Garbage classification method based on multi-sensor fusion

Abstract

The invention relates to the technical field of garbage classification, in particular to a garbage classification method based on multi-sensor fusion, which comprises the following steps: collecting physical characteristic information of garbage through multiple sensors, carrying out preprocessing and characteristic extraction on signals by utilizing a data processing module, and judging garbage types by a classification decision module according to rules. And precise throwing is completed by the execution mechanism. According to the method, data of optical, weight and shape sensors are combined, environmental interference is eliminated through filtering, amplification and feature value extraction, and the classification and delivery precision is improved by adopting priority judgment and position correction. Meanwhile, the fault detection module monitors the equipment state in real time, and stable operation of the system is ensured. The accuracy and efficiency of garbage classification can be effectively improved, and the mistaken throwing phenomenon is reduced.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent environmental protection, and specifically relates to a garbage classification method based on multi-sensor fusion. BACKGROUND

[0002] Garbage classification usually relies on manual sorting or single sensor detection, such as optical sensors, weight sensors, etc., to realize the identification and classification of different types of garbage. At the same time, in order to improve the classification efficiency, some methods introduce machine learning algorithm to analyze the characteristics of garbage, so as to optimize the classification result.

[0003] In the prior art, single sensor combined with simple threshold judgment is used to distinguish garbage types, and then mechanical device is used to put garbage into corresponding recycling container. However, in actual application, single sensor is easily disturbed by environmental factors, such as light change, garbage surface contamination or irregular shape, etc., which may cause the recognition accuracy to decrease.

[0004] In the above method, in order to improve the classification accuracy, additional pretreatment steps are usually added, such as cleaning the surface of garbage or adjusting the position of sensor. However, these operations may prolong the overall classification time, and due to the limited information obtained by single sensor, some similar material garbage may still be misjudged, affecting the classification effect.

[0005] In addition, when multiple sensors are introduced, if the data between sensors is not effectively fused, it may cause information conflict or redundancy, further reducing the reliability of the system. In this case, the classification system may need to be repeatedly calibrated or the algorithm needs to be redesigned to ensure the consistency and accuracy of the data of each sensor. SUMMARY

[0006] The application relates to the technical field of garbage classification, and specifically relates to a garbage classification method based on multi-sensor fusion. It is mentioned in the background that the prior art has the problems of single sensor being easily disturbed by environment, low classification accuracy and multi-sensor data not being effectively fused in the garbage classification process.

[0007] To solve the above problems, the application provides a garbage classification method based on multi-sensor fusion, which comprises the following steps:

[0008] Step 1: Collecting physical characteristic information of garbage through multi-sensor, including optical characteristics, weight characteristics and shape characteristics;

[0009] Step 2: Inputting the collected multiple sets of characteristic information into a data processing module, using a signal modulation unit to pretreat the data of each sensor, eliminating noise interference and extracting key characteristic values;

[0010] Step three: transmit the processed feature value to the classification decision module, and the classification decision module preliminarily judges the garbage type according to the preset rule;

[0011] Step four: send the preliminary judgment result to the execution mechanism, and the execution mechanism completes the garbage throwing action and records the throwing position and garbage type information.

[0012] Further, in step two, the data processing module includes a signal modulation unit and a feature extraction unit. The signal modulation unit is provided with two groups corresponding to the output signals of the optical sensor and the weight sensor, and the signal modulation unit is internally provided with a filter circuit and an amplifier circuit. The filter circuit is used to remove high-frequency noise, and the amplifier circuit is used to enhance the signal strength; the feature extraction unit is located behind the signal modulation unit, and the feature extraction unit extracts the key feature values of the garbage such as reflectivity peak value, weight distribution curve and edge profile parameter through algorithm analysis.

[0013] Further, the feature extraction unit includes a master control chip and a storage module. The master control chip adopts a multi-core processor, the input end of the master control chip is connected with the signal modulation unit, and the output end of the master control chip is connected with the storage module; the storage module is used to save the extracted feature values, and the feature values are transmitted to the classification decision module through an access interface. The storage module is internally provided with a cache area with a capacity of 16MB, which is used to temporarily store real-time data streams.

[0014] Further, it further includes a data calibration structure, which is used to adjust the filter parameters and amplification multiples of the signal modulation unit to adapt to the output characteristics of different sensors. The data calibration structure includes a knob adjuster and a digital display screen. The knob adjuster is connected with the filter circuit and the amplifier circuit through a mechanical transmission mechanism, and the digital display screen is used to display the current adjustment parameters.

[0015] Further, the classification decision module includes a logic operation unit and a rule base. The logic operation unit receives the feature values from the feature extraction unit, and the logic operation unit is internally provided with a plurality of parallel computing channels, each channel corresponding to a garbage type; the rule base stores the preset classification rules, and the rule base supports the rule input of the new garbage type through a dynamic updating mechanism. The logic operation unit and the rule base are connected through a high-speed data bus, and the data bus bandwidth is 100Mbps.

[0016] Further, the classification decision module further includes a priority judgment unit. The priority judgment unit is located behind the logic operation unit, and the priority judgment unit sorts the classification results according to the confidence score of the garbage feature value. The garbage with a confidence score lower than a preset threshold is marked as a re-inspection object.

[0017] Further, the execution mechanism includes a conveyor belt and a sorting arm. The conveyor belt is arranged below the classification decision module, and the surface of the conveyor belt is provided with anti-skid texture with a depth of 0.5-1 mm for preventing displacement of the garbage during conveying; the sorting arm is installed at the end of the conveyor belt and is driven by a servo motor with a rotating speed range of 500-1500 rpm, and the end of the sorting arm is provided with a clamp with an adjustable opening width in a range of 20-100 mm.

[0018] Further, a position correction unit is arranged between the sorting arm and the classification decision module, which is used to detect the actual position of the garbage on the conveyor belt and feed the position information to the sorting arm control system to compensate for the position deviation during the operation of the conveyor belt. The position correction unit includes a photoelectric sensor for capturing the edge profile of the garbage and an encoder for recording the running distance of the conveyor belt.

[0019] Further, a fault detection module is further included and arranged above the execution mechanism for real-time monitoring of the working state of the sorting arm. The fault detection module includes a temperature sensor for detecting the temperature rise of the servo motor and a vibration sensor for detecting abnormal vibration during the operation of the sorting arm. The fault detection module transmits the monitoring data to the central control unit through a wireless communication module, and the wireless communication module adopts Zigbee protocol with a communication distance of 30 meters.

[0020] Further, the central control unit includes a main control board and an alarm device. The main control board receives the monitoring data from the fault detection module and judges the equipment operating state according to the preset threshold value; the alarm device is arranged on one side of the main control board and includes an LED indicator and a buzzer, the LED indicator is used to display the equipment operating state, and the buzzer is used to emit an alarm sound.

[0021] The present application has the following advantages: the garbage classification method based on multi-sensor fusion eliminates the problem that a single sensor is easily disturbed by the environment by synchronously collecting and preprocessing the multi-path data of the optical sensor, the weight sensor and the shape sensor; the accuracy of garbage feature analysis and the classification efficiency are improved through the cooperative work of the feature extraction unit and the classification decision module; the precise control of the garbage dropping position is realized through the cooperation of the execution mechanism and the position correction unit, and the occurrence of misdropping is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 The step flow chart of the present application is shown in the figure;

[0023] Figure 2 The internal structure diagram of the data processing module is shown in the figure;

[0024] Figure 3A structure schematic diagram of the classification decision module;

[0025] Figure 4 A structure schematic diagram of the actuator;

[0026] Figure 5 A working principle schematic diagram of the position correction unit;

[0027] Figure 6 A structure schematic diagram of the fault detection module;

[0028] Figure 7 A structure schematic diagram of the central control unit.

[0029] The reference signs are as follows: 1, data processing module; 2, signal modulation unit; 3, feature extraction unit; 4, classification decision module; 5, logical operation unit; 6, rule base; 7, priority judgment unit; 8, actuator; 9, conveying belt; 10, sorting arm; 11, clamp; 12, position correction unit; 13, photoelectric sensor; 14, encoder; 15, fault detection module; 16, temperature sensor; 17, vibration sensor; 18, wireless communication module; 19, central control unit; 20, main control board; 21, alarm device. DETAILED DESCRIPTION

[0030] The application provides a garbage classification method based on multi-sensor fusion, and the specific implementation is described in detail in combination with the drawings. Figure 1 A system overall structure block diagram is shown, which shows the main modules of the multi-sensor fusion garbage classification method and the connection relationship. The system comprises a data processing module 1, a classification decision module 4, an actuator 8, a position correction unit 12, a fault detection module 15 and a central control unit 19. These modules are connected with each other through signal transmission lines, and form a complete garbage classification system.

[0031] The data processing module 1 is one of the core parts of the whole system, and the internal structure is as shown in Figure 2As shown, mainly includes signal modulation unit 2 and feature extraction unit 3. Signal modulation unit 2 is provided with two groups, corresponding to the output signal of optical sensor and weight sensor respectively. Each group of signal modulation unit 2 is internally provided with filter circuit and amplifier circuit, filter circuit is used for removing high frequency noise, amplifier circuit is used for enhancing signal strength. Signal modulation unit 2 is connected with external sensor through wire, the collected original signal is transmitted to feature extraction unit 3 after pretreatment. Feature extraction unit 3 is located behind signal modulation unit 2, containing main control chip and storage module. The main control chip adopts multi-core processor, the input end is connected with signal modulation unit 2 through data bus, and the output end is connected with storage module through access interface. The storage module is internally provided with a cache area with a capacity of 16MB, used for temporarily storing real-time data stream, and the extracted feature value is transmitted to classification decision module 4 through high-speed data bus.

[0032] The structure of classification decision module 4 is as Figure 3 shown, including logic operation unit 5, rule base 6 and priority judgment unit 7. Logic operation unit 5 receives feature value from feature extraction unit 3, which is internally provided with a plurality of parallel computing channels, each channel corresponding to a kind of garbage type. Rule base 6 supports the rule entry of new garbage type through dynamic updating mechanism, and the rule base is stored in nonvolatile memory and connected with logic operation unit 5 through high-speed data bus, and the data bus bandwidth is 100Mbps. Priority judgment unit 7 is located behind logic operation unit 5, and the classification results are sorted through confidence score. Priority judgment unit 7 judges the reliability of the classification results according to the confidence score of the garbage feature value, if the confidence score is lower than the preset threshold, the garbage is marked as the object to be rechecked, and is fed back to the central control unit 19 through the signal line.

[0033] The structure of execution mechanism 8 is as Figure 4 shown, including conveying belt 9 and sorting arm 10. Conveying belt 9 is arranged below classification decision module 4, and the surface is provided with anti-skid texture, the texture depth is 0.5-1mm, which is used for preventing displacement of garbage during conveying. Conveying belt 9 is driven by motor, and the running speed can be adjusted according to the garbage disposal demand. Sorting arm 10 is installed at the end of conveying belt 9, driven by servo motor, and the rotating speed range of servo motor is 500-1500rpm. The end of sorting arm 10 is provided with clamp 11, the opening width of clamp 11 is adjustable, the adjustment range is 20-100mm, to adapt to garbage of different sizes. Sorting arm 10 is connected with servo motor through mechanical transmission mechanism, and is connected with classification decision module 4 through control signal line, receives the classification results and performs the corresponding delivery action.

[0034] The working principle of position correction unit 12 is as Figure 5As shown, including photoelectric sensor 13 and encoder 14. Photoelectric sensor 13 is installed above the conveyor belt 9, used to capture the garbage edge profile and generate position signal. Encoder 14 is installed on the drive shaft of the conveyor belt 9, used to record the running distance of the conveyor belt 9. Photoelectric sensor 13 and encoder 14 are connected to the control system of the sorting arm 10 through the signal line, the detected position information is fed back to the sorting arm 10 to compensate the position deviation during the running of the conveyor belt 9. The position correction unit 12 adjusts the action parameters of the sorting arm 10 in real time through high-precision algorithm, ensures the accuracy of the garbage throwing position.

[0035] The structure of the fault detection module 15 is shown in Figure 6 As shown, including temperature sensor 16 and vibration sensor 17. Temperature sensor 16 is installed on the servo motor shell, used to detect the temperature rise of the servo motor. Vibration sensor 17 is installed on the base of the sorting arm 10, used to detect the abnormal vibration during the running of the sorting arm 10. Temperature sensor 16 and vibration sensor 17 are connected to the wireless communication module 18 through the signal line, the wireless communication module 18 uses Zigbee protocol, the communication distance is 30 meters, the monitoring data is transmitted to the central control unit 19.

[0036] The structure of the central control unit 19 is shown in Figure 7 As shown, including main control board 20 and alarm device 21. The main control board 20 receives the monitoring data from the fault detection module 15, and judges the equipment running state according to the preset threshold value. Alarm device 21 is located on one side of the main control board 20, including LED indicator and buzzer. LED indicator is used to display the equipment running state, buzzer is used to send alarm sound. The main control board 20 is connected with the actuator 8 and other modules through the control signal line, realizes the centralized management and real-time monitoring of the whole system.

[0037] In practical application, when the garbage enters the system, the physical characteristic information is first collected by optical sensor, weight sensor and shape sensor. These information are transmitted to the data processing module 1, the original signal is preprocessed by the signal modulation unit 2, the noise interference is eliminated and the signal strength is enhanced. Then, the feature extraction unit 3 extracts the key characteristic value through algorithm analysis, such as reflectivity peak value, weight distribution curve and edge profile parameter, and stores the extracted characteristic value in the cache area. After receiving the characteristic value, the classification decision module 4 preliminarily determines the garbage type according to the classification rules in the rule base 6. The priority judgment unit 7 scores the confidence degree of the classification result, if the score is lower than the preset threshold value, it is marked as the object to be rechecked.

[0038] The preliminary determination result is transmitted to the execution mechanism 8, and the conveyor belt 9 transports the garbage to the working area of the sorting arm 10. The position correction unit 12 detects the actual position of the garbage through the photoelectric sensor 13 and the encoder 14, and feeds back the position information to the control system of the sorting arm 10. The sorting arm 10 adjusts the opening width of the clamp 11 according to the classification result and the position information, completes the garbage throwing action, and records the throwing position and garbage type information. During the whole process, the fault detection module 15 monitors the working state of the sorting arm 10 in real time, and if an abnormal condition is detected, transmits the data to the central control unit 19 through the wireless communication module 18. The central control unit 19 judges the equipment running state according to the monitoring data, and sends an alarm through the alarm device 21.

[0039] The above embodiments describe the specific operation process of the present application and the cooperation relationship between the modules in detail, which ensures that the system can efficiently and accurately complete the garbage classification task.

[0040] In order to better enable those skilled in the art to fully understand and implement the present application, the specific implementation principles of the present application are further supplemented in the following in combination with a specific application scenario.

[0041] When the garbage is put into the system, the physical characteristic information of the garbage is first collected by the optical sensor, the weight sensor and the shape sensor. Taking a plastic bottle as an example, the optical sensor detects the surface reflectivity peak value, the weight sensor measures the weight distribution curve, and the shape sensor captures the edge contour parameters. These original signals are transmitted to the data processing module 1 through wires, and the signal modulation unit 2 pre-processes the signals. The filter circuit removes the high-frequency noise introduced by the change of ambient light or the vibration of equipment running, and the amplification circuit enhances the signal strength, ensuring the accuracy of subsequent feature extraction. Subsequently, the main control chip in the feature extraction unit 3 extracts key feature values through algorithm analysis, such as the high reflectivity peak value, the light weight distribution curve and the regular cylindrical contour parameters of the plastic bottle. These feature values are temporarily stored in the cache area of the storage module, and are transmitted to the classification decision module 4 through the high-speed data bus.

[0042] After the feature values are received by the classification decision module 4, the logical operation unit 5 makes a preliminary determination of the type of garbage according to the classification rules in the rule base 6. Assuming that the classification rules for plastic bottles have been entered into the rule base 6, including their typical reflectivity range, weight interval, and shape characteristics. The logical operation unit 5 compares the current garbage feature values with the standard values in the rule base through multiple parallel computing channels. If the feature values match well, a preliminary classification result is generated; if the matching degree is low, the priority judgment unit 7 sorts the classification results according to the confidence score. Assuming that a feature value such as the weight distribution curve has a slight deviation, resulting in a confidence score of 70%, which is lower than the preset threshold of 80%, the garbage will be marked as a re-inspection object and fed back to the central control unit 19 through the signal line.

[0043] The preliminary classification result is transmitted to the actuator 8, and the conveyor belt 9 transports the garbage to the working area of the sorting arm 10. In this process, the position correction unit 12 detects the actual position of the garbage in real time through the photoelectric sensor 13 and the encoder 14. The photoelectric sensor 13 captures the edge profile of the garbage and generates a position signal, and the encoder 14 records the running distance of the conveyor belt 9. The data of the two is fed back to the control system of the sorting arm 10 through the signal line to compensate for the position deviation during the running of the conveyor belt. For example, when the garbage deviates slightly during the conveying process, the photoelectric sensor 13 detects that its edge profile deviates from the center line, and the encoder 14 records the actual running distance, and the sorting arm 10 adjusts the action parameters of the clamp 11 according to the feedback to ensure the accuracy of the drop position.

[0044] The sorting arm 10 adjusts the opening width of the clamp 11 according to the classification result. For medium-sized garbage such as plastic bottles, the opening width of the clamp 11 is adjusted to 50mm to accommodate their diameter. The servo motor drives the sorting arm 10 to complete the garbage drop action, and records the drop position and garbage type information to the system. Assuming that the plastic bottle is dropped into the recyclable garbage container, the system will update its drop record simultaneously for subsequent statistics and analysis.

[0045] During the entire garbage classification process, the fault detection module 15 monitors the working state of the sorting arm 10 in real time. The temperature sensor 16 detects the temperature rise of the servo motor housing, and if the temperature rise exceeds the preset threshold (such as 60°C), the data is transmitted to the central control unit 19 through the wireless communication module 18. The vibration sensor 17 detects abnormal vibration during the operation of the sorting arm 10, and if the vibration amplitude is too large (such as more than 0.5mm), an alarm is sent through the wireless communication module 18. After receiving the monitoring data, the central control unit 19 judges the equipment running state according to the preset threshold. If an abnormality is detected, the LED indicator light in the alarm device 21 flashes and is accompanied by a buzzer alarm sound, prompting the operator to maintain the equipment in time.

[0046] The above embodiments elaborate the operation principle of the application in specific application scenarios. Through synchronous acquisition and preprocessing of multi-sensor data, the problem of single sensor being easily disturbed by the environment is eliminated; through the cooperative work of the feature extraction unit and the classification decision module, the accuracy of garbage feature analysis and the classification efficiency are improved; through the cooperation of the actuator and the position correction unit, the precise control of the garbage throwing position is realized, and the occurrence of misplacement is reduced. At the same time, the real-time monitoring function of the fault detection module and the central control unit ensures the stable operation and efficient management of the system.

Claims

1. A garbage classification method based on multi-sensor fusion, characterized in that: The following steps are involved: Step 1: Collect physical characteristics of garbage through multiple sensors, including optical characteristics, weight characteristics and shape characteristics; Step 2: Input the collected multiple sets of feature information into the data processing module (1), and use the signal modulation unit (2) to pre-process the data of each sensor to eliminate noise interference and extract key feature values; Step 3: The processed feature values ​​are transmitted to the classification decision module (4), and the classification decision module (4) makes a preliminary judgment on the type of garbage according to the preset rules; Step 4: Send the preliminary determination result to the execution mechanism (8), which completes the garbage placement action and records the placement location and garbage type information.

2. The garbage classification method based on multi-sensor fusion according to claim 1 is characterized by: In step 2, the data processing module (1) includes a signal modulation unit (2) and a feature extraction unit (3). The signal modulation unit (2) is provided with two groups, corresponding to the output signals of the optical sensor and the weight sensor respectively. The signal modulation unit (2) is provided with a filtering circuit and an amplifying circuit inside. The filtering circuit is used to remove high-frequency noise, and the amplifying circuit is used to enhance the signal strength. The feature extraction unit (3) is located behind the signal modulation unit (2). The feature extraction unit (3) extracts key feature values ​​of the garbage through algorithm analysis, including reflectivity peak, weight distribution curve and edge profile parameters.

3. The garbage classification method based on multi-sensor fusion according to claim 2 is characterized in that: The feature extraction unit (3) includes a main control chip and a storage module. The main control chip adopts a multi-core processor. The input end of the main control chip is connected to the signal modulation unit (2), and the output end of the main control chip is connected to the storage module. The storage module is used to store the extracted feature values ​​and transmit the feature values ​​to the classification decision module (4) through an access interface. A cache area is provided inside the storage module. The cache area has a capacity of 16MB and is used to temporarily store real-time data streams.

4. The garbage classification method based on multi-sensor fusion according to claim 2 is characterized in that: The invention also includes a data calibration structure, which is used to adjust the filtering parameters and amplification factors of the signal modulation unit (2) to adapt to the output characteristics of different sensors. The data calibration structure includes a knob adjuster and a digital display screen. The knob adjuster is connected to the filtering circuit and the amplification circuit through a mechanical transmission mechanism, and the digital display screen is used to display the current adjustment parameters.

5. The garbage classification method based on multi-sensor fusion according to claim 1, characterized in that: The classification decision module (4) includes a logic operation unit (5) and a rule base (6). The logic operation unit (5) receives the feature value from the feature extraction unit (3). The logic operation unit (5) is provided with a plurality of parallel computing channels, each channel corresponding to a type of garbage. The rule base (6) stores preset classification rules. The rule base (6) supports the entry of rules for newly added types of garbage through a dynamic update mechanism. The logic operation unit (5) and the rule base (6) are connected via a high-speed data bus, and the bandwidth of the data bus is 100Mbps.

6. The garbage classification method based on multi-sensor fusion according to claim 5 is characterized in that: The classification decision module (4) further includes a priority judgment unit (7), which is located behind the logic operation unit (5). The priority judgment unit (7) sorts the classification results according to the confidence score of the garbage feature value, and garbage with a confidence score lower than a preset threshold is marked as an object to be re-inspected.

7. The garbage classification method based on multi-sensor fusion according to claim 1, characterized in that: The actuator (8) includes a conveyor belt (9) and a sorting arm (10). The conveyor belt (9) is arranged below the classification decision module (4). The surface of the conveyor belt (9) is provided with an anti-slip texture, and the depth of the anti-slip texture is 0.5mm to 1mm. The sorting arm (10) is installed at the end of the conveyor belt (9). The sorting arm (10) is driven by a servo motor, and the speed range of the servo motor is 500rpm to 1500rpm. A clamp (11) is provided at the end of the sorting arm (10), and the opening width of the clamp (11) is adjustable, and the adjustment range is 20mm to 100mm.

8. The garbage classification method based on multi-sensor fusion according to claim 7, characterized in that: A position correction unit (12) is provided between the sorting arm (10) and the classification decision module (4). The position correction unit (12) is used to detect the actual position of the garbage on the conveyor belt (9) and feed back the position information to the sorting arm (10) control system to compensate for the position deviation of the conveyor belt (9) during operation. The position correction unit (12) includes a photoelectric sensor (13) and an encoder (14). The photoelectric sensor (13) is used to capture the edge contour of the garbage, and the encoder (14) is used to record the running distance of the conveyor belt (9).

9. The garbage classification method based on multi-sensor fusion according to claim 1, characterized in that: The invention also includes a fault detection module (15), which is located above the actuator (8) and is used to monitor the working state of the sorting arm (10) in real time. The fault detection module (15) includes a temperature sensor (16) and a vibration sensor (17). The temperature sensor (16) is used to detect the temperature rise of the servo motor, and the vibration sensor (17) is used to detect abnormal vibration during the operation of the sorting arm (10). The fault detection module (15) transmits the monitoring data to the central control unit (19) through the wireless communication module (18). The wireless communication module (18) adopts the Zigbee protocol and has a communication distance of 30 meters.

10. The garbage classification method based on multi-sensor fusion according to claim 9, characterized in that: The central control unit (19) includes a main control board (20) and an alarm device (21). The main control board (20) receives monitoring data from the fault detection module (15) and determines the operating state of the equipment according to a preset threshold value. The alarm device (21) is located on one side of the main control board (20). The alarm device (21) includes an LED indicator light and a buzzer. The LED indicator light is used to display the operating state of the equipment, and the buzzer is used to emit an alarm sound.

Images