Solid waste resourceful treatment method for intelligent environmental protection

Abstract

The invention relates to the technical field of waste treatment, and particularly discloses an intelligent and environment-friendly solid waste resourceful treatment method which comprises the following steps: feeding mixed solid waste into a magnet crushing device for vibrating and crushing, and in the crushing process, acquiring material crushing state images at a first time point and a second time point by using a high-speed camera; meanwhile, an electromagnet working current signal between two time points is collected through a current sensor, and then a deep learning algorithm is introduced to extract material crushing state change characteristics from the material crushing state images at the two time points; and intelligent adjustment of the crushing force is realized on the basis of a nonlinear interaction response relationship between a working current time sequence fluctuation mode of the electromagnet and the change of a material crushing state. And after crushing is completed, generated gas, liquid and solid waste particles are classified, treated and recycled. In this way, the dynamic change of the solid waste in the crushing process can be mastered in time, and the crushing force can be rapidly adjusted to optimize the crushing effect.

Description

Technical Field

[0001] This application relates to the technical field of waste treatment, and more specifically, to a method for the resource treatment of solid waste for intelligent environmental protection. Background Art

[0002] With the acceleration of the urbanization process and the improvement of the consumption level, the output and complexity of mixed solid waste have increased significantly. Traditional landfill and incineration treatment methods can no longer meet the requirements of resource utilization and environmental protection. In the prior art, the invention patent with the publication number of CN114247735B proposes a method for the resource treatment of mixed solid waste. After the solid waste entering the device is crushed by a magnet crushing device, the harmful gases in the device are absorbed by the device, and the solid waste is sorted by a large particle treatment part. After treatment, it mainly includes plastic substances, iron substances, and other metal materials, etc. The solid waste is treated by a small particle treatment part, and the treated substances are mainly composed of glass substances and sediment, etc. At the same time, the wastewater generated during the treatment process is recycled to reduce water resource waste, achieving resource recovery and utilization while treating solid waste and avoiding secondary pollution.

[0003] However, during the crushing process of solid waste, the prior art uses a dynamic clustering algorithm to classify and optimize the crushing force of solid waste. By initially setting the crushing force of a conventional solid waste sample, and then dynamically clustering all the sample crushing force data, the crushing force is dynamically optimized and adjusted through iterative calculations. However, in actual operation, the selection of the initial crushing force condensation point highly depends on manual experience, and the dynamic clustering algorithm needs to standardize and repeatedly iterate the sample data, which is easily affected by the initial setting value and outliers, resulting in uncertainty in the optimization result. At the same time, if the sample data volume is large or the dimension is high, it will greatly increase the computational complexity, reduce the processing efficiency, and easily cause response delays in crushing scenarios with high real-time requirements, making it difficult to meet the continuous production requirements. For example, when treating a mixed material of fluorescent tubes and construction waste, the strength difference between brittle glass fragments and metal hinges can reach 3 - 5 times. The clustering algorithm needs to go through multiple iterations to converge, which may lead to a lag in the adjustment of the crushing force behind the actual working conditions, affecting the crushing effect and resource utilization efficiency.

[0004] Therefore, there is an expectation for a method and system for the resource treatment of solid waste for intelligent environmental protection. Summary of the Invention

[0005] To solve the above technical problems, this application is proposed. An embodiment of this application provides a method for resource treatment of solid waste for intelligent environmental protection. The method feeds mixed solid waste into a magnet crushing device for vibrating crushing. During the crushing process, a high-speed camera is used to collect images of the material crushing state at the first time point and the second time point. At the same time, an electromagnetic working current signal between the two time points is collected through a current sensor. Then, a deep learning algorithm is introduced to extract the characteristics of the change in the material crushing state from the images of the material crushing state at the two time points. Based on the non-linear interaction response relationship between the electromagnetic working current time series fluctuation pattern and the change in the material crushing state, intelligent adjustment of the crushing force is realized. After the crushing is completed, the generated gas, liquid, and solid waste particles are classified and recycled, so as to achieve efficient resource utilization of solid waste. In this way, the dynamic changes of solid waste during the crushing process can be grasped in a timely manner, and the crushing force can be quickly adjusted to optimize the crushing effect.

[0006] According to one aspect of this application, there is provided a method for resource treatment of solid waste for intelligent environmental protection, which includes:

[0007] Feeding the mixed solid waste into a magnet crushing device for vibrating crushing;

[0008] During the crushing process, a high-speed camera is used to collect images of the material crushing state at the first time point and the second time point, and at the same time, an electromagnetic working current signal of the magnet crushing device between the first time point and the second time point is collected through a current sensor;

[0009] Inputting the image of the material crushing state at the first time point, the image of the material crushing state at the second time point, and the electromagnetic working current signal into a crushing force dynamic control module based on visual feedback to obtain a crushing force adjustment instruction;

[0010] Based on the crushing force adjustment instruction, adjusting the electromagnetic current of the magnet crushing device;

[0011] After the crushing is completed, the gas generated in the magnet crushing device is hermetically sucked, stored after being adsorbed by activated carbon, the generated liquid is recycled and stored after multi-stage filtration and chemical treatment, the large particle solid waste generated is classified and recycled after being vibrated, floated, magnetically separated, and heated, and the small particle solid waste generated is washed, filtered, and collected.

[0012] Compared with the prior art, the solid waste resource treatment method for intelligent environmental protection provided by this application sends mixed solid waste into a magnet crushing device for vibrating and crushing. During the crushing process, high-speed cameras are used to collect images of the material crushing state at the first time point and the second time point. At the same time, an electromagnetic working current signal between the two time points is collected through a current sensor. Then, a deep learning algorithm is introduced to extract the characteristics of the material crushing state change from the material crushing state images at the two time points, and based on the non-linear interaction response relationship between the electromagnetic working current time series fluctuation pattern and the material crushing state change, the intelligent adjustment of the crushing force is realized. After the crushing is completed, the generated gas, liquid, and solid waste particles are classified and recycled, so as to realize the efficient resource utilization of solid waste. In this way, the dynamic changes of solid waste during the crushing process can be grasped in time, and the crushing force can be quickly adjusted to optimize the crushing effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other objects, features, and advantages of the present application will become more obvious. The accompanying drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0014] Figure 1 It is a flowchart of the solid waste resource treatment method for intelligent environmental protection according to the embodiment of the present application.

[0015] Figure 2 It is a flowchart of sub-step S3 of the solid waste resource treatment method for intelligent environmental protection according to the embodiment of the present application.

[0016] Figure 3 It is a schematic diagram of data flow of sub-step S3 of the solid waste resource treatment method for intelligent environmental protection according to the embodiment of the present application.

[0017] Figure 4 It is a flowchart of sub-step S31 of the solid waste resource treatment method for intelligent environmental protection according to the embodiment of the present application.

[0018] Figure 5 It is a flowchart of sub-step S33 of the solid waste resource treatment method for intelligent environmental protection according to the embodiment of the present application.

[0019] Figure 6 It is a flowchart of sub-step S333 of the solid waste resource treatment method for intelligent environmental protection according to the embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] As shown in this application and the claims, unless the context clearly indicates otherwise, words such as "a", "an", "one" and / or "the" are not specifically singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0021] Although this application makes various references to certain modules in the system according to the embodiments of this application, however, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0022] Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the operations before or below do not necessarily need to be executed precisely in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more steps can be removed from these processes.

[0023] Next, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited by the exemplary embodiments described here.

[0024] It is worth noting that in this application, all actions of obtaining data are carried out on the premise of complying with the corresponding data protection regulations and policies of the country where it is located and obtaining the authorization given by the owner of the corresponding device.

[0025] As mentioned in the above background art, patent CN114247735B proposes a method and system for ensuring network transaction security. After the solid waste entering the device is crushed by a magnet crushing device, the harmful gases in the device are absorbed by the device, and the solid waste is sorted by a large particle treatment part. After treatment, it mainly includes plastic substances, iron substances and other metal materials, etc. The solid waste is treated by a small particle treatment part. The treated substances are mainly composed of glass substances and sediment, etc. At the same time, the wastewater generated during the treatment process is recycled to reduce water resource waste, realizing resource recovery and utilization while treating solid waste and avoiding secondary pollution.

[0026] However, the existing technology uses a dynamic clustering algorithm to optimize the crushing force of solid waste. However, the initial crushing force condensation point depends on manual experience. The algorithm requires standardization and repeated iterative calculations, is easily interfered by initial values and outliers, and the results are uncertain. When the amount of data is large or the dimension is high, the computational complexity is high, the processing efficiency is low, and the real-time performance is poor, making it difficult to meet the requirements of continuous production. For example, when processing a mixture of fluorescent tubes and construction waste, the strength difference between glass and metal is large. The algorithm requires multiple iterations, and the adjustment of the crushing force lags behind, affecting the crushing effect and resource utilization efficiency. To address the above technical problems, the present application proposes a method for resource utilization of mixed solid waste, which feeds the mixed solid waste into a magnet crushing device for vibrating crushing. During the crushing process, a high-speed camera is used to collect images of the material crushing state at the first time point and the second time point, and at the same time, an electromagnetic working current signal between the two time points is collected through a current sensor. Then, a deep learning algorithm is introduced to extract the characteristics of the change in the material crushing state from the images of the material crushing state at the two time points, and based on the non-linear interaction response relationship between the electromagnetic working current time series fluctuation pattern and the change in the material crushing state, intelligent adjustment of the crushing force is achieved. After the crushing is completed, the generated gas, liquid, and solid waste particles are classified and processed for recycling, thereby realizing the efficient resource utilization of solid waste. In this way, the dynamic changes of solid waste during the crushing process can be timely grasped, and the crushing force can be quickly adjusted to optimize the crushing effect.

[0027] Figure 1 FIG. is a flowchart of a method for resource utilization of solid waste for intelligent environmental protection according to an embodiment of the present application. As Figure 1 shown, the method for resource utilization of solid waste for intelligent environmental protection includes the steps of: S1, feeding the mixed solid waste into a magnet crushing device for vibrating crushing; S2, during the crushing process, using a high-speed camera to collect images of the material crushing state at the first time point and the second time point, and at the same time collecting the electromagnetic working current signal of the magnet crushing device between the first time point and the second time point through a current sensor; S3, inputting the image of the material crushing state at the first time point, the image of the material crushing state at the second time point, and the electromagnetic working current signal into a crushing force dynamic control module based on visual feedback to obtain a crushing force adjustment instruction; S4, based on the crushing force adjustment instruction, adjusting the electromagnetic current of the electromagnet of the magnet crushing device; S5, after the crushing is completed, hermetically sucking the generated gas in the magnet crushing device, storing it after being adsorbed by activated carbon, recycling and storing the generated liquid after multi-stage filtration and chemical treatment, classifying and recycling the generated large particle solid waste after vibrating, flotation, magnetic separation, and heat treatment in sequence, and flushing, filtering, and collecting the generated small particle solid waste.

[0028] In the above-mentioned method for the resource treatment of solid waste for intelligent environmental protection, in step S1, the mixed solid waste is fed into a magnet crushing device for vibrating and crushing. Specifically, the magnet crushing device is composed of a magnet, a sliding wheel, a track, a metal crushing plate, and a spring. When the electromagnet is energized, it attracts the metal crushing plate, and the materials are crushed when the two are in contact. The function of the spring is to return the crushing plate to its original position. Through the periodic adsorption-separation action between the electromagnet and the metal crushing plate, mechanical impact force is generated, which can quickly crush large solid wastes, provide uniform particle sizes for subsequent sorting, and improve the resource recovery rate.

[0029] Specifically, when the mixed solid waste enters the magnet crushing device, the operation process starts. At the moment when the electromagnet is energized, a strong magnetic field immediately attracts the metal crushing plate to move towards it until the two are in close contact. During this process, due to the strong attraction of the magnet, any solid waste located on the path of the metal crushing plate will be subjected to huge pressure, which is sufficient to cause deformation or even fracture of most solid substances. Especially for materials with large differences in physical properties, such as the brittle glass fragments in fluorescent tubes and the metal hinges in construction waste, by precisely controlling the current intensity of the electromagnet, the pressure applied to the solid waste can be adjusted to adapt to materials with different hardnesses and toughnesses. It should be noted that at the moment when the electromagnet is de-energized, the originally tightly adsorbed metal crushing plate will quickly bounce back to its original position due to the action of the spring, forming an effective vibrating action. Such a design not only improves the crushing efficiency but also reduces the wear of the equipment caused by long-term high pressure.

[0030] As the solid waste is continuously fed into the magnet crushing device, the adsorption-separation cycle between the electromagnet and the metal crushing plate is continuously repeated. In each cycle, a high-speed camera captures the images of the material crushing states at the first time point and the second time point. At the same time, a current sensor records the working current signal of the electromagnet during this period. These data are then input into the dynamic control module of the crushing force based on visual feedback. By analyzing the images of the material crushing states at the two time points and identifying the change pattern of the working current signal of the electromagnet, it is possible to accurately judge the change characteristics of the material crushing state under the current crushing conditions and generate corresponding crushing force adjustment instructions accordingly. Specifically, if the analysis result shows that the current crushing force is not sufficient to effectively crush all types of solid waste, the system will automatically increase the energizing current of the electromagnet, thereby enhancing the force between the magnet and the metal crushing plate; otherwise, the current intensity will be reduced to avoid unnecessary energy consumption or equipment loss caused by over-crushing.

[0031] During the entire crushing process, each component inside the magnet crushing device collaborated closely to effectively crush the mixed solid waste. The electromagnet, through periodic energization and de-energization operations, drove the metal crushing plate to complete a series of adsorption-separation actions. Each contact meant that a part of the solid waste was effectively crushed. The presence of the spring ensured that the metal crushing plate could quickly return to its original state after each impact, ready to receive the next round of impacts. At the same time, the combined use of the sliding wheel and the track made the movement of the metal crushing plate smoother and more stable, reducing the energy loss caused by friction or jamming.

[0032] In the above-mentioned method for resource treatment of solid waste for intelligent environmental protection, in step S2, during the crushing process, a high-speed camera is used to collect the material crushing state images at the first time point and the second time point, and at the same time, an electromagnet working current signal of the magnet crushing device between the first time point and the second time point is collected through a current sensor. It should be understood that the magnitude of the magnetic force generated by the electromagnet is positively correlated with the current intensity passed through the electromagnet. The greater the current intensity, the stronger the magnetic force and the greater the crushing force on the material. The material crushing state images at different time points reflect the dynamic changes of the material during the crushing process. In this application, by using a camera to capture the material crushing state images at a predetermined frequency and synchronously monitoring the electromagnet current intensity, the morphological changes of the material under different crushing forces can be captured in a timely manner, providing an effective basis for the subsequent dynamic adjustment of the crushing force. In a specific example of this application, the first time point is the current time point, and the second time point is the time point before the first time point and has a preset time interval with the first time point. For example, the first time point is the current time point T, and the second time point is T - ΔT, where ΔT is the preset time interval. In practical applications, the camera collects real-time material crushing state images at a predetermined frequency, and at the same time, based on the preset time interval ΔT, retrieves the material crushing state image at the historical moment T - ΔT corresponding to the current time point T, and uses the two as the material crushing state images at the first time point and the second time point respectively, so as to facilitate the comparative analysis of the state changes of the material during the crushing process.

[0033] Specifically, when the electromagnet starts to be energized, the magnetic force gradually increases until it is strong enough to attract the metal crushing plate. At this time, the high-speed camera starts the shooting process according to the preset frequency and captures the image of the material crushing state at the current time point T. This time point is defined as the first time point. At the same time, the current sensor also starts to work synchronously, accurately measuring the working current signal of the electromagnet within a preset time interval ΔT forward from the current time point T. The current intensity data within this time period directly reflects the magnitude of the magnetic force generated by the electromagnet and its changing trend over time. It should be noted that the selection of the preset time interval ΔT is crucial. It is necessary to ensure sufficient data for subsequent analysis while avoiding missing important crushing moments due to too long an interval. Therefore, in the specific implementation process, the optimal value of ΔT is usually determined according to the specific crushing task requirements and material characteristics.

[0034] Compared with the first time point, the second time point represents an earlier state, forming a temporal comparison relationship between the two. This comparison in the time series not only helps to observe the physical changes that occur to the material during the crushing process but also provides an intuitive basis for understanding how the working current signal of the electromagnet affects these changes. In addition, by comparing the material crushing state images between the two time points, it is possible to identify which areas have undergone significant deformation or fracture, thereby evaluating the effectiveness of the existing crushing strategy.

[0035] Throughout the crushing cycle, the high-speed camera and the current sensor always maintain close cooperation. On the one hand, due to its high frame rate advantage, the high-speed camera ensures that each frame of the image clearly shows the specific form of the material at a specific moment. This includes not only the changes in the overall contour but also fine features such as the direction of crack propagation and the distribution of fragments at the local level. On the other hand, the current sensor focuses on recording the working current signal of the electromagnet. Since the magnitude of the magnetic force generated by the electromagnet is positively correlated with the magnitude of the applied current, by analyzing the change curve of the current signal, it is possible to directly infer the magnitude of the crushing force applied to the material at different moments. Considering these two types of information together, a complete model for describing the material crushing process can be established. To further improve the quality and efficiency of data acquisition, special attention needs to be paid to selecting suitable high-speed cameras and current sensors during the hardware selection phase. An ideal high-speed camera should have high resolution, low latency, and good environmental adaptability so that it can operate stably under various complex working conditions. Similarly, the current sensor also needs to have high sensitivity and accuracy and be able to work normally in a strong magnetic field environment without being interfered.

[0036] In the specific implementation process, considering the complex and diverse composition of mixed solid waste and the significant differences in the physical properties of each substance, the data acquisition system must be flexible enough to handle various situations. For example, when dealing with a mixture containing a large amount of fragile glass fragments and hard metal hinges, the high-speed camera needs to be able to quickly capture the moment when the glass breaks and the moment when the metal deforms. At the same time, the current sensor also needs to accurately reflect the current fluctuations of the electromagnet during this period, because even a small current adjustment may have a significant impact on the final crushing effect. By continuously optimizing the working parameters of the high-speed camera and the current sensor and combining advanced data analysis techniques, the overall efficiency of the crushing process and the resource recovery rate can be greatly improved.

[0037] In the above method for resource treatment of solid waste for intelligent environmental protection, in step S3, the material crushing state image at the first time point, the material crushing state image at the second time point, and the electromagnet working current signal are input into the dynamic control module for crushing force based on visual feedback to obtain a crushing force adjustment instruction. That is, in order to overcome the problems of sensitivity to outliers, high computational complexity of multiple iterations, and response lag in the dynamic clustering algorithm in the prior art, this application introduces a deep learning algorithm to establish an association model between the solid waste crushing state and the current intensity, and dynamically adjusts the current magnitude based on the visual feedback of the solid waste crushing state to achieve adaptive control of the crushing force. In this way, the adaptability to complex mixed materials can be effectively improved, and the crushing efficiency and resource recovery effect can be enhanced. Among them, Figure 2 FIG. is a flowchart of sub-step S3 of the method for resource treatment of solid waste for intelligent environmental protection according to an embodiment of the present application. Figure 3 FIG. is a schematic diagram of data flow of sub-step S3 of the method for resource treatment of solid waste for intelligent environmental protection according to an embodiment of the present application. As Figure 2 and Figure 3 shown, step S3 includes the steps of: S31, extracting the material crushing state change characteristics from the material crushing state image at the first time point and the material crushing state image at the second time point to obtain a material crushing state change characteristic map; S32, extracting the current intensity fluctuation characteristics from the electromagnet working current signal to obtain an electromagnet working current time series feature coding vector; S33, performing current intensity-material crushing state feature deep-level interaction on the material crushing state change characteristic map and the electromagnet working current time series feature vector to obtain a current intensity-material crushing state cross-modal significant interaction response coding vector; S34, determining the crushing force adjustment instruction based on the current intensity-material crushing state cross-modal significant interaction response coding vector.

[0038] Figure 4It is a flowchart of sub-step S31 of the solid waste resource treatment method for intelligent environmental protection according to an embodiment of the present application. As Figure 4 shown, the step S31 includes steps: S311, performing image feature extraction based on the MobileNet model on the material crushing state image at the first time point and the material crushing state image at the second time point to obtain a material crushing state feature map at the first time point and a material crushing state feature map at the second time point; S312, calculating the position-wise difference between the material crushing state feature map at the first time point and the material crushing state feature map at the second time point to obtain the material crushing state change feature map.

[0039] More specifically, in the step S311, image feature extraction based on the MobileNet model is performed on the material crushing state image at the first time point and the material crushing state image at the second time point to obtain a material crushing state feature map at the first time point and a material crushing state feature map at the second time point. It should be understood that during the solid waste crushing process, the amount of original material crushing state image data is huge and there is a lot of redundant information. In order to extract representative key features from the material crushing state images at the first time point and the second time point, the present application uses the MobileNet model based on deep learning to perform feature extraction on both. The MobileNet model, with its lightweight and efficient characteristics, can quickly process image data and extract effective image feature information. Specifically, based on its depthwise separable convolution structure, the MobileNet model decomposes the traditional standard convolution operation into two steps: depthwise convolution and pointwise convolution. The depthwise convolution performs convolution operations on each channel of the input image respectively to extract the spatial features within each channel; the pointwise convolution then linearly combines the feature maps output by the depthwise convolution through a 1×1 convolution kernel to achieve information fusion between channels. In this way, the computational amount and the number of parameters of the model can be effectively reduced, while ensuring the effective extraction of the material crushing state image features, capturing visual features such as the size, shape, and crushing degree of the material particles, and generating a material crushing state feature map at the first time point and a material crushing state feature map at the second time point, thus providing an important basis for the subsequent dynamic adjustment of the crushing force.

[0040] More specifically, in step S312, calculate the position-wise difference between the material crushing state feature map at the first time point and the material crushing state feature map at the second time point to obtain the material crushing state change feature map. It should be understood that since the dynamic evaluation of the solid waste crushing effect requires quantifying the morphological changes of the material within the first to second time windows. Based on this, the present application calculates the position-wise difference between the material crushing state feature map at the first time point and the material crushing state feature map at the second time point to capture the morphological differences of the material at different time points, generate the material crushing state change feature map, and thereby reveal the transition of the solid waste from one state to another during the crushing process, including the reduction of particle size, the change of shape, and the deepening of the crushing degree, etc.

[0041] Specifically, in a specific example of the present application, the step S32 includes: extracting the current intensity fluctuation features of the electromagnet working current signal based on a temporal convolutional neural network to obtain the temporal feature encoding vector of the electromagnet working current. It should be understood that since the electromagnet working current signal is temporal data that changes over time, it contains the fluctuation information of the electromagnet working intensity, and traditional signal processing methods are difficult to effectively capture the complex fluctuation patterns and long-term dependence relationships in the temporal signal. Therefore, the present application further uses a temporal convolutional neural network (TCN), which performs excellently in temporal data analysis, to extract features from the electromagnet working current signal. The TCN model can effectively capture the local features and global dependence relationships in the current signal without destroying the order of the temporal data through a multi-layer causal convolution structure. Specifically, the causal convolution layer in the TCN model ensures that the output of each layer depends only on the input of the previous layer at the same time step or earlier time steps, thus ensuring the causality of temporal data processing. At the same time, by stacking multiple convolutional layers and applying the dilated convolution technique, the TCN model can gradually expand the receptive field and capture the long-term dependence relationships in the current signal. In addition, the TCN model also adopts strategies such as residual connections and weight normalization to alleviate the problems of gradient vanishing and gradient explosion in the training of deep networks and improve the stability and performance of the model. By performing TCN-based feature extraction on the electromagnet working current signal, the fluctuation features of the current intensity changing over time can be effectively extracted, and the temporal feature encoding vector of the electromagnet working current can be generated, providing key information for establishing the association model between the solid waste crushing state and the current intensity. In addition, through in-depth analysis of the temporal features of the current intensity, the changing trend of the electromagnet working intensity can be effectively understood, avoiding the sensitivity to outliers and noise signals in traditional control methods, and further improving the stability and controllability of the crushing process.

[0042] Specifically, in step S33, current intensity - material crushing state feature deep - level interaction is performed on the material crushing state change feature map and the electromagnet working current time - series feature vector to obtain a current intensity - material crushing state cross - modal significant interaction response coding vector. It should be understood that the optimization of the crushing force of the magnet crushing device needs to comprehensively consider the adaptability of the material crushing state to the working intensity of the electromagnet, while traditional multi - modal fusion methods (such as feature splicing) ignore the deep interaction relationship between modalities. Therefore, in order to deeply explore the potential connection between the current intensity and the material crushing state, this application proposes a deep - level interaction method for cross - modal data. By compressing and exciting the information of the material crushing state change feature map and the electromagnet working current time - series feature vector, they are mapped into the same high - dimensional feature space. Through a hierarchical feature interaction mechanism and a recursive sequence modeling strategy, the dynamic association pattern between the material crushing state change feature and the current intensity is deeply mined, generating a current intensity - material crushing state cross - modal significant interaction response coding vector, so as to more accurately characterize the interaction mechanism between the current intensity and the material crushing state during the material crushing process, and realize the dynamic optimization and adjustment of the crushing force. Among them, Figure 5 is a flowchart of sub - step S33 of the solid waste resource treatment method for intelligent environmental protection according to an embodiment of the present application. As Figure 5 shown, step S33 includes steps: S331, performing information excitation based on de - convolutional coding on the electromagnet working current time - series feature vector to obtain an electromagnet working current time - series feature excitation coding vector; S332, compressing the information of the material crushing state change feature map to obtain a set of material crushing state change local feature compression coding vectors; S333, performing hierarchical cross - modal interaction recursive coding on the set of material crushing state change local feature compression coding vectors and the electromagnet working current time - series feature excitation coding vector to obtain the current intensity - material crushing state cross - modal significant interaction response coding vector.

[0043] More specifically, step S331 is represented by the formula:

[0044] v1 = Deconv(V1; W deconv ; b deconv )

[0045] where V1 represents the electromagnet working current time - series feature vector, W deconv represents the de - convolutional coding weight matrix, b deconv represents the de - convolutional coding bias vector, Deconv(·) represents the de - convolutional coding function, and v1 represents the electromagnet working current time - series feature excitation coding vector.

[0046] That is, through the non-linear mapping mechanism of deconvolution coding, the current time-series feature vector is transformed from the original low-dimensional feature space to a high-dimensional feature space, thereby enhancing the feature expression ability of the key fluctuation patterns in the current signal. This information excitation process is not a simple upsampling operation, but rather a spatial reconstruction of the electromagnet working current time-series feature vector through parametric learning of the deconvolution kernel, strengthening the significant features related to the crushing force adjustment decision in the signal while suppressing noise interference. The generated electromagnet working current time-series feature excitation coding vector can more clearly represent the potential coupling relationship between the current intensity and the material crushing dynamics, providing a current feature expression with higher discrimination for subsequent cross-modal interaction analysis.

[0047] More specifically, in a specific example of the present application, the step S332 includes: First, perform feature dissociation on the material crushing state change feature map along the channel dimension to obtain a set of material crushing state change local feature coding matrices, which can be expressed by the formula:

[0048]

[0049] where F2 represents the material crushing state change feature map, Partition(·) represents the feature dissociation operation, and respectively represent the 1st, 2nd, i-th, and n-th material crushing state change local feature coding matrices in the set of material crushing state change local feature coding matrices.

[0050] That is, due to the significant differences in the mechanical properties of different materials during the crushing process, the coupled channel features will mask the independent change patterns of key physical properties. Therefore, in the present application, by performing feature dissociation on the material crushing state change feature map along the channel dimension, the crushing state features mixed between channels can be structurally decomposed, converting the original composite feature expression into a set of material crushing state change local feature coding matrices with independent physical meanings. Through this dissociation operation, the parsing granularity of the model for the dynamic changes of the crushing state can be improved, thereby providing a set of feature primitives with high interpretability for subsequent cross-modal interaction.

[0051] Then, combining the feature distribution space characteristics of each material crushing state change local feature coding matrix in the set of material crushing state change local feature coding matrices, perform information compression on each material crushing state change local feature coding matrix based on dilated convolution coding to obtain a set of material crushing state change local feature compression coding vectors, which can be expressed by the formula:

[0052]

[0053] where ||·||F denotes the Frobenius norm, DilatedConv(·) denotes the dilated convolution operation, and x i represents the i-th local feature compression coding vector of the set of local feature compression coding vectors of the material crushing state change.

[0054] That is, by introducing the adaptive coding mechanism of dilated convolution, according to the characteristic distribution space characteristics of each local feature coding matrix of the material crushing state change, the dilation rate is dynamically adjusted to expand the receptive field, so as to capture the cross-regional physical correlation while maintaining the resolution of the material crushing state change feature map, and realize the efficient semantic concentration of feature information. Specifically, through the spatial-aware context modeling of the local feature coding matrix of the material crushing state change, it helps to refine the physical state characteristics of the material, form the local feature compression coding vector of the material crushing state change containing the key mechanical response mode, accurately reflect the physical essence of the mixed material crushing state, and provide a highly discriminative feature basis for subsequent cross-modal interaction.

[0055] Figure 6 is a flowchart of sub-step S333 of the solid waste resource treatment method for intelligent environmental protection according to an embodiment of the present application. As Figure 6 shown, the step S333 includes steps: S3331, performing first-level cross-modal interaction coding on the electromagnetic working current timing feature excitation coding vector and each local feature compression coding vector of the material crushing state change in the set of local feature compression coding vectors of the material crushing state change to obtain a set of current intensity-material crushing state local interaction coding vectors; S3332, inputting the set of current intensity-material crushing state local interaction coding vectors into a second-level feature interaction recursive unit based on the LSTM model to obtain the current intensity-material crushing state cross-modal significant interaction response coding vector.

[0056] Specifically, in a preferred example of the present application, the step S3331 includes: First, based on the characteristic manifold orientation constraint of the local feature coding matrix of the material crushing state change, optimizing the topological structure features of the local feature compression coding vector of the material crushing state change to obtain an optimized local feature compression coding vector of the material crushing state change, which is expressed by the formula:

[0057]

[0058] where α represents the characteristic manifold orientation constraint factor of the local feature coding matrix of the material crushing state change, β represents the topological invariant of the local feature coding matrix of the material crushing state change, and r i represents The corresponding local feature convolution encoding vector of the material crushing state change, sigmoid(·) j represents the j-th eigenvalue of the vector after being activated by the sigmoid function, m represents the length of the local feature convolution encoding vector of the material crushing state change, and γ represents the additional field orientation factor of r i x' i represents the first-level optimized local feature compression encoding vector of the material crushing state change, x" i represents the optimized local feature compression encoding vector of the material crushing state change.

[0059] That is, by introducing the feature manifold orientation constraint, the local feature encoding matrix of the material crushing state change is regarded as the discrete sampling of the composite topological manifold, and the geometric regularization of the feature space is carried out by using the manifold compact orientation operation, so as to avoid the topological structure distortion caused by conventional compression. Specifically, through the topological optimization based on manifold orientation, the local feature compression encoding vector of the material crushing state change can accurately depict the cross-scale dynamic structure evolution during the crushing of the mixed material, and the discrimination degree of the crushing characteristics of heterogeneous components is enhanced by the additive representation in the sub-space, so as to provide a high-fidelity feature base with physical interpretability for subsequent cross-modal interaction.

[0060] Then, input the electromagnetic working current timing feature excitation encoding vector and the optimized local feature compression encoding vector of the material crushing state change into the first-level feature interaction unit to obtain the current intensity-material crushing state local interaction encoding vector, which is expressed by the formula:

[0061]

[0062] Y = {y1, y2,..., y i ..., y n}

[0063] where Concat(·;·) represents the concatenation function, b T represents the interaction bias term, W T represents the interaction weight matrix, ⊙ represents the dot product, represents the point subtraction, σ(·) represents the activation function, f interact1 (·,·) represents the first-level feature interaction unit, Y represents the set of current intensity-material crushing state local interaction encoding vectors, and y1, y2, y i and y n represent the 1st, 2nd, i-th, and n-th current intensity-material crushing state local interaction encoding vectors in the set of current intensity-material crushing state local interaction encoding vectors respectively.

[0064] That is, through the parallel local interaction mechanism of the first-level feature interaction unit, it is possible to independently perform spatio-temporal alignment between the local feature compression coding vector of each optimized material crushing state change and the excitation coding vector of the electromagnet working current time series feature, so as to establish a preliminary causal correlation mapping between the electromagnet working state and the physical response of material crushing while preserving the independence of heterogeneous component crushing features. Specifically, by local interaction, the microscopic physical correlation between cross-modal features is captured, providing a physically interpretable basic correlation unit for subsequent hierarchical interactions. This preliminary interaction based on independent local units not only avoids feature confusion caused by global fusion, but also strengthens the direct mapping ability between current adjustment requirements and specific material crushing states through local alignment, enabling the generated current intensity - material crushing state local interaction coding vector to accurately reflect the real-time physical coupling relationship between the crushing process of heterogeneous components in the mixed material and the working state of the electromagnet.

[0065] In a specific example of this application, the step S3332 is expressed by the formula:

[0066] v f = LSTM(Y)

[0067] where LSTM(·) represents the LSTM model, and v f represents the cross-modal significant interaction response coding vector of current intensity - material crushing state.

[0068] That is, by using the hidden state memory mechanism of the LSTM model, the cross-modal action laws hidden in the set of local interaction coding vectors of current intensity - material crushing state are deeply mined. Key time series nodes are screened through the gating mechanism, and long-range context information is fused to construct a cross-modal global interaction semantic expression covering the complete crushing cycle. Specifically, the recursive nature of the LSTM model enables the generated cross-modal significant interaction response coding vector of current intensity - material crushing state to dynamically characterize the co-evolution law of heterogeneous material components and current adjustment requirements during the crushing process, and realizes the anticipatory optimization of the crushing force adjustment instruction by capturing sequence dependence relationships, significantly improving the adaptability of the instruction generation system to the complex dynamics of mixed material crushing.

[0069] Specifically, in a specific example of the present application, step S34 includes: inputting the cross-modal significant interaction response encoding vector of current intensity - material crushing state into a classifier-based crushing force adjustment decision module to obtain the crushing force adjustment instruction, where the crushing force adjustment instruction is used to indicate whether the electromagnet energizing current value at the current time point should be increased, decreased, or maintained. It should be understood that the classifier has a powerful classification and decision-making ability and can map the encoded high-dimensional feature representation to specific class labels. In the present application, a classifier (such as a support vector machine or a deep neural network) is used to perform classification and decision-making on the cross-modal significant interaction response encoding vector of current intensity - material crushing state to determine whether the current electromagnet energizing current value needs to be adjusted and the adjustment direction, so as to output a clear crushing force adjustment instruction. Specifically, the classifier-based crushing force adjustment decision module is pre-trained using a large number of labeled sample data to establish a mapping relationship between the interaction response pattern between current intensity and material crushing state and the crushing force adjustment strategy. When an unlabeled cross-modal significant interaction response encoding vector of current intensity - material crushing state is input, the classifier parses its features to map it to a predefined crushing force adjustment action. For example, when the cross-modal significant interaction response encoding vector of current intensity - material crushing state indicates a significant change in the material crushing state and the electromagnet working current intensity is moderate, the classifier may output an instruction to maintain the current current value; if the change in the material crushing state is not obvious or the crushing effect is not good, it may indicate increasing the current intensity to enhance the crushing force; conversely, if the material is close to the ideal crushing state, it may indicate decreasing the current intensity to gradually stop crushing, so as to avoid over-crushing. In this way, the crushing force can be dynamically adjusted according to the real-time working conditions to ensure the efficiency and stability of the crushing process.

[0070] In the above method for intelligent environmental protection solid waste resource treatment, in step S4, based on the crushing force adjustment instruction, the electromagnet energizing current of the magnet crushing device is adjusted. That is, the generated crushing force adjustment instruction is sent to the electromagnet control system to achieve instant adjustment of the crushing force. Since the magnetic force of the electromagnet is proportional to the energizing current, by adjusting the energizing current, the force of the electromagnet attracting the metal crushing plate can be changed, thereby changing the crushing force on the material, so as to ensure that the crushing device can provide an appropriate crushing force according to the actual situation of the material, thereby improving the crushing effect, reducing energy waste, and enhancing the resource treatment efficiency.

[0071] In the above method for resource treatment of solid waste for intelligent environmental protection, in step S5, after crushing, the gas generated in the magnet crushing device is hermetically sucked, adsorbed by activated carbon and then stored. The generated liquid is subjected to multi-stage filtration and chemical agent treatment and then recycled and stored. The large particle solid waste generated is subjected to vibration, flotation, magnetic separation and heat treatment in sequence and then classified and recycled, and the small particle solid waste generated is subjected to flushing, filtration and collection treatment. It should be understood that since harmful gases (such as mercury vapor in fluorescent tubes) and waste water may be released during the crushing process, closed-loop treatment is required to avoid secondary pollution. Specifically, for gas treatment, first, the waste gas is introduced into the activated carbon adsorption tower through a closed air extraction system to remove volatile organic compounds (VOCs) and heavy metal vapors, and then the adsorbed gas is stored in a pressure-resistant tank and regularly handed over to a professional institution for treatment. For liquid treatment, the waste water is first subjected to multi-stage filtration (the pore size is gradually reduced to 0.1 μm) to remove suspended particles, and then a coagulant (such as PAC) and an oxidant (such as H2O2) are added to degrade organic matter, and after purification, it is recycled for the flushing process of solid waste particles. For the treatment of large particle solid waste, first, light plastics and heavy metals are separated by vibration; then, plastic films and rubber are sorted by flotation using density differences; then, ferromagnetic metals (such as hinges, screws) are adsorbed by an electromagnet; the plastic (PE / PP) is melted by an electric heating rod, and the liquid plastic is recycled after being formed by a mold. For the treatment of small particle solid waste, first, flushing treatment is carried out, and glass debris and sediment are separated by high-pressure water flow; then, a filtering operation is carried out, and the vibrating screen classifies by particle size (such as glass > 2 mm, sediment < 0.5 mm); furthermore, the glass is sent to a furnace for regeneration, and the sediment is used as subgrade filler.

[0072] In summary, the method for resource treatment of solid waste for intelligent environmental protection based on the embodiments of the present application is clarified. The mixed solid waste is sent into a magnet crushing device for vibration crushing. During the crushing process, the material crushing state images at the first time point and the second time point are collected by a high-speed camera, and at the same time, the electromagnet working current signals between the two time points are collected by a current sensor. Then, a deep learning algorithm is introduced to extract the material crushing state change characteristics from the material crushing state images at the two time points, and based on the non-linear interaction response relationship between the electromagnet working current time series fluctuation pattern and the material crushing state change, the intelligent adjustment of the crushing force is realized. After crushing, the generated gas, liquid and solid waste particles are classified and treated and recycled, so as to realize the efficient resource utilization of solid waste. In this way, the dynamic changes of solid waste during the crushing process can be grasped in time, and the crushing force can be quickly adjusted to optimize the crushing effect.

[0073] Example 1

[0074] In this embodiment, the above-mentioned solid waste resource treatment method for intelligent environmental protection is used to treat urban construction waste. Specifically, the collected construction waste is sent into a magnet crushing device, and during the crushing process, the crushing force is dynamically adjusted. The crushing test results are shown in Table 1 below:

[0075] Table 1

[0076] Parameter Example 1 Example 2 Breaking force control strategy Dynamic clustering algorithm Deep learning algorithm Single batch processing time (1 ton) 45 minutes 35 minutes Glass breaking uniformity Standard deviation ±1.2 cm Standard deviation ±0.5 cm Energy consumption (kWh / ton) 1.25 0.88

[0077] As can be seen from Table 1, compared with the existing methods, the intelligent control strategy of the crushing force provided by this application has significant improvements in terms of processing efficiency, resource recovery rate, and energy consumption. Therefore, using the solid waste resource treatment method for intelligent environmental protection provided by this application can effectively improve the comprehensive benefits of solid waste resource treatment and achieve efficient recovery and utilization of resources.

[0078] The basic principles of the present invention have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, benefits, effects, etc. mentioned in the present invention are only examples and not limitations. It cannot be considered that these advantages, benefits, effects, etc. are essential for each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and easy understanding, rather than limitations. The above details do not limit the present invention to necessarily adopt the above specific details to be implemented.

[0079] In the above embodiments, the descriptions of each embodiment have their own focuses. For parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative. For example, the unit division is only a logical function division, and there can be other division methods in actual implementation. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0080] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above-described exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference signs in the claims should not be construed as limiting the claims concerned.

[0081] In addition, it is obvious that the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. The multiple elements stated in the system claims can also be implemented by one element through software or hardware.

[0082] Finally, it should be noted that the above description has been given for purposes of illustration and description. In addition, the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for the resource treatment of solid waste for intelligent environmental protection, characterized in that, Including: Feeding the mixed solid waste into a magnet crushing device for vibrating and crushing; During the crushing process, using a high-speed camera to collect the material crushing state images at the first time point and the second time point, and simultaneously collecting the electromagnet working current signals of the magnet crushing device between the first time point and the second time point through a current sensor; Inputting the material crushing state image at the first time point, the material crushing state image at the second time point, and the electromagnet working current signal into a crushing force dynamic control module based on visual feedback to obtain a crushing force adjustment instruction; Based on the crushing force adjustment instruction, adjusting the electromagnet energizing current of the magnet crushing device; After the crushing is completed, hermetically sucking the gas generated in the magnet crushing device, storing it after being adsorbed by activated carbon, subjecting the generated liquid to multi-stage filtration and chemical treatment and then recycling and storing it, subjecting the generated large-particle solid waste to vibration, flotation, magnetic separation and heating treatment in sequence and then classifying and recycling it, and subjecting the generated small-particle solid waste to flushing, filtering and collection treatment.

2. The method for resource treatment of solid waste for intelligent environmental protection according to claim 1, characterized in that, Inputting the material crushing state image at the first time point, the material crushing state image at the second time point, and the electromagnet working current signal into a crushing force dynamic control module based on visual feedback to obtain a crushing force adjustment instruction, including: Extracting the material crushing state change features from the material crushing state image at the first time point and the material crushing state image at the second time point to obtain a material crushing state change feature map; Extracting the current intensity fluctuation features from the electromagnet working current signal to obtain an electromagnet working current time series feature coding vector; Performing current intensity-material crushing state feature deep-level interaction on the material crushing state change feature map and the electromagnet working current time series feature vector to obtain a current intensity-material crushing state cross-modal significant interaction response coding vector; Based on the current intensity-material crushing state cross-modal significant interaction response coding vector, determining the crushing force adjustment instruction.

3. The method for resource treatment of solid waste for intelligent environmental protection according to claim 2, characterized in that, Extracting the material crushing state change features from the material crushing state image at the first time point and the material crushing state image at the second time point to obtain a material crushing state change feature map, including: Performing image feature extraction based on the MobileNet model on the material crushing state image at the first time point and the material crushing state image at the second time point to obtain a material crushing state feature map at the first time point and a material crushing state feature map at the second time point; Calculating the position-wise difference between the material crushing state feature map at the first time point and the material crushing state feature map at the second time point to obtain the material crushing state change feature map.

4. The method for resource treatment of solid waste for intelligent environmental protection according to claim 3, characterized in that Extracting the current intensity fluctuation features from the electromagnet working current signal to obtain an electromagnet working current time series feature coding vector, including: Performing current intensity fluctuation feature extraction based on a temporal convolutional neural network on the electromagnet working current signal to obtain the electromagnet working current time series feature coding vector.

5. The method for resource treatment of solid waste for intelligent environmental protection according to claim 4, characterized in that, Perform current intensity - material crushing state feature deep - level interaction on the material crushing state change feature map and the electromagnet working current time - series feature vector to obtain a current intensity - material crushing state cross - modal significant interaction response coding vector, including: Perform information excitation based on de - convolution coding on the electromagnet working current time - series feature vector to obtain an electromagnet working current time - series feature excitation coding vector; Perform information compression on the material crushing state change feature map to obtain a set of material crushing state change local feature compression coding vectors; Perform hierarchical cross - modal interaction recursive coding on the set of material crushing state change local feature compression coding vectors and the electromagnet working current time - series feature excitation coding vector to obtain the current intensity - material crushing state cross - modal significant interaction response coding vector.

6. The method for resource treatment of solid waste for intelligent environmental protection according to claim 5, wherein Perform information compression on the material crushing state change feature map to obtain a set of material crushing state change local feature compression coding vectors, including: Perform feature dissociation on the material crushing state change feature map along the channel dimension to obtain a set of material crushing state change local feature coding matrices; Combine the feature distribution space characteristics of each material crushing state change local feature coding matrix in the set of material crushing state change local feature coding matrices, and perform information compression based on dilated convolution coding on each material crushing state change local feature coding matrix to obtain the set of material crushing state change local feature compression coding vectors.

7. The method for resource treatment of solid waste for intelligent environmental protection according to claim 6, characterized in that, Perform hierarchical cross - modal interaction recursive coding on the set of material crushing state change local feature compression coding vectors and the electromagnet working current time - series feature excitation coding vector to obtain the current intensity - material crushing state cross - modal significant interaction response coding vector, including: Perform first - level cross - modal interaction coding on the electromagnet working current time - series feature excitation coding vector and each material crushing state change local feature compression coding vector in the set of material crushing state change local feature compression coding vectors respectively to obtain a set of current intensity - material crushing state local interaction coding vectors; Input the set of current intensity - material crushing state local interaction coding vectors into a second - level feature interaction recursive unit based on the LSTM model to obtain the current intensity - material crushing state cross - modal significant interaction response coding vector.

8. The method for resource treatment of solid waste for intelligent environmental protection according to claim 7, characterized in that, Perform first - level cross - modal interaction coding on the electromagnet working current time - series feature excitation coding vector and each material crushing state change local feature compression coding vector in the set of material crushing state change local feature compression coding vectors respectively to obtain a set of current intensity - material crushing state local interaction coding vectors, including: Based on the feature manifold orientation constraint of the material crushing state change local feature coding matrix, perform topological structure feature optimization on the material crushing state change local feature compression coding vector to obtain an optimized material crushing state change local feature compression coding vector; Input the excitation coding vector of the working current timing characteristics of the electromagnet and the compressed coding vector of the optimized local characteristics of the material crushing state change into the first-level feature interaction unit to obtain the local interaction coding vector of the current intensity - material crushing state.

9. The method for resource treatment of solid waste for intelligent environmental protection according to claim 8, characterized in that, Based on the cross-modal significant interaction response coding vector of the current intensity - material crushing state, determine the crushing force adjustment instruction, including: Input the cross-modal significant interaction response coding vector of the current intensity - material crushing state into the crushing force adjustment decision module based on the classifier to obtain the crushing force adjustment instruction, and the crushing force adjustment instruction is used to indicate whether the energizing current value of the electromagnet at the current time point should be increased, decreased or maintained.

Images