KR102268240B1 - Operation system and method of automated waste collection bins based on deep learning based artificial intelligence - Google Patents

Abstract

In a system for operating an artificial intelligence based on deep learning of an automated waste collection box, the system according to an embodiment of the present invention comprises: each of automated waste collection boxes; an AI device; and a deep learning-based AI operation server. Therefore, the present invention is capable of allowing an artificial intelligence system to be operated efficiently.

Description

OPERATION SYSTEM AND METHOD OF AUTOMATED WASTE COLLECTION BINS BASED ON DEEP LEARNING BASED ARTIFICIAL INTELLIGENCE

The present invention relates to the operation of an automated waste collection box, and more particularly, to a deep learning-based artificial intelligence operating system and method of an automated waste collection box.

Waste is mainly collected in one place and disposed of in a way that a waste collection vehicle collects it collectively. Waste bins installed on roadsides, public places, residential areas, etc. load various wastes (eg, general waste, recycling waste, industrial waste, medical waste, or food waste). The general waste, recycling waste, industrial waste, or food waste will be collectively referred to as “waste” hereinafter.

In general, it happens that waste is collected before the bins are full. In such a case, the maintenance of the waste collection box is more expensive than necessary. In addition, even though the waste bin is full, there are cases where the collection time does not come, and the waste comes out of the container. In such a case, there is a problem that an odor is generated and the aesthetics is impaired.

In order to solve the above problems, various automated waste collection boxes have been introduced. In this case, even if the operator does not directly visit the various automated waste collection boxes, an operation method using artificial intelligence is required to automatically operate the various automated waste collection boxes in a short or long distance.

The present invention provides an artificial intelligence system and method for efficiently operating various automated waste collection boxes.

A system according to an embodiment of the present invention, in a deep learning-based artificial intelligence operating system of an automated waste collection box, compresses or generally accommodates waste, generates detection information of each sensor, and the type and state of each automated waste collection box Each automated waste collection box for transmitting first information related to, the detection information, and second information indicating the location of each automated waste collection box to an AI (Artificial Intelligence) device; Receiving the first information, the sensing information, and the second information, selecting at least a plurality of parameters from among a plurality of parameters included in the sensing information to generate a first parameter set, and a recursive neural network (RNN: The first set of parameters is input to the first deep learning algorithm of Long Short Term Memory (LSTM) among Recurrent Neural Network) techniques, and the first collection schedule and first operating elements for each automated waste collection box output, and when the detection information includes a large amount of data, deep unprocessed detection information, the first information, and the second information including at least one parameter that has not been processed by the AI device among the plurality of parameters The AI device for transmitting to a learning-based AI operation server; receiving the first information, the unprocessed detection information, and the second information, and selecting a plurality of parameters from among a plurality of parameters included in the unprocessed detection information and a plurality of preset parameters to obtain a second parameter A set is generated, and the second parameter set is input to a long and short-term memory second deep learning algorithm to output a second collection schedule and a second operating element for each automated waste collection box, and the first information and the second information based on the second collection schedule and the collection vehicles available for the schedule matching the second operation factor, determine the relevant collection vehicle and manpower suitable for the characteristics of the waste and collection box, and instruct the collection vehicle and the terminal of the person in charge and a deep learning-based AI operation server, wherein the first operation element and the second operation element include at least one of service of the automated waste collection box, troubleshooting, full pre-removal, additional deployment, and evacuation of the deployment area. .

In the method according to an embodiment of the present invention, in the deep learning-based artificial intelligence operation method of the automated waste collection box, in the AI (Artificial Intelligence) device, detection information of each sensor generated in each automated waste collection box, each automated waste Receiving first information related to the type and state of the collection box and second information indicating the location of each automated waste collection box; generating, in the AI device, a first parameter set by selecting at least a plurality of parameters from among a plurality of parameters included in the detection information; In the AI device, each of the automated wastes is collected by inputting the first set of parameters to the first deep learning algorithm of Long Short Term Memory (LSTM) among Recurrent Neural Network (RNN) techniques. outputting a first collection schedule and a first operating element for ; In the AI device, when the detection information includes a large amount of data, unprocessed detection information including at least one parameter not processed by the AI device among the plurality of parameters, the first information, and the second information The process of sending to a deep learning-based AI operation server; Receiving the first information, the unprocessed sensing information, and the second information in the deep learning-based AI operation server; generating, in the deep learning-based AI operation server, a second parameter set by selecting a plurality of parameters from among a plurality of parameters included in the unprocessed detection information and a plurality of preset parameters; In the deep learning-based AI operation server, inputting the second parameter set to a second deep learning algorithm for long-term memory and outputting a second collection schedule and a second operation element for each automated waste collection box; and in the deep learning-based AI operation server, based on the first information and the second information, according to the characteristics of waste and collection boxes among the collection vehicles available for the second collection schedule and a schedule matching the second operation element Determining the collection vehicle and manpower, and instructing the collection vehicle and the terminal of the person in charge, wherein the first operating element and the second operating element are service of the automated waste collection box, troubleshooting, and full-time player It includes at least one of moving, additional deployment, and evacuation of the deployment area.

The present invention enables the integrated management of each automated waste bin.

The present invention enables the construction of a reliable database, big dataization, and analysis thereof.

The present invention can become policy data that can be applied to reflect future waste collection policies by analyzing the accumulated database.

The present invention can minimize the social exposure of pathogens by minimizing the contact between waste and humans and animals.

The present invention has the effect of reducing the collection cost by efficiently managing the collection of waste.

The present invention can establish a minimum operating manpower and equipment.

The present invention can provide a flexible and highly reliable automated waste collection box collection control system.

The present invention can prevent receivables for waste collection costs by utilizing an automated waste collection equipment capable of advance payment.

The present invention may enable remote maintenance and operation support through the Internet.

The present invention can prevent the external exposure of the generated waste by increasing the accuracy of the waste collection schedule.

1 is a block diagram of a collection control system of an automated waste collection box according to an embodiment of the present invention.
2 is a block diagram illustrating an edge computing gateway according to an embodiment of the present invention.
3 is a block diagram of a deep learning-based AI operation server according to an embodiment of the present invention.
4 is a flowchart illustrating the operation of a deep learning control system of an automated waste collection box and an edge computing gateway according to an embodiment of the present invention.
5 is a flowchart illustrating a detailed operation of an edge computing gateway according to an embodiment of the present invention.
6 is a flowchart illustrating detailed operations of big data analysis, deep learning algorithm function, and operation function control of a deep learning-based AI operation server according to an embodiment of the present invention.
7 is a flowchart illustrating a detailed operation of a deep learning-based AI module in a deep learning-based AI operation server according to an embodiment of the present invention.
8 is a flowchart illustrating a detailed operation of determining the operation result of the deep learning-based AI module and determining the selection and application rate of the artificial intelligence neural network operation parameter for correcting the operation result according to another embodiment of the present invention.

Hereinafter, various embodiments of the present invention will be described with reference to the accompanying drawings. However, this is not intended to limit the technology described in the present invention to specific embodiments, and it should be understood that various modifications, equivalents, and/or alternatives of the embodiments of the present invention are included. . In connection with the description of the drawings, like reference numerals may be used for like components.

In the present invention, expressions such as “have”, “may have”, “include”, or “may include” indicate the presence of a corresponding characteristic (eg, a numerical value, function, operation, or component such as a part). and does not exclude the presence of additional features.

In the present invention, expressions such as “A or B”, “at least one of A or/and B”, or “one or more of A or/and B” may include all possible combinations of the items listed together. . For example, "A or B", "at least one of A and B", or "at least one of A or B" means (1) includes at least one A, (2) includes at least one B; Or (3) it may refer to all cases including both at least one A and at least one B.

Expressions such as "first", "second", "first", or "second" used in the present invention may modify various elements, regardless of order and/or importance, and may modify one element to another. It is used only to distinguish it from the components, and does not limit the components. For example, the first user equipment and the second user equipment may represent different user equipment regardless of order or importance. For example, without departing from the scope of the rights described in the present invention, the first component may be named as the second component, and similarly, the second component may also be renamed as the first component.

A component (eg, a first component) is "coupled with/to (operatively or communicatively)" to another component (eg, a second component); When referring to "connected to", it should be understood that any of the above-described components may be directly connected to the above-described other component or may be connected through another component (eg, a third component). On the other hand, when it is said that a component (eg, a first component) is "directly connected" or "directly connected" to another component (eg, a second component), a component different from a component It may be understood that no other component (eg, a third component) exists between the elements.

The expression "configured to (or configured to)" used in the present invention depends on the context, for example, "suitable for", "having the capacity to" It can be used interchangeably with "," "designed to", "adapted to", "made to", or "capable of". The term “configured (or configured to)” may not necessarily mean only “specifically designed to” in hardware. Instead, in some circumstances, the expression “a device configured to” may mean that the device is “capable of” with other devices or parts. For example, the phrase “a processor configured (or configured to perform) A, B, and C” refers to a dedicated processor (eg, an embedded processor) for performing the operations, or by executing one or more software programs stored in a memory device. , may mean a generic-purpose processor (eg, a CPU or an application processor (AP)) capable of performing corresponding operations.

Terms used in the present invention are only used to describe specific embodiments, and may not be intended to limit the scope of other embodiments. The singular expression may include the plural expression unless the context clearly dictates otherwise. Terms used herein, including technical or scientific terms, may have the same meaning as commonly understood by one of ordinary skill in the art described in the present invention. Among the terms used in the present invention, terms defined in general dictionary may be interpreted with the same or similar meaning as the meaning in the context of the related art, and unless explicitly defined in the present invention, ideal or excessively formal meanings is not interpreted as In some cases, even terms defined in the present invention cannot be construed to exclude embodiments of the present invention.

In this specification, the terms "type of waste collection device", "type of waste collection equipment" and "information related to the type of automated waste collection box" have the same meaning, and therefore will be used interchangeably. In addition, in this specification, "type and state of waste collection equipment", "type and state of waste collection equipment" and "information related to the type and state of automated waste collection box" are used interchangeably because they have the same meaning.

Hereinafter, a preferred embodiment according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram of a collection control system of an automated waste collection box according to an embodiment of the present invention.

Referring to FIG. 1 , a deep learning-based AI operation control system of an automated waste collection box will be described.

The deep learning-based AI operation control system of the automated waste collection box includes the automated waste collection box (110-1, 110-2, ... , 110-N), the terminal 120, and the deep learning-based AI (Artificial Intelligence) operation server ( 130), the edge computing gateway 140, and the like.

AI is a field of computer science and information technology that studies how computers can do the thinking, learning, and self-development that human intelligence can do. device that makes it possible. The AI device includes, for example, the edge computing gateway 140, but is not limited thereto. The AI device will be referred to as an edge computing gateway 140 hereinafter. The edge computing gateway 140 may learn a vast amount of information (eg, big data) by implementing a deep learning neural network having an appropriate graphic processing unit (GPU) and a central processing unit (CPU).

Additionally, the operation of the edge computing gateway 140 according to an embodiment of the present invention is a Hadoop Distributed File System (HDFS) that analyzes and processes large amounts of data quickly at a low cost and distributes and stores files in multiple servers. ) can be applied to

The automated waste collection boxes (110-1, 110-2, ..., 110-N) may start to operate with a non-contact sensor (Untact sensor, contactless sensor, or AI Camera) in consideration of user convenience and safety. In addition, users can be recognized by recognition methods such as credit cards, radio frequency (RF) cards, T-money cards, 3D facial recognition, QR (Quick Response) codes, or barcodes, and waste disposal costs when discharging waste It can be used as a means of payment of expenses such as credit points or compensation for money when discharging certain wastes such as recyclables.

The automated waste collection boxes (110-1, 110-2, ..., 110-N) can be paid or received according to the characteristics of waste and collection boxes. And the automated waste collection box (110-1, 110-2, ... , 110-N) after payment, the user puts the waste into the automated waste collection box (110-1, 110-2, ... , 110-N) Upon input, the waste is compressed or generally received. Automated waste collection boxes (110-1, 110-2, ... , 110-N) may be collection boxes that do not require payment according to laws and ordinances of the country or local government.

The automated waste bins 110-1, 110-2, ..., 110-N may be configured to transmit location information of the automated waste bin and information related to the type of automated waste bin to the edge computing gateway 140. . Here, the information related to the type of the automated waste collection box may include an ID of the automated waste collection box.

Additionally, the automated waste collection box (110-1, 110-2, ..., 110-N) generates detection information by at least one or more sensors installed in advance when receiving waste, and transmits the generated detection information to the edge computing gateway 140 It may be configured to transmit The detection information includes information on the loading amount of waste (eg weight or volume), compression pressure (compression information) of the cylinder, tilt information, overturn information, temperature information of waste, and toxic substances harmful to humans such as VOC (Volatile Organic Compounds). may include information on the amount of generation of

The automated waste collection bins 110-1, 110-2, ... , 110-N are, for example, the automated waste collection bins 1 (110-1), ..., and the automated waste collection bin 6 (110-6) to be described below. may include, but is not limited to. In more detail, each automated waste collection box (110-1, 110-2, ..., 110-N) will be described as follows.

Referring to FIG. 1 , an automated waste collection box 1 110 - 1 may include an Internet of Things (IoT)-based waste compression container (replaceable type). When the equipment is operated by opening the non-contact shutter and recognizing the operation start button and card, waste is put into the automated waste collection box 1 (110-1), the weight is measured, and the cost of discharging waste is determined according to the measured weight. , After paying the indicated amount, with the waste acceptance, the payment card is discharged, and the waste is discharged to the compression unit.

The automated waste collection box 1 (110-1) does not require separate production or use of a plastic bag for discharging waste, and can reduce collection personnel and vehicles by up to about 80%. In addition, the volume of the automated waste collection box 1 (110-1) can be reduced to about 1/5 by compressing the waste. Automated waste collection box 1 (110-1) is equipped with an odor sensor to remove odors by spraying ozone and odor-removing chemical agents, etc., or by using a blower fan equipped with a filter for odor removal when odors are generated from waste. Odors can be emitted to the outside.

The automated waste collection box 2 110 - 2 may be, for example, a collection box provided with a loading amount detection sensor, such as an ultrasonic sensor, a camera, and a UV (Ultraviolet) sensor, on the upper part of the waste collection box. By equipping the automatic waste collection box 2 (110-2) with a loading tank detection sensor, the change in the amount of waste generated in each collection box classified according to each waste is predicted in advance, and the inclination of the equipment which is the state value of the waste collection box, the occurrence of fire, humidity , rain and snow conditions can be detected and used as operating parameters for operation and control. In addition, the automated waste collection box 2 (110-2) predicts the need for collection through the deep learning-based AI algorithm of the deep learning-based AI operation server to realize the optimal movement of the collection vehicle, thereby reducing waste collection costs.

The automated waste collection box 3 (110-3) may include, for example, a smart waste vertical compression collection box to which a solar panel and a collection amount detection sensor are attached. Automated waste collection box 3 (110-3) is installed on beaches, parks, city streets, university campuses and residential areas, and collects and compresses waste thrown away by passers-by and tourists, and then transmits the sensing information through the Internet to the edge computing gateway (140). can be sent to In addition, the automated waste collection box 3 (110-3) is equipped with a solar panel and a battery that can store the generated electricity, so the automated waste collection box produces and stores its own energy and uses it when necessary, so it can be easily moved or added according to the amount of generation. As a result, it can be an efficient means of collecting waste.

When the user's recognition is used as a payment method such as a credit card, the automated waste collection box 4 (110-4) can measure the weight of food waste, for example, and then charge or pay using a credit card or RF card. And, the automated waste collection box 4 (110-4) can apply the hot air fermentation drying method by using the Fin Heater heat source to the accommodated food waste, and semi-permanently deodorize the generated odor using a platinum catalyst. The automated waste collection box 4 (110-4) can maximize energy efficiency by generating hot air, circulation, and recycling.

The automated waste collection box 5 (110-5) can accommodate up to about 5 to 7 times more waste than conventional trash cans, can be applied with a hands-free inlet equipped with a motion sensor, and can indicate the loading status with a status indicator. have. The automated waste collection box 5 (110-5) can be neat and hygienic because the user does not have to directly touch the waste collection box by applying a hands-free inlet equipped with a motion sensor. In addition, the automatic waste collection box 5 (110-5) can easily check the loading status of the waste with a status indicator light. For example, the automated waste collection box 5 ( 110 - 5 ) may be displayed in green when it operates normally and is ready for use.

The automated waste collection box 6 ( 110 - 6 ) may include an underground storage device for a waste collection box equipped with an automated waste collection box 1 ( 110 - 1 ) and an automated waste collection box 2 ( 110 - 2 ). The automated waste collection box 6 (110-6) may be a waste collection system that can be applied when there is a shortage of waste collection space on the ground. The automated waste collection box 6 (110-6) discharges the waste to an external chute in which the scale and the payment system are combined, thereby minimizing the generation of odor and dust and the decay of the waste.

Here, the automated waste collection box 2 (110-2) to the automated waste collection box 6 (110-6) are not a replacement type, but a collection type, unlike the automated waste collection box 1 (110-1).

Meanwhile, the terminal 120 is configured to include a collection vehicle terminal, a driver terminal, and a service engineer terminal, and hereinafter, the terminal 120 will be referred to as a collection vehicle terminal, a driver terminal, and a service engineer terminal 120 . The collection vehicle terminal, the driver terminal, and the service driver terminal 120 transmit the location information of the collection vehicle, the loading amount information and the loading information of the waste loaded in the collection vehicle to the edge computing gateway 140 or a deep learning-based AI operation server ( 130) can be transmitted. The collection vehicle terminal, the driver's terminal and the service engineer's terminal 120 according to an embodiment of the present invention are, for example, a smartphone, a tablet personal computer (PC), a mobile phone, a video phone, e-book reader (e-book reader), desktop PC (desktop personal computer), laptop PC (laptop personal computer), netbook computer (netbook computer), workstation (workstation), PDA (personal digital assistant), PMP (portable multimedia) player), and a wearable device.

The edge computing gateway 140 may receive automated waste collection bins, waste collection vehicles and other sensing information from each automated waste collection bin 110-1, 110-2, ..., 110-N, and perform deep learning processing. It can be configured to In addition, the edge computing gateway 140 may be configured to transmit unprocessed data (eg, detection information) to the deep learning-based AI operation server 130 so that the deep learning-based AI operation server can process it. By processing the primary data in the edge computing gateway 140, the load of the deep learning-based AI operation server 130 can be distributed.

The automated waste collection box (110-1, 110-2, ..., 110-N) is an edge computing gateway 140 and TCP (Transmission Control Protocol), LoRa (Long Range: LoRa) or Wi-Fi (Wi-Fi) etc. can be connected.

The edge computing gateway 140 is located near the automated waste collection boxes 110-1, 110-2, ..., 110-N, and may be inserted into an access point (AP). The AP is not simply used as a signal repeater, but can be used as an edge device or AI device (edge computing gateway 140) that processes core data in real time.

The deep learning-based AI operation server 130 is each automated waste collection box (110-1, 110-2, 110-1, 110-2, for the integrated management of each automated waste collection box (110-1, 110-2, ..., 110-N)) ... , 110-N) status can be checked and operated in real time.

In addition, the deep learning-based AI operation server 130 receives detection information from the edge computing gateway 140 and performs big data analysis based on the detection information. Thereafter, the deep learning-based AI operation server 130 may apply the big data processing result to a deep learning algorithm. Here, the deep learning-based AI operation server 130 may omit the big data analysis operation.

By applying the big data analysis result and additional detection data to a deep learning algorithm with a parameter set to be described later, the operation judgment result of artificial intelligence can be obtained.

The deep learning-based AI operation server 130 is based on the result of applying the big data processing result and the additional detection data to the deep learning algorithm, the automated waste collection box (110-1, 110-2, ... , 110-N) It may be configured to determine whether collection, collection schedule, or collection order, and to determine collection vehicles available for collection expected schedule. That is, whether the automated waste collection box is collected, a collection schedule, or a collection order may be determined based on a result of applying a big data processing result to a deep learning algorithm.

The deep learning-based AI operation server 130 collects the future collection schedule of the automated waste collection boxes (110-1, 110-2, ..., 110-N) determined to be collected, the vehicle terminal, the driver terminal, and the service driver terminal ( 120), and may be configured to select from collection vehicles running on the road for emergency collection and transmit a collection instruction message to the corresponding collection vehicle terminal, driver terminal, and service engineer terminal 120 . In addition, information on the repair service may also be transmitted to the collection vehicle terminal, the driver terminal, and the service engineer terminal 120 indicating the terminal of the repair service person by dividing the deep learning-based emergency on-site repair and the repair after collection.

The collection instruction message may include location information of the corresponding automated waste collection boxes (110-1, 110-2, ..., 110-N), and the location information is provided to the collection vehicle terminal, driver terminal, and service engineer terminal 120 . It may be directly applied to a navigation application and configured to provide an optimal route guidance service of the corresponding location information.

In summary, each automated waste collection box according to an embodiment of the present invention compresses or generally accommodates waste, generates detection information of each sensor, and first information related to the type and state of each automated waste collection box, the detection information , and second information indicating the location of each of the automated waste collection bins.

The AI device receives the first information, the sensed information, and the second information, and uses the sensed information in a Recurrent Neural Network (RNN) technique of Long Short Term Memory (LSTM) first deep learning (deep learning) is applied to the algorithm, and based on the result applied to the first deep learning algorithm, the first collection schedule and the first operating element of each automated waste collection bin are determined, and if there is unprocessed sensing information, the The first information, the unprocessed sensing information, and the second information are transmitted to a deep learning-based AI (Artificial Intelligence) operation server.

The deep learning-based AI operation server receives the first information, the unprocessed sensed information, and the second information, performs big data analysis based on the unprocessed sensed information, and displays the big data analysis result Among the Recurrent Neural Network (RNN) techniques, Long Short Term Memory (LSTM) is applied to the second deep learning algorithm, and the automation is based on the result applied to the second deep learning algorithm. A second collection schedule and a second operating element of the waste container are determined, and based on the first and second information, a corresponding one suitable for the characteristics of the waste and the container among the collection vehicles available for the schedule consistent with the determination result It determines the collection vehicle and manpower, and gives a task instruction to the collection vehicle and the terminal of the person in charge.

Here, the first operating element and the second operating element include at least one of service of the automated waste collection box, troubleshooting, full pre-removal, additional deployment, and evacuation of the deployment area, and a neural network for application of the deep learning algorithm Each parameter of is represented by a parameter set, the parameter set is composed of a combination of at least two or more parameters, and the parameter set is determined by the learning result of the long-term memory deep learning.

2 is a block diagram illustrating an edge computing gateway according to an embodiment of the present invention.

The edge computing gateway 130 according to an embodiment of the present invention may include a control module 141 , a communication module 142 , an edge computing module 143 , a sensor module 144 , and an equipment module 145 . can

The control module 141 may control each of the communication module 142 , the edge computing module 143 , the sensor module 144 , and the equipment module 145 to perform an AI operation.

The communication module 142 is an automated waste collection box (110-1, 110-2, ..., 110-N) from each automated waste collection box (110-1, 110-2, ..., 110-N) ), or to transmit the location information and detection information of the automated waste collection boxes (110-1, 110-2, ..., 110-N) to the deep learning-based AI operation server 130 when necessary It can be configured to

The edge computing module 143 receives sensing information from each automated waste collection box (110-1, 110-2, ..., 110-N) and applies a deep learning-based AI algorithm to determine system operation. You can create a parameter set and extract deep learning results based on the created parameter set. Data that has not been processed by the edge computing module 143 may be configured to be transmitted to the deep learning-based AI operation server 130 to be processed by the deep learning-based AI operation server 130 .

The constant sensor module 144 is equipped with one or more sensors, and is configured to collect detection information of each automated waste collection box 110-1, 110-2, ..., 110-N using the one or more sensors. can be

The equipment module 145 may be configured to store information related to the type and state of the automated waste collection box.

3 is a block diagram of a deep learning-based AI operation server according to an embodiment of the present invention.

The deep learning-based AI operation server 130 is a central operation control module 131, a communication module 132, an operator terminal / monitoring system module 133, a collection vehicle, a collection agent, a service sector information module 134, ㅡbig It may be configured to include a data analysis module 135 , a deep learning-based AI module 136 , and a database (hereinafter, referred to as “DB”) 137 , and the like.

The central operation control module 131 controls the collection of each automated waste collection box 110-1, 110-2, ..., 110-N for each component, that is, the communication module 132, the operator terminal/monitoring system. The module 133 , the collection vehicle, the collection agent, the service sector information module 134 , the big data analysis module 135 , the deep learning-based AI module 136 , and the DB 137 may be controlled, respectively. The central operation control module 131 uses the result applied to the deep learning-based AI algorithm of the big data analysis result to determine whether each automated waste collection box (110-1, 110-2, ..., 110-N) is collected or not. It may be configured to determine a schedule, or collection order and service availability.

The communication module 132 may be configured to receive location information and information related to the type and state of the automated waste collection box from the edge computing gateway 140 via the Internet. Also, the communication module 131 may be configured to receive sensing information from the edge computing gateway 140 through the Internet. In addition, the communication module 132 may be configured to receive the location information of the collection vehicle and information on the amount of waste loaded in the collection vehicle from the collection vehicle terminal, the driver terminal, and the service technician terminal 120 . Additionally, the communication module 131 may be configured to transmit a collection instruction and a service message to the collection vehicle terminal, the driver terminal, and the service engineer terminal 120 .

The operator terminal/monitoring system module 133 may be configured to display the operating status of the system and each device to the user, and to reflect the interference in the system operation.

The collection vehicle, collection agent, and service part information module 134 includes the current location information of the collection vehicle received through the communication module 131 from the collection vehicle terminal, the driver terminal, and the service engineer terminal 120, and the information on the current location of the collection vehicle loaded on the collection vehicle. It may be configured to store information on the amount of the waste and whether information about whether or not to load. In addition, the collection vehicle, collection agent, service part information module 134 is determined by the deep learning-based AI module 136, and determined, each automated waste collection box (110-1, 110-2, ..., 110-N) It may be configured to select a collection vehicle for the automated waste collection box as a collection target based on information stored in the collection vehicle, collection agent, and service part information module 134 .

Deep learning-based AI module 136 provides location information of each automated waste collection box (110-1, 110-2, ..., 110-N), automated waste collection box (110-1, 110-2, ... , 110-N), it is possible to perform deep learning analysis of the regression neural network method using information and detection information related to the type and state, and store and transmit the analysis result. The deep learning-based AI module 136 is more specifically the location information of each automated waste collection box (110-1, 110-2, ..., 110-N), the automated waste collection box (110-1, 110-2, . .., 110-N) type and state related information and detection information are applied to the LSTM (Long Short Term Memory) deep learning algorithm among RNN (Recurrent Neural Network) technology to create an automated waste collection box. It may be configured to determine a collection schedule and an operational element including at least one of service, breakdown repair, full evacuation, additional deployment, and evacuation of the deployment area. The deep learning analysis may be performed through deep learning-based AI algorithm execution.

RNN has a problem in that when the distance between related information and the point where the information is used is long, the gradient gradually decreases during backpropagation and the learning ability is greatly reduced. To solve this problem, LSTM is used. LSTM is a structure in which cell-state is added to the hidden state of RNN. The LSTM in an embodiment of the present invention may be configured to generate a parameter set based on input sensing information (input value or input data). For reference, the LSTM may include an input gate for an input value, an output gate for an output value, a forget gate for a forget value, and the like, and It is possible to store an input value input through the method, adjust a forgetting value of the forgetting gate, and output an output value to the output gate.

The big data analysis module 135 may analyze the analysis result of the deep learning-based AI module 136 using a big data analysis technique and output the result or store the result in the DB 137 .

The DB 137 may be configured to store the processing result of the deep learning-based AI module 136 or to store the analysis result of the big data analysis module 135 .

The movement of the collection vehicle is optimized by the artificial intelligence system according to an embodiment of the present invention, and collection efficiency can be increased.

4 is a flowchart illustrating the operation of a deep learning control system of an automated waste collection box and an edge gateway according to an embodiment of the present invention.

The automated waste collection box is started by a user's direct hand signal or operation in step 401 and may be operated by recognizing a card such as a credit card. Any method by which an automated waste bin can be operated is available, as well as this recognition method.

In step 402 , the automated waste bin receives each designated waste according to the characteristics of each type of the automated waste bin. After receiving the waste, if necessary, the automated waste collection box measures the weight of the waste, recognizes the input with an AI camera, etc., and charges using a payment method such as a credit card or RF card, or accumulates monetary benefits and points .

The automated waste collection box may generate detection information related to the received waste in step 403 , and transmit the generated detection information to the edge computer gateway 140 . In this case, the automated waste collection box may transmit information related to the type and state of the automated waste collection box and the location information of the automated waste collection box to the edge computer gateway 140 in advance before transmitting the detection information to the edge computer gateway 140 .

5 is a flowchart illustrating a detailed operation of an edge computing gateway according to an embodiment of the present invention.

The edge computing gateway 140 receives data in step 501 .

The edge computing gateway 140 generates a parameter set for system determination based on the data input in step 502 .

The edge computing gateway 140 extracts the deep learning result based on the parameter set generated in step 503 .

The operations of steps 501 to 503 may be executed in the edge computer module 143 of FIG. 2 .

The result of applying the input data to the deep learning-based artificial intelligence AI algorithm is shown as a parameter set for determining the collection schedule, etc., the parameter set is composed of a combination of at least two or more parameters, and the parameter set is for learning is determined by

Here, the parameters are the temperature, humidity, weight or volume of the waste collection box, the frequency of malfunctions, fire, the type of collection equipment, the type of waste, the amount of collection and batch truck work, the increase rate of the existing waste generation in the area where the device is installed, season, regional Characteristics, odor detection sensor value, VOC (volatile organic compounds) detection sensor value, operation period after previous collection or operation period after the first waste input, replacement schedule for worn parts, frequency of complaints, accidents during operation, civil complaints, administrative tasks It may include one or more or all of operator intervention, traffic statistics, day of the week, weather, and weather forecast. The edge computing gateway 140 gives each parameter a weight through analysis of data accumulated up to now after the normal operation of the system, or LSTM among RNN (Recurrent Neural Network - Recurrent Neural Network) technology that prevents recurrent of a specific parameter. By applying the deep learning technology of (Long Short Term Memory), it is possible to determine the priority of collection of waste and equipment.

Meanwhile, the edge computing gateway 140 determines whether there is data to be additionally processed or unprocessed data in step 504 . Since the edge computing gateway 140 processes low-capacity data, it cannot process large-capacity data. Therefore, data that cannot be processed by the edge computing gateway 140 can be processed by the deep learning-based AI operation server 130 .

If there is data to be additionally processed or unprocessed data, the edge computing gateway 140 transmits additional sensed data or unprocessed sensed data to the deep learning-based AI server in step 505 .

However, if there is no data to be additionally processed or unprocessed data, the edge computing gateway 140 checks the deep learning result in step 506 .

6 is a flowchart illustrating a detailed operation of a deep learning-based AI operation server according to an embodiment of the present invention.

The deep learning-based AI operation server may receive an automated waste collection box, a waste collection vehicle, and other sensing information from the edge computing gateway 140 in step 601 . Although not shown in FIG. 6 , the deep learning-based AI operation server may receive information related to the type and state of the automated waste container and location information of the automated waste container before receiving the detection information.

The deep learning-based AI operation server inputs detection information, etc. to the big data analysis module 135 in step 602 . The deep learning-based AI operation server 130 performs big data analysis in step 603 .

The deep learning-based AI operation server 130 inputs the big data analysis result to the deep learning-based AI module 136 in step 604 .

The deep learning-based AI operation server 130 analyzes the big data analysis result as deep learning-based AI in step 605 . The deep learning-based AI operation server 130 may output the result of applying the big data analysis result to the artificial intelligence deep learning neural network or display it through a graphic interface not shown in the drawing. The deep learning-based AI operation server 130 uses the determination result of applying the big data analysis result to the artificial intelligence deep learning neural network in step 606 to each automated waste collection box (110-1, 110-2, ... , 110-N). ), that is, whether to collect, determine the collection schedule, or determine the order of collection, make repair decisions, dismantle, install additional collection devices, and notify the module and operating personnel.

The deep learning-based AI operation server 130 may determine available vehicles and personnel for the collection schedule determined in step 607 . The collection vehicle may be determined based on an optimal driving route of the collection vehicle, a collection schedule of the collection vehicle, information on a loading amount of the collection vehicle, information on whether the collection vehicle is loaded, and the like.

Additionally, the deep learning-based AI operation server can determine the appropriate collection vehicle for the type of automated waste collection box. If the type of the automated waste collection box is a replaceable type (eg, automated waste collection box 1), the collection vehicle may determine that the collection vehicle has a capacity higher than a threshold value. However, when the type of the automated waste collection box is a collection type (eg, automated waste collection boxes 2 to 6), the collection vehicle may be determined to be a collection vehicle having a capacity of the collection vehicle equal to or less than a threshold value. If a collection vehicle with an appropriate capacity is determined according to the type of automated waste collection box, there is an effect of cost reduction.

The deep learning-based AI operation server may be configured to send a collection instruction message to the corresponding collection vehicle driver terminal of the designated collection vehicle for the collection of the automated waste collection box determined to be collected in step 608 .

7 is a flowchart illustrating a detailed operation of a deep learning-based AI module according to an embodiment of the present invention.

The deep learning-based AI module starts the analysis and judgment process by inputting data into the deep learning-based neural network module in step 701.

The deep learning-based AI module creates a parameter set of the neural network for various decisions of system operation in step 702. The parameter set is determined through deep learning learning, and may be created by combining the following two or more parameters for system operation decision. The parameter set can be additionally updated or added depending on the situation around the automated waste collection box. For reference, the learning rate, the number of repetitions, etc. may be preset before determining the parameter set.

The parameters for selecting the collection order are as follows.

A. Temperature of automated waste bins

B. Moisture in automated waste bins

C. Weight or Volume of Waste Generated

D. Failure, Periodic Inspection and Replacement - Failure of waste collection device parts and periodic inspection and replacement of consumables

E. Increasing trend of waste generation in the installation area

F. Fire in equipment

G. Types of Collection Equipment - Automated Waste Bin 1 (eg iTainer), Automated Waste Bin 3 (eg iSolar), Automated Waste Bin 4 (eg iDry B), etc. The classification of waste is also different.

H. Type of waste

I. Work load of collection vehicles and personnel - Collection rate of 80% or more If there are not enough collection trucks or workers at the time of expected loading, collection is carried out in advance

J. Historic Waste Increase Trend at the location - Rate of increase compared to existing waste generation

K. Season; For priority collection targets such as beaches in summer and famous mountains in spring and autumn, additional points are added when determining the priority of collection and increase in the amount of equipment placed in the parameters.

L. Regional Features Categories

i. rural area

ii.City Area

iii.Beach area

iv.mountainous areas

v.populated areas

vi. Population Rare Areas

vii. Estimated number of users or households

viii. industrial area

ix.Commercial Area - Priority dispatch

M. Odor detection sensor value

N.VOC (volatile organic compounds) detection sensor value

O. Storage period of the first waste - The maximum storage period is 10 days in summer and 30 days in winter.

P. Consumable parts replacement schedule - hydraulic oil replacement, grease addition, bearing replacement, etc.

Q.Occurrence of complaints: Actual occurrence and frequency of existing local complaints

R. Traffic volume statistics - In the case of Friday, at least dispatch due to congestion, etc.

S. Weekdays: On Saturdays and Sundays, the priority of dispatching is lowered as much as possible due to increased operating costs.

T. Weather Factor

i.Prediction of rainfall and typhoon - Frequent collection conditions

ii. Prediction of snowfall and extreme cold - condition of collection as late as possible

U. Operator intervention:

i. Intervention to collect the maximum amount of waste prior to regular maintenance of the waste treatment facility

ii. In case of intervention in collecting the minimum amount of waste during regular maintenance of waste treatment facilities, intervention in the pre-collection of waste at the event site, etc.

V. Population ratio by gender and age

W. Others can be added and deleted

An example of the judgment of the deep learning-based neural network of the deep learning-based AI module is that relatively young people live in densely populated areas and there is a large amount of waste. On the other hand, in areas with a sparse population, a relatively large number of elderly people live and the amount of waste generated is relatively small. In the case of rare population areas, it may take a long time for waste to accumulate in the automatic waste collection box, so even if the deep learning-based AI server is only 30% full, after a predetermined period, it can be collected unconditionally regardless of the full load rate. At this time, the parameter set may be a population rarity area, period, weight, or volume.

On the other hand, the deep learning-based AI module 136 may set the collection standard of the automatic waste collection box shorter in summer and longer in winter in consideration of the predicted load amount of waste according to the result of big data analysis, for example. At this time, the parameter set can be a factor of season and weather information.

Accordingly, an embodiment of the present invention can prevent the waste from overflowing in the automated waste collection boxes 110-1, 110-2, ..., 110-N, and optimal selection and collection is possible.

The deep learning-based AI module extracts the operation judgment result of deep learning-based AI based on the parameter set created in step 703 and proceeds accordingly.

8 is a flowchart illustrating a detailed operation of determining the operation result of the deep learning-based AI module and determining the selection and application rate of the artificial intelligence neural network operation parameter for correcting it.

The deep learning-based AI module can check the processing result of the deep learning-based artificial intelligence neural network in step 801. Here, the big data analysis result is represented as a parameter set, the parameter set is composed of a combination of at least two or more parameters, and the parameter set may be changed and determined by learning. The analysis result of the input data applied to the artificial intelligence algorithm of the deep learning-based neural network can be used as input data for determining the operation decision such as the collection schedule, and the processing result of the deep learning-based artificial intelligence is, for example, It can be expressed in terms of the load (weight or volume) and the ratio of the increase in the existing collection. Hereinafter, the loading amount (weight or volume) and the existing collection amount increase ratio are referred to as a first parameter set.

The deep learning-based AI module may check the first parameter set at a specific time t1 in step 802 .

The deep learning-based AI module may determine the offset value as the difference between the first parameter set and the threshold value at a specific time t1 in step 803 .

The deep learning-based AI module may apply the offset value determined in step 507 to the predicted value of the first parameter set at t2 in step 804.

The deep learning-based AI module may determine the second parameter set to which the offset value at t2 is applied in step 805 .

In an embodiment of the present invention, as shown in FIG. 8 , by configuring a parameter set by applying an offset value, it is possible to more accurately determine whether an automated waste collection box is collected, a collection schedule, or a collection order.

An embodiment of the present invention may be implemented as a module or device capable of performing each of the above-described methods or each function. In this case, the module may be implemented as a configuration of software and/or hardware. In addition, according to an embodiment of the present invention, the above-described method or function may be implemented as processor-readable code on a medium in which a program is recorded. Those of ordinary skill in the art to which the present invention pertains to the above-described contents can be modified and modified without departing from the essential characteristics of the present invention. Accordingly, the exemplary embodiments of the present invention are for explanation rather than limiting the technical idea, and the scope of the technical idea of the present invention is not limited by these embodiments. The protection scope of the present invention should be construed by the following claims, and all technical ideas within the equivalent range should be construed as being included in the scope of the present invention.

Claims (10)

In the deep learning-based artificial intelligence operating system of the automated waste collection box,
Compression or general acceptance of waste, generating detection information of each sensor, and AI with first information related to the type and state of each automated waste bin, the detection information, and second information indicating the location of each automated waste bin (Artificial Intelligence) each of the automated waste collection bins for transmission to the device;
Receiving the first information, the sensing information, and the second information, selecting at least a plurality of parameters from among a plurality of parameters included in the sensing information to generate a first parameter set, and a recursive neural network (RNN: A first collection schedule and a first operating element for each automated waste collection box by inputting the first parameter set to a long short-term memory (LSTM) first deep learning algorithm among Recurrent Neural Network) techniques output, and when the detection information includes a large amount of data, deep unprocessed detection information, the first information, and the second information including at least one parameter that has not been processed by the AI device among the plurality of parameters The AI device for transmitting to a learning-based AI operation server;
A second parameter set is received by receiving the first information, the unprocessed detection information, and the second information, and selecting a plurality of parameters from among a plurality of parameters included in the unprocessed detection information and a plurality of preset parameters. generates, and inputs the second parameter set to a long-term memory second deep learning algorithm to output a second collection schedule and a second operation element for each automated waste collection box, and the first information and the second information Based on the second collection schedule and the collection vehicles available in the schedule corresponding to the second operation factor, the collection vehicle and manpower suitable for the characteristics of the waste and collection box are determined, and the operation is given to the collection vehicle and the terminal of the person in charge. Including a deep learning-based AI operation server,
The first operating element and the second operating element are based on deep learning of the automated waste collection box, characterized in that it includes at least one of service of the automated waste collection box, troubleshooting, full removal, additional placement, and evacuation of the deployment area. artificial intelligence operating system.
According to claim 1,
A number of parameters included in the detection information are the temperature, humidity, total amount of waste collection weight or volume, malfunction frequency, fire occurrence, odor sensor value, VOC (volatile organic compounds) detection sensor value, and the value of the first generated waste. at least a plurality of storage periods and types of waste;
The predetermined number of parameters include the type of collection equipment, the daily workload of the collection truck, the increase rate of the collection amount to date at the collection location, the season, regional characteristics, the replacement schedule for worn parts, the frequency of complaints and existing complaints, traffic volume statistics, day of the week A deep learning-based artificial intelligence operating system of an automated waste collection bin, characterized in that it includes at least a plurality of , weather, and weather forecasts.
3. The method of claim 2,
The collection vehicle is an automated waste collection box, characterized in that it is determined based on an optimal driving route of the collection vehicle, a collection schedule of the collection vehicle, information on the amount of loading of the collection vehicle, and information on whether or not the collection vehicle is loaded. artificial intelligence operating system.
delete delete In the deep learning-based artificial intelligence operation method of the automated waste collection box,
In the AI (Artificial Intelligence) device, sensing information of each sensor generated in each automated waste bin, first information related to the type and state of each automated waste bin, and second information indicating the location of each automated waste bin are received process;
generating, in the AI device, a first parameter set by selecting at least a plurality of parameters from among a plurality of parameters included in the detection information;
In the AI device, each of the automated wastes is collected by inputting the first parameter set to the first deep learning algorithm of Long Short Term Memory (LSTM) among Recurrent Neural Network (RNN) techniques. outputting a first collection schedule and a first operating element for ;
In the AI device, when the detection information includes a large amount of data, unprocessed detection information including at least one parameter not processed by the AI device among the plurality of parameters, the first information, and the second information The process of sending to a deep learning-based AI operation server;
Receiving the first information, the unprocessed sensing information, and the second information in the deep learning-based AI operation server;
generating, in the deep learning-based AI operation server, a second parameter set by selecting a plurality of parameters from among a plurality of parameters included in the unprocessed detection information and a plurality of preset parameters;
outputting a second collection schedule and a second operating element for each automated waste collection box by inputting the second parameter set to a second deep learning algorithm of long and short-term memory in the deep learning-based AI operation server; and
In the deep learning-based AI operation server, based on the first information and the second information, among the collection vehicles available for the second collection schedule and the schedule corresponding to the second operation element, the waste and the corresponding characteristic of the collection box Determining the collection vehicle and manpower, including the process of instructing the collection vehicle and the terminal of the person in charge,
The first operating element and the second operating element are based on deep learning of the automated waste collection box, characterized in that it includes at least one of service of the automated waste collection box, troubleshooting, full removal, additional placement, and evacuation of the deployment area. How artificial intelligence operates.
7. The method of claim 6,
A number of parameters included in the detection information are the temperature, humidity, total amount of waste collection weight or volume, malfunction frequency, fire occurrence, odor sensor value, VOC (volatile organic compounds) detection sensor value, and the value of the first generated waste. at least a plurality of storage periods and types of waste;
The predetermined number of parameters include the type of collection equipment, the daily workload of the collection truck, the increase rate of the collection amount to date at the collection location, the season, regional characteristics, the replacement schedule for worn parts, the frequency of complaints and existing complaints, traffic volume statistics, day of the week A deep learning-based AI operation method of an automated waste collection bin, comprising at least a plurality of , weather, and weather forecast.
8. The method of claim 7,
Deep learning of the automated waste collection box, characterized in that the collection vehicle is determined based on the optimal driving route of the collection vehicle, the collection schedule of the collection vehicle, information on the loading amount of the collection vehicle, and information on whether the collection vehicle is loaded How to operate based artificial intelligence.
delete delete

Images