodiment, shown in FIG. 2, the mechanical deformation system consists of a crusher CR. In this embodiment the crusher consists of an actuator with a solid plane pressing the object to the sides of the chamber K. The device is fitted with a feeder P, which feeds objects one by one, deformation chamber K emptying assembly OP, and a sorting device SO.

(15) b) Creating Acoustic Signal Model Based on the Analysis of Model Objects

(16) An acoustic signal model was obtained in the following way: a thousand of each kind of objects made of HDPE and PET are gathered and constitute a training set. Objects from the training set are fed to the machine which operates in the previously described manner. The feeder P, which feeds objects one by one, feeds the objects to the deformation chamber K, where they are mechanically deformed, in this case crushed in a specified time period. The acoustic signal created during crushing is recorded using a microphone 1 and a signal recording assembly 2, and next, the computer 3 calculates a set of acoustic signal parameters both temporal and spectral which are known from the signal processing domain. In this case two parameters characteristic for HDPE and PET were chosen: zero crossing density in the time domain zed, and second order normalised spectral moment, interpreted as a square of signal bandwidth Muc2. FIG. 3 presents the fragment of the recorded acoustic signal, with a marked fragment which was chosen for the analysis of the model object. After calculating the signal features for all objects from the model set, classifier training takes place, that is, adjusting the statistical algorithm to the model set to create the analysis model for the model objects. After recording the classification model in the device memory, it is possible to identify objects of unknown type.

(17) c) Testing the Analysis Model for Model Objects

(18) In order to test the analysis model for model objects, a mixed set of objects containing 111 HDPE objects and 120 objects from PET were subjected to the analysis. Objects from the testing set were fed to the machine which was operating according to the method for creating a classification model described previously. Next, after zed and Muc2 parameters were calculated for the objects from the test set, the classification algorithm assigned them to one of two types of identified objects.

(19) The diagram from FIG. 4 presents the results of the test group object analysis described above. The sets corresponding to the PET and HDPE plastics are linearly separable and they can be differentiated on the basis of the model established from the model object set, represented as a line.

(20) According to the model, on the graph from FIG. 4, all the points located under the line are from the set of HDPE objects, and points located over the line—from PET.

(21) d) Identification and Segregation of Packages and Packaging Waste

(22) Packages or packaging waste are fed individually, through the feeder P which feeds objects one by one, to the mechanical deformation chamber K. In this embodiment of the invention the deformation step is carried by crushing the objects accomplished by a crusher CR. Crushing takes place under physical conditions specified in the model, i.e. the exact same force and time of crushing is maintained. The acoustic signal created during crushing is recorded by the microphone 1 and the signal recording assembly 2. Next, the computer 3 calculates the characteristic parameters: zero crossing density in the time domain zed and second order normalised spectral moment, interpreted as a square of signal bandwidth Muc2. The computer 3 compares these parameters with the model and sends the information about the material type to the sorting device SO. The object is removed from the deformation chamber K by the chamber emptying assembly OP, and is directed to the sorting device SO, which performs the sorting process based on the information received from the computer 3.

EXAMPLE 2

Identification of Polystyrene (PS) and Polyvinyl Chloride (PVC) Objects

(23) In the embodiment of the invention, the device is consisted as described in example I, except that it is fitted with a dampened deformation chamber K, and the model is developed as described in the example 1, except that the acoustic signal model of PS and PVC objects is created, 1000 pieces each, and the chosen and calculated parameters regarding the acoustic signals created during the object crushing process, which are known to a person skilled in the signal processing domain, are different. In order to create the model, two parameters characteristic to PS and PVC were chosen: spectral flatness measure SFM and spectral slope (spslope). The model was tested on 115 objects made of polystyrene PS and 90 polyvinyl chloride PVC objects. The diagram from FIG. 5 presents the results of model testing. According to the model, in the graph from FIG. 5, all the points located under the line are from the set of PVC objects, and points located over the line—of PS. Identification of the tested objects takes place similarly to the example 1.

EXAMPLE 3

Identification of Aluminium Objects in the Stream of Mixed Cans and Packages as Well as Plastic Packaging Waste

(24) In the embodiment of the invention the device is constructed as described in the example 1 while the acoustic signal model is developed in similar way to that described in example 1, except that the set of model objects consists of 1000 objects of aluminium cans and 1000 objects made of plastics (PET, HDPE, LDPE, PP, PS and PVC), and a different set of parameters was chosen from the acoustic signal parameters calculated for acoustic signals created during crushing objects, known to a person skilled in the signal processing domain. In order to develop the model, two parameters characteristic for the tested object groups were chosen from the set of calculated parameters: third order normalised central spectral moment—signal spectral skewness Muc3 and first order normalised spectral moment—power spectrum centroid Mu1. The model was tested on 100 aluminium can objects and 650 plastic package objects. The diagram from FIG. 5 presents the results of model testing. According to the model, on the graph from FIG. 5 all the points located under the line are from the set of plastic objects, and points located over the line from the set of aluminium cans. The identification of the tested objects takes place similarly to the example 1.

EXAMPLE 4

Identification of Material Tape of Spatially Formed Waste Objects by Putting in Motion by Direct Impact of Compressed Air

(25) a) Construction of the Device

(26) In the embodiment of the FIG. 7 device for identifying of material type of spatially formed objects, especially plastic waste objects, comprises the deformation chamber K, the mechanical deformation system F, the object feeder P, the emptying system OP, sorting device SO, electro-acoustic transducer in the form of a microphone 1, circuitry for registering the acoustic signal 2 and a data processing unit in the form of a computer 3 with an installed model acquired from analyzing exemplary objects. In the embodiment of the invention the mechanical deformation system F comprises a stiff barrier PR, and the drive WZ for setting the objects in movement with the use of directly applied compressed air.

(27) b) Establishing a Model from Exemplary Objects

(28) The model of analysis of exemplary objects was acquired in the following way: 2000 objects of type HDPE and PET and 2000 objects of different types, in particular 500 objects of each of the following types: ALU, PS, PP, Tetra-pack, are collected, constituting the training set. The objects from the training set are fed into the mentioned device. The system for object feeding P feeds the objects into the system for putting objects in motion WZ, which accelerates them to a velocity not smaller than 0.5 m/s by applying compressed air. Next, the objects collide with a stiff barrier PR inside the deformation chamber K. The signal emitted as a result of the collision is registered with the microphone 1 and the circuitry for registering signals 2. Subsequently, employing a computer 3 the set of signal features is calculated. The set of signal features includes temporal and spectral features known to the person trained in the technical domain, two of which are particularly distinctive for PET and HDPE types: signal energy en and 1-st order normalized spectral moment Mu1 understood as the spectral centroid of the signal. After calculating the features from all objects from the training set, the training of the classifier is performed, i.e. fitting of a statistical algorithm to the set of known patterns, which allows for establishing an acoustic model of exemplary objects. The model is written in the computer's memory.

(29) c) Testing the Acoustic Model of Exemplary Objects.

(30) In the process of model validation a test set, comprising 200 objects of type HDPE, 200 objects of type PET and 200 objects of other types, including 50 objects of each of the types: ALU, PP, PS, Tetra-pack, is analyzed. The objects are fed into the device described in section a). Next, after calculating the features en and Mu1 for objects from the test set, the classification algorithm assigns them to one of the two types of recognized objects.

(31) The distribution of signal features for different material types in the multidimensional feature space allows for recognizing the type of material. It is illustrated in FIG. 10 where the distributions of parameters en and Mu1 of the signals emitted by colliding the objects with a stiff barrier are depicted.

(32) d) Identification and Segregation of Packaging Waste Objects.

(33) The packaging waste objects A are fed into the deformation chamber K with the feeder P and the drive for putting objects in motion WZ, accelerated to a velocity not smaller than 0.5 m/s and not greater than 10 m/s, and collided with a stiff barrier PR. The signal emitted as a result of the collision is registered by a microphone 1 and circuitry for signal registration 2. Next, the signals are analyzed with a computer 3, in which the signal features are calculated, including signal energy en and 1-st order normalized spectral moment Mu1 understood as the spectral centroid of the signal. The classification algorithm installed in the computer 3 compares the features to the acoustic model of exemplary objects, assigns them to one of the known types and sends a signal to the sorting device SO. The object A is removed from the chamber K by an emptying system UP. Subsequently, it is dropped into the sorting device SO, which sorts the object accordingly to the signal received from the computer 3.

EXAMPLE 5

Identification of Material Type of Spatially Formed Waste Objects by Putting in Motion by Gravity

(34) The recognition process is carried out by the device described in example 4, in which the objects A are put into motion by dropping from 1 meter height.

(35) The model described in the example 4 is employed accordingly.

EXAMPLE 6

Identification of Material Type of Spatially Formed Waste Objects Based on Deformation Carried Out by Colliding Objects with a Moveable Solid Body Mounted on a Piston of a Pneumatic Actuator

(36) a) The Recognizing Device

(37) In the embodiment of FIG. 8, the device for identification of material type of spatially formed objects, especially plastic waste objects, comprises the deformation chamber K, object feeder P, the mechanical deformation system, the emptying system OP, sorting device SO, electro-acoustic transducer in the form of a microphone 1, circuitry for registering the acoustic signal 2 and a data processing unit in the form of a computer 3 with an installed model acquired from analyzing exemplary objects. In this example and embodiment the mechanical deformation system consists of an actuator WR, whose piston has a solid body CS attached to it.

(38) b) Acquiring the Model of Exemplary Objects

(39) The model is acquired accordingly to the example 4, the deformation step is carried in such a way that the objects are hit with a solid body moving at a velocity of at least 0.5 m/s and not greater than 10 m/s.

(40) c) Validation of the Model of Exemplary Objects

(41) The validation is carried out according to example 4 c), given the device described in FIG. 8 is employed.

(42) d) Recognition and Segregation of Packaging Waste Objects

(43) The recognition and segregation is performed according to the example 4, given the object A is hit with a solid body CS driven into motion by means in the form of an actuator whose piston moves at a velocity of at least 0.5 m/s. Subsequently, the acoustic signal emitted during the collision is registered. The signal is then analyzed according to the example 4.

EXAMPLE 7

Identification of Material Type of Spatially Formed Waste Objects by Hitting them with a Stream of Sand

(44) a) Recognizing Device

(45) In the embodiment shown FIG. 9, the device for identifying material type of spatially formed objects, especially plastic waste objects, comprises the deformation chamber K, object feeder P, the mechanical deformation system F, the emptying system OP, sorting device SO, electro-acoustic transducer in the form of a microphone 1, circuitry for registering the acoustic signal 2 and a data processing unit in the form of a computer 3 with an installed model acquired from analyzing exemplary objects. In this example, the mechanical deformation system F consists of a nozzle supplying a stream of sand D.

(46) b) Acquiring the Model of Exemplary Objects

(47) The model is acquired accordingly to example 4, given the objects are hit with a stream of sand flowing at a velocity of at least 0.5 m/s and not greater than 10 m/s.

(48) c) Validation of the Model of Exemplary Objects

(49) The validation is carried out according to example 4 c), given the device described in FIG. 9 is employed.

(50) d) Recognition and Segregation of Packaging Waste Objects

(51) The identification and segregation is performed according to example 1, given the object A is hit with a stream of sand flowing at a velocity of at least 0.5 m/s and not greater than 10 m/s. Subsequently, the acoustic signal emitted during the collision is registered. The signal is then analyzed according to the example 4.

(52) The invention has been described in detail with particular embodiments, but it will be understood that variations and modifications can be effected within the spirit and scope of the invention.