Kitchen waste grease separation process collaborative optimization method based on dynamic game theory

Abstract

The invention relates to the technical field of big data analysis, in particular to a kitchen waste grease separation process collaborative optimization method based on a dynamic game theory. Acquiring initial state data of a to-be-processed resource flow, and identifying and quantifying influence factors in the to-be-processed resource flow through an asset risk assessment model; based on the identification result of the influence factor, sorting the to-be-processed resource flow to obtain an optimized resource flow; before the optimized resource flow enters the processing equipment, acquiring component data of the optimized resource flow; after the optimized resource flow is processed by the processing equipment, collecting process feature data of the optimized resource flow; establishing a dynamic game optimization decision model to solve the obtained component data, and determining optimal operation data; and carrying out feedback calibration on the dynamic game optimization decision model by utilizing the collected process characteristic data. According to the invention, through analysis processing of the to-be-processed resource flow data, adaptive regulation and control of the processing equipment are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of big data analysis, and in particular to a collaborative optimization method for a kitchen waste grease separation process based on dynamic game theory. Background Art

[0002] In production enterprises that deal with dynamic and variable resources, the core operation management goal is to optimize workflows to maximize overall business value.

[0003] The current management paradigm often relies on static, standard operating procedures based on historical averages. This model fails to effectively analyze the real-time state of resource flows, resulting in delayed adjustments to operational strategies. Furthermore, it fails to dynamically analyze cost-benefit to determine the optimal resource allocation and workload at a specific point in time. Decision-makers lack quantitative tools to predict the combined impact of an operational adjustment on both short-term productivity and long-term operating costs.

[0004] Business profitability depends on balancing multiple, conflicting key performance indicators (KPIs). Existing systems lack built-in predictive analytics and mathematical modeling capabilities to provide managers with forward-looking decision-making insights. Consequently, operational decisions often remain reactive rather than proactive, data-driven, and predictive.

[0005] To this end, a collaborative optimization method for the kitchen waste grease separation process based on dynamic game theory was proposed. Summary of the Invention

[0006] The purpose of the present invention is to provide a collaborative optimization method for the food waste grease separation process based on dynamic game theory, which realizes adaptive regulation of the processing equipment through the dynamic balance of crushing benefits and emulsification costs in the grease separation process.

[0007] To achieve the above object, the present invention provides the following technical solutions: The collaborative optimization method of the food waste grease separation process based on dynamic game theory includes: Obtaining initial state data of the resource flow to be processed through a sensing device; Based on the initial state data, the asset risk assessment model is used to identify and quantify the factors that affect the operational risk of the processing equipment in the resource flow to be processed; Based on the identification results of influencing factors, the resource flow to be processed is sorted and processed to obtain the optimized resource flow; Before the optimized resource flow enters the processing equipment, the component data of the optimized resource flow is obtained; after the optimized resource flow is processed by the processing equipment, the process characteristic data of the optimized resource flow is collected; A dynamic game optimization decision model is pre-established with the operating parameters of the processing equipment as decision variables and a built-in mathematical function that weighs the crushing benefits and emulsification costs. During real-time operation, the acquired component data is solved according to the dynamic game optimization decision model to determine the optimal operating data. The collected process characteristic data is used to perform feedback calibration on the dynamic game optimization decision model to achieve adaptive control of the processing equipment.

[0008] Utilizing deployed image sensors, image data of the resource stream to be processed is continuously captured to form an image data stream; The dynamic weighing sensor installed below the resource flow to be processed can obtain the quality data of the resource flow to be processed in real time, and the density distribution characteristics of the resource flow to be processed can be calculated by combining the volume measured by the laser sensor. The image data stream, quality data and density distribution characteristics are timestamp aligned and data fused to construct the initial state data of the resource stream to be processed.

[0009] The asset risk assessment model includes image data analysis, abnormal area judgment and influencing factor identification; The image data analysis is to identify the image data stream in the initial state data based on the target detection model of the convolutional neural network, analyze the stacking structure, color distribution and texture information of the resource stream to be processed, and obtain image feature data; The abnormal area judgment layer identifies, based on the image feature data and the corresponding density distribution features, the resource flow area to be processed where target detection anomalies and density anomalies exist, as the abnormal area; The influencing factor identification is to identify the influencing factors of the abnormal area, obtain the identification results of the influencing factors according to the size, shape and density of the influencing factors, and output the operation risk index based on the operational interference caused by similar objects on the processing equipment marked in the historical data.

[0010] Deploy a near-infrared spectrometer at the feed inlet of the processing equipment to scan the optimized resource flow entering the processing equipment through the near-infrared spectrometer to obtain component data of the optimized resource flow; the component data includes moisture content, oil content data, protein content data, and cellulose content data; An online viscometer and a laser particle size analyzer are deployed at the outlet of the processing equipment to continuously monitor and optimize the physical properties of the resource flow after processing, including viscosity data obtained from the online viscometer and particle size data obtained from the laser particle size analyzer.

[0011] The construction process of the dynamic game optimization decision model includes: Obtaining component data, process characteristic data, equipment operating parameters, crushing benefits, and emulsification costs for the complete processing of the processing equipment; the operating parameters include equipment speed and equipment spacing; Constructing crushing benefit function and emulsification cost function based on neural network model; The crushing benefit function takes the component data and decision variables as input and takes the crushing benefit as output; the emulsification cost function takes the component data and decision variables as input and takes the emulsification cost as output; The fragmentation benefit function and the emulsification cost function are solidified within the dynamic game optimization decision model; wherein the difference between the fragmentation benefit function and the emulsification cost function is defined as the net benefit objective function for the final decision.

[0012] The adaptive control is an instruction-based closed-loop execution process; The optimal operation data calculated by the dynamic game optimization decision model is formatted into control instructions and sent to the underlying programmable logic controller of the processing equipment via the industrial Ethernet protocol; The programmable logic controller accurately adjusts the equipment speed and equipment spacing through the frequency converter according to the instructions received; At the same time, the effect of equipment processing is continuously monitored through process characteristic data, and the data is fed back to the dynamic game optimization decision model; The dynamic game optimization decision model uses feedback as the input for the next round of decision-making to correct its internal functions and the strategy output at the next moment.

[0013] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention converts fuzzy operational risks into precise numerical values ​​by establishing a quantifiable risk assessment model based on the characteristic identification and analysis of kitchen waste. It also realizes the automated and high-precision identification and removal of harmful impurities, thus ensuring the safe and stable operation of subsequent crushing equipment from the source and significantly reducing the downtime and maintenance costs caused by stalling or wear.

[0014] 2. This invention transforms the complex production trade-off problem into a clear mathematical optimization problem. By simulating the gains and losses of both parties in the game, it calculates in real time the equipment operating parameters that are optimal in terms of energy efficiency and output under the current material composition, achieving a leap from empirical operation to data-driven refined decision-making.

[0015] 3. The present invention constitutes a complete "decision-making-execution-feedback-learning" closed-loop control system; the online calibration mechanism enables the model to continuously learn and automatically adapt to model drift caused by seasonal changes in raw materials, batch differences, and equipment wear, ensuring the long-term accuracy and robustness of optimization decisions and realizing adaptive regulation of the entire processing process. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Schematic diagram of the process of collaborative optimization of kitchen waste grease separation based on dynamic game theory of the present invention; Figure 2 It is a structural diagram of the asset risk assessment model of the present invention; Figure 3 It is a logical diagram of the adaptive regulation of the oil separation process of the present invention. DETAILED DESCRIPTION

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0018] Example 1: The present invention proposes a collaborative optimization method for the separation process of kitchen waste grease based on dynamic game theory, the process of which is as follows: Figure 1 Shown, including: Obtaining initial state data of the resource flow to be processed through a sensing device; Based on the initial state data, the asset risk assessment model is used to identify and quantify the factors that affect the operational risk of the processing equipment in the resource flow to be processed; Based on the identification results of influencing factors, the resource flow to be processed is sorted and processed to obtain the optimized resource flow; Before the optimized resource flow enters the processing equipment, the component data of the optimized resource flow is obtained; after the optimized resource flow is processed by the processing equipment, the process characteristic data of the optimized resource flow is collected; A dynamic game optimization decision model is pre-established with the operating parameters of the processing equipment as decision variables and a built-in mathematical function that weighs the crushing benefits and emulsification costs. During real-time operation, the acquired component data is solved according to the dynamic game optimization decision model to determine the optimal operating data. The collected process characteristic data is used to perform feedback calibration on the dynamic game optimization decision model to achieve adaptive control of the processing equipment.

[0019] The processing equipment is a kitchen waste crushing equipment; the resource flow to be processed is kitchen waste to be processed.

[0020] Preferably, the deployed image sensor is used to continuously capture image data of the resource flow to be processed to form an image data stream; The dynamic weighing sensor installed below the resource flow to be processed can obtain the quality data of the resource flow to be processed in real time, and the density distribution characteristics of the resource flow to be processed can be calculated by combining the volume measured by the laser sensor. The image data stream, quality data and density distribution characteristics are timestamp aligned and data fused to construct the initial state data of the resource stream to be processed.

[0021] The present invention obtains image data of kitchen waste to be processed based on image sensor recognition, obtains real-time quality data and volume data based on the sensor, and obtains density distribution characteristics; the image data stream, quality data and density distribution characteristics are timestamp aligned and data fused to accurately reflect the initial state of the kitchen waste.

[0022] Preferably, the structure of the asset risk assessment model is as follows: Figure 2 As shown, it includes image data analysis, abnormal area judgment and influencing factor identification; The image data analysis is to identify the image data stream in the initial state data based on the target detection model of the convolutional neural network, analyze the stacking structure, color distribution and texture information of the resource stream to be processed, and obtain image feature data; The abnormal area judgment layer identifies, based on the image feature data and the corresponding density distribution features, the resource flow area to be processed where target detection anomalies and density anomalies exist, as the abnormal area; The influencing factor identification is to identify the influencing factors of the abnormal area, obtain the identification results of the influencing factors according to the size, shape and density of the influencing factors, and output the operation risk index based on the operational interference caused by similar objects on the processing equipment marked in the historical data.

[0023] The image data analysis uses a convolutional neural network model based on the YOLOv5 architecture. This model, trained on a large number of food waste images, can accurately identify several specific influencing factors, including: "large bones," "metal cutlery (knife, fork, spoon)," "ceramic fragments," "plastic packaging bags," and "long strips of fabric (rags, ropes)."

[0024] Furthermore, multimodal feature fusion judgment can be adopted, that is, instead of looking at density values ​​or image features in isolation, density features and image features are more deeply integrated. By allowing features from different sources to corroborate each other, the misjudgment rate can be significantly reduced, making the judgment more accurate.

[0025] For example, the model can learn that areas with high density and metallic reflections or regular edges in the image have a much higher risk weight than areas with high density but a texture similar to "meat fibers." By allowing features from different sources to corroborate each other, the false positive rate can be significantly reduced, leading to more accurate judgments.

[0026] The logic for determining abnormal areas is that if the YOLOv5 model identifies any of the above-mentioned specific influencing factors in the image, the area is marked as a target detection anomaly. At the same time, after calculating the average density of the material, if the density of a certain area calculated by the current laser sensor and weighing sensor fluctuates compared with the average value, it is marked as a density anomaly area.

[0027] The process of determining the operational risk index is to construct a risk knowledge base of influencing factors; the risk knowledge base is collected based on historical data, including the size, shape, density and risk labels of the influencing factors; the risk labels are determined based on the difference between the equipment state of the processing equipment when processing the influencing factors and the standard equipment state; wherein the standard equipment state is the equipment state of the processing equipment when there are no influencing factors; the equipment state is determined based on the vibration frequency of the processing equipment, and can also be determined based on the degree of wear of the equipment.

[0028] The operational risk index is determined based on the association between the size, shape, density, and other factors of influence and risk labels. This can be determined using a neural network model. Specifically, this involves constructing a three-layer feedforward neural network model; the model's input layer receives three normalized parameters: the size, density, and shape of the influencing factor; and the output layer is a single neuron that outputs a risk index ranging from 0 to 1. The model is trained using historical data. Each sample in the database contains the size, density, and shape, as well as a risk label corresponding to the peak value of equipment vibration caused by the sample during processing. This label is a value between 0 and 1 obtained by normalizing or grading the vibration peak.

[0029] After identifying areas with high operational risk factors, the system sends the location coordinates and timestamp information of these areas to a sorting device deployed downstream of the conveyor belt. Based on the received instructions, the sorting device converts the target position in the image sensor coordinate system into a grasping point in the robot coordinate system, removing the factors influencing the target.

[0030] The present invention transforms fuzzy operational risks into precise numerical values ​​by establishing a quantifiable risk assessment model based on the feature identification and analysis of kitchen waste. It realizes the automated and high-precision identification and removal of harmful impurities, thus ensuring the safe and stable operation of subsequent crushing equipment from the source and significantly reducing the downtime and maintenance costs caused by blockage or wear.

[0031] Furthermore, an unsupervised or semi-supervised anomaly detection model is introduced into the asset risk assessment model. This model specifically learns the visual and density characteristics of "normal" food waste. During real-time processing, any area that significantly deviates from the learned "normal" pattern, regardless of the specific object, is marked as an anomaly. This approach can work in parallel with the existing YOLOv5 model, forming a complementary effect: YOLOv5 is responsible for efficiently and accurately detecting known high-frequency risk objects, while the anomaly detection model complements it by identifying all "abnormal" unknown risks, thereby significantly improving the comprehensiveness of risk assessment.

[0032] Preferably, a near-infrared spectrometer is deployed at the feed inlet of the processing equipment, and the optimized resource flow entering the processing equipment is scanned by the near-infrared spectrometer to obtain component data of the optimized resource flow; the component data includes moisture content data, oil content data, protein content data and cellulose content data; An online viscometer and a laser particle size analyzer are deployed at the outlet of the processing equipment to continuously monitor and optimize the physical properties of the resource flow after processing, including viscosity data obtained from the online viscometer and particle size data obtained from the laser particle size analyzer.

[0033] The present invention obtains viscosity and particle size data after optimized resource processing by continuously monitoring and optimizing the physical properties of the resource flow; realizes online, real-time monitoring of materials entering the processing equipment, and provides key feedforward information for subsequent optimization decision models, enabling the system to actively adjust strategies according to immediate changes in materials rather than responding laggingly.

[0034] Preferably, the process of constructing the dynamic game optimization decision model includes: Obtain component data, process characteristic data, equipment operating parameters, crushing benefits, and emulsification costs for the complete processing of the processing equipment; the operating parameters include equipment speed and equipment spacing; the equipment speed is the speed of the processing equipment tool; the equipment spacing is the aperture size of the crusher screen.

[0035] A crushing benefit function and an emulsification cost function are constructed based on a neural network model; the crushing benefit function takes component data and decision variables as input and takes crushing benefit as output; the emulsification cost function takes component data and decision variables as input and takes emulsification cost as output; The fragmentation benefit function and the emulsification cost function are solidified within the dynamic game optimization decision model; wherein the difference between the fragmentation benefit function and the emulsification cost function is defined as the net benefit objective function for the final decision.

[0036] The present invention regards the process of separating grease from kitchen waste as a game, where the two parties in the game are the crushing income and emulsification cost of the kitchen waste after crushing. The crushing benefit refers to the economic benefits of separating food waste during the complete oil separation process, as increased rotation speed and reduced spacing allow for finer comminution, thereby releasing as much fat as possible. The emulsification cost refers to the cost of dealing with emulsification during the oil separation process, as excessive crushing can lead to the formation of a stable emulsion of fat, water, and protein. This increases viscosity, making subsequent oil-water separation extremely difficult and costly, while also increasing equipment energy consumption. The crushing benefit represents the economic benefits of the separated fat during the complete oil separation process, while the emulsification cost represents the cost of dealing with emulsification during the oil separation process.

[0037] Both the crushing benefit function and the emulsification cost function were constructed using a feedforward neural network with three hidden layers (128 neurons per layer, using the Reluctant Unit (ReLU) activation function). Experiments were conducted with different food waste compositions and various operating parameter combinations. The corresponding input vectors, along with the actual oil release rate and outlet viscosity, were collected to form a training dataset. The network was trained using the Adam optimizer, using mean squared error as the loss function, until the model converged.

[0038] During real-time operation, the acquired component data is solved using a dynamic game optimization decision-making model, and a particle swarm optimization algorithm is used to search for the maximum value of the objective function within the speed and spacing range. This involves generating multiple combinations of equipment speeds and spacings. For each combination, the expected crushing revenue and emulsification cost are calculated, and then the cost is subtracted from the revenue to obtain the net profit. Ultimately, the "speed-pitch" combination that produces the highest net profit is the optimal operating data for the current operating conditions.

[0039] This invention transforms the complex production trade-off problem into a clear mathematical optimization problem. By simulating the gains and losses of the two parties in the game, it calculates in real time the equipment operating parameters that are optimal in terms of energy efficiency and output under the current material composition, achieving a leap from empirical operation to data-driven refined decision-making.

[0040] Preferably, the adaptive control is an instruction-based closed-loop execution process, and its logic is as follows: Figure 3 As shown: The optimal operation data calculated by the dynamic game optimization decision model is formatted into control instructions and sent to the underlying programmable logic controller of the processing equipment via the industrial Ethernet protocol; The programmable logic controller accurately adjusts the equipment speed and equipment spacing through the frequency converter according to the instructions received; At the same time, the effect of equipment processing is continuously monitored through process characteristic data, and the data is fed back to the dynamic game optimization decision model; The dynamic game optimization decision model uses feedback as the input for the next round of decision-making to correct its internal functions and the strategy output at the next moment.

[0041] Feedback calibration is achieved through an online learning mechanism. After each decision cycle, a new training sample is constructed using the optimized resource flow composition data, the actual operating parameters used, and the processed and collected process characteristic data. This sample is then used to perform a small-step gradient descent update on the crushing revenue and emulsification cost neural network models. This continuous fine-tuning enables the model to adapt to gradual changes in raw materials and equipment wear, achieving adaptive control.

[0042] Furthermore, the residuals between the predicted values ​​of the two core functions and the actual values ​​collected by the outlet sensors are continuously compared. When the processing residuals of a certain material or a certain type of material deviate from the normal range continuously and systematically over a period of time (for example, the model always overestimates or underestimates the viscosity in a specific situation), the system will trigger a deep retraining process of the model.

[0043] Data from the period before and after the triggering event is cached as a high-value set of difficult samples. Using this set of difficult samples, the original neural network is fine-tuned or specific neurons are added through transfer learning to specifically handle these difficult operating conditions. This enables the model to autonomously identify knowledge blind spots beyond the initial model's cognition, greatly improving the model's adaptability and robustness to rare operating conditions and changes in the nature of raw materials.

[0044] The present invention constitutes a complete "decision-making-execution-feedback-learning" closed-loop control system; the online calibration mechanism enables the model to continuously learn and automatically adapt to model drift caused by seasonal changes in raw materials, batch differences, and equipment wear, ensuring the long-term accuracy and robustness of optimization decisions and realizing adaptive regulation of the entire processing process.

[0045] Example 2: The present invention proposes a collaborative optimization method for the kitchen waste grease separation process based on dynamic game theory, comprising: Obtaining initial state data of the resource flow to be processed through a sensing device; Based on the initial state data, the asset risk assessment model is used to identify and quantify the factors that affect the operational risk of the processing equipment in the resource flow to be processed; Based on the identification results of influencing factors, the resource flow to be processed is sorted and processed to obtain the optimized resource flow; Before the optimized resource flow enters the processing equipment, the component data of the optimized resource flow is obtained; after the optimized resource flow is processed by the processing equipment, the process characteristic data of the optimized resource flow is collected; A dynamic game optimization decision model is pre-established with the operating parameters of the processing equipment as decision variables and a built-in mathematical function that weighs the crushing benefits and emulsification costs. During real-time operation, the acquired component data is solved according to the dynamic game optimization decision model to determine the optimal operating data. The collected process characteristic data is used to perform feedback calibration on the dynamic game optimization decision model to achieve adaptive control of the processing equipment.

[0046] Utilizing deployed image sensors, image data of the resource stream to be processed is continuously captured to form an image data stream; The dynamic weighing sensor installed below the resource flow to be processed can obtain the quality data of the resource flow to be processed in real time, and the density distribution characteristics of the resource flow to be processed can be calculated by combining the volume measured by the laser sensor. The image data stream, quality data and density distribution characteristics are timestamp aligned and data fused to construct the initial state data of the resource stream to be processed.

[0047] In this embodiment, the specific data acquisition includes: The food waste to be processed is transported at a speed of 0.5 m / s via a 1.2 m wide horizontal conveyor belt. The data acquisition system is deployed in the middle of the conveyor belt and has the following structure: Image data acquisition: An industrial camera is mounted vertically 1.5 meters above the conveyor belt with a resolution of 1920x1080 pixels and a frame rate of 30fps to continuously capture bird's-eye views of the waste flow.

[0048] 3D Profile and Density Data Collection: A line laser profile sensor is installed 20 cm downstream of the camera. Its scanning line is perpendicular to the conveyor belt's direction of motion and has a scanning frequency of 200 Hz. This sensor is used to obtain the precise height and volume distribution of the material. Directly below this laser sensor, an array of dynamic weighing sensors is integrated at the bottom of the conveyor belt. With a weighing area of ​​0.5 meters and an accuracy of ±10 grams, it measures the mass of the corresponding volume of material in real time.

[0049] Data synchronization and fusion: The entire acquisition system is time-synchronized to ensure that timestamps for image, contour, and mass data are within 1 millisecond. The system fuses image, volume, and mass data within a 0.1-second window, constructing initial state data with precise spatiotemporal labels that accurately reflect the accumulation morphology, color, texture, and density distribution of food waste.

[0050] Through multi-sensor fusion and precise timing synchronization, comprehensive and accurate quantification of the physical state of the heterogeneous and dynamically changing food waste flow is achieved, providing high-fidelity raw data for subsequent risk assessment and composition analysis.

[0051] The asset risk assessment model includes image data analysis, abnormal area judgment and influencing factor identification; The image data analysis is to identify the image data stream in the initial state data based on the target detection model of the convolutional neural network, analyze the stacking structure, color distribution and texture information of the resource stream to be processed, and obtain image feature data; The abnormal area judgment layer identifies, based on the image feature data and the corresponding density distribution features, the resource flow area to be processed where target detection anomalies and density anomalies exist, as the abnormal area; The influencing factor identification is to identify the influencing factors of the abnormal area, obtain the identification results of the influencing factors according to the size, shape and density of the influencing factors, and output the operation risk index based on the operational interference caused by similar objects on the processing equipment marked in the historical data.

[0052] Deploy a near-infrared spectrometer at the feed inlet of the processing equipment to scan the optimized resource flow entering the processing equipment through the near-infrared spectrometer to obtain component data of the optimized resource flow; the component data includes moisture content, oil content data, protein content data, and cellulose content data; An online viscometer and a laser particle size analyzer are deployed at the outlet of the processing equipment to continuously monitor and optimize the physical properties of the resource flow after processing, including viscosity data obtained from the online viscometer and particle size data obtained from the laser particle size analyzer.

[0053] Before entering the crushing equipment, the sorted optimized resource flow passes through a near-infrared spectrometer deployed above the feed inlet. This analyzer scans the material in real time, analyzing the spectral absorption characteristics of the 900-1700nm band. Combined with a pre-established chemometric model (partial least squares regression), it calculates the core composition data of the optimized resource flow in real time, including moisture content (%), oil content (%), protein content (%), and cellulose content (%). The data refresh rate is 1Hz.

[0054] The construction process of the dynamic game optimization decision model includes: Obtaining component data, process characteristic data, equipment operating parameters, crushing benefits, and emulsification costs for the complete processing of the processing equipment; the operating parameters include equipment speed and equipment spacing; Constructing crushing benefit function and emulsification cost function based on neural network model; The crushing benefit function takes the component data and decision variables as input and takes the crushing benefit as output; the emulsification cost function takes the component data and decision variables as input and takes the emulsification cost as output; The fragmentation benefit function and the emulsification cost function are solidified within the dynamic game optimization decision model; wherein the difference between the fragmentation benefit function and the emulsification cost function is defined as the net benefit objective function for the final decision.

[0055] The adaptive control is an instruction-based closed-loop execution process; The optimal operation data calculated by the dynamic game optimization decision model is formatted into control instructions and sent to the underlying programmable logic controller of the processing equipment via the industrial Ethernet protocol; The programmable logic controller accurately adjusts the equipment speed and equipment spacing through the frequency converter according to the instructions received; At the same time, the effect of equipment processing is continuously monitored through process characteristic data, and the data is fed back to the dynamic game optimization decision model; The dynamic game optimization decision model uses feedback as the input for the next round of decision-making to correct its internal functions and the strategy output at the next moment.

[0056] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A collaborative optimization method for the separation process of kitchen waste grease based on dynamic game theory, characterized by: include: Obtaining initial state data of the resource flow to be processed through a sensing device; Based on the initial state data, the asset risk assessment model is used to identify and quantify the factors that affect the operational risk of the processing equipment in the resource flow to be processed; Based on the identification results of influencing factors, the resource flow to be processed is sorted and processed to obtain the optimized resource flow; obtaining composition data of the optimized resource stream before the optimized resource stream enters the processing equipment; After the optimized resource flow is processed by the processing equipment, process characteristic data of the optimized resource flow is collected; A dynamic game optimization decision model is pre-established with the operating parameters of the processing equipment as decision variables and a built-in mathematical function that weighs the crushing benefits and emulsification costs. During real-time operation, the acquired component data is solved according to the dynamic game optimization decision model to determine the optimal operating data. The collected process characteristic data is used to perform feedback calibration on the dynamic game optimization decision model to achieve adaptive control of the processing equipment.

2. The collaborative optimization method for kitchen waste grease separation process based on dynamic game theory according to claim 1 is characterized in that: Utilizing deployed image sensors, image data of the resource stream to be processed is continuously captured to form an image data stream; The dynamic weighing sensor installed below the resource flow to be processed can obtain the quality data of the resource flow to be processed in real time, and the density distribution characteristics of the resource flow to be processed can be calculated by combining the volume measured by the laser sensor. The image data stream, quality data and density distribution characteristics are timestamp aligned and data fused to construct the initial state data of the resource stream to be processed.

3. The collaborative optimization method for kitchen waste grease separation process based on dynamic game theory according to claim 1 is characterized in that: The asset risk assessment model includes image data analysis, abnormal area judgment and influencing factor identification; The image data analysis is to identify the image data stream in the initial state data based on the target detection model of the convolutional neural network, analyze the stacking structure, color distribution and texture information of the resource stream to be processed, and obtain image feature data; The abnormal area judgment layer identifies, based on the image feature data and the corresponding density distribution features, the resource flow area to be processed where target detection anomalies and density anomalies exist, as the abnormal area; The influencing factor identification identifies the influencing factors of the abnormal area, obtains the identification results of the influencing factors based on the size, shape and density of the influencing factors, combines the operation interference caused by similar objects on the processing equipment marked in historical data, outputs the operation risk index, and sorts the resource flow to be processed according to the operation risk index.

4. The collaborative optimization method for kitchen waste grease separation process based on dynamic game theory according to claim 1, characterized in that: Deploy a near-infrared spectrometer at the feed inlet of the processing equipment to scan the optimized resource flow entering the processing equipment through the near-infrared spectrometer to obtain component data of the optimized resource flow; the component data includes moisture content, oil content data, protein content data, and cellulose content data; Through the online viscometer and laser particle size analyzer, the process characteristic data after resource flow processing is continuously monitored and optimized, including viscosity data obtained from the online viscometer monitoring and particle size data obtained from the laser particle size analyzer monitoring.

5. The collaborative optimization method for kitchen waste grease separation process based on dynamic game theory according to claim 1 is characterized in that: The construction process of the dynamic game optimization decision model includes: Obtaining component data, process characteristic data, equipment operating parameters, crushing benefits, and emulsification costs for the complete processing of the processing equipment; the operating parameters include equipment speed and equipment spacing; Constructing crushing benefit function and emulsification cost function based on neural network model; The crushing benefit function takes the component data and decision variables as input and takes the crushing benefit as output; the emulsification cost function takes the component data and decision variables as input and takes the emulsification cost as output; The fragmentation benefit function and the emulsification cost function are solidified within the dynamic game optimization decision model; wherein the difference between the fragmentation benefit function and the emulsification cost function is defined as the net benefit objective function for the final decision.

6. The collaborative optimization method for kitchen waste grease separation process based on dynamic game theory according to claim 1 is characterized in that: The adaptive control is an instruction-based closed-loop execution process; The optimal operation data calculated by the dynamic game optimization decision model is formatted into control instructions and sent to the underlying programmable logic controller of the processing equipment via the industrial Ethernet protocol; The programmable logic controller accurately adjusts the equipment speed and equipment spacing through the frequency converter according to the instructions received; At the same time, the state of resource flow after equipment processing is continuously monitored and optimized through process characteristic data, and the data is fed back to the dynamic game optimization decision model; The dynamic game optimization decision model takes feedback as the input for the next round of decision-making, and corrects its internal function and the strategy output at the next moment.

Images