Solid waste detection method and system

Abstract

The invention relates to the technical field of environmental protection, in particular to a solid waste detection method and system.The method comprises the steps that solid waste samples with disclosed categories, components and contents and Raman spectrum data corresponding to the solid waste samples are obtained from the Internet to serve as a training set, and the training set is used for training an R-CNN-based target detection model; after training is completed, the flooding ability and performance of the model are evaluated through cross validation; after verification is completed, solid waste samples are prepared and subjected to sample pretreatment; after the sample pretreatment is completed, scanning the sample by using a Raman spectrometer to obtain Raman spectrum data of the sample, and then carrying out data pretreatment; after data preprocessing is completed, Raman spectrum data are input into the verified R-CNN-based target detection model, and a detection result is output. According to the method, the information of the solid waste is accurately detected through a Raman spectrum and machine learning combined method, technical support is provided for management of the solid waste, environmental pollution is reduced, and the ecological environment is protected.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental protection, and in particular to a solid waste detection method and system. Background Art

[0002] With the continuous development of industrialization and social economy, the amount of solid waste generated has increased year by year, including industrial solid waste, domestic garbage, medical waste, construction waste, electronic waste and municipal sludge. Among them, industrial solid waste, medical waste and electronic waste contain a large amount of harmful substances, such as heavy metals, organic pollutants and discarded medicines, which pose a serious threat to the environment and people's safety. Traditional solid waste detection methods have problems such as low efficiency, high cost and poor accuracy, which are difficult to meet the growing amount of solid waste and the needs of modern environmental protection work. Therefore, it is of great practical significance to develop an efficient and accurate solid waste detection method. Summary of the invention

[0003] In view of the above-mentioned technical deficiencies, the present invention provides a solid waste detection method and system.

[0004] The present invention is achieved through the following technical solutions:

[0005] A solid waste detection method is provided, the method comprising the following steps:

[0006] Step S10: obtaining solid waste samples of publicly disclosed categories, components and contents and their corresponding Raman spectrum data from the Internet as a training set;

[0007] Step S20: Use the training set to train the R-CNN-based target detection model. During the training process, the generalization ability and performance of the model are evaluated through cross-validation, and the model is optimized and the model version is determined based on the evaluation results.

[0008] Step S30: after the model version is determined, a solid waste sample is prepared and sample preprocessing is performed on the solid waste sample;

[0009] Step S40: After the sample preprocessing is completed, the sample is scanned using a Raman spectrometer to obtain one-dimensional Raman spectrum data of the solid waste sample and then perform data preprocessing;

[0010] Step S50: After the data preprocessing is completed, the Raman spectrum data is input into a determined version of the R-CNN-based target detection model to output the detection result;

[0011] Wherein, the channels for obtaining the solid waste samples and their corresponding Raman spectrum data of the public categories, components and contents from the Internet in step S10 include government agency websites, scientific research institution websites, environmental protection organization websites and public data packages published by enterprises on the Internet;

[0012] Wherein, the solid waste samples of publicly disclosed categories, components and contents obtained from the Internet in step S10 include industrial solid waste, domestic waste, medical waste, construction waste, electronic waste and municipal sludge;

[0013] Among them, the output detection result of the target detection model based on R-CNN in step S50 includes the category, composition and content of the detected solid waste sample.

[0014] Preferably, the R-CNN-based target detection model training step in step S20 includes:

[0015] Training set preparation: The one-dimensional Raman spectrum data of the solid waste samples were converted into image data. The image size was set to 227×227, and the category and composition information of each sample were annotated to obtain a labeled data set as the training set.

[0016] Model construction: After obtaining the training set, the model is constructed using the AlexNet network architecture, including the input layer, convolution layer, pooling layer, fully connected layer, and output layer. The cross entropy loss function is used to calculate the classification loss, and the smooth L1 loss function is used to calculate the bounding box regression loss. The weighted sum of the classification loss and regression loss is used as the total loss function. The Adam optimizer is selected as the optimizer.

[0017] Model training: After the model is built, use the training set to train the model and adjust the model parameters to optimize performance.

[0018] Preferably, in the step S20, the generalization ability and performance of the model are evaluated by cross-validation during the training process, using K-fold cross-validation, including:

[0019] Dataset partitioning: The dataset is divided into a training set and a validation set, where the dataset is randomly divided into K equal subsets, each subset is used as a validation set in turn, and the remaining subsets are used as training sets;

[0020] Model training and validation: When each subset is used as a validation set, use other subsets as training sets to train the model, use the validation set to evaluate the model performance, and record the results of each validation;

[0021] Performance evaluation: Calculate the weighted average of all validation results to obtain the final generalization ability and performance evaluation of the model;

[0022] Model optimization: When the final generalization ability and performance evaluation of the model do not meet expectations, adjust the model parameters and retrain until the optimal model parameter combination is obtained.

[0023] Preferably, the sample pretreatment of the solid waste sample in step S30 is physical treatment of the sample, including crushing, grinding and stirring.

[0024] Preferably, after obtaining the one-dimensional Raman spectrum data of the solid waste sample in step S40, data preprocessing is performed, and the data preprocessing step includes:

[0025] De-noising: Fourier transform is used to remove noise from the one-dimensional Raman spectroscopy data of solid waste samples;

[0026] Baseline correction: The polynomial fitting method is used to remove the fluorescence background interference in the one-dimensional Raman spectroscopy data of solid waste samples;

[0027] Normalization: Normalize the 1D Raman spectroscopy data of solid waste samples to the same range;

[0028] Data conversion: Convert the one-dimensional Raman spectrum data of one-dimensional solid waste samples into two-dimensional image data.

[0029] The method for converting the one-dimensional Raman spectrum data of the one-dimensional solid waste sample into two-dimensional image data in the data conversion is to combine the Raman spectrum data into a color image, and the specific steps include:

[0030] Reshape: Reshape the one-dimensional Raman spectral data of length N into a two-dimensional matrix of M×M, where M is the integer closest to the square root of N. <M 2 When N=M, zero is added at the end to make N=M 2 ; When N is not a perfect square and N>M 2 When truncated at the end, N = M 2 ; When N is not a perfect square and N=M 2 No zero padding or truncation is performed;

[0031] Combination: Select every 3 M×M two-dimensional matrices, corresponding to the 3 channels of RGB respectively, and combine the 3 M×M two-dimensional matrices into an M×M×3 three-dimensional matrix according to the channels to obtain color image data. When the remaining M×M two-dimensional matrices cannot be combined into 3, use a two-dimensional matrix with 0 elements to fill it.

[0032] Preferably, after the test results are output in step S50, a test report is automatically generated and the type, composition and content of the solid waste are analyzed; when there is an error between the test results and the actual situation, an error analysis is performed.

[0033] In addition, to achieve the above-mentioned purpose, the present invention also provides a solid waste detection device, the solid waste detection device comprising:

[0034] Data acquisition module: used to obtain solid waste samples with publicly disclosed categories, components and contents and their corresponding Raman spectrum data from the Internet as training sets;

[0035] Model training and evaluation module: used to train the R-CNN-based target detection model using the training set. During the training process, the generalization ability and performance of the model are evaluated through cross-validation, and the model is optimized and the model version is determined based on the evaluation results.

[0036] Solid waste sample preparation and preprocessing module: used to prepare solid waste samples and perform sample preprocessing on solid waste samples after the model version is determined;

[0037] Raman spectrum scanning module: used to scan the sample using a Raman spectrometer after the sample pretreatment is completed, and obtain the one-dimensional Raman spectrum data of the solid waste sample and then perform data preprocessing;

[0038] Solid waste detection module: After data preprocessing, it inputs Raman spectroscopy data into a certain version of the R-CNN-based target detection model to output the detection results;

[0039] The channels for obtaining the solid waste samples and their corresponding Raman spectrum data of the public categories, components and contents from the Internet in the data acquisition module include government agency websites, scientific research institution websites, environmental protection organization websites and public data packages published by enterprises on the Internet;

[0040] The data acquisition module acquires solid waste samples of publicly disclosed categories, components and contents from the Internet, including industrial solid waste, domestic waste, medical waste, construction waste, electronic waste and municipal sludge;

[0041] The output detection results of the R-CNN-based target detection model in the solid waste detection module include the category, composition and content of the detected solid waste samples.

[0042] In addition, to achieve the above-mentioned purpose, the present invention also proposes a solid waste detection device, which includes: a memory, a processor, and programs such as an R-CNN target detection algorithm stored in the memory and run on the processor, and the programs such as the R-CNN target detection algorithm are steps to implement the solid waste detection method described above.

[0043] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer program product, which includes programs based on the R-CNN target detection algorithm and the like. When the programs based on the R-CNN target detection algorithm and the like are executed by the processor, the solid waste detection method as described above is implemented.

[0044] The advantages and effects of the present invention are:

[0045] The present invention proposes a solid waste detection method and system, which use Raman spectroscopy combined with machine learning to accurately detect the type, composition and content of solid waste, and greatly shorten the detection cycle, thereby realizing automated, efficient and accurate solid waste detection, reducing manual operation links and reducing labor costs; providing technical support for the effective management and disposal of solid waste, helping to reduce environmental pollution and protect the ecological environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 The present invention is a flow chart of a solid waste detection method.

[0048] Figure 2 It is a structural schematic diagram of a solid waste detection device of the present invention.

[0049] Figure 3 The present invention is a schematic block diagram of the structure of a solid waste detection electronic device. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] The present invention provides a solid waste detection method, such as Figure 1 As shown, the following steps are included:

[0052] Step S10: obtaining solid waste samples of publicly disclosed categories, components and contents and their corresponding Raman spectrum data from the Internet as a training set.

[0053] Among them, the channels for obtaining solid waste samples with publicly disclosed categories, components and contents and their corresponding Raman spectral data from the Internet include government agency websites, scientific research institution websites, environmental protection organization websites and public data packages released by enterprises on the Internet; solid waste samples with publicly disclosed categories, components and contents obtained from the Internet include industrial solid waste, domestic waste, medical waste, construction waste, electronic waste and municipal sludge.

[0054] When the Raman spectra of some solid wastes cannot be obtained from the Internet, relevant solid waste samples are collected for sample preprocessing, and then the solid waste samples are scanned using a Raman spectrometer to obtain their Raman spectral data, and their categories and composition information are annotated to obtain labeled Raman spectral data to be added to the training set.

[0055] Step S20: Use the training set to train the R-CNN-based target detection model. During the training process, the generalization ability and performance of the model are evaluated through cross-validation, and the model is optimized and the model version is determined based on the evaluation results.

[0056] The training steps of the R-CNN-based target detection model include:

[0057] Training set preparation: The one-dimensional Raman spectrum data of the solid waste samples were converted into image data. The image size was set to 227×227×3, that is, each image had 227×227 pixels and 3 RGB color channels. The category and composition information of each sample were annotated to obtain a labeled data set as the training set.

[0058] Model construction: After obtaining the training set, the model is constructed using the AlexNet network architecture, including the input layer, convolution layer, pooling layer, fully connected layer and output layer. The cross entropy loss function is used to calculate the classification loss. The cross entropy loss calculation of a single sample is shown in formula (1):

[0059]

[0060] Where C is the number of solid waste sample categories, y i is the real solid waste sample Raman spectrum data vector, is the Raman spectrum data vector of solid waste samples output by the model;

[0061] The smooth L1 loss function is used to calculate the bounding box regression loss. The formula of the smooth L1 loss function is shown in formula (2):

[0062]

[0063] Where x represents the difference between the output value and the true value, |x| represents the absolute value of x; when |x|<1, 0.5x is used. 2 As the loss function, a larger gradient at small errors helps converge faster. When |x| ≥ 1, use |x|-0.5 as the loss function, so that the gradient is constant at large errors to avoid gradient explosion.

[0064] The weighted sum of classification loss and regression loss is used as the total loss function, and the optimizer selects the Adam optimizer;

[0065] Model training: After the model is built, use the training set to train the model and adjust the model parameters to optimize performance;

[0066] In the model construction, the image size received by the input layer is 227×227×3, that is, each image has 227×227 pixels and 3 RGB color channels; the convolution layer includes 6 sub-convolution layers, and each sub-convolution layer uses the ReLU activation function for nonlinear transformation. The number of parameters and the amount of calculation of the convolution layer vary according to the size, step size and padding of the convolution kernel. For example, the first convolution layer uses an 11×11 convolution kernel, a step size of 4, a padding of 0, and an output feature map size of 55×55×96; the pooling layer uses The maximum pooling method is used to reduce the size of the feature map while retaining important features. The window size is set to 3×3 with a step size of 2. The fully connected layer contains 3 sub-fully connected layers, which are used to integrate the features extracted by the previous convolutional layer and output the final classification results. The number of neurons in the fully connected layer is 4096, 4096 and the number of solid waste categories, respectively. In addition, Dropout regularization is used to reduce the overfitting phenomenon of the model. The output layer uses the softmax activation function to map the output of the fully connected layer to a category probability distribution.

[0067] Specifically, in step S20, during the training process, the generalization ability and performance of the model are evaluated through cross-validation, K-fold cross-validation is adopted, and the model is optimized according to the evaluation results to determine the model version, including:

[0068] Dataset partitioning: The dataset is divided into a training set and a validation set. The dataset is randomly divided into K equal subsets. The value of K is usually 5 or 10. Each subset is used as a validation set in turn, and the remaining subsets are used as training sets.

[0069] Model training and validation: When each subset is used as a validation set, use other subsets as training sets to train the model, use the validation set to evaluate the model performance, and record the results of each validation;

[0070] Performance evaluation: Calculate the weighted average of all validation results to obtain the final generalization ability and performance evaluation of the model;

[0071] Model optimization: When the final generalization ability and performance evaluation of the model do not meet expectations, adjust the model parameters and retrain until the optimal model parameter combination is obtained.

[0072] Step S30: After the model version is determined, a solid waste sample is prepared and sample preprocessing is performed on the solid waste sample.

[0073] Sample pretreatment of solid waste samples mainly involves physical treatment of the samples, including crushing, grinding and stirring, so that a Raman spectrometer can be used to scan the solid waste samples in the next step to obtain Raman spectral data.

[0074] Step S40: After the sample preprocessing is completed, the sample is scanned using a Raman spectrometer to obtain one-dimensional Raman spectrum data of the solid waste sample and then perform data preprocessing.

[0075] The steps for scanning a solid waste sample using Raman spectroscopy include:

[0076] Determine instrument parameters: According to the characteristics of the sample and experimental requirements, select appropriate laser wavelength, power, detector and other parameters. The selection of these parameters will directly affect the clarity and stability of the spectral data;

[0077] Instrument calibration: Calibrate the spectrometer before scanning, including wavenumber calibration and intensity calibration. Wavenumber calibration is to ensure the accuracy of the obtained spectral data. Standard samples can be used for wavenumber calibration. Intensity calibration is to ensure that the light signals between different samples can be accurately compared. White samples (such as transparent glass sheets) are usually used for intensity calibration.

[0078] Collect sample spectrum: Place the sample on the sample holder for testing. Single-point testing or multi-point scanning testing can be performed to obtain more accurate spectral data.

[0079] Specifically, after obtaining the one-dimensional Raman spectrum data of the solid waste sample in step S40, data preprocessing is performed, and the data preprocessing steps include:

[0080] De-noising: Fourier transform is used to remove noise from the one-dimensional Raman spectroscopy data of solid waste samples;

[0081] Baseline correction: The polynomial fitting method is used to remove the fluorescence background interference in the one-dimensional Raman spectroscopy data of solid waste samples;

[0082] Normalization: Normalize the 1D Raman spectroscopy data of solid waste samples to the same range;

[0083] Data conversion: Convert the one-dimensional Raman spectrum data of one-dimensional solid waste samples into two-dimensional image data.

[0084] The method for converting the one-dimensional Raman spectrum data of the one-dimensional solid waste sample into two-dimensional image data in the data conversion is to combine the Raman spectrum data into a color image, and the specific steps include:

[0085] Reshape: Reshape the one-dimensional Raman spectral data of length N into a two-dimensional matrix of M×M, where M is the integer closest to the square root of N. <M 2 When N=M, zero is added at the end to make N=M 2 ; When N is not a perfect square and N>M 2When truncated at the end, N = M 2 ; When N is not a perfect square and N=M 2 No zero padding or truncation is performed;

[0086] Combination: Select every 3 M×M two-dimensional matrices, corresponding to the 3 channels of RGB respectively, and combine the 3 M×M two-dimensional matrices into an M×M×3 three-dimensional matrix according to the channels to obtain color image data. When the remaining M×M two-dimensional matrices cannot be combined into 3, use a two-dimensional matrix with 0 elements to fill it.

[0087] Step S50: After the data preprocessing is completed, the Raman spectrum data is input into a determined version of the R-CNN-based target detection model to output the detection result.

[0088] The output detection result of the target detection model based on R-CNN in step S50 includes the category, composition and content of the solid waste sample detected; after the detection result is output, a detection report is automatically generated and the category, composition and content of the solid waste are analyzed; when there is an error between the detection result and the actual situation, an error analysis is performed, including:

[0089] Analyze the types of errors: such as misclassification, missed detection, and false detection, to understand in which categories the model performs poorly;

[0090] Visualize errors: Use tools such as confusion matrix and ROC curve to visualize errors and help understand the performance bottleneck of the model;

[0091] Case analysis: Check specific error samples, analyze their characteristics, and find out the causes of the errors;

[0092] In addition, by collecting more training data, especially data from categories where the model performs poorly, you can also use data augmentation techniques (such as rotation, scaling, flipping, etc.) to increase the diversity of training data, further train the model, and improve the model's detection capabilities.

[0093] In addition, the present invention also provides a solid waste detection device, please refer to Figure 2 , the solid waste detection device comprises:

[0094] Data acquisition module: used to obtain solid waste samples with publicly disclosed categories, components and contents and their corresponding Raman spectrum data from the Internet as training sets;

[0095] Model training and evaluation module: used to train the R-CNN-based target detection model using the training set. During the training process, the generalization ability and performance of the model are evaluated through cross-validation, and the model is optimized and the model version is determined based on the evaluation results.

[0096] Solid waste sample preparation and preprocessing module: used to prepare solid waste samples and perform sample preprocessing on solid waste samples after the model version is determined;

[0097] Raman spectrum scanning module: used to scan the sample using a Raman spectrometer after the sample pretreatment is completed, and obtain the one-dimensional Raman spectrum data of the solid waste sample and then perform data preprocessing;

[0098] Solid waste detection module: After data preprocessing, it inputs Raman spectroscopy data into a certain version of the R-CNN-based target detection model to output the detection results;

[0099] In the data acquisition module, the channels for obtaining the publicly disclosed categories, components and contents of solid waste samples and their corresponding Raman spectral data from the Internet include government agency websites, scientific research institution websites, environmental protection organization websites and public data packages published by enterprises on the Internet;

[0100] In the data acquisition module, solid waste samples with publicly available categories, components and contents are obtained from the Internet, including industrial solid waste, domestic waste, medical waste, construction waste, electronic waste and municipal sludge;

[0101] The output detection results of the R-CNN-based target detection model in the solid waste detection module include the category, composition and content of the detected solid waste samples.

[0102] The solid waste detection device provided by the present application adopts the solid waste detection method in the above embodiment, which can solve the technical problems of low accuracy and efficiency of traditional solid waste inspection methods. Compared with the prior art, the beneficial effects of the solid waste detection device provided by the present application are the same as the beneficial effects of the solid waste detection method provided by the above embodiment, and the other technical features in the solid waste detection device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0103] The present application provides a solid waste detection device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the solid waste detection method in the above-mentioned embodiment one.

[0104] Reference below Figure 3, which shows a schematic diagram of the structure of a solid waste detection device suitable for implementing the embodiment of the present application. The solid waste detection device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The solid waste detection device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0105] Figure 3 The solid waste detection device shown may include a processing device 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM) 1004. In RAM 1004, various programs and data required for the operation of the solid waste detection device are also stored. The processing device 1001, ROM 1002, and RAM 1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the solid waste detection device to communicate with other devices wirelessly or by wire to exchange data. Although the solid waste detection device with various systems is shown in the figure, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0106] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0107] The solid waste detection device provided by the present application adopts the solid waste detection method in the above embodiment, which can solve the technical problems of low accuracy and efficiency of traditional solid waste inspection methods. Compared with the prior art, the beneficial effects of the solid waste detection device provided by the present application are the same as the beneficial effects of the solid waste detection method provided by the above embodiment, and the other technical features in the solid waste detection device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0108] The various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0109] The present application also provides a computer program product, including a computer program, which implements the steps of the solid waste detection method as described above when executed by a processor.

[0110] The computer program product provided by the present application can solve the technical problems of low accuracy and efficiency of traditional solid waste inspection methods. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as the beneficial effects of the solid waste detection method provided by the above embodiment, which will not be repeated here.

[0111] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A solid waste detection method, characterized in that: The method comprises the following steps: Step S10: obtaining solid waste samples of publicly disclosed categories, components and contents and their corresponding Raman spectrum data from the Internet as a training set; Step S20: Use the training set to train the R-CNN-based object detection model. During the training process, the generalization ability and performance of the model are evaluated through cross-validation, and the model is optimized and the model version is determined based on the evaluation results. Step S30: after the model version is determined, a solid waste sample is prepared and sample preprocessing is performed on the solid waste sample; Step S40: After the sample preprocessing is completed, the sample is scanned using a Raman spectrometer to obtain one-dimensional Raman spectrum data of the solid waste sample and then perform data preprocessing; Step S50: After the data preprocessing is completed, the Raman spectrum data is input into a target detection model based on R-CNN of a determined version and the detection result is output; The channels for obtaining the solid waste samples and their corresponding Raman spectrum data of the public categories, components and contents from the Internet in step S10 include government agency websites, scientific research institution websites, environmental protection organization websites and public data packages published by enterprises on the Internet; The solid waste samples of publicly disclosed categories, compositions and contents obtained from the Internet in step S10 include industrial solid waste, domestic waste, medical waste, construction waste, electronic waste and municipal sludge; The output detection result of the R-CNN-based target detection model in step S50 includes the category, composition and content of the detected solid waste sample.

2. A solid waste detection method according to claim 1, characterized in that: The target detection model training step based on R-CNN in step S20 includes: Training set preparation: The one-dimensional Raman spectrum data of the solid waste samples were converted into image data. The image size was set to 227×227, and the category and composition information of each sample were annotated to obtain a labeled data set as the training set. Model construction: After obtaining the training set, the model is constructed using the AlexNet network architecture, including the input layer, convolution layer, pooling layer, fully connected layer, and output layer. The cross entropy loss function is used to calculate the classification loss, and the smooth L1 loss function is used to calculate the bounding box regression loss. The weighted sum of the classification loss and regression loss is used as the total loss function. The Adam optimizer is selected as the optimizer. Model training: After the model is built, use the training set to train the model and adjust the model parameters to optimize performance.

3. A solid waste detection method according to claim 1, characterized in that: In the step S20, the generalization ability and performance of the model are evaluated by cross-validation during the training process, using K-fold cross-validation, including: Dataset partitioning: The dataset is divided into a training set and a validation set, where the dataset is randomly divided into K equal subsets, each subset is used as a validation set in turn, and the remaining subsets are used as training sets; Model training and validation: When each subset is used as a validation set, use other subsets as training sets to train the model, use the validation set to evaluate the model performance, and record the results of each validation; Performance evaluation: Calculate the weighted average of all validation results to obtain the final generalization ability and performance evaluation of the model; Model optimization: When the final generalization ability and performance evaluation of the model do not meet expectations, adjust the model parameters and retrain until the optimal model parameter combination is obtained.

4. A solid waste detection method according to claim 1, characterized in that: The sample pretreatment of the solid waste sample in step S30 is physical treatment of the sample, including crushing, grinding and stirring.

5. A solid waste detection method according to claim 1, characterized in that: After obtaining the one-dimensional Raman spectrum data of the solid waste sample in step S40, data preprocessing is performed, and the data preprocessing step includes: De-noising: Fourier transform is used to remove noise from the one-dimensional Raman spectroscopy data of solid waste samples; Baseline correction: The polynomial fitting method is used to remove the fluorescence background interference in the one-dimensional Raman spectroscopy data of solid waste samples; Normalization: Normalize the 1D Raman spectroscopy data of solid waste samples to the same range; Data conversion: Convert the one-dimensional Raman spectrum data of solid waste samples into two-dimensional image data.

6. A solid waste detection method according to claim 5, characterized in that: The method for converting the one-dimensional Raman spectrum data of the one-dimensional solid waste sample into two-dimensional image data in the data conversion is to combine the Raman spectrum data into a color image, and the specific steps include: Reshape: Reshape the one-dimensional Raman spectral data of length N into a two-dimensional matrix of M×M, where M is the integer closest to the square root of N. <M 2 When N=M, zero is added at the end to make N=M 2 ; When N is not a perfect square and N>M 2 When truncated at the end, N = M 2 ; When N is not a perfect square and N = M 2 No zero padding or truncation is performed; Combination: Select every 3 M×M two-dimensional matrices, corresponding to the 3 channels of RGB respectively, and combine the 3 M×M two-dimensional matrices into an M×M×3 three-dimensional matrix according to the channels to obtain color image data. When the remaining M×M two-dimensional matrices cannot be combined into 3, use a two-dimensional matrix with 0 elements to fill it.

7. A solid waste detection method according to claim 1, characterized in that: After the test results are output in step S50, a test report is automatically generated and the type, composition and content of the solid waste are analyzed; when there is an error between the test results and the actual situation, an error analysis is performed.

8. A solid waste detection system, characterized in that: The solid waste detection system comprises: Data acquisition module: used to obtain solid waste samples with publicly disclosed categories, components and contents and their corresponding Raman spectrum data from the Internet as training sets; Model training and evaluation module: used to train the R-CNN-based target detection model using the training set. During the training process, the generalization ability and performance of the model are evaluated through cross-validation, and the model is optimized and the model version is determined based on the evaluation results. Solid waste sample preparation and preprocessing module: used to prepare solid waste samples and perform sample preprocessing on solid waste samples after the model version is determined; Raman spectrum scanning module: used to scan the sample using a Raman spectrometer after the sample pretreatment is completed, and obtain the one-dimensional Raman spectrum data of the solid waste sample and then perform data preprocessing; Solid waste detection module: After data preprocessing, it inputs Raman spectroscopy data into a certain version of the R-CNN-based target detection model to output the detection results; The channels for obtaining the solid waste samples and their corresponding Raman spectrum data of the public categories, components and contents from the Internet in the data acquisition module include government agency websites, scientific research institution websites, environmental protection organization websites and public data packages published by enterprises on the Internet; The data acquisition module acquires solid waste samples of publicly disclosed categories, components and contents from the Internet, including industrial solid waste, domestic waste, medical waste, construction waste, electronic waste and municipal sludge; The output detection results of the R-CNN-based target detection model in the solid waste detection module include the category, composition and content of the detected solid waste samples.

9. A solid waste detection device, characterized in that: The solid waste detection equipment comprises: A memory, a processor, and a solid waste detection program stored in the memory and executable on the processor, wherein the solid waste detection program, when executed by the processor, implements the solid waste detection method according to any one of claims 1 to 7.

10. A computer program product, characterized in that The computer program product comprises a solid waste detection program, and when the solid waste detection program is executed by a processor, the solid waste detection method according to any one of claims 1 to 7 is implemented.

Images