Biomass pyrolysis feeding process pre-screening system based on computer vision

Abstract

The invention relates to a pre-screening system for a biomass pyrolysis feeding process based on computer vision, which is provided with a visual detector, a data memory, an image processing device, a classification evaluation device and a decision control device. Image information of biomass raw materials in a feeder is shot from multiple angles through a visual detector in the biomass feeding pre-screening process, appearance feature image data of the biomass raw materials are obtained, and then the appearance feature image data are preprocessed through an image processing device; biomass features are extracted from the preprocessing result through a deep learning algorithm, and then the biomass features are compared with experimental data through a classification evaluation device so as to obtain the probability that the biomass raw material belongs to each biomass category in the experimental data; and finally, an electric control adjusting part is controlled through a decision control device according to the probability and purity information to adjust the feeding amount of the feeder, so that the screening precision can be improved, and the pyrolysis efficiency and quality of the biomass raw materials are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of biomass pyrolysis, and in particular to a biomass pyrolysis feed process pre-screening system based on computer vision. Background Art

[0002] The biomass feed pre-screening system combines computer vision algorithms with component analysis cross-validation technology to optimize the raw material screening during biomass pyrolysis, thereby improving overall economic benefits. This type of system receives and transports biomass to the analysis area through a feeder, and then accurately screens the biomass sample through an intelligent robotic arm.

[0003] However, the biomass feed pre-screening system in the prior art has the problems of low raw material screening efficiency and low accuracy in optimizing the biomass pyrolysis process, which leads to the technical defect that the pre-screening of the biomass pyrolysis feed process is difficult to effectively carry out. Summary of the invention

[0004] The technical problem to be solved by the present invention is how to solve the technical problem existing in the prior art that the biomass feed pre-screening system has low raw material screening efficiency and low accuracy in the process of optimizing biomass pyrolysis. In order to overcome the above problems of the prior art, the present invention provides a biomass pyrolysis feed process pre-screening system based on computer vision.

[0005] The present invention provides a computer vision-based biomass pyrolysis feeding process pre-screening system, comprising a feeder and an electric control regulating member for regulating the feeding amount of the feeder, and further comprising:

[0006] A visual detector is configured to capture image information of the biomass raw material in the feeder from multiple different angles to obtain image data of the appearance characteristics of the biomass raw material;

[0007] A data storage device for storing experimental data; the experimental data includes physical characteristic information, chemical composition information and pyrolysis potential information of various biomass types obtained in the experiment;

[0008] An image processing device is configured to preprocess the shape feature image data and then extract biomass features from the preprocessing results using a deep learning algorithm; the biomass features include the morphology, color, and surface texture of the biomass;

[0009] a classification evaluation device configured to compare the biomass characteristics with the experimental data to obtain a probability that the biomass raw material belongs to each biomass category in the experimental data;

[0010] A decision control device, configured to control the electric control adjustment member to adjust the feed amount of the feeder according to the probability that the biomass raw material belongs to each biomass category in the experimental data and the purity represented by the chemical composition information of the biomass raw material;

[0011] Among them, the image processing device is electrically connected to the visual detector, the classification and evaluation device is electrically connected to the data storage device and the image processing device at the same time, and the decision control device is electrically connected to the classification and evaluation device and the electric control adjustment component at the same time.

[0012] The computer vision-based biomass pyrolysis feeding process pre-screening system disclosed in the present invention can realize that in the biomass feed pre-screening process, the visual detector is used to capture the image information of the biomass raw material in the feeder from multiple different angles, thereby obtaining the appearance feature image data of the biomass raw material, and then the appearance feature image data is pre-processed by the image processing device, and the biomass features are extracted from the pre-processing results by a deep learning algorithm, and then the biomass features are compared with the experimental data by the classification evaluation device to obtain the probability that the biomass raw material belongs to each biomass category in the experimental data, and finally the decision control device controls the electric control adjustment part according to these probabilities and purity information to adjust the feed amount of the feeder, thereby realizing the multi-angle collection of the image information of the biomass raw material, performing image processing to finally improve the screening accuracy, and improving the pyrolysis efficiency and quality of the biomass raw material, and the complexity of the algorithm used in the classification evaluation and control decision process is relatively small, which can save materials and energy, simplify the control procedure, and finally achieve environmental protection and sustainable effects.

[0013] In a possible implementation, the visual detector includes:

[0014] A plurality of high-definition cameras are distributed in multiple positions of the feeding area of ​​the feeder to capture image information of the biomass raw materials in the feeder from multiple different angles;

[0015] An image integration unit, configured to integrate all image information obtained by the plurality of high-definition cameras at the same sampling time into an image list to obtain the appearance feature image data;

[0016] Wherein, the image integration unit is electrically connected to all the high-definition cameras and the image processing device at the same time;

[0017] This solution uses multiple high-definition cameras to capture biomass images from different angles simultaneously, combined with image processing technology to ensure more accurate identification of the morphology, color and texture characteristics of biomass.

[0018] In a possible implementation, the image processing device includes:

[0019] A preprocessing module, configured to preprocess the shape feature image data to obtain a preprocessing result;

[0020] A feature extraction module is configured to obtain the biomass feature by performing multiple convolution and multiple downsampling operations on the preprocessing result;

[0021] Wherein, the preprocessing module is electrically connected to the image integration unit, and the feature extraction module is electrically connected to the preprocessing module.

[0022] In a possible implementation, the preprocessing module is configured to perform the following steps:

[0023] A1: De-noising the shape feature image data by using a Gaussian blur algorithm to obtain a de-noising result;

[0024] A2: enhancing the edges and details of the biomass in the denoising result by contrast adjustment or histogram equalization to obtain an enhanced result;

[0025] A3: extracting the biomass contour from the enhancement processing result by using the Kenny edge detection algorithm to obtain contour feature information;

[0026] A4: performing color space conversion on the contour feature information by color transformation mapping to obtain the preprocessing result;

[0027] Since the biomass image acquisition process may be affected by noise, this solution first uses Gaussian blur technology to remove noise, and then uses contrast adjustment and histogram equalization technology to enhance the edges and details of the biomass in the image, making the features more obvious. The Canny edge detection algorithm used subsequently can identify the outline of the biomass, providing a basis for subsequent morphological analysis. The final color space conversion helps to distinguish different types of biomass.

[0028] In a possible implementation, in step A4, the color space conversion of the contour feature information is performed as follows:

[0029] The contour feature information is transformed from a color space form in which the color space form is a combination of red, green, and blue to a color space form in which the color space form is a combination of hue, saturation, and brightness;

[0030] This scheme not only helps to distinguish different types of biomass, especially biomass raw materials with a higher cellulose content, but also because this type of biomass may have a different color distribution from other types, these features can be more effectively identified using color space conversion.

[0031] In a possible implementation, the feature extraction module is a network module formed by connecting multiple convolution units, multiple downsampling units and a fully connected layer unit in series; this solution automatically classifies and evaluates the quality of images through a convolutional neural network of a deep learning algorithm, so that the system can quickly and accurately screen out suitable biomass, thereby improving production efficiency.

[0032] In a possible implementation, the classification evaluation device is configured to perform the following steps:

[0033] B1: Calculate the probability that the biomass raw material belongs to each biomass category in the experimental data according to the biomass characteristics through the Softmax function;

[0034] B2: Retrieving the maximum probability value according to the calculation result of step B1, and determining the biomass category in the experimental data corresponding to the maximum probability value;

[0035] This solution evaluates the ownership of biomass raw materials by calculating probabilities, which can achieve orderly and efficient operation of decision-making and control.

[0036] In a possible implementation, the decision control device is configured to perform the following steps:

[0037] C1: Determine whether the maximum probability is not less than a threshold;

[0038] If yes, the biomass category in the experimental data corresponding to the maximum probability value is used as the matching category of the biomass raw material, and then the next step is executed;

[0039] If not, the electric control adjustment member is operated to close the feeding of the feeder, and a first warning is issued, and then step C4 is executed;

[0040] C2: Determine whether the quality of the biomass raw material meets the pyrolysis conditions;

[0041] If yes, proceed to the next step;

[0042] If not, the electric control adjustment member is operated to close the feeding of the feeder, and a second warning is issued, and then step C4 is executed;

[0043] C3: judging whether the purity requirement is met according to the chemical composition information of the matching category;

[0044] If yes, the electric control adjustment member is operated to maintain the initial feeding speed of the feeder, and then the next step is performed;

[0045] If not, the electric control adjustment member is operated to reduce the feeding speed of the feeder, and a third warning is issued, and then the next step is executed;

[0046] C4: performing near infrared spectroscopy analysis on the biomass raw material to obtain the cellulose content of the biomass;

[0047] This solution compares the results with the biomass data in the experimental data. If the classification results are highly matched with the data in the database and the quality of the biomass meets the pyrolysis requirements, the system will allow the feed to continue. If the purity of the biomass is low, the system will issue a warning and take corresponding control measures (reducing the feed rate). In addition, near-infrared spectroscopy (NIR) analysis is used to verify the cellulose content of the biomass, thereby further ensuring the quality and pyrolysis potential of the incoming raw materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A schematic diagram of the structure of a biomass pyrolysis feed process pre-screening system based on computer vision disclosed in an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of the operation flow of the preprocessing module disclosed in the embodiment of the present invention;

[0050] Figure 3 The present invention is a schematic diagram of the operation flow of the decision control device disclosed in the embodiment of the present invention. DETAILED DESCRIPTION

[0051] First, those skilled in the art should understand that these implementations are only used to explain the technical principles of the embodiments of the present application, and are not intended to limit the protection scope of the embodiments of the present application. Those skilled in the art can make adjustments to them as needed to adapt to specific application scenarios.

[0052] In the description of the embodiments of the present application, it should be noted that, unless otherwise clearly specified and limited, the terms "electrical connection" and "electrical connection relationship" should be understood in a broad sense, that is, referring to a connection method with an electrical relationship, for example, it can be a circuit connection through a wire, or it can be an electrical connection through a radio signal channel (channel), or a combination of the two. In addition, "electrical connection" and "electrical connection relationship" can be based on mechanical connection (such as a wire set in a connection key); it can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in the embodiments of the present application can be understood according to specific circumstances.

[0053] In the embodiments of the present application, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may mean that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, a first feature being "above", "above" or "above" a second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. A first feature being "below", "below" or "below" a second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.

[0054] The present application is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] See also Figure 1 to Figure 3 As shown, the present application embodiment discloses a biomass pyrolysis feed process pre-screening system based on computer vision, see Figure 1 As shown, the pre-screening system includes a feeder, an electric control adjustment member, a visual detector, a data storage device, an image processing device, a classification evaluation device and a decision control device, wherein the electric control adjustment member is arranged on the feeding channel of the feeder, and is used to adjust the feeding amount of the feeder; the image processing device is electrically connected to the visual detector, the classification evaluation device is electrically connected to the data storage device and the image processing device at the same time, and the decision control device is electrically connected to the classification evaluation device and the electric control adjustment member at the same time. The data storage device is used to store experimental data, and the experimental data includes physical characteristic information, chemical composition information and pyrolysis potential information of various biomass categories obtained in the experiment.

[0056] In the pre-screening system, the visual detector is configured to capture image information of the biomass raw material in the feeder from multiple different angles to obtain image data of the appearance characteristics of the biomass raw material. Figure 1 The visual detector includes a plurality of high-definition cameras and an image integration unit, wherein the image integration unit is electrically connected to all high-definition cameras and image processing devices at the same time. The plurality of high-definition cameras are distributed in multiple positions in the feeding area of ​​the feeder to capture image information of the biomass raw materials in the feeder from multiple different angles; the image integration unit is configured to integrate all image information obtained by the plurality of high-definition cameras at the same sampling time into an image list to obtain shape feature image data.

[0057] In the pre-screening system, the image processing device is configured to pre-process the shape feature image data, and then extract the biomass features from the pre-processing results through a deep learning algorithm; the biomass features include the morphology, color, and surface texture of the biomass. Figure 1In this embodiment, the image processing device includes a preprocessing module and a feature extraction module, wherein the preprocessing module is electrically connected to the image integration unit, and the feature extraction module is electrically connected to the preprocessing module. The preprocessing module is configured to preprocess the shape feature image data to obtain a preprocessing result; the feature extraction module is configured to obtain the biomass feature by performing multiple convolution and multiple downsampling operations on the preprocessing result.

[0058] See also Figure 2 In this embodiment, the preprocessing module is configured to perform the following steps: A1: Denoise the shape feature image data by using the Gaussian blur algorithm to obtain a denoising result; A2: Enhance the edges and details of the biomass in the denoising result by contrast adjustment or histogram equalization to obtain an enhanced result; A3: Extract the biomass contour from the enhanced result by using the Kenny edge detection algorithm to obtain contour feature information; A4: Perform color space conversion on the contour feature information by color transformation mapping to obtain a preprocessing result. In step A4, the color space conversion of the contour feature information is performed by converting the contour feature information from a color space form of a combination of red, green, and blue to a color space form of a combination of hue, saturation, and brightness.

[0059] The preprocessing module of this embodiment first uses Gaussian blur technology to remove noise, and then enhances the edges and details of the biomass in the image through contrast adjustment and histogram equalization technology to make the features more obvious. The Kenny edge detection algorithm used subsequently can identify the outline of the biomass, providing a basis for subsequent morphological analysis. The final color space conversion helps to distinguish different types of biomass.

[0060] In this embodiment, the feature extraction module is a network module formed by connecting m convolution units, m downsampling units and a fully connected layer unit in series, and its operation order is: convolution unit 1, downsampling unit 1, convolution unit 2, downsampling unit 2, ..., convolution unit m, downsampling unit m, fully connected layer unit. The convolution unit extracts local features such as edges, corners, etc. through convolution operations. Each convolution unit uses multiple convolution kernels (filters) to detect different image features. The downsampling unit is mainly used for downsampling to reduce the size of the image while retaining important features. The downsampling method used in this embodiment is maximum pooling, which reduces the dimension of the feature map by taking the maximum value in the pooling window. After being processed by multiple convolution layers and pooling layers, it is finally processed by the fully connected layer unit, and the fully connected layer unit is used to map the extracted features to the final category prediction. In biomass classification, the fully connected layer unit matches the extracted features with the biomass categories in the database. Through this network model, the pre-screening system can automatically learn how to extract the most recognizable features from biomass images, thereby making accurate judgments during the classification process.

[0061] In the pre-screening system, the classification evaluation device is configured to compare the biomass characteristics with the experimental data to obtain the probability that the biomass raw material belongs to each biomass category in the experimental data. In this embodiment, the classification evaluation device is configured to perform the following steps: B1: Calculate the probability that the biomass raw material belongs to each biomass category in the experimental data according to the biomass characteristics by using the Softmax function; B2: Retrieve the maximum probability value according to the calculation result of step B1, and determine the biomass category in the experimental data corresponding to the maximum probability value.

[0062] In the pre-screening system, the decision control device is configured to control the electronic control element to adjust the feed amount of the feeder according to the probability that the biomass raw material belongs to each biomass category in the experimental data and the purity represented by the chemical composition information of the biomass raw material.

[0063] See also Figure 3 In this embodiment, the decision control device is configured to perform the following steps:

[0064] C1: Determine whether the maximum probability is not less than a threshold value; if so, take the biomass category in the experimental data corresponding to the maximum probability value as the matching category of the biomass raw material, and then execute the next step; if not, manipulate the electronic control adjustment member to close the feed of the feeder, issue a first warning, and then execute step C4;

[0065] C2: Determine whether the quality of the biomass raw material meets the pyrolysis conditions; if yes, execute the next step; if no, operate the electric control adjustment member to close the feed of the feeder, issue a second warning, and then execute step C4;

[0066] C3: judging whether the purity requirement is met according to the chemical composition information of the matching category; if so, manipulating the electric control regulating member to maintain the initial feeding speed of the feeder, and then executing the next step; if not, manipulating the electric control regulating member to reduce the feeding speed of the feeder, and issuing a third warning, and then executing the next step;

[0067] C4: Near infrared spectroscopy analysis of biomass raw materials to obtain the cellulose content of biomass; this process obtains the spectral data of biomass through sensors and verifies it with the spectral features in the database to further ensure the quality and pyrolysis potential of the incoming raw materials.

[0068] In this embodiment, the decision control device compares the results with the biomass data in the experimental data. If the classification results are highly matched with the data in the database and the quality of the biomass meets the pyrolysis requirements, the system will allow the feed to continue. If the purity of the biomass is low, the system will issue a warning and take corresponding control measures (reduce the feed rate). In addition, near-infrared spectroscopy is used to verify the cellulose content of the biomass, thereby further ensuring the quality and pyrolysis potential of the feed raw materials. An application scenario of the pre-screening system will be further disclosed below.

[0069] This pre-screening system is used in a pyrolysis plant specializing in the treatment of wood waste to achieve efficient screening of biomass and optimization of the pyrolysis process. The main goal of the plant is to convert biomass such as wood chips into renewable energy through pyrolysis. The pre-screening system combines visual detectors, data storage, image processing devices, classification evaluation devices and decision control devices to ensure the quality of biomass feed and the stability of the pyrolysis process. The following is a specific operation example.

[0070] First, biomass feed is screened and quality assessed. In the factory, wood chips are the main feed material of biomass, and their moisture content is about 10%. The pre-screening system first uses a visual detector to take a full-scale photo of each batch of feed wood chips. Multiple high-definition cameras of the visual detector simultaneously image the wood chips from different angles. The image data is processed by an image processing device to remove noise and enhance the edge and texture features of the wood chips to ensure that the shape, size and distribution of the wood chips can be accurately identified.

[0071] Subsequently, the image processing device uses a deep learning algorithm to analyze the collected image data to identify the type, quality and pyrolysis potential of the sawdust. Through automatic classification using a deep learning algorithm, the pre-screening system can efficiently distinguish different types of sawdust and score their pyrolysis efficiency. If a batch of sawdust is of poor quality (for example, it contains too many impurities or too high a moisture content), the system will automatically mark it and adjust the screening criteria.

[0072] The classification and evaluation device is then used to perform real-time data comparison and verification. Before pyrolysis, the classification and evaluation device further verifies the quality of the sawdust by comparing it with the historical data in the data storage device. The image data of each batch of sawdust is compared with the standard data stored in the experimental data. The pre-screening system automatically optimizes the screening criteria based on the comparison results and provides real-time feedback to ensure that each batch of sawdust meets the quality standards before entering the pyrolysis system. If the system detects that the quality of a batch of sawdust does not meet the standards, the system will trigger a warning and stop feeding. At the same time, it will analyze the historical data in the database and provide recommended improvement plans. For example, a batch of sawdust may be automatically marked as unqualified due to excessive moisture content. The system will recommend adjusting the feed rate or waiting for the sawdust to dry before processing.

[0073] While screening the biomass feed, the pre-screening system also uses near-infrared spectroscopy (NIR) to verify the composition of the sawdust. The NIR sensor collects the spectrum of the sawdust, and the system quickly evaluates the content of key components such as cellulose and lignin in the sawdust by comparing the real-time collected spectral data with the spectral features in the database (Technical Point 4). If the NIR verification results show that the composition of the sawdust does not meet the pyrolysis requirements, the pre-screening system will automatically stop feeding and issue a warning. For example, in one experiment, the system found that the sawdust had a low cellulose content and poor pyrolysis potential. Through NIR analysis, the system confirmed this and automatically stopped feeding, avoiding inefficient pyrolysis reactions.

[0074] When the quality of the sawdust is verified to be qualified, the pre-screening system performs automatic feeding operation through the coordinated control of the electronic control unit and the feeder. The electronic control unit is called to adjust the feeding speed of the sawdust according to the image data and real-time feedback. If the system detects that a batch of sawdust is of low quality or contains impurities, it will automatically reduce the feeding speed or stop feeding if necessary to ensure the stable operation of the pyrolysis system.

[0075] For example, during a batch of sawdust feeding, the system detects through image recognition and real-time feedback that some sawdust may contain larger impurities. The robotic arm then reduces the feeding speed to ensure that unqualified sawdust is not fed into the reactor.

[0076] Finally, there is the self-learning and continuous optimization of the system. As the plant operation time increases, the data storage of the pre-screening system will continue to accumulate new data, and perform self-learning and model optimization. Through the continuous accumulation of image and component data, the system's recognition model is constantly updated and optimized, thereby improving screening accuracy and pyrolysis efficiency. For example, the pre-screening system can optimize wood chip classification and feeding strategies based on historical data, allowing the system to cope with biomass from different sources and types.

[0077] In summary, the computer vision-based biomass pyrolysis feeding process pre-screening system disclosed in this embodiment, by setting a visual detector, a data storage device, an image processing device, a classification evaluation device and a decision control device, can realize that in the biomass feed pre-screening process, the visual detector is used to capture the image information of the biomass raw material in the feeder from multiple different angles, thereby obtaining the appearance feature image data of the biomass raw material, and then the appearance feature image data is preprocessed by the image processing device, and the biomass features are extracted from the preprocessing results by a deep learning algorithm, and then the biomass features are compared with the experimental data by the classification evaluation device to obtain the probability that the biomass raw material belongs to each biomass category in the experimental data, and finally the decision control device controls the electric control adjustment member according to these probabilities and purity information to adjust the feed amount of the feeder, so that the image information of the biomass raw material can be collected from multiple angles, and the image processing is performed to finally improve the screening accuracy, improve the pyrolysis efficiency and quality of the biomass raw material, and the complexity of the algorithm used in the classification evaluation and control decision process is relatively small, which can save materials and energy, simplify the control procedure, and finally achieve environmental protection and sustainable effects.

[0078] In the description of the embodiments of the present application, it should be noted that in the description of the present application, terms such as "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is only for the convenience of description, and does not indicate or imply that the device or component must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present application.

[0079] In the description of the present application, the description with reference to the terms "one embodiment", "some embodiments", "in the present embodiment", "specific example", or "some examples" etc. means that the specific features, mechanisms, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, mechanisms, materials or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0080] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A computer vision-based biomass pyrolysis feed process pre-screening system, comprising a feeder and an electric control regulating member for regulating the feed amount of the feeder, characterized in that: Also includes: A visual detector is configured to capture image information of the biomass raw material in the feeder from multiple different angles to obtain image data of the appearance characteristics of the biomass raw material; A data storage device for storing experimental data; the experimental data includes physical characteristic information, chemical composition information and pyrolysis potential information of various biomass types obtained in the experiment; An image processing device is configured to preprocess the shape feature image data and then extract biomass features from the preprocessing results using a deep learning algorithm; the biomass features include the morphology, color, and surface texture of the biomass; a classification evaluation device configured to compare the biomass characteristics with the experimental data to obtain a probability that the biomass raw material belongs to each biomass category in the experimental data; A decision control device, configured to control the electric control adjustment member to adjust the feed amount of the feeder according to the probability that the biomass raw material belongs to each biomass category in the experimental data and the purity represented by the chemical composition information of the biomass raw material; Among them, the image processing device is electrically connected to the visual detector, the classification and evaluation device is electrically connected to the data storage device and the image processing device at the same time, and the decision control device is electrically connected to the classification and evaluation device and the electric control adjustment component at the same time.

2. The computer vision-based biomass pyrolysis feed process pre-screening system according to claim 1, characterized in that: The visual detector comprises: A plurality of high-definition cameras are distributed in multiple positions of the feeding area of ​​the feeder to capture image information of the biomass raw materials in the feeder from multiple different angles; An image integration unit, configured to integrate all image information obtained by the plurality of high-definition cameras at the same sampling time into an image list to obtain the appearance feature image data; Wherein, the image integration unit is electrically connected to all the high-definition cameras and the image processing device at the same time.

3. The computer vision-based biomass pyrolysis feed process pre-screening system according to claim 2, characterized in that: The image processing device comprises: A preprocessing module, configured to preprocess the shape feature image data to obtain a preprocessing result; A feature extraction module is configured to obtain the biomass feature by performing multiple convolution and multiple downsampling operations on the preprocessing result; Wherein, the preprocessing module is electrically connected to the image integration unit, and the feature extraction module is electrically connected to the preprocessing module.

4. The computer vision-based biomass pyrolysis feed process pre-screening system according to claim 3, characterized in that: The pre-processing module is configured to perform the following steps: A1: De-noising the shape feature image data by using a Gaussian blur algorithm to obtain a de-noising result; A2: enhancing the edges and details of the biomass in the denoising result by contrast adjustment or histogram equalization to obtain an enhanced result; A3: extracting the biomass contour from the enhancement processing result by using the Kenny edge detection algorithm to obtain contour feature information; A4: Performing color space conversion on the contour feature information through color transformation mapping to obtain the preprocessing result.

5. The computer vision-based biomass pyrolysis feed process pre-screening system according to claim 4, characterized in that: In step A4, the color space conversion of the contour feature information is performed as follows: The contour feature information is transformed from a color space form in which the color space form is a combination of red, green and blue to a color space form in which the color space form is a combination of hue, saturation and brightness.

6. The computer vision-based biomass pyrolysis feed process pre-screening system according to any one of claims 3 to 5, characterized in that: The feature extraction module is a network module formed by connecting multiple convolution units, multiple downsampling units and a fully connected layer unit in series.

7. The computer vision-based biomass pyrolysis feed process pre-screening system according to claim 6, characterized in that: The classification evaluation device is configured to perform the following steps: B1: Calculate the probability that the biomass raw material belongs to each biomass category in the experimental data according to the biomass characteristics through the Softmax function; B2: According to the calculation result of step B1, a maximum probability value is retrieved, and the biomass category in the experimental data corresponding to the maximum probability value is determined.

8. The computer vision-based biomass pyrolysis feed process pre-screening system according to claim 7, characterized in that: The decision control device is configured to perform the following steps: C1: Determine whether the maximum probability is not less than a threshold; If yes, the biomass category in the experimental data corresponding to the maximum probability value is used as the matching category of the biomass raw material, and then the next step is executed; If not, the electric control adjustment member is operated to close the feeding of the feeder, and a first warning is issued, and then step C4 is executed; C2: Determine whether the quality of the biomass raw material meets the pyrolysis conditions; If yes, proceed to the next step; If not, the electric control adjustment member is operated to close the feeding of the feeder, and a second warning is issued, and then step C4 is executed; C3: judging whether the purity requirement is met according to the chemical composition information of the matching category; If yes, the electric control adjustment member is operated to maintain the initial feeding speed of the feeder, and then the next step is performed; If not, the electric control adjustment member is operated to reduce the feeding speed of the feeder, and a third warning is issued, and then the next step is executed; C4: performing near infrared spectroscopy analysis on the biomass raw material to obtain the cellulose content of the biomass.

Images