Intelligent management system and method for hazardous waste storage

Abstract

The invention relates to the technical field of hazardous waste management, in particular to an intelligent management system and method for hazardous waste storage. According to the method, a data set is constructed based on image features, sensor data features and spectral features of the hazardous wastes, features in the data set are screened, features related to the danger of the hazardous wastes are screened out, the screened data set is adopted to train a machine learning algorithm model, and the risk of the hazardous wastes is obtained. The method comprises the following steps: constructing a hazardous waste danger level evaluation model, evaluating the danger of hazardous wastes by adopting the hazardous waste danger level evaluation model, defining hazardous wastes with high, medium and low dangers, allocating a vehicle dispatching scheme according to the danger of the hazardous wastes, and allocating a high-safety vehicle dispatching scheme to the hazardous wastes with high dangers. The low-risk hazardous waste is distributed to a low-safety vehicle dispatching scheme; management of the hazardous waste is rationalized, and economic losses caused by leakage of the hazardous waste and the like are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of hazardous waste management, and in particular to an intelligent management system and method for hazardous waste storage. Background Art

[0002] With the acceleration of industrialization, the amount of hazardous waste generated has increased year by year. These wastes seriously threaten the environment and human health due to their toxicity, corrosiveness, flammability, reactivity and infectivity. At present, the treatment and assessment of hazardous waste mainly rely on manual judgment and laboratory analysis, which is not only time-consuming and labor-intensive, but also prone to misjudgment. Existing technologies such as image recognition, sensor fusion and spectral analysis each provide certain support for hazardous waste identification, but a single technology is difficult to fully and accurately assess the hazard of waste. Although image recognition technology can quickly obtain surface features of objects, it is difficult to deeply analyze the composition of materials; sensor fusion technology can integrate information from multiple sensors, but there are still challenges in data integration and analysis; spectral analysis technology can provide information on the chemical composition of substances, but it is limited by equipment complexity and cost in field applications. Therefore, how to effectively integrate these technologies to achieve a comprehensive assessment of hazardous waste is the main challenge in the current field.

[0003] In addition, at present, the management of hazardous waste mainly involves multiple links such as transportation, warehousing, outbound and storage. In these links, how to manage hazardous waste efficiently and safely is a major challenge. Current technologies are mostly focused on optimizing a single link, such as optimizing transportation routes or maximizing the use of storage space. However, there is a lack of a comprehensive management strategy that comprehensively considers the correlation between various links, resulting in inefficiency, waste of resources and safety hazards in actual operations. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention aims to provide an intelligent management system and method for hazardous waste storage, so as to achieve comprehensive assessment and intelligent management of the hazard of hazardous waste.

[0005] In order to solve the above problems, the present invention adopts the following technical solutions: On the one hand, the present invention provides an intelligent management system for hazardous waste storage, including an image feature extraction module, a sensor data extraction module, a spectral feature extraction module, a hazard level assessment module and a hazardous material management module.

[0006] The image feature extraction module is used to extract the appearance image of the hazardous waste using a high-resolution camera, and to extract features from the appearance image of the hazardous waste using a convolutional neural network to obtain image features of the hazardous waste.

[0007] The sensor data extraction module is used to collect the physical and chemical characteristic parameters of hazardous waste and the environmental parameters of the environment in which the hazardous waste is located through sensors, and perform data fusion on the physical and chemical characteristic parameters and the environmental parameters to obtain sensor data characteristics of the hazardous waste.

[0008] The spectral feature extraction module is used to collect spectral data of hazardous waste samples through a spectrometer, and extract characteristic peaks in the spectral data to obtain spectral features of the hazardous waste.

[0009] The hazard level assessment module is used to construct a data set based on the image features, sensor data features and spectral features of hazardous waste, and perform feature screening on the features in the data set to screen out features related to the hazard of hazardous waste. The screened data set is used to train the machine learning algorithm model to construct a hazardous waste hazard level assessment model, and the hazardous waste hazard level assessment model is used to assess the hazard of hazardous waste and define hazardous wastes with high, medium and low hazard levels.

[0010] The hazardous waste management module is used to allocate vehicle dispatching plans according to the hazard level of hazardous wastes, allocating high-hazard hazardous wastes to high-safety vehicle dispatching plans and allocating low-hazard hazardous wastes to low-safety vehicle dispatching plans.

[0011] As an implementable method, it also includes a vehicle dispatching plan safety assessment module; the vehicle dispatching plan safety assessment module is used to assess the safety level of the vehicle dispatching plan based on the work concentration index of the transportation personnel, the vehicle safety index and the transportation route safety index.

[0012] As an implementable method, the safety level of the vehicle dispatching scheme is calculated by the following formula: W = α × Wp + β × Wv + γ × Wr Among them, W is the safety level of the vehicle dispatch plan, Wp is the work concentration index of the transport personnel; Wv is the vehicle safety index; Wr is the transportation route safety index; α, β, and γ are the weight coefficients of each index respectively.

[0013] As an implementable method, the image features of the hazardous waste include: A high-resolution camera is used to capture the appearance of hazardous waste.

[0014] The extracted appearance images are normalized, resized, and augmented.

[0015] The processed appearance image is input into the convolutional neural network, the feature map of the appearance image is extracted through the convolutional layer of the convolutional neural network, the dimension of the feature map is reduced through the pooling layer, the features of the feature map are comprehensively analyzed through the fully connected layer, and the image features of hazardous waste are output through the Softmax layer.

[0016] As an implementable method, the data fusion of the physical and chemical characteristic parameters and the environmental parameters includes: The physical and chemical characteristic parameters of hazardous waste and the environmental parameters of the environment in which the hazardous waste is located are collected regularly through sensors at the same time.

[0017] The physical and chemical characteristic parameters of hazardous wastes collected at the same time and the environmental parameters of the environment in which the hazardous wastes are located are fused by weighted averaging, Kalman filtering or Bayesian estimation. The mathematical model of data fusion is expressed as follows:

[0018] in, is the estimated value after fusion, x i is the measurement value of the i-th sensor, w i is the weight corresponding to the sensor, satisfying .

[0019] As an implementable method, the spectral characteristics of the hazardous waste include: The spectral data of hazardous waste samples are collected by a spectrometer, with wavelength as the horizontal axis and light intensity as the vertical axis.

[0020] The spectral data were baseline corrected, smoothed and denoised.

[0021] Characteristic peaks are extracted from the processed spectral data, and the composition and concentration of hazardous waste are determined by the position and intensity of the characteristic peaks.

[0022] Partial least squares regression was used to establish a quantitative relationship between the spectral data and the concentration of the components:

[0023] Among them, C is the concentration matrix of the components, X is the spectral data matrix, B is the regression coefficient matrix, and E is the error matrix. The value of B is determined by the training sample set to achieve quantitative analysis.

[0024] As an implementable method, the machine learning algorithm model includes a random forest and a support vector machine; and a cross-validation method is used to evaluate the generalization ability of the model:

[0025] Among them, f(x) is the decision function, α iis the weight coefficient, K(xi,x) is the kernel function, and b is the bias.

[0026] As an implementable method, the hazardous waste management module is used to match the weighed in and out quantities according to the hazardousness of the hazardous waste; and to match the storage area of ​​the hazardous waste according to the hazardousness of the hazardous waste.

[0027] In another aspect, the present invention provides an intelligent management method for hazardous waste storage, comprising: A high-resolution camera is used to extract the appearance image of hazardous waste, and a convolutional neural network is used to extract features of the appearance image of hazardous waste to obtain the image features of hazardous waste.

[0028] The physical and chemical characteristic parameters of hazardous wastes and the environmental parameters of the environment in which the hazardous wastes are located are collected through sensors, and the physical and chemical characteristic parameters and environmental parameters are fused to obtain the sensor data characteristics of the hazardous wastes.

[0029] The spectral data of the hazardous waste samples are collected by a spectrometer, and the characteristic peaks in the spectral data are extracted to obtain the spectral characteristics of the hazardous waste.

[0030] A data set is constructed based on the image features, sensor data features and spectral features of hazardous waste, and the features in the data set are screened to screen out the features related to the hazard of hazardous waste. The screened data set is used to train the machine learning algorithm model, and a hazardous waste hazard level assessment model is constructed. The hazardous waste hazard level assessment model is used to assess the hazard of hazardous waste and define hazardous wastes with high, medium and low hazard levels.

[0031] The dispatching plan is determined based on the hazard level of hazardous wastes. High-risk hazardous wastes are assigned a high-safety dispatching plan, while low-risk hazardous wastes are assigned a low-safety dispatching plan.

[0032] As an implementable method, it also includes matching the weighed quantities entering and leaving the warehouse according to the hazardousness of the hazardous waste; and matching the storage area of ​​the hazardous waste according to the hazardousness of the hazardous waste.

[0033] The beneficial effects of the present invention are as follows: the present invention utilizes image recognition to collect the appearance characteristics of waste, sensors to collect the physical and chemical properties of waste, and spectroscopy to collect the photosensitive chemical characteristics of waste, and comprehensively analyzes the hazard of hazardous waste; based on its hazard, different specifications are made for vehicle dispatch scenarios, warehousing, outbound warehousing and partitioned storage, to build an intelligent hazardous waste storage management system, rationalize the management of hazardous waste, and reduce the economic losses caused by leakage of hazardous waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is a schematic diagram of an intelligent management system for hazardous waste storage according to the present invention.

[0035] Figure 2 This is a flow chart of an intelligent management method for hazardous waste storage according to the present invention. DETAILED DESCRIPTION

[0036] The present invention is further described in detail below in conjunction with specific embodiments.

[0037] It should be noted that these embodiments are only used to illustrate the present invention rather than to limit the present invention. Simple improvements to the method based on the concept of the present invention all fall within the scope of protection claimed by the present invention.

[0038] See also Figure 1 , which is an intelligent management system for hazardous waste storage, including an image feature extraction module 100, a sensor data extraction module 200, a spectral feature extraction module 300, a hazard level assessment module 400 and a hazardous material management module 500.

[0039] The image feature extraction module 100 is used to extract the appearance image of the hazardous waste using a high-resolution camera, and to extract features from the appearance image of the hazardous waste using a convolutional neural network to obtain image features of the hazardous waste.

[0040] This embodiment provides an implementation method of the above process: A high-resolution camera is used to capture the appearance of hazardous waste.

[0041] The implementation of image recognition relies on the deployment of high-resolution cameras to obtain images of the appearance of hazardous waste. These cameras should have sufficient resolution and sensitivity to ensure that the subtle features of the waste can be captured. Image data can be collected under different lighting conditions to ensure the robustness of the recognition algorithm.

[0042] The extracted appearance images are normalized, resized, and data augmented to improve the generalization ability of the model.

[0043] The processed appearance image is input into the convolutional neural network CNN, the feature map of the appearance image is extracted through the convolutional layer of the convolutional neural network, the dimension of the feature map is reduced through the pooling layer, the features of the feature map are comprehensively analyzed through the fully connected layer, and the image features of the hazardous waste are output through the Softmax layer.

[0044] Through its multi-layer structure, CNN can effectively extract local and global features in images. Commonly used network structures include ResNet, VGG, etc. These networks can be optimized through transfer learning to adapt to the specific identification tasks of hazardous waste.

[0045] The sensor data extraction module 200 is used to collect the physical and chemical characteristic parameters of hazardous wastes and the environmental parameters of the environment in which the hazardous wastes are located through sensors, and perform data fusion on the physical and chemical characteristic parameters and the environmental parameters to obtain sensor data characteristics of the hazardous wastes.

[0046] This embodiment provides an implementation method of the above process: The physical and chemical characteristic parameters of hazardous waste and the environmental parameters of the environment in which the hazardous waste is located are collected regularly through sensors at the same time.

[0047] Using a variety of sensors, multiple physical and chemical parameters of hazardous waste are monitored to provide comprehensive environmental perception and data support. The module integrates devices such as temperature sensors, gas sensors, and humidity sensors, each of which is responsible for collecting specific types of data. For example, temperature sensors are used to detect temperature changes in waste, gas sensors are used to detect harmful gases that may be released, and humidity sensors monitor changes in ambient humidity.

[0048] The physical and chemical characteristic parameters of hazardous wastes collected at the same time and the environmental parameters of the environment in which the hazardous wastes are located are fused by weighted averaging, Kalman filtering or Bayesian estimation. The mathematical model of data fusion is expressed as follows:

[0049] in, is the estimated value after fusion, x i is the measurement value of the i-th sensor, w i is the weight corresponding to the sensor, satisfying By selecting weights reasonably, the fused results can more accurately reflect the real situation of hazardous waste.

[0050] The spectral feature extraction module 300 is used to collect spectral data of hazardous waste samples through a spectrometer, and extract characteristic peaks in the spectral data to obtain spectral features of the hazardous waste.

[0051] This embodiment provides an implementation method of the above process: Place the hazardous waste sample to be analyzed in the measurement area of ​​the spectrometer to ensure the homogeneity and representativeness of the sample.

[0052] The spectral data of hazardous waste samples are collected by a spectrometer, with wavelength as the horizontal axis and light intensity as the vertical axis.

[0053] The spectral data were baseline corrected, smoothed and denoised.

[0054] Characteristic peaks are extracted from the processed spectral data, and the composition and concentration of hazardous waste are determined by the position and intensity of the characteristic peaks.

[0055] Partial least squares regression was used to establish a quantitative relationship between the spectral data and the concentration of the components:

[0056] Among them, C is the concentration matrix of the components, X is the spectral data matrix, B is the regression coefficient matrix, and E is the error matrix. The value of B is determined by the training sample set to achieve quantitative analysis.

[0057] The hazard level assessment module 400 is used to construct a data set based on the image features, sensor data features and spectral features of hazardous waste, and perform feature screening on the features in the data set to screen out features related to the hazard of hazardous waste, use the screened data set to train the machine learning algorithm model, build a hazardous waste hazard level assessment model, and use the hazardous waste hazard level assessment model to assess the hazard of hazardous waste and define hazardous wastes with high, medium and low hazard levels.

[0058] Feature extraction includes: Extract features such as shape, color, and texture from image data.

[0059] Extract sensor time series features from sensor data.

[0060] Extract characteristic peaks of key chemical components from spectral data.

[0061] Apply statistical analysis and machine learning methods (such as LASSO regression) to select the features most relevant to risk. Use feature importance ranking methods (such as information gain) to filter out redundant features.

[0062] Machine learning algorithm models include random forest and support vector machine.

[0063] Random forests are suitable for processing high-dimensional data and nonlinear relationships, and can provide importance ranking of features.

[0064] Support vector machines (SVMs) are suitable for classification tasks, especially when the data distribution is uneven.

[0065] And the cross-validation method is used to evaluate the generalization ability of the model:

[0066] Among them, f(x) is the decision function, α i is the weight coefficient, K(xi,x) is the kernel function, and b is the bias.

[0067] The output results of each module are comprehensively evaluated to form the final hazard level. For example, through feature extraction, morphological characteristics (such as area, perimeter), temperature (25°C), pH value (7.2) and main chemical components (such as phenol concentration of 3mg / L) are obtained. Using the pre-trained hazardous waste hazard level assessment model, the hazard coefficient of the waste is calculated to be 0.85 (in the range of 0 to 1), and its hazard level is determined to be "high" based on the classification threshold. By setting the hazard index score, it corresponds to high, medium and low hazards.

[0068] The hazardous waste management module 500 is used to allocate vehicle dispatching plans according to the hazardousness of hazardous wastes, allocating high-hazard hazardous wastes to high-safety vehicle dispatching plans and allocating low-hazard hazardous wastes to low-safety vehicle dispatching plans.

[0069] For example, according to the hazard index, based on the personnel work concentration index, vehicle safety index and transportation route safety index, the vehicle dispatch decision plan is matched. For example, if a person has been working for 4 hours, his work concentration is average, the planned vehicle is a low-safety vehicle, and the route is a low-safety route, then the low-hazard factor waste will be matched.

[0070] The amount of weighing for inbound and outbound storage is also matched according to the hazard level of hazardous waste. For example, for highly hazardous waste, during the outbound and inbound process, the single weighing, transportation and transfer must be within certain indicators to ensure safety.

[0071] The storage area of ​​hazardous waste is also matched according to the hazard of the hazardous waste. For example, high-hazard waste should be placed in a high-hazard level partition and equipped with relevant safety measures.

[0072] The system also includes a vehicle dispatching scheme safety assessment module 600; the vehicle dispatching scheme safety assessment module 60 is used to assess the safety level of the vehicle dispatching scheme based on the work concentration index of the transport personnel, the vehicle safety index and the transport route safety index.

[0073] The safety level of the vehicle dispatching plan is calculated by the following formula: W = α × Wp + β × Wv + γ × Wr Among them, W is the safety level of the vehicle dispatch plan, Wp is the work concentration index of the transport personnel; Wv is the vehicle safety index; Wr is the transportation route safety index; α, β, and γ are the weight coefficients of each index respectively.

[0074] Transport personnel’s work concentration indicators: Working hours: The start and end times of transport personnel’s work can be recorded through a time clock system or GPS positioning device, so as to calculate their cumulative working hours.

[0075] Working status: Through wearable devices such as smart bracelets, the physiological indicators of transport personnel, such as heart rate and fatigue, are monitored to assess their current status. These data can be transmitted to the central database in real time via wireless networks. Work concentration indicators are measured based on working hours and working status.

[0076] Vehicle safety indicators: Vehicle status: Install sensors on the vehicle to detect the status of key components of the vehicle in real time, such as the brake system, tire pressure, engine performance, etc. This data can be obtained through the on-board diagnostic system (OBD) interface.

[0077] Historical maintenance records: Extract the vehicle's historical maintenance records from the vehicle maintenance management system, including recent repair and maintenance information, to assess the overall safety of the vehicle.

[0078] Vehicle safety indicators are measured based on vehicle condition and historical maintenance records.

[0079] Transport route safety indicators: Route environment: Use geographic information systems (GIS) and traffic monitoring data to analyze the surrounding environment of transportation routes, including residential density, road conditions, traffic flow, etc.

[0080] Real-time traffic conditions: Through the data interface with the transportation department, obtain real-time traffic information, such as traffic congestion, road construction, weather conditions, etc., to ensure the safety of transportation routes.

[0081] Transportation route safety indicators are measured by route environment and real-time road conditions.

[0082] Assume that a transporter has been working for 4 hours straight, and the current heart rate monitoring shows that he is in a general fatigue state. The vehicle's sensor data shows that the brake system is worn out and the last maintenance has exceeded the recommended cycle. At the same time, the planned transport route passes through multiple residential areas, and real-time traffic data shows that there are multiple construction sites along the route and that it is transported at night.

[0083] Based on the above data, the system calculates: Wp=0.6 (average work concentration) Wv=0.4 (vehicle safety is low) Wr=0.3 (low route safety) Through the calculation formula, the safety level W of the vehicle dispatching plan is calculated as: W=0.4×0.6+0.3×0.4+0.3×0.3=0.48.

[0084] According to the safety level W of the dispatching scheme, the system matches it to a low-risk hazardous waste transportation task to ensure safety. The weights of the above schemes can be defined based on historical data schemes.

[0085] See also Figure 2 , which is an intelligent management method for hazardous waste storage, including: S100, using a high-resolution camera to extract the appearance image of the hazardous waste, and using a convolutional neural network to extract features of the appearance image of the hazardous waste to obtain image features of the hazardous waste.

[0086] This embodiment provides an implementation method of the above process: A high-resolution camera is used to capture the appearance of hazardous waste.

[0087] The implementation of image recognition relies on the deployment of high-resolution cameras to obtain images of the appearance of hazardous waste. These cameras should have sufficient resolution and sensitivity to ensure that the subtle features of the waste can be captured. Image data can be collected under different lighting conditions to ensure the robustness of the recognition algorithm.

[0088] The extracted appearance images are normalized, resized, and data augmented to improve the generalization ability of the model.

[0089] The processed appearance image is input into the convolutional neural network CNN, the feature map of the appearance image is extracted through the convolutional layer of the convolutional neural network, the dimension of the feature map is reduced through the pooling layer, the features of the feature map are comprehensively analyzed through the fully connected layer, and the image features of the hazardous waste are output through the Softmax layer.

[0090] Through its multi-layer structure, CNN can effectively extract local and global features in images. Commonly used network structures include ResNet, VGG, etc. These networks can be optimized through transfer learning to adapt to the specific identification tasks of hazardous waste.

[0091] S200, collecting physical and chemical characteristic parameters of hazardous waste and environmental parameters of the environment in which the hazardous waste is located through sensors, and fusing the physical and chemical characteristic parameters and environmental parameters to obtain sensor data characteristics of the hazardous waste.

[0092] This embodiment provides an implementation method of the above process: The physical and chemical characteristic parameters of hazardous waste and the environmental parameters of the environment in which the hazardous waste is located are collected regularly through sensors at the same time.

[0093] Using a variety of sensors, multiple physical and chemical parameters of hazardous waste are monitored to provide comprehensive environmental perception and data support. The module integrates devices such as temperature sensors, gas sensors, and humidity sensors, each of which is responsible for collecting specific types of data. For example, temperature sensors are used to detect temperature changes in waste, gas sensors are used to detect harmful gases that may be released, and humidity sensors monitor changes in ambient humidity.

[0094] The physical and chemical characteristic parameters of hazardous wastes collected at the same time and the environmental parameters of the environment in which the hazardous wastes are located are fused by weighted averaging, Kalman filtering or Bayesian estimation. The mathematical model of data fusion is expressed as:

[0095] in, is the estimated value after fusion, x i is the measurement value of the i-th sensor, w i is the weight corresponding to the sensor, satisfying By selecting weights reasonably, the fused results can more accurately reflect the real situation of hazardous waste.

[0096] S300, collecting spectral data of a hazardous waste sample through a spectrometer, and extracting characteristic peaks in the spectral data to obtain spectral characteristics of the hazardous waste.

[0097] This embodiment provides an implementation method of the above process: Place the hazardous waste sample to be analyzed in the measurement area of ​​the spectrometer to ensure the homogeneity and representativeness of the sample.

[0098] The spectral data of hazardous waste samples are collected by a spectrometer, with wavelength as the horizontal axis and light intensity as the vertical axis.

[0099] The spectral data were baseline corrected, smoothed and denoised.

[0100] Characteristic peaks are extracted from the processed spectral data, and the composition and concentration of hazardous waste are determined by the position and intensity of the characteristic peaks.

[0101] Partial least squares regression was used to establish a quantitative relationship between the spectral data and the concentration of the components:

[0102] Among them, C is the concentration matrix of the components, X is the spectral data matrix, B is the regression coefficient matrix, and E is the error matrix. The value of B is determined by the training sample set to achieve quantitative analysis.

[0103] S400. Construct a data set based on the image features, sensor data features and spectral features of hazardous waste, and perform feature screening on the features in the data set to screen out features related to the hazard of hazardous waste. Use the screened data set to train the machine learning algorithm model, construct a hazardous waste hazard level assessment model, and use the hazardous waste hazard level assessment model to assess the hazard of hazardous waste, and define high, medium and low hazardous wastes.

[0104] Feature extraction includes: Extract features such as shape, color, and texture from image data.

[0105] Extract sensor time series features from sensor data.

[0106] Extract characteristic peaks of key chemical components from spectral data.

[0107] Apply statistical analysis and machine learning methods (such as LASSO regression) to select the features most relevant to risk. Use feature importance ranking methods (such as information gain) to filter out redundant features.

[0108] Machine learning algorithm models include random forest and support vector machine.

[0109] Random forests are suitable for processing high-dimensional data and nonlinear relationships, and can provide importance ranking of features.

[0110] Support vector machines (SVMs) are suitable for classification tasks, especially when the data distribution is uneven.

[0111] And the cross-validation method is used to evaluate the generalization ability of the model:

[0112] Among them, f(x) is the decision function, α i is the weight coefficient, K(xi,x) is the kernel function, and b is the bias.

[0113] The output results of each module are comprehensively evaluated to form the final hazard level.

[0114] S500. Distribute vehicle dispatch plans according to the hazard level of hazardous wastes, allocate highly hazardous wastes to a highly secure vehicle dispatch plan, and allocate low hazardous wastes to a low secure vehicle dispatch plan.

[0115] The amount of weighing for inbound and outbound storage is also matched according to the hazard level of hazardous waste. For example, for highly hazardous waste, during the outbound and inbound process, the importance of single weighing, transportation and transfer must be within certain indicators to ensure safety.

[0116] The storage area of ​​hazardous waste is also matched according to the hazard of the hazardous waste. For example, high-hazard waste should be placed in a high-hazard level partition and equipped with relevant safety measures.

[0117] The method of the present invention also includes evaluating the safety level of the vehicle dispatching plan based on the work concentration index of the transport personnel, the vehicle safety index and the transport route safety index.

[0118] The safety level of the vehicle dispatching plan is calculated by the following formula: W = α × Wp + β × Wv + γ × Wr Among them, W is the safety level of the vehicle dispatch plan, Wp is the work concentration index of the transport personnel; Wv is the vehicle safety index; Wr is the transportation route safety index; α, β, and γ are the weight coefficients of each index respectively.

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described with reference to the preferred embodiments of the present invention, it should be understood by those skilled in the art that various changes may be made in form and details without departing from the spirit and scope of the present invention as defined in the appended claims.

Claims

1. An intelligent management system for hazardous waste storage, characterized in that: It includes an image feature extraction module, a sensor data extraction module, a spectral feature extraction module, a hazard level assessment module and a hazardous material management module; The image feature extraction module is used to extract the appearance image of the hazardous waste using a high-resolution camera, and to extract features of the appearance image of the hazardous waste using a convolutional neural network to obtain image features of the hazardous waste; The sensor data extraction module is used to collect the physical and chemical characteristic parameters of the hazardous waste and the environmental parameters of the environment in which the hazardous waste is located through sensors, and perform data fusion on the physical and chemical characteristic parameters and the environmental parameters to obtain the sensor data characteristics of the hazardous waste; The spectral feature extraction module is used to collect spectral data of hazardous waste samples through a spectrometer, and extract characteristic peaks in the spectral data to obtain spectral features of the hazardous waste; The hazard level assessment module is used to construct a data set based on the image features, sensor data features and spectral features of hazardous waste, and perform feature screening on the features in the data set to screen out features related to the hazard of hazardous waste, use the screened data set to train the machine learning algorithm model, build a hazardous waste hazard level assessment model, and use the hazardous waste hazard level assessment model to assess the hazard of hazardous waste and define hazardous wastes with high, medium and low hazard levels; The hazardous waste management module is used to allocate vehicle dispatching plans according to the hazard level of hazardous wastes, allocating high-hazard hazardous wastes to high-safety vehicle dispatching plans and allocating low-hazard hazardous wastes to low-safety vehicle dispatching plans.

2. The intelligent management system for hazardous waste storage according to claim 1 is characterized in that: It also includes a vehicle dispatching plan safety assessment module; the vehicle dispatching plan safety assessment module is used to assess the safety level of the vehicle dispatching plan based on the work concentration index of the transport personnel, the vehicle safety index and the transport route safety index.

3. The intelligent management system for hazardous waste storage according to claim 2 is characterized in that: The safety level of the vehicle dispatching scheme is calculated by the following formula: W = α × Wp + β × Wv + γ × Wr Among them, W is the safety level of the vehicle dispatch plan, Wp is the work concentration index of the transport personnel; Wv is the vehicle safety index; Wr is the transportation route safety index; α, β, and γ are the weight coefficients of each index respectively.

4. The intelligent management system for hazardous waste storage according to claim 3 is characterized in that: The image features of hazardous waste include: Use high-resolution cameras to capture images of the appearance of hazardous waste; Normalize, resize and perform data augmentation on the extracted appearance images; The processed appearance image is input into the convolutional neural network, the feature map of the appearance image is extracted through the convolutional layer of the convolutional neural network, the dimension of the feature map is reduced through the pooling layer, the features of the feature map are comprehensively analyzed through the fully connected layer, and the image features of hazardous waste are output through the Softmax layer.

5. The intelligent management system for hazardous waste storage according to claim 4 is characterized in that: The data fusion of the physical and chemical characteristic parameters and the environmental parameters includes: Regularly collect the physical and chemical characteristic parameters of hazardous waste and the environmental parameters of the environment where the hazardous waste is located at the same time through sensors; The physical and chemical characteristic parameters of hazardous wastes collected at the same time and the environmental parameters of the environment in which the hazardous wastes are located are fused by weighted averaging, Kalman filtering or Bayesian estimation. The mathematical model of data fusion is expressed as follows: in, is the estimated value after fusion, x i is the measurement value of the i-th sensor, w i is the weight corresponding to the sensor, satisfying .

6. The intelligent management system for hazardous waste storage according to claim 5 is characterized in that: The spectral characteristics of the hazardous waste include: The spectrometer is used to collect spectral data of hazardous waste samples, with wavelength as the horizontal axis and light intensity as the vertical axis; Baseline correction, smoothing and denoising were performed on the spectral data; Extract characteristic peaks from the processed spectral data, and determine the composition and concentration of hazardous waste by the position and intensity of the characteristic peaks; Partial least squares regression was used to establish a quantitative relationship between the spectral data and the concentration of the components: Among them, C is the concentration matrix of the components, X is the spectral data matrix, B is the regression coefficient matrix, and E is the error matrix. The value of B is determined by the training sample set to achieve quantitative analysis.

7. The intelligent management system for hazardous waste storage according to claim 6 is characterized in that: The machine learning algorithm model includes random forest and support vector machine; and the cross-validation method is used to evaluate the generalization ability of the model: Among them, f(x) is the decision function, α i is the weight coefficient, K(xi,x) is the kernel function, and b is the bias.

8. The intelligent management system for hazardous waste storage according to claim 7 is characterized in that: The hazardous waste management module is used to match the weighed in and out quantities according to the hazardousness of the hazardous waste; and to match the storage area of ​​the hazardous waste according to the hazardousness of the hazardous waste.

9. An intelligent management method for hazardous waste storage, characterized in that: include: A high-resolution camera is used to extract the appearance image of hazardous waste, and a convolutional neural network is used to extract features of the appearance image of hazardous waste to obtain image features of hazardous waste; The physical and chemical characteristic parameters of hazardous waste and the environmental parameters of the environment in which the hazardous waste is located are collected through sensors, and the physical and chemical characteristic parameters and environmental parameters are fused to obtain sensor data characteristics of hazardous waste; Collecting spectral data of hazardous waste samples through a spectrometer, and extracting characteristic peaks in the spectral data to obtain spectral characteristics of the hazardous waste; A data set is constructed based on the image features, sensor data features, and spectral features of hazardous waste, and the features in the data set are screened to screen out features related to the hazard of hazardous waste. The screened data set is used to train the machine learning algorithm model, and a hazardous waste hazard level assessment model is constructed. The hazardous waste hazard level assessment model is used to assess the hazard of hazardous waste and define hazardous wastes with high, medium, and low hazard levels. The dispatching plan is determined based on the hazard level of hazardous wastes. High-risk hazardous wastes are assigned a high-safety dispatching plan, while low-risk hazardous wastes are assigned a low-safety dispatching plan.

10. The intelligent management method for hazardous waste storage according to claim 9, characterized in that: It also includes matching the weighed quantities entering and leaving the warehouse according to the hazardous nature of the hazardous waste, and matching the storage areas of hazardous waste according to the hazardous nature of the hazardous waste.

Images