Refining processing route scheduling optimization method and system based on benefit evaluation

Abstract

The invention discloses a refining and chemical processing route scheduling optimization method and system based on benefit evaluation, which not only are innovated on the basis of a traditional benefit calculation model and predicts energy consumption and output by introducing a neural network, but also enables a raw material distribution optimization process to be more efficient and intelligent in combination with a particle swarm optimization algorithm. The method comprises the following steps: naming and defining each processing route, and determining the technological process of each processing route; constructing a benefit function for evaluating a specified raw material distribution amount for each processing route; establishing a total benefit objective function of multiple processing routes; and carrying out optimization solution on the raw material distribution amount of each processing route so as to maximize the total benefit objective function. According to the method, more accurate production scheduling and benefit evaluation can be realized, so that a more competitive optimization tool is provided for refinery enterprises when the refinery enterprises face complex production environments and continuously changing market requirements, and finally, the maximization of production benefits is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of petroleum refining, and particularly relates to a method and system for optimizing the scheduling of a refining and processing route based on benefit evaluation. Background Art

[0002] The petroleum refining industry is one of the important basic industries in the national economy, and undertakes the important task of providing raw materials for multiple industries such as chemical industry and energy. With the transformation of the global energy structure and the increasingly strict environmental protection requirements, petroleum refining enterprises are facing more and more challenges. These challenges are not only reflected in the changes in crude oil supply and market demand, but also include the balance problem between economic benefits and production efficiency in the refining process. Especially under the conditions of multiple raw materials and complex processes, how to optimize resource allocation and improve benefits through fine production scheduling and processing route selection has become an urgent problem for enterprises to solve.

[0003] In actual production, refining enterprises often need to process raw materials from different sources. These raw materials have large property differences, which may affect the subsequent processing efficiency and product quality. For the same raw material, there are usually multiple different processing routes. For example, vacuum residue can choose routes such as fluid catalytic cracking (FCC) or deep catalytic cracking (DCC). The selection of these processing routes directly affects the economic benefits of the enterprise and the feasibility of the production plan. However, the benefit differences brought by different processing routes are huge, and there are also significant differences in the costs, product quality, processing speed, etc. generated. Therefore, how to scientifically and reasonably select and schedule processing routes has become a key issue in the refining industry. At present, although some refining enterprises select routes and arrange production through manual experience and simple scheduling models, this method is not only inefficient, but also difficult to adapt to market fluctuations, changes in raw material properties, and dynamic adjustments of unit states. Problems such as inefficiency and inaccuracy that may exist in the manual decision-making process often lead to waste of production resources and failure to maximize benefits. Therefore, the traditional production scheduling mode can no longer meet the requirements of modern refining industry for efficient and refined management.

[0004] With the development of information technology and optimization algorithms, more and more intelligent production scheduling systems have begun to enter the refining industry. By establishing a more accurate unit material balance model and combining real-time data, it has become possible to conduct refined benefit evaluation and production scheduling. However, the existing technologies can usually only handle the benefit evaluation problem of a single route and are difficult to cope with complex optimization tasks with multiple routes and multiple variables. More importantly, the existing optimization methods often lack sufficient flexibility and cannot be dynamically adjusted under the conditions of market demand fluctuations, changes in raw material quality, and different unit states, resulting in the final optimization results being difficult to fully meet the actual production requirements.

[0005] Therefore, refining enterprises urgently need a new method that can combine multiple factors such as market conditions, raw material properties, and plant capabilities, and provide a scientific decision-making basis for the selection of multiple processing routes and production scheduling through efficient calculation and optimization means. Such a technical solution can not only help enterprises improve production efficiency, but also reduce energy consumption, cut costs, optimize resource utilization, and thus maximize benefits. This is not only crucial for the sustainable development of enterprises, but also of far-reaching significance for enhancing the overall competitiveness of the industry and promoting the green transformation of the energy industry. Summary of the Invention

[0006] To solve the above technical problems, the present invention provides a refining processing route scheduling and optimization method and system based on benefit evaluation, which calculates and optimally solves the benefits for different processing routes to achieve the purpose of optimizing resource allocation and plant production scheduling and obtaining the maximum benefit.

[0007] The technical solution provided by the present invention is as follows:

[0008] A refining processing route scheduling and optimization method based on benefit evaluation, comprising the steps of:

[0009] S1. Name and define each processing route, and clarify the process flow of each processing route;

[0010] S2. Construct a benefit function for each processing route to evaluate the benefits under a specified raw material allocation;

[0011] S3. Establish a total benefit objective function for multiple processing routes;

[0012] S4. Optimally solve the raw material allocation of each processing route to maximize the total benefit objective function.

[0013] Further, the benefit function in step S2 is:

[0014]

[0015] where E i represents the benefit obtained by the i-th processing route, S i is the cost corresponding to the raw material quantity X i allocated to the i-th processing route; P ij represents the output value of the j-th device on the i-th processing route; F ij represents the operating cost of the j-th device on the i-th processing route.

[0016] Further, the output value P of the j-th device on the i-th processing route ij, which is equal to the product of the quantities of various economic products of the device output by the energy consumption / output prediction model and their corresponding market unit prices; the operating cost F of the jth device on the ith processing route ij , which is equal to the product of the energy consumption of the device output by the energy consumption / output prediction model and its corresponding market unit price;

[0017] The construction and application process of the energy consumption / output prediction model includes the steps:

[0018] For each processing route, collect the historical production data of the processing route, including the raw material input quantity and its composition index for each production, as well as the energy consumption and economic product quantities of each device on the corresponding processing route;

[0019] Construct an energy consumption / output prediction model for each processing route respectively. The energy consumption / output prediction model adopts a multi-layer feedforward neural network, including an input layer, an output layer and multiple hidden layers. Among them, the input layer is used to input the raw material input quantity and its various composition indexes, the output layer is used to output the energy consumption and economic product quantities corresponding to each device on the processing route, and the hidden layer is used to capture the mapping relationship between the raw material input and the energy consumption output;

[0020] Use the historical production data of each processing route collected to train the corresponding energy consumption / output prediction model of each processing route respectively until the model converges;

[0021] For the given raw material input quantity and its composition index, input it into the energy consumption / output prediction model corresponding to the trained processing route to obtain the energy consumption and various economic product quantities of each device on the processing route.

[0022] Furthermore, the benefit function in step S2 is:

[0023] E i (X i ) = w1P(X i ) - w2En(X i ) - w3Ec(X i ) - w4U(X i ) - w5Q(X i )

[0024] Among them, P(X i ) is the output value of the ith processing route, which depends on the raw material quantity X i allocated to the processing route; En(X i ) is the energy consumption of processing route i when the input raw material quantity is X i ; Ec(X i ) is the production cost of processing route i; U(X i) is the equipment utilization rate of processing route i, which is used to measure the load level of the equipment; Q(X i ) is the product quality of processing route i, which is used to reflect the influence of raw material quality on the final product quality; w1 to w5 are the weight coefficients of each item, which are used to dynamically adjust the benefit function according to the production period.

[0025] Furthermore, the total benefit objective function in step S3 is:

[0026]

[0027] where m is the total number of processing routes, and E i (X i ) is the benefit function of the i-th processing route under the given raw material allocation X i ; the total raw material allocation of all processing lines cannot exceed the total available raw material, and the raw material allocation of each processing line cannot exceed its processing capacity.

[0028] Furthermore, in step S4, the particle swarm optimization algorithm is used to optimize and solve the raw material allocation of each processing route, including the steps:

[0029] S401. Particle representation: Each particle is represented as a position vector (X1, X2,..., X m ) containing m dimensions, where X m represents the amount of raw material allocated to the m-th processing route;

[0030] S402. Particle swarm initialization: Each dimension of each particle is randomly initialized from a preset range to ensure that the total amount constraint of the raw material and the processing capacity of the processing route are satisfied; and the velocity vector of each particle is initialized to zero;

[0031] S403. Fitness function calculation: Calculate the fitness of each particle based on the total benefit objective function, that is, the fitness fitness j of the j-th particle is determined by the total benefit of its corresponding raw material allocation plan:

[0032] S404. Respectively take the position vector of each particle that can achieve the maximum fitness up to the current iteration round as the individual best position; at the same time, take the position vector of the particle swarm that can achieve the maximum fitness up to the current iteration round as the global best position;

[0033] S405. Position update: where, and are the position vectors of the j-th particle in the k-th and k+1-th iteration rounds respectively, and They are the velocity vectors of the j-th particle at the k-th and (k + 1)-th iterations respectively; pbest j is the individual best position of the j-th particle up to the k-th iteration, and gbest is the global best position in the particle swarm up to the k-th iteration; c1 and c2 are acceleration constants, r1 and r2 are random numbers used to increase search diversity, and w is the velocity inertia weight that controls the convergence of the particle to the global best solution;

[0034] S406. Iteration termination and result output: Repeat S403 - S405 until the preset maximum number of iterations is reached, or when the change in the fitness value is less than the given threshold, stop the iteration. The final optimization result is the raw material allocation plan corresponding to the global best position gbest in the particle swarm.

[0035] A refining and processing route scheduling optimization system based on the above method includes the following functional modules:

[0036] Processing route construction module: It is used to select the devices included in the processing route through a graphical operation method of component dragging, and is used to set the parameters of each device and the flow relationship of materials between devices;

[0037] Benefit function construction module: It is used to define the benefit function of each processing route and the total benefit objective function of multiple processing routes;

[0038] Raw material allocation optimization module: It is used to optimize and solve the raw material allocation amounts of each processing route to maximize the total benefit objective function.

[0039] Furthermore, the benefit function construction module further includes a prediction model construction and application module, which is used to construct an energy consumption / output prediction model for each processing route respectively. The energy consumption / output prediction model is used to output the energy consumption amounts and various economic product amounts of each device on the processing route according to the input raw material input amount and its component indicators; the benefit function construction module further includes a data collection and storage module, which is used to collect and store the historical production data of each processing route, including the raw material input amount and its component indicators for each production, as well as the energy consumption amounts and economic product amounts of each device on the processing route.

[0040] Furthermore, the benefit function construction module further includes a market condition data setting module, which is used to set the market price data involved in benefit calculation, including raw material prices, energy consumption prices, and refined product prices.

[0041] Furthermore, the raw material allocation optimization module uses the particle swarm optimization algorithm to optimize and solve the raw material allocation amounts of each processing route, including an initial parameter setting module, an iteration termination condition setting module, and a result output module.

[0042] Compared with the prior art, the present invention has significant advantages and innovations, especially in the aspects of raw material distribution and benefit evaluation for multiple processing routes:

[0043] First of all, by establishing a detailed material balance model and a device operation cost model, the present invention can accurately depict the operation of each device in each processing route, and then comprehensively evaluate the benefits of the processing route. This structured method based on the material balance and cost model makes the benefit calculation of different processing routes more systematic and efficient.

[0044] In addition, since fixed mathematical models are often difficult to adapt to changes in the quantity and quality of raw materials, resulting in a lack of flexibility and accuracy in predicting the benefits of processing routes, the present invention introduces a neural network model to predict energy consumption and output. Without relying entirely on traditional mathematical formulas, it can more accurately capture the complex relationship between raw material input and device output. Especially when the relationship between energy consumption / output and raw material input is usually highly nonlinear and complex, the neural network can automatically identify these complex nonlinear patterns through learning historical data, thereby improving the accuracy of benefit prediction. This neural network model can predict the energy consumption and output of each device according to different raw material input quantities and quality characteristics, not only improving the calculation accuracy but also significantly enhancing the adaptability of the method, and being able to handle various complex situations that may be encountered in the actual production of refining enterprises.

[0045] On the other hand, the present invention uses the particle swarm optimization algorithm to optimize raw material distribution. Considering multiple processing routes and complex constraints, it can intelligently search for the optimal raw material distribution plan. By simulating the movement of a particle swarm in the search space, the PSO algorithm can find the optimal solution while ensuring the constraints. Compared with traditional optimization methods, the PSO has a stronger global search ability, avoiding the risk of falling into local optimal solutions and being able to better handle multi-objective and multi-constraint optimization problems. Combined with a dynamically adjusted benefit function, the PSO can flexibly optimize raw material distribution according to different production demands (such as peak and trough production periods), improving the overall benefits of the enterprise.

[0046] In summary, the present invention not only innovates on the basis of traditional benefit calculation models. By introducing neural network prediction of energy consumption and output, it further improves the prediction accuracy and model adaptability. Moreover, combined with the particle swarm optimization algorithm, the raw material distribution optimization process becomes more efficient and intelligent. The present invention can achieve more accurate production scheduling and benefit evaluation, thus providing a more competitive optimization tool for refining enterprises in the face of complex production environments and changing market demands, and ultimately maximizing production benefits. Brief Description of the Drawings

[0047] The accompanying drawings are used to provide a further understanding of the present invention and form a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention.

[0048] Figure 1 is a schematic flowchart of an optimization method for refining and processing route scheduling provided by an embodiment of the present invention;

[0049] Figure 2 is a schematic diagram of the module composition of a processing route scheduling optimization system provided by an embodiment of the present invention. Detailed implementation manners

[0050] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0051] Embodiment 1

[0052] This embodiment provides an optimization method for refining and processing route scheduling based on benefit evaluation. As Figure 1 shown, the method mainly includes the following steps:

[0053] 1. Define the processing route

[0054] Name and define each processing route. For example, the fluid catalytic cracking (FCC) route, the deep catalytic cracking (DCC) route, etc. Sort out the process flow of each processing route, clarify the input, output, and their interrelationships of each device on the processing route. For example, the output of "pyrolysis gasoline" from FCC flows to the aromatics extraction device.

[0055] 2. Build a benefit model

[0056] Establish a material balance model: Establish a detailed material balance model for each device on each processing route to describe the conversion relationship between the input raw materials and various products of each device. For the DCC device, its material balance may include the conversion ratios of hydrogenated heavy oil, pyrolysis gasoline, etc.

[0057] Establish a device operating cost model: The energy consumption of each device is usually closely related to the amount of input raw materials and the processing efficiency. To establish an operating cost model, the following operating cost formula can be used: C = αM in + βM out + γ, where C represents the operating cost, M in and M outThey represent the raw material input amount and the output material amount respectively, and α, β, and γ are empirical parameters of the equipment and process.

[0058] Based on the material balance model and the device operation cost model, a benefit function is constructed for each processing route:

[0059]

[0060] Among them, E i represents the benefit obtained from the i-th processing route, S i is the cost corresponding to the raw material amount X i allocated to the i-th processing route. P ij represents the output value of the j-th device on the i-th processing route, which can be calculated based on the output material amount of the device and its corresponding market price. F ij represents the operation cost of the j-th device on the i-th processing route, including fuel consumption, power cost, auxiliary material consumption, operation cost, etc., and can be calculated using the above operation cost formula based on empirical parameters.

[0061] In some embodiments, the refining and chemical enterprise not only considers the output value, energy consumption, and production cost, but also needs to consider the influence of product quality, device utilization rate, and raw material characteristics. Based on this, a comprehensive benefit function including multiple dimensions is provided:

[0062] E i (X i ) = w1P(X i ) - w2En(X i ) - w3Ec(X i ) - w4U(X i ) - w5Q(X i )

[0063] Among them, P(X i ) is the output value of the i-th processing route, which depends on the raw material amount X i allocated to this processing route; En(X i ) is the energy consumption of the processing route i when the input raw material amount is X i ; Ec(X i ) is the production cost of the processing route i, including raw material cost, energy consumption cost, operation cost, etc.; U(X i ) is the device utilization rate of the processing route i, which measures the load level of the device; Q(X i) is the product quality of processing route i, reflecting the impact of raw material quality on the final product quality; w1 to w5 are the weight coefficients of each item. To control benefits more precisely, the benefit function can be dynamically adjusted at different times (such as day, night, or production peak and trough periods). For example, during the production peak period, maximizing production capacity may be given priority, while during the trough period, minimizing energy consumption and cost control are emphasized.

[0064] 3. Establish the optimization objective function for multiple processing routes

[0065] In the optimization process of multiple processing routes, the main objective is to maximize the total benefit by reasonably allocating raw materials to different processing routes. The mathematical expression of the total benefit function is as follows:

[0066]

[0067] where m is the total number of processing routes, and E i (X i ) is the benefit function of the i-th processing route under the given raw material allocation X i . In the actual production process, in addition to the objective function (total benefit function), the optimization problem must also satisfy some constraints. In this embodiment, the total raw material allocation of all processing lines cannot exceed the total available raw materials, and the raw material allocation of each processing line cannot exceed its processing capacity. The devices on each processing route may be limited by capacity when processing raw materials, so it is necessary to ensure that the load of the device does not exceed its maximum processing capacity.

[0068] 4. Obtain the optimal allocation plan through particle swarm optimization

[0069] The Particle Swarm Optimization (PSO) algorithm is a heuristic global optimization algorithm that searches for the optimal solution by simulating the flight process of particles in the search space. In this embodiment, PSO will be used to search for the raw material allocation plan to maximize the total benefit.

[0070] Specifically, it includes the steps:

[0071] Particle representation: Each particle represents a raw material allocation plan. Assuming there are m processing routes, then each particle can be represented as a position vector with m dimensions: X = (X1, X2, …, X m ).

[0072] Generation of the initial particle swarm: Randomly generate a particle swarm, and each dimension of each particle (i.e., the raw material amount of each processing route) is randomly initialized from a preset range to ensure that the total amount constraint of the raw materials is satisfied.

[0073] Xi min ≤X i ≤X i max

[0074] wherein, X i min and X i max are respectively the minimum value and the maximum value of the raw material allocation amount of the i-th processing route.

[0075] Fitness function calculation: The position of each particle corresponds to a benefit value, which is called fitness. The fitness calculation method is based on the aforementioned total benefit function. That is, the fitness j of the j-th particle is determined by the total benefit of the corresponding raw material allocation plan:

[0076] Updating the individual best solution and the global best solution: Each particle will update its individual best position according to its own fitness, that is, the position vector that can obtain the maximum fitness of this particle up to the current iteration round is used as the individual best position; at the same time, the global best solution in the population will also be continuously updated with the optimization process of the particle swarm, that is, the position vector that can obtain the maximum fitness among all particles in the particle swarm up to the current iteration round is used as the global best position.

[0077] Velocity and position update: In the particle swarm, each particle has two important parameters: position and velocity. The position represents the current solution of the particle (i.e., the raw material allocation amount), and the velocity determines the direction of the particle searching for the best solution. The velocity and position update formulas of the particle are as follows:

[0078]

[0079]

[0080] wherein, and are respectively the position vectors (i.e., the raw material allocation amounts) of the j-th particle at the k-th and k + 1-th iteration rounds, and are respectively the velocity vectors of the j-th particle at the k-th and k + 1-th iteration rounds; pbest j is the individual best position of the j-th particle up to the k-th iteration round, and gbest is the global best position in the particle swarm up to the k-th iteration round; c1 and c2 are acceleration constants, and r1 and r2 are random numbers used to increase the diversity of the search. The inertia weight w in the velocity update formula controls the convergence speed of the particle to the global best solution. A larger w value is helpful for global search, while a smaller w value is helpful for local search.

[0081] Iteration Termination and Result Output: Repeat the above calculations of the fitness function ~ speed and position updates until the preset maximum number of iterations is reached, or when the change in the fitness value is less than the given threshold, stop the iteration. The final optimized result is the raw material allocation plan corresponding to the global best solution gbest in the particle swarm.

[0082] Through particle swarm optimization, under the premise of meeting the constraints, raw materials can be reasonably allocated to different processing routes, thereby maximizing the overall benefits of the refining enterprise. Although in the implementation process of the above method, the benefits of each processing route under a specific raw material input can be characterized by constructing a benefit model, its accuracy still needs to be improved. The benefit model with fixed mathematical expressions often has difficulty accurately describing the impact of the quality and input volume of raw materials on the energy consumption and output of each device on the processing route, and this relationship is often non-linear and complex.

[0083] Therefore, this embodiment also provides a neural network-based solution to solve this problem. The neural network can accurately predict the energy consumption and output of each processing route under different raw material input volumes and qualities by learning the patterns and trends in historical data, and then evaluate the benefits that can be obtained for the entire processing route based on the market prices of raw materials, energy consumption, and output products.

[0084] Specifically, it includes the following steps:

[0085] (1) Data Collection and Processing

[0086] For each processing route, collect the historical production data of this processing route, including the raw material input volume and its component indicators (such as sulfur content, metal content, density, carbon residue, and the content of four components (alkanes, aromatics, resins, asphaltenes)) for each production, as well as the corresponding energy consumption and economic product volume of each device on the processing route. Then, clean the collected historical production data to remove outliers. If there are missing values, interpolation, mean filling, or machine learning algorithms can be used for filling.

[0087] (2) Construction and Training of the Prediction Model

[0088] Construct an energy consumption / output prediction model for each processing route. The energy consumption / output prediction model uses a multi-layer feedforward neural network, including an input layer, an output layer, and multiple hidden layers. The input layer is used to input the raw material input volume and its various component indicators, the output layer is used to output the corresponding energy consumption and economic product volume of each device on this processing route, and the hidden layer is used to capture the complex mapping relationship between raw material input and energy consumption output.

[0089] Use the collected historical production data of each processing route to train the corresponding energy consumption / output prediction model for each processing route until the model converges.

[0090] (3) Model Application and Benefit Calculation

[0091] For a given new raw material input quantity and quality index, input them into the energy consumption / output prediction model of a certain processing route that has been trained, and the predicted values of the energy consumption and output of each device on this processing route can be obtained. Furthermore, calculate the benefits that can be achieved by this processing route based on the market prices of raw materials, energy consumption, and economic products.

[0092] That is, for the benefit function described above The output value P of the jth device on the ith processing route ij , is equal to the product of the quantities of various economic products of this device output by the energy consumption / output prediction model and their corresponding market unit prices; the operating cost F of the jth device on the ith processing route ij , is equal to the product of the energy consumption quantity of this device output by the energy consumption / output prediction model and its corresponding market unit price.

[0093] Example 2

[0094] Based on the above method, this example provides a refining and chemical processing route scheduling optimization system, as Figure 2 shown. This system mainly includes the following modules:

[0095] (1) Processing Route Construction Module: Used to select the devices included in the processing route through a graphical operation method of component dragging. For example, the DCC route includes devices such as catalytic cracking DCC, 3# desulfurization and mercaptan removal, ethylene recovery, pyrolysis gasoline hydrogenation, and aromatics extraction. This module is also used to set the parameters of each device and the flow relationship of materials between devices. For example, the "pyrolysis gasoline" produced by the catalytic cracking DCC device enters the pyrolysis gasoline hydrogenation device, and the "hydrogenated C6-C8" produced by the pyrolysis gasoline hydrogenation enters the aromatics extraction device, etc.

[0096] (2) Benefit Function Construction Module: Used to define the benefit function of each processing route and the total benefit objective function of multiple processing routes. It also includes a market condition data setting module, used to set the market price data involved in benefit calculation, including raw material prices, energy consumption prices, refining product prices, etc.

[0097] In some embodiments, the benefit function construction module also includes a prediction model construction and application module, used to construct an energy consumption / output prediction model for each processing route respectively. The energy consumption / output prediction model is used to output the energy consumption quantity and various economic product quantities of each device on this processing route according to the input raw material input quantity and its composition index; correspondingly, it also needs to include a data collection and storage module, used to collect and store the historical production data of each processing route, including the raw material input quantity and its composition index for each production, as well as the energy consumption quantity and economic product quantities of each device on the processing route.

[0098] (3) Raw material allocation optimization module: used to optimize and solve the raw material allocation amounts for each processing route so as to maximize the total benefit objective function. In this embodiment, the PSO algorithm is used to optimize and solve the raw material allocation amounts for each processing route, and it further specifically includes an initial parameter setting module, an iteration termination condition setting module, and a result output module.

[0099] The above system or product can execute the refining and processing route scheduling optimization method described in Embodiment 1, and has the corresponding functional modules and beneficial effects of the method. For the technical details not described in detail in this embodiment, reference can be made to the refining and processing route scheduling optimization method provided in Embodiment 1 of the present invention.

[0100] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be an individual computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0101] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; under the idea of the present invention, the technical features in the above embodiments or different embodiments can also be combined, and the steps can be implemented in any order, and there are many other changes in different aspects of the present invention as described above. For the sake of brevity, they are not provided in detail; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of each embodiment of the present application.

Claims

1. A scheduling optimization method for refining and processing routes based on benefit evaluation, characterized in that Including the steps: S1. Name and define each processing route, and clarify the process flow of each processing route; S2. Construct a benefit function for each processing route to evaluate the benefits under a specified raw material allocation; S3. Establish the total benefit objective function for multiple processing routes; S4. Optimize and solve the raw material allocation for each processing route to maximize the total benefit objective function.

2. The refining and processing route scheduling optimization method according to claim 1, wherein The benefit function described in step S2 is: Among them, E i represents the benefit obtained from the i-th processing route, and S i is the amount of raw materials allocated to the i-th processing route, and X i is the corresponding cost; P ij represents the output value of the j-th device on the i-th processing route; F ij represents the operating cost of the j-th device on the i-th processing route.

3. The refining and processing route scheduling optimization method according to claim 2, characterized in that, The output value P of the jth device on the ith processing route ij , is equal to the product of the quantities of various economic products of the device output by the energy consumption / output prediction model and their corresponding market unit prices; The operating cost F of the j-th device on the i-th processing route ij , which is equal to the product of the energy consumption of the device output by the energy consumption / output prediction model and its corresponding market unit price; The construction and application process of the energy consumption / output prediction model includes the steps: For each processing route, collect the historical production data of this processing route, including the raw material input and its component indicators for each production, as well as the energy consumption and economic product output of each device on the corresponding processing route; Construct an energy consumption / output prediction model for each processing route respectively. The energy consumption / output prediction model adopts a multi-layer feedforward neural network, including an input layer, an output layer and multiple hidden layers. Among them, the input layer is used to input the raw material input and its various component indicators, the output layer is used to output the energy consumption and economic product output corresponding to each device on this processing route, and the hidden layer is used to capture the mapping relationship between the raw material input and the energy consumption output; Use the collected historical production data of each processing route to train the corresponding energy consumption / output prediction model for each processing route respectively until the model converges; For the given raw material input and its component indicators, input them into the trained energy consumption / output prediction model corresponding to the processing route to obtain the energy consumption and various economic product outputs of each device on this processing route.

4. The refining and processing route scheduling optimization method according to claim 1, characterized in that The benefit function described in step S2 is: E i (X i ) = w1P(X i ) - w2En(X i ) - w3Ec(X i ) - w4U(X i ) - w5Q(X i ) where P(X i ) is the output value of the i-th processing route, which depends on the amount of raw materials X i allocated to this processing route; En(X i ) is the energy consumption of the processing route i when the input raw material amount is X i ; Ec(X i ) is the production cost of the processing route i; U(X i ) is the equipment utilization rate of the processing route i, which is used to measure the load level of the equipment; Q(X i ) is the product quality of the processing route i, which is used to reflect the influence of raw material quality on the final product quality; w1 to w5 are the weight coefficients of each item, which are used to dynamically adjust the benefit function according to the production period.

5. The refining and processing route scheduling optimization method according to any one of claims 2 to 4, characterized in that, The total benefit objective function described in step S3 is: where m is the total number of processing routes, and E i (X i ) is the benefit function of the i-th processing route under the given raw material allocation X i ; the total raw material allocation of all processing routes cannot exceed the total available raw materials, and the raw material allocation of each processing route cannot exceed its processing capacity.

6. The refining and processing route scheduling optimization method according to claim 5, wherein, In step S4, the particle swarm optimization algorithm is used to optimize and solve the raw material allocation for each processing route, including the steps: S401. Particle representation: Each particle is represented as a position vector (X1, X2, …, X m ) with m dimensions, where X m represents the amount of raw material allocated to the m-th processing route; S402. Particle swarm initialization: Each dimension of each particle is randomly initialized from a preset range to ensure that the total amount constraint of the raw material and the processing capacity of the processing route are met; and the velocity vector of each particle is initialized to zero; S403. Fitness function calculation: Calculate the fitness of each particle based on the total benefit objective function, that is, the fitness fitness of the j-th particle j is determined by the total benefit of its corresponding raw material allocation plan: S404. Respectively take the position vector of each particle that can achieve the maximum fitness up to the current iteration round as the individual best position; at the same time, take the position vector of the particle swarm that can achieve the maximum fitness up to the current iteration round as the global best position; S405. Position update: wherein and are the position vectors of the j-th particle at the k-th and (k + 1)-th iterations respectively, and are the velocity vectors of the j-th particle at the k-th and (k + 1)-th iterations respectively; pbest j is the individual best position of the j-th particle up to the k-th iteration, and gbest is the global best position in the particle swarm up to the k-th iteration; c1 and c2 are acceleration constants, r1 and r2 are random numbers used to increase search diversity, and w is the velocity inertia weight that controls the convergence of the particle to the global best solution; S406. Iteration termination and result output: Repeat S403 - S405 until the preset maximum number of iterations is reached, or when the change in the fitness value is less than the given threshold, stop the iteration. The final optimization result is the raw material allocation plan corresponding to the global best position gbest in the particle swarm.

7. A refining and processing route scheduling optimization system based on the method described in claim 6, characterized in that, Including the following functional modules: Processing route construction module: Used to select the devices included in the processing route through a graphical operation method of component dragging, and used to set the parameters of each device and the flow relationship of materials between each device; Benefit function construction module: Used to define the benefit function of each processing route and the total benefit objective function of multiple processing routes; Raw material allocation optimization module: Used to optimize and solve the raw material allocation for each processing route to maximize the total benefit objective function.

8. The refining and processing route scheduling optimization system according to claim 7, wherein The benefit function construction module further includes a prediction model construction and application module, which is used to construct an energy consumption / output prediction model for each processing route respectively. The energy consumption / output prediction model is used to output the energy consumption of each device and the quantities of various economic products on the processing route according to the input raw material input quantity and its composition index. The benefit function construction module further includes a data collection and storage module, which is used to collect and store the historical production data of each processing route, including the raw material input quantity and its composition index for each production, as well as the energy consumption and economic product quantities of each device on the processing route.

9. The refining and processing route scheduling optimization system according to claim 7, wherein The benefit function construction module further includes a market quotation data setting module, which is used to set the market price data involved in benefit calculation, including raw material price, energy consumption price, and refining product price.

10. The refining and processing route scheduling optimization system according to claim 7, wherein The raw material allocation optimization module uses the particle swarm optimization algorithm to optimize and solve the raw material allocation quantity of each processing route, including an initial parameter setting module, an iteration termination condition setting module, and a result output module.

Images