JP7301783B2 - Composition analysis method for electronic/electrical equipment parts scrap, electronic/electrical equipment parts scrap processing method, electronic/electrical equipment parts scrap composition analysis device, and electronic/electrical equipment parts scrap processing equipment - Google Patents

Description

The present invention relates to a method for analyzing the composition of electronic/electrical equipment parts scrap, a method for processing electronic/electrical equipment parts scrap, a composition analysis apparatus for electronic/electrical equipment parts scrap, and an electronic/electrical equipment parts scrap processing apparatus.

In recent years, from the viewpoint of resource conservation, it has become increasingly popular to recover valuable metals from electronic and electrical device parts scraps such as waste home appliances, PCs, mobile phones, and the like. In recent years, the amount of electronic and electrical equipment parts scraps to be processed has been increasing, and efficient collection methods have been studied and proposed.

For example, in Japanese Patent Application Laid-Open No. 2015-123418 (Patent Document 1), after incinerating electrical and electronic device component scraps containing copper, they are pulverized to a predetermined size or less, and the pulverized electrical and electronic device component scraps are smelted with copper. Furnace processing is described.

However, due to the increase in the processing amount of electronic and electrical equipment parts scraps, some of the substances contained in the electronic and electrical equipment parts scraps are unfavorable substances (smelting inhibitors) for the subsequent copper smelting process. It will be put in more than before. If the amount of smelting inhibitors charged into such a copper smelting process increases, there arises a situation where the amount of electronic/electrical equipment parts scraps charged must be restricted.

For example, electronic and electrical equipment parts scraps include parts scraps of various shapes and types, and the raw material composition varies depending on the difference of the supplier. In order to properly sort out the raw materials that are put into the copper smelting process, currently, the raw material composition of electronic and electrical equipment parts scraps is evaluated in advance by visual judgment and chemical analysis, and the results are used in the operation management and operating conditions of the sorting process. are reflected in the settings of

However, in the method of visually determining the raw material composition, evaluation results vary depending on individual experience and skills, and quantitative evaluation is not possible. Chemical analysis and manual sorting to identify raw material composition also take time.

JP 2015-123418 A

The present inventors examined the use of image recognition technology as a method for grasping raw material composition other than visual observation, and conducted various studies. However, the target electronic and electrical equipment parts scraps are mixed with various parts, and the composition varies greatly depending on the situation of the supplier, and the shape is also various. Individual identification may not be performed with high accuracy.

In view of the above problems, the present disclosure is an electronic device that can improve image recognition accuracy and can efficiently analyze the composition of scrap parts in electronic and electrical equipment scraps in a short time regardless of individual experience and skills. Provide a method for analyzing the composition of electrical equipment parts scrap, a method for processing electronic/electrical equipment parts scrap, an apparatus for analyzing the composition of electronic/electrical equipment parts scrap, and a processing equipment for electronic/electrical equipment parts scrap.

In one embodiment of the method for analyzing the composition of electronic/electrical device parts scrap according to the embodiment of the present invention, from among the captured images of raw materials containing a plurality of electronic/electrical device parts scraps including a plurality of component types, Using image recognition processing using a machine learning system, electronic and electrical equipment parts scraps are extracted and composition analysis is performed. If the information is reflected, the certainty of the machine learning system is set to the first threshold, and if the information of the raw material to be subjected to composition analysis is not reflected in the learning data, it is lower than the first threshold. A composition analysis method for electronic/electrical equipment parts scrap including extracting electronic/electrical equipment parts scrap by setting a second threshold value.

In one embodiment of the method for processing electronic and electrical equipment component scraps according to the embodiment of the present invention, a selection step of selecting a specific component type from among a plurality of component types based on the composition analysis result is included. This is a method for processing scrap electronic and electrical equipment parts.

In one embodiment of the electronic/electrical device component scrap composition analysis apparatus according to the embodiment of the present invention, from the captured image of raw materials containing a plurality of electronic/electrical device component scraps including a plurality of component types, A composition analysis device for electronic/electrical equipment parts scraps that extracts electronic/electrical equipment parts scraps and performs composition analysis using image recognition processing using a machine learning system, which is used for learning of the machine learning system. If the data reflects the information of the raw material that is the target of composition analysis, set the confidence level of the machine learning system to the first threshold, and the information of the raw material that is the target of composition analysis is not reflected in the learning data. In the case, it is an electronic/electrical equipment parts waste composition analysis apparatus provided with a processing device for extracting electronic/electrical equipment parts scraps by setting a second threshold value lower than the first threshold value.

In one embodiment of the electronic/electrical device parts waste processing apparatus according to the embodiment of the present invention, an imaging device that images a plurality of electronic/electrical device parts waste including a plurality of component types, and from the captured image, A composition analysis device for electronic/electrical equipment parts scraps that extracts electronic/electrical equipment parts scraps and performs composition analysis using image recognition processing using a machine learning system, which is used for learning of the machine learning system. If the data reflects the information of the raw material that is the target of composition analysis, set the confidence level of the machine learning system to the first threshold, and the information of the raw material that is the target of composition analysis is not reflected in the learning data. In the case, a composition analysis device equipped with a processing device that extracts electronic / electrical equipment parts scrap by setting a second threshold value lower than the first threshold value, and based on the composition analysis result analyzed by the composition analysis device - It is a processing apparatus for electronic/electrical equipment parts scraps, which is provided with a sorter for sorting out specific parts scraps from electric equipment parts scraps.

According to the present disclosure, an electronic/electrical device capable of improving image recognition accuracy and efficiently analyzing the composition of component scraps in electronic/electrical device component scraps in a short time regardless of individual experience and skills. It is possible to provide a component scrap composition analysis method, electronic/electrical equipment component scrap processing method, electronic/electrical equipment component scrap composition analysis apparatus, and electronic/electrical equipment component scrap processing apparatus.

BRIEF DESCRIPTION OF THE DRAWINGS It is a block diagram which shows the processing apparatus of the electronic/electrical equipment components waste which concerns on embodiment of this invention. FIG. 2(a) is a photograph showing an example of an image in which a recognition frame is given to an electronic/electrical device component waste (substrate) present in a captured image, and FIG. 2(b) is an electronic・It is a photograph showing an example of arranging electric equipment parts scraps for each part type (substrate/plastic). It is a flow chart which shows an example of image analysis processing of electronic/electric equipment parts scrap. FIG. 10 is a table showing an example of comparison between the results of estimating the areas of substrates and plastics respectively extracted as electronic/electric devices from the captured image and the measured values of the areas; FIG. It is a table|surface showing the example of a comparison of the recognition result with respect to the measured value at the time of extracting electronic/electrical equipment component waste from a captured image using the machine-learning system which concerns on embodiment of this invention.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described with reference to the drawings. The embodiments shown below are examples of devices and methods for embodying the technical idea of the present invention. not something to do.

As shown in FIG. 1, the electronic/electrical device parts scrap processing apparatus according to the embodiment of the present invention acquires a captured image of a raw material containing a plurality of electronic/electrical device parts scraps including a plurality of types of parts. A composition analysis apparatus 10 for extracting electronic/electrical equipment parts scraps and performing composition analysis using an imaging device 12 capable of capturing electronic/electrical equipment parts and image recognition processing using a machine learning system, and a composition analyzing apparatus A sorter 13 for sorting out specific parts scrap from the electronic/electrical equipment parts scrap based on the composition analysis result analyzed by 10 .

"Electronic/electrical device parts scrap" in the present embodiment refers to scraps obtained by crushing electronic/electrical devices such as waste home appliances, PCs, mobile phones, etc. After being collected, they are crushed to an appropriate size. point to something In the present embodiment, the crushing to obtain electronic/electrical device parts scraps may be performed by the processor himself/herself, or the crushed parts purchased in the market may be used.

The crushing method is not limited to a specific device, and may be a shearing method or an impact method. Therefore, it does not include equipment belonging to the category of grinders whose purpose is to grind finely.

Electronic and electrical equipment parts scrap consists of multiple types of parts such as plastics (synthetic resins) used for substrates, housings, etc., metal pieces, copper wire scraps, capacitors, IC chips, and others. can be classified in more detail. Although it is not limited to the following, in the present embodiment, it is possible to suitably process electronic and electrical equipment component scraps that have been crushed to a particle size of 50 mm or less. Although the lower limit of the particle size is not particularly limited, it is 5 mm or more, more typically 10 mm or more, and further 15 mm or more.

The composition analysis apparatus 10 includes a processing device 100 that performs image analysis processing of captured images according to the composition analysis algorithm according to the present embodiment, a storage device 110 that stores information necessary for various processes, an input device 120, and a display device 130. and The composition analysis device 10 is configured to be able to transmit the results of processing by the processing device 100 via the network 11 to the server 15 or the sorter 14 connected via the network 11 .

The processing device 100 performs image recognition processing on the captured image by using a machine learning system using learning data, and performs composition analysis of the electronic/electrical device component waste in the captured image. For machine learning, various analysis software for image recognition using deep learning or the like can be used. The storage device 110 stores information necessary for processing by the processing device 100 .

For example, in the storage device 110, from the image information of the electronic/electrical equipment parts scrap, a plurality of parts types, namely substrates, plastics, metal pieces, copper wire scraps, capacitors, IC chips, and others (connectors, film scraps) , coated wire scraps, etc.) at least 2 types, preferably 7 types or more, further 10 types or more, extraction data for giving a recognition frame to the extracted electronic / electrical equipment parts scrap Recognition frame data, component type area ratio data for measuring or estimating the area of electronic/electrical device parts scrap within the recognition frame, analysis data necessary for composition analysis of electronic/electrical device parts scrap enclosed by the recognition frame, etc. Various composition analysis information including .

The extraction information used in the extraction process includes, for example, shape information for extracting geometric shapes such as outlines of electronic/electrical equipment parts scraps, color information for extracting colors of electronic/electrical equipment parts scraps, electronic・Includes background information for extracting background information such as the color of the background image around electrical equipment parts scraps, shadows due to unevenness, etc., and rotating images of electronic/electrical equipment parts scraps.

The processing device 100 extracts a plurality of electronic/electrical device component scraps from the captured image by machine learning using the learning data stored in the storage device 110, and classifies them by distinguishing them by a plurality of component types. do. The learning data includes various information for detecting the characteristics of electronic and electrical equipment parts scraps based on past image analysis results. By updating (reflecting) the learning data used for , the recognition accuracy of the electronic/electrical equipment parts scrap is improved.

However, electronic and electrical equipment parts scraps are often crushed before being brought in as raw materials, do not have a fixed shape, and may contain various parts, so all of these are used as learning data It may be difficult to reflect For example, it is often difficult to reflect electronic and electrical equipment scraps from new acquisition routes in the learning data. If the learning data used for the learning of the machine learning system does not reflect the information of the raw material to be subjected to the composition analysis, the accuracy will not be high even if the extraction process is performed, and the recognition error rate will increase sharply. Specifically, when the misrecognition rate exceeds a certain value, for example, when the misrecognition rate exceeds 15% in the present embodiment, or further exceeds 10%, the learning data used for learning of the machine learning system is subjected to composition analysis. It can be judged as a case where the information of the raw material is not reflected.

When a new raw material to be sorted is transported, the processing device 100 extracts electronic and electrical equipment parts scrap from the captured image of the new raw material. Used for learning of the machine learning system. When the information of the raw material to be subjected to composition analysis is reflected in the learning data obtained, the certainty of the machine learning system is set to the first threshold, and the information of the raw material to be subjected to composition analysis is reflected in the learning data. If not, set a second threshold that is lower than the first threshold.

If the information of raw materials to be subjected to composition analysis is not reflected in the learning data, by setting the second threshold lower than the first threshold, it is possible to expand the recognition target of electronic/electrical device parts scrap. Therefore, the recognition target range can be relaxed. As a result, it is possible to improve the extraction rate of electronic/electrical equipment parts scraps even for new raw materials, and to improve the accuracy of composition analysis of the raw materials. In addition, the degree of certainty in the present embodiment indicates the degree of certainty of the extraction result of the electronic/electrical device parts scrap by the machine learning system provided in the processing apparatus 100 of FIG. 1 .

In the case of extraction processing of electronic and electrical equipment parts scraps, if the certainty is set high, the recognition rate for extracting electronic / electrical equipment parts scraps can be increased, but the number of extractions decreases, so it is difficult to judge the composition of raw materials. It is not suitable as an extraction process for On the other hand, if the recognition rate is set too low, erroneous recognition increases, so it is not suitable as an extraction process for judging the composition of raw materials. In this embodiment, the first threshold is preferably 0.2 to 0.5, more preferably 0.25 to 0.4. The second threshold is preferably 0.01 to 0.1, more preferably 0.02 to 0.09. As a result, it is possible to extract 70% or more of the electronic/electric component scraps from the photographed image of the new raw material compared to the actual value by visual inspection, thereby improving the accuracy of the composition analysis.

The processing device 100 provides a recognition frame to the electronic/electrical equipment parts scrap extracted according to the set values of the first and second thresholds of the certainty. The recognition frame may be a minimum frame that outlines the outline of the electronic/electrical equipment parts scrap, or a recognition frame that includes the electronic/electrical equipment parts scrap and the background image around the electronic/electrical equipment parts scrap. good too.

When providing a recognition frame including a background image around electronic/electrical device parts scrap, it is preferable to provide a circumscribing figure that circumscribes the outline of the electronic/electrical device parts scrap as a recognition frame. The shape of the circumscribing graphic may be any shape such as a circumscribing rectangle, a circumscribing circle, or a circumscribing polygon.

Recognition frames are given with different characteristics (color, line shade, line type (thickness, dotted line/solid line, etc.)) for each part type so that the operator can easily distinguish between multiple pieces of scrap parts. is desirable. FIG. 2(a) is a photograph showing an example of a case where substrates are extracted from a captured image as electronic/electrical equipment component scraps. FIG. 2(b) shows an example of a substrate in the upper part of the paper and a plastic extract in the lower part of the paper.

The processing device 100 further measures or estimates the area of the electronic/electrical device component waste present within the given recognition frame. An area measurement tool capable of measuring the area based on the captured image can be used to measure the area of the electronic/electrical device parts scrap. When estimating the area of electronic/electrical equipment parts scrap using a machine learning system, the processing device 100 stores at least information on the area ratio of the electronic/electrical equipment parts scrap with respect to the recognition frame, Using the part type area ratio data, the total area of the electronic/electrical equipment parts scrap with recognition frames is estimated for each of a plurality of part types. The part type area ratio data includes the outline shape information of the electronic/electrical equipment parts scrap, the area of the recognition frame, the area of the electronic/electrical equipment parts scrap within the recognition frame and the area of the background, and the electronic/electrical equipment relative to the recognition frame. Various information such as information on the area ratio of scrap parts and position information can be included.

The processing device 100 further multiplies the result of estimating the total area of the electronic and electrical equipment parts scrap for each part type by the presumed weight per unit area for each of a plurality of predetermined part types, and divides the plurality of part types into By analyzing the weight of the included electronic/electrical device parts scrap, the weight ratio of a plurality of component types is analyzed, thereby analyzing the composition of the electronic/electrical device parts scrap present in the captured image.

The assumed weight per unit area of a plurality of component types can be set in advance by the operator via the input device 120 or the like according to the operation results. Although not limited to the following, for example, when electronic and electrical equipment parts scraps are classified into three types of substrates, plastics, and other parts, the assumed weight of substrate scraps is assumed to be 2.0 g/cm ² , and the assumed weight of plastics is 2.0 g/cm 2 . The weight can be set at 1.5 g/cm ² and the other parts at 1.0 g/cm ² .

In addition to the information on the area described above, the processing unit 100 calculates the number of parts constituting the part type for each extracted part type, the result of estimating the area, and the number of parts. It is also possible to analyze other physical properties such as the average particle diameter and the weight ratio of elements constituting each part type and output them to the display device 130 or the like.

Preferably, the processing device 100 further generates information on operating conditions of a sorter for sorting out a plurality of component types based on the analysis results of the weight ratios of the plurality of component types. As the sorting machine, there are various sorting machines such as picking, color sorter, metal sorter, eddy current sorter, wind sorter, and sieving machine. For example, the processor 100 generates operating conditions for a color sorter that sorts substrates and plastics, for example, based on the analysis result of the component type of wind speed, and stores the generated operating conditions in the storage device 110 . The operating conditions stored in the storage device 110 are output to the sorting machines 13 and 14, and the sorting machines 13 and 14 can perform sorting processing according to the output operating conditions.

The processing device 100 may acquire the position information of each of the electronic/electrical equipment component scraps extracted from the captured image of the electronic/equipment component scraps. Then, the position information may be output to specific sorters 13 and 14 for extracting the positions of specific electronic/electrical equipment component scraps and sorting them out. For example, substrates and metal pieces cannot be separated by specific sorters 13 and 14 such as metal sorters, but if individual position information can be obtained from image information, they can be sorted by sorters 13 and 14 having a picking function. will be able to

The processing device 100 can further learn learning data necessary for the extraction processing and the estimation processing by machine learning. For example, the processing apparatus 100 can determine the characteristics of each of a plurality of component types such as substrates, plastics, metal pieces, and copper wire scraps. With respect to information such as relationships, it is possible to learn the characteristics based on the input of hundreds to tens of thousands of data, and learn so as to improve the accuracy of extraction processing and estimation processing.

In addition, in the event of misrecognition of electronic/electrical equipment parts scrap, etc., or in the event of failure to extract electronic/electrical equipment parts scrap, Preferably, the characteristics are configured to be further learned by the machine learning system of processing device 100 . For example, the processing device 100 may include information on the outline of electronic/electrical device parts scrap not extracted by the electronic/electrical equipment parts scrap extraction process, electronic/electrical equipment parts scrap not extracted by the extraction process, and background images. A new learning model can be created by machine learning or the like according to the input of information on a new recognition frame including , information on the area ratio of electronic/electrical equipment parts scrap to the recognition frame, and the like.

According to the apparatus for analyzing the composition of electronic/electrical device parts scrap according to the embodiment of the present invention, it is possible to quantify and evaluate the weight ratio for each part type of electronic/electrical device parts scrap by image processing of the captured image. can. This makes it possible to more efficiently select sorting conditions such as whether the raw material should be physically sorted or incinerated in a kiln or the like for the composition of the raw material. Based on the recognition result of the captured image, for example, if the raw material contains a large amount of plastic, it is possible to review the purchase conditions based on the weight ratio of plastic and adjust the heat load of the kiln furnace. .

An example of a method for processing electronic/electrical device component scraps using the electronic/electrical device component scrap processing apparatus shown in FIG. 1 will be described with reference to the flowchart of FIG. First, in step S100, a captured image is obtained. The captured image may be a captured image input via the input device 120 or the network 11, or a captured image by the imaging device 12 may be used. In step S101, it is determined whether or not the information of the raw material to be subjected to composition analysis, which is included in the captured image, has already been reflected in learning data used for learning of the machine learning system. The determination may be made manually by the operator via the input device 120 , or may be made automatically by the processing device 100 referring to the data in the storage device 110 .

When the information of the raw material to be subjected to composition analysis captured in the captured image is reflected in the learning data used for learning of the machine learning system, the process proceeds to step S102, and the certainty of the machine learning system reaches the first threshold. is set to If the information of the raw material to be subjected to the composition analysis is not reflected in the learning data, the process proceeds to step S103, where a second threshold lower than the first threshold is set, and the process proceeds to step S104.

In steps S104 to S107, image recognition processing using a machine learning system is performed. Based on the set certainty of the machine learning system, the processing device 100 classifies electronic and electrical equipment scraps present in the captured image by component type (for example, substrates, plastics, metal pieces, copper wire scraps, capacitors , IC chips, and other component types) and extracted. The classification result can be displayed by the display device 130 or the like.

In step S105, the processing device 100 assigns a recognition frame including a background image around the electronic/electrical device parts scrap for each of the plurality of component types to the plurality of electronic/electrical device component scraps extracted from the captured image. do. In step S106, the processing device 100, based on the component type area ratio data having the information of the area ratio of the electronic/electrical equipment parts scrap to the recognition frame, for each of the plurality of component types, the electronic/electrical device to which the recognition frame is attached. Estimate the area (total area) of electrical equipment parts scrap. In step S107, the processing device 100 multiplies the area estimation result by the assumed weight per unit area of the plurality of component types, and analyzes the weight ratios of the plurality of component types, thereby obtaining the electronic and electronic components in the captured image. Analyze the composition of electrical equipment parts scrap. For example, by multiplying the total area of each part type in the image by the weight per unit area, the total weight of scrap parts can be roughly calculated. By comparing the total weight of the parts scrap for each part type, the composition of the electronic/electrical equipment parts scrap contained in the captured image can be analyzed. The analysis result is output in step S109.

According to the composition analysis method and composition analysis apparatus for electronic/electrical device parts scrap according to the embodiment of the present invention, the weight ratio of each component type of electronic/electrical device parts scrap present in the captured image is determined by image analysis. can ask. This allows the raw material composition to be rapidly deduced regardless of individual experience and skill. As a result, the operator can make decisions on raw material purchase conditions and raw material selection methods at an early stage, so that the entire factory can be operated more efficiently.

The background image around the electronic/electrical equipment parts scrap may differ depending on the conditions of the imaging destination, and in some cases, the machine learning system may not be able to appropriately extract the electronic/electrical equipment parts scrap. In the present embodiment, the machine learning system includes background information and information of a new recognition frame obtained by synthesizing an image obtained by combining the edged image of the electronic/electrical equipment parts scrap, including the information of the electronic/electrical equipment parts scrap extraction process. By newly learning learning data for , it is possible to obtain a more flexible composition analysis apparatus that can cope with various conditions in the captured image. In addition, if the misrecognition rate is high even when using this system, the contours, colors, area ratios to the recognition frame, and background of electronic and electrical equipment parts scraps contained in raw materials that are not reflected in the learning data By having the machine learning system learn at least one of the information, it is possible to improve the recognition accuracy of the electronic/electrical device parts scrap.

FIG. 4 shows an example of comparison between the result of estimating the area of a board and a plastic as an electronic/electrical device, respectively, extracted from a captured image according to the analysis method according to the embodiment of the present invention and the measured value. It is a table representing According to the method for analyzing the composition of electronic and electrical equipment component scraps according to the embodiment of the present invention, appropriate evaluation can be made with a measurement error of less than 5.0% for substrates and with a measurement error of less than 10% for plastics. A recognition rate of 90% or higher was achieved when two types of parts were included. Therefore, it can be said that this method is sufficiently effective in numerically grasping the composition of raw materials.

FIG. 5 is a table showing an example of comparison of recognition results with actual measurement values when electronic/electrical device component waste is extracted from captured images using the machine learning system according to the embodiment of the present invention. Example 10 shows the measured value of the extracted number, the recognized (extracted) number, the number of misrecognitions, the recognition rate, and the misrecognition rate when the characteristics are reflected in the learning data used for learning of the machine learning system. Confidence in case was set to 0.3. Example 2 and Comparative Example 2 show examples in which the raw material information is not reflected in the learning data. The certainty factor of Example 2 was set to 0.07, and the certainty factor of Comparative Example 2 was set to 0.3. As can be seen from the results of Example 2 and Comparative Example 1, lowering the confidence factor slightly increases the number of misrecognitions by the machine learning system, but lowering the confidence factor results in a better recognition rate. It is It can be said that this method is sufficiently effective in numerically grasping the composition of raw materials.

Although the present invention has been described by the above embodiments, the statements and drawings forming part of this disclosure should not be understood to limit the present invention. That is, the present invention is not limited to each embodiment, and can be embodied by modifying the constituent elements without departing from the spirit of the present invention. Moreover, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in each embodiment. For example, some components may be deleted from all components shown in the embodiments. Furthermore, components of different embodiments may be combined as appropriate.

For example, the processing device 100 quantifies and analyzes the average area, number, average particle diameter, weight ratio, etc. of each of a plurality of component types from the captured image, thereby enabling manual sorting by electronic and The raw material composition can be quantified and grasped remarkably faster than evaluating the raw material composition of electrical equipment parts scrap.

Furthermore, based on the raw material analysis result analyzed by the processing device 100, information on the operating conditions of a sorter that sorts out a specific part type from among a plurality of component types, for example, selection of a sorter for sorting raw materials Then, operating conditions such as sorting conditions and sorting order can be determined, and sorting processing can be performed based on the operating conditions.

For example, to increase the weight ratio of substrates and plastics in the processed materials after sorting by sorting electronic and electrical equipment parts scraps into light and heavy items by wind sorting using a wind sorter. can be processed. In this case, in the processing by the sorters 13 and 14, the air volume of the wind sorter can be adjusted according to the average grain size of a plurality of component types based on the raw material analysis results analyzed by the machine learning system. The air volume can be, for example, about 5 to 20 m/s, more preferably about 5 to 12 m/s, furthermore about 5 to 10 m/s. Wind sorting can be repeated two or more times depending on the raw material analysis results.

Alternatively, by performing a picking process using a picking device before carrying out the above wind sorting, it is possible to perform a picking process for removing clumps of copper wire scraps. In this picking process, the position information of the scrap copper wire stored in the storage device 110 is output to the picking device as the sorter 13, and the picking device can remove the scrap copper wire according to the output result. This copper wire scrap can be sent, for example, to a valuable metal recovery process.

When the wind sorting is repeated two or more times, a sorting process using a sieving machine can be performed between the first wind sorting and the second wind sorting. In this case, a sieving machine is adopted as the sorting machine 13, and the size of the sieve mesh of the sieving machine for sorting out the characteristic part type based on the average particle size of a plurality of part types obtained from the raw material analysis result can be changed.

In addition to the methods described above, the present invention is applied to the sorting machine 13 used in the magnetic sorting process, the eddy current sorting process, and the optical sorting process for optically sorting metallic objects and non-metallic objects. By utilizing the composition analysis results obtained by the composition analysis apparatus according to the embodiment, the raw material composition can be analyzed by analyzing the image data in real time while continuously photographing electronic and electrical device parts scraps being transported. can be done.

Conventionally, the raw material composition of electronic and electrical equipment parts scrap was evaluated by manual sorting, and the results were reflected in the operation management of the sorting process and the setting of operating conditions. In the method of grasping , it was not possible to perform rapid processing.

According to the embodiment of the present invention, the composition of electronic/electrical device parts whose composition changes from moment to moment is instantaneously determined by image analysis and separation based on predetermined classification data. Since it can be quantified, it is possible to quickly sort a large amount of electronic and electrical equipment parts scrap under more appropriate conditions.

Furthermore, by image analysis of the raw material composition of scrap parts before and after processing by the sorters 13 and 14, the sorting efficiency (performance) of the sorters 13 and 14 can be evaluated based on the amount of change in the scrap parts. By identifying the raw material composition of electronic and electrical equipment parts scraps and extracting their position information, and by interlocking with sorting machines 13 and 14 such as picking devices, color sorters, and metal sorters, it becomes easy to separate individual parts types. . In addition, since the analysis results are displayed on the display device 130 with a frame of a different color for each raw material type, the operator can easily recognize the analysis results, so that erroneous detection by the composition analysis apparatus can be easily recognized.

DESCRIPTION OF SYMBOLS 10... Composition-analysis apparatus 11... Network 12... Imaging apparatus 13, 14... Sorting machine 15... Server 100... Control apparatus 110... Storage device 120... Input device 130... Display device

Claims (7)

Using image recognition processing using a machine learning system from among captured images of raw materials containing a plurality of electronic/electrical equipment parts scraps including a plurality of component types, extracting the electronic/electrical equipment parts scraps, performing a compositional analysis;
When the learning data used for learning of the machine learning system reflects the information of the raw material to be subjected to composition analysis, the certainty of the machine learning system is set to the first threshold, and the composition is to be analyzed. If the raw material information is not reflected in the learning data, extracting the electronic / electrical equipment parts scrap by setting a second threshold lower than the first threshold ,
Extracting the electronic and electrical equipment parts scrap and performing composition analysis,
A recognition frame including an image of the electronic/electrical device parts scrap and the background image around the electronic/electrical device parts scrap is given to the electronic/electrical device parts scrap extracted based on the set certainty. ,
Based on the component type area ratio data having at least the information of the area ratio of the electronic/electrical device parts scrap to the recognition frame stored in the storage device, the recognition frame is attached to each of the plurality of component types. Estimate the total area of electronic and electrical equipment parts scrap,
By multiplying the estimation result of the total area by the assumed weight per unit area for each of the plurality of component types, and analyzing the weight ratio of the electronic/electrical device parts scrap for each of the plurality of component types, the Performing a composition analysis of the electronic/electrical device parts scrap in the captured image
A composition analysis method for electronic and electrical equipment parts scraps , comprising : 2. The method for analyzing the composition of electronic and electrical equipment component scraps according to claim 1, comprising setting the first threshold to 0.2 to 0.5 and setting the second threshold to 0.01 to 0.1. Make the machine learning system learn at least one of the outline, color, area ratio with respect to the recognition frame, and background of the electronic/electrical device parts scrap contained in the raw material that is not reflected in the learning data. 3. The composition analysis method for electronic/electrical equipment parts scrap according to claim 1 or 2, comprising: 4. The method for analyzing the composition of electronic/electrical device scraps according to claim 1, wherein the plurality of component types include at least substrates and plastics. Based on the composition analysis result according to any one of claims 1 to 4 , electronic and electrical equipment parts scrap characterized by including a sorting step of selecting a specific part type from among the plurality of parts types How to handle. Using image recognition processing using a machine learning system from among captured images of raw materials containing a plurality of electronic/electrical equipment parts scraps including a plurality of component types, extracting the electronic/electrical equipment parts scraps, A composition analysis device for electronic and electrical equipment parts scrap for composition analysis,
Giving a recognition frame including an image of the electronic/electrical device parts and the background surrounding the electronic/electrical device parts to the electronic/electrical device parts scrap extracted based on the set certainty,
Based on the component type area ratio data having at least the information of the area ratio of the electronic/electrical device parts scrap to the recognition frame stored in the storage device, the recognition frame is attached to each of the plurality of component types. Estimate the total area of electronic and electrical equipment parts scrap,
By multiplying the estimation result of the total area by the assumed weight per unit area for each of the plurality of component types, and analyzing the weight ratio of the electronic/electrical device parts scrap for each of the plurality of component types, the Perform composition analysis of the electronic and electrical equipment parts scrap in the captured image,
When the learning data used for learning of the machine learning system reflects the information of the raw material to be subjected to composition analysis, the certainty of the machine learning system is set to the first threshold, and the composition analysis target A processing device for extracting the electronic/electrical equipment parts scrap by setting a second threshold value lower than the first threshold value when the raw material information is not reflected in the learning data. Composition analysis equipment for electronic and electrical equipment parts scraps. an imaging device for imaging a plurality of electronic/electrical device component scraps including a plurality of component types;
A composition analysis apparatus for electronic/electrical equipment parts scraps that extracts the electronic/electrical equipment parts scraps from captured images using image recognition processing using a machine learning system and performs composition analysis,
Giving a recognition frame including an image of the electronic/electrical device parts and the background surrounding the electronic/electrical device parts to the electronic/electrical device parts scrap extracted based on the set certainty,
Based on the component type area ratio data having at least the information of the area ratio of the electronic/electrical device parts scrap to the recognition frame stored in the storage device, the recognition frame is attached to each of the plurality of component types. Estimate the total area of electronic and electrical equipment parts scrap,
By multiplying the estimation result of the total area by the assumed weight per unit area for each of the plurality of component types, and analyzing the weight ratio of the electronic/electrical device parts scrap for each of the plurality of component types, the Perform composition analysis of the electronic and electrical equipment parts scrap in the captured image,
When the learning data used for learning of the machine learning system reflects the information of the raw material to be subjected to composition analysis, the certainty of the machine learning system is set to the first threshold, and the composition analysis target When the information of the raw material to be used is not reflected in the learning data, the composition analysis device equipped with a processing device that extracts the electronic / electrical equipment parts scrap by setting a second threshold lower than the first threshold ,
A processing apparatus for electronic/electrical equipment parts scraps, comprising: a sorter for sorting out specific parts scraps from the electronic/electrical equipment parts scraps based on the composition analysis result analyzed by the composition analysis device.

Images