A Cogeneration Load Allocation Method Based on Improved Gray Wolf Algorithm
By improving the Grey Wolf algorithm and utilizing chaotic sequence initialization and population hierarchical strategies, the problems of thermal performance differences and electric-thermal load coupling among multiple units in a combined heat and power system were solved, achieving the search for the global optimal solution and improving energy efficiency.
Patent Information
- Application Number
- CN202610023029.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-08
- Publication Date
- 2026-05-26
AI Technical Summary
Traditional algorithms struggle to effectively balance the differences in thermodynamic performance and the coupling characteristics of electric and heat loads among multiple units in combined heat and power systems, resulting in insufficient economic efficiency and renewable energy absorption capacity. Furthermore, they are difficult to find the globally optimal solution in high-dimensional multi-objective conflict scenarios.
An improved gray wolf algorithm is adopted, which generates a chaotic sequence to initialize the population through Logistic mapping and maps it to the solution space of a cogeneration power plant. The population is divided into elite wolves, potential wolves and ordinary wolves, and different position update methods are used for each. Combined with exponential weighted moving average and random perturbation term, a multi-objective comprehensive trade-off is achieved.
It significantly increases the probability of finding the global optimal solution in a combined heat and power system, reduces coal consumption, improves energy utilization efficiency, and solves the local optimum dilemma of traditional algorithms in complex scenarios.
Smart Images
Figure CN122088904A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of optimization control technology for cogeneration systems, and specifically to a method for cogeneration load allocation in cogeneration power plants based on an improved gray wolf algorithm. Background Technology
[0002] Against the backdrop of the global energy low-carbon transition and my country's "dual-carbon" goals, renewable energy sources such as wind power and photovoltaics are being connected to the grid on a large scale, and regional heating demand is also continuing to grow. Cogeneration systems have thus become a core hub for balancing energy supply and demand. However, the combined operation of multiple cogeneration units faces a series of pressing problems.
[0003] On the one hand, the thermal performance of each unit differs, and the electric and thermal loads have coupling characteristics. How to formulate the optimal allocation strategy under the total load command becomes the key to maximizing the system's economic benefits and renewable energy absorption capacity. The traditional automatic generation control (AGC) mode for a single unit, because it does not take into account the performance differences of units in the same plant, makes it difficult to balance economy and environmental protection during operation.
[0004] On the other hand, plant-level AGC (Automatic Generation Control) models present multi-objective optimization problems, such as minimizing coal consumption, reducing carbon emissions, and maximizing renewable energy utilization. This places higher demands on the flexibility and optimization capabilities of algorithms. The actual operating conditions of cogeneration plants are extremely complex and variable, with frequent fluctuations in system parameters. Traditional gradient-based optimization algorithms, such as Newton's method, mixed-integer programming, and convex optimization, rely on the continuous differentiability of the objective function, making it difficult to construct accurate mathematical models in such nonlinear and strongly coupled scenarios.
[0005] Furthermore, there are multi-objective conflict problems in the process of solving high-dimensional solution spaces. For example, there is a contradiction between the optimal allocation of electricity and heat loads and the electricity spot market. The allocation of electricity and heat loads needs to take into account factors such as unit performance and energy consumption to achieve economic operation. However, the price of electricity spot market fluctuates frequently. When optimizing the allocation of electricity and heat loads, it is difficult to simultaneously meet the requirements of market economic benefits and internal system operation optimization, which further exacerbates the optimization difficulty.
[0006] Existing research shows that metaheuristic algorithms, with their random search mechanism, exhibit unique advantages in nonlinear and strongly coupled scenarios such as combined heat and power systems. However, existing algorithms still have shortcomings when dealing with high-dimensional multi-objective conflicts and complex constraints. Some algorithms rely on static optimization tables, resulting in poor real-time performance and an inability to adjust promptly according to changes in actual operating conditions. In multi-objective collaborative optimization, the trade-off mechanism for object conflicts needs improvement, making it difficult to effectively balance the relationships between various objectives. In high-dimensional solution spaces, the balance between local search accuracy and global exploration capability also needs strengthening, as local optima are easily encountered, failing to find the global optimum and thus affecting the overall optimization effect of the system.
[0007] Therefore, we propose a cogeneration load allocation method based on the improved gray wolf algorithm. Summary of the Invention
[0008] The purpose of this invention is to provide a cogeneration load allocation method for cogeneration power plants based on an improved gray wolf algorithm, so as to solve the problems mentioned in the background art.
[0009] To achieve the above objectives, the present invention adopts the following technical solution: This invention provides a method for cogeneration load allocation in cogeneration plants based on an improved gray wolf algorithm, comprising the following steps: S1: Set parameters; S2: Use Logistic mapping to generate chaotic sequences, map the chaotic variables in the chaotic sequences to the solution space of the cogeneration load allocation problem, and obtain the initial population individuals; S3: Calculate the fitness value based on the optimization objective function of the cogeneration power plant and select the best individuals; S4: Based on the fitness calculated in step S3, the population is stratified into different levels, and different position update methods are adopted according to the characteristics of different levels. S5: Iteration and Output Results.
[0010] Furthermore, the parameters in step S1 include the number of gray wolf individuals, the maximum number of iterations, the problem dimension, the proportion of elite wolves, and the proportion of potential wolves.
[0011] Furthermore, the specific process of step S2 is as follows: Using the Logistic mapping, the mathematical model is as follows: , in, For control parameters; It is a chaotic variable, and its initial value is randomly selected in the interval (0, 1); The solution space of the cogeneration load allocation problem in a cogeneration power plant is: ,in, This represents the lower bound of the decision variable; This represents the upper limit of the decision variable; Mapping chaotic variables in chaotic sequences to the solution space of the combined heat and power (CHP) load allocation problem in cogeneration power plants. Thus, the initial population of individuals is obtained; , in, It is a chaotic variable; This represents the value taken by an individual in the corresponding solution space.
[0012] Furthermore, in step S3, the optimization objective function of the cogeneration power plant is based on the total coal consumption of the entire plant. The minimum target is considered, taking into account the content of cogeneration power plants. The objective function for the combined heat and power (CHP) unit is: , in, Indicates the first The coal consumption function of the unit is obtained based on the unit's energy consumption curve; Indicates the first Power generation of the unit; Indicates the first Heating supply of the unit; The objective function value calculated using the objective function is the fitness value. The three individuals with the best fitness values are selected as the top three individuals. Wolf, wolves and Wolves, the rest of the individuals in the population are classified as Wolf.
[0013] Furthermore, in step S3, when calculating fitness, constraints need to be considered simultaneously: the internal constraint is that the energy supply of a single unit has upper and lower limits, that is:
[0014] in, Indicates the first The lower limit of power generation capacity of the unit; Indicates the first The maximum power generation capacity of the unit; Indicates the first The lower limit of the heating capacity of the unit; Indicates the first The maximum heat supply capacity of the unit; The global constraint is that the total energy supply of the combined heat and power plant must meet the electrical and thermal load demands of the power grid and the heating network, that is: , in, This represents the electrical load demand of the power grid and heating network; This represents the heat load demand of the power grid and heating network.
[0015] Furthermore, the specific process of step S4 is as follows: the entire population is arranged according to the fitness calculated in step S3. Using a preset elite wolf ratio as the screening standard, individuals whose fitness ranking is within this ratio are elite wolves. Using a preset potential wolf ratio as the screening standard, individuals whose fitness has decreased twice consecutively are selected from the remaining individuals in the population to become potential wolves. The number of potential wolves is at most a preset upper limit. The remaining wolves in the population are ordinary wolves.
[0016] Furthermore, in step S4, the mathematical model for the attack process of the elite wolf is as follows: Computational Elite Wolf Tracking Wolf, wolves and The process of the wolf: , in, Indicates the position of the elite wolf; This represents the number of iterations. They represent Wolf, wolves and The wolf's current position vector; These represent other elite wolf individuals. and Wolf, wolves and The distance of the wolves; It is a random vector; Calculate the individual elite wolf Wolf, wolves and The process of the wolf approaching: , in, This indicates that individual wolves in the elite group are... Wolf, wolves and The step size for moving closer; It is a random vector; Final update on the final location of the elite wolf individual: , in, Represented as the individual's historical best position; The weighted value represents the individual's historical best position, and its value ranges from (0, 1). Indicates the final location of an individual.
[0017] Furthermore, in step S4, the mathematical model for the attack process of the potential wolf is as follows: Calculate potential wolf tracking Wolf, wolves and The process of the wolf: , in, Indicates the potential wolf's position; This represents the number of iterations. They represent Wolf, wolves and The wolf's current position vector; These represent other potential wolf individuals. and Wolf, wolves and The distance of the wolves; It is a random vector; Calculating the potential of individual wolves Wolf, wolves and The process of the wolf approaching: , in, This indicates the potential of individual wolves. Wolf, wolves and The wolf's approaching stride length; It is a random vector; Final update on the final position of the potential wolf individual: , in, This is a trend factor with a value range of (0, 1), representing the degree to which an individual maintains its self-movement inertia; It is an exponentially weighted shift vector; This is the movement vector for the current iteration. The attenuation coefficient has a value range of (0, 1) and represents the weight of the historical movement vector.
[0018] Furthermore, in step S4, the mathematical model for the attack process of a common wolf is as follows: Calculate ordinary wolf tracking Wolf, wolves and The process of the wolf: , in, Indicates the position of a regular wolf; This represents the number of iterations. They represent Wolf, wolves and The wolf's current position vector; These represent other ordinary wolf individuals. and Wolf, wolves and The distance of the wolves; It is a random vector; Calculate the individual characteristics of a common wolf Wolf, wolves and The process of the wolf approaching: , in, This indicates that an individual wolf is... Wolf, wolves and The wolf's approaching stride length; It is a random vector; Final update of the final location of the ordinary wolf individual: , in, This is a random disturbance term.
[0019] Furthermore, the specific process of step S5 is as follows: if the maximum number of iterations has not been reached, repeat steps S3 to S4; otherwise, output the optimal individual in the population, which is the algorithm iteration result. This optimal individual is the heat and power load allocation strategy that minimizes the total coal consumption of the entire plant under various constraints of the cogeneration power plant.
[0020] Compared with the prior art, the present invention has the following technical effects: 1. The allocation method provided by this invention employs chaotic mapping to initialize the population. Leveraging the randomness, ergodicity, and regularity of chaotic systems, a seemingly random sequence is generated within a specific range according to certain rules via Logistic mapping. This sequence is then mapped onto the problem search space to obtain the initial population. The generated initial population is evenly distributed throughout the search space. Compared to traditional random initialization, this effectively overcomes the drawback of population clustering in local areas, which is prone to occur in traditional random initialization. This allows the algorithm to have a rich variety of search starting points from the startup phase, greatly expanding the search range and enabling the algorithm to better adapt to real-time changes in the system, improving real-time performance and significantly increasing the probability of finding the global optimal solution in complex nonlinear and strongly coupled problems such as cogeneration systems.
[0021] 2. The method provided by this invention cleverly divides a gray wolf population into elite wolves, potential wolves, and ordinary wolves. Elite wolves consist of the gray wolves with the best fitness values. Utilizing a weighted average of individual historical optimal position information and the group's optimal position, the population is precisely guided towards the currently perceived optimal region, focusing on executing refined local search tasks. In the optimization of combined heat and power systems, this mechanism can precisely adjust the unit's power generation and heating parameters, thereby effectively reducing coal consumption. Potential wolves employ an exponentially weighted moving average algorithm, which retains some historical movement trend information while focusing on recent position changes, thereby uncovering entirely new potential search areas. This significantly enhances the algorithm's global exploration capability in high-dimensional solution spaces and multi-objective conflict scenarios, effectively preventing the algorithm from getting trapped in local optima. Ordinary wolves, by retaining the original attack logic and adding random perturbation terms, further increase the probability of the population escaping local optima, ensuring the algorithm's stable operation under various complex conditions. These three different levels of wolves—elite, potential, and ordinary—achieve a comprehensive trade-off between multiple objectives through different strategies.
[0022] 3. The method provided by this invention balances these two capabilities through the collaborative cooperation of elite wolves and potential wolves. Elite wolves focus on refined local searches, improving local search accuracy; potential wolves explore new potential search areas in a relatively broader space, enhancing global exploration capabilities. By dynamically adjusting the algorithm's global and local search capabilities in this way, the algorithm can quickly converge to near a better solution while maintaining sufficient exploration capability to escape local optima when solving complex optimization problems. This improves the overall performance of the algorithm, effectively solves optimization problems in complex scenarios of combined heat and power (CHP) power plants, and can accurately formulate more reasonable CHP load allocation strategies, effectively reducing unnecessary energy losses, thereby significantly reducing coal consumption and significantly improving energy utilization efficiency, bringing substantial energy-saving benefits to CHP systems. Attached Figure Description
[0023] Figure 1 This is a schematic diagram of the movement process in the traditional gray wolf optimization algorithm. Figure 2 This is a schematic diagram of the movement process of the elite wolf according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the potential wolf's movement process according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the movement process of a common wolf according to an embodiment of the present invention; Figure 5 This is a flowchart of a cogeneration power plant heat and power load allocation method according to an embodiment of the present invention. Detailed Implementation
[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of the present invention.
[0025] In this article, terms such as "left," "right," "up," "down," "front," and "back" are established based on the positional relationships shown in the attached drawings. Depending on the attached drawings, the corresponding positional relationships may also change. Therefore, they should not be interpreted as an absolute limitation on the scope of protection.
[0026] Figure 1 This demonstrates the movement process of the traditional Grey Wolf optimization algorithm. In the simulation logic of the Grey Wolf algorithm, the optimal solution is defined as... The wolf, the suboptimal solution, and the tertiary solution are defined as follows: wolves and Wolves, the remaining gray wolves are all classified as Wolves, the hunting (optimization) process is... Wolf, wolves and Wolves are dominant. In attacking prey, Wolves often don't know the location (optimal solution) of their prey; rather, they rely on... Wolf, wolves and Wolf-led, The wolf follows, gradually approaching its prey. The mathematical model of this encirclement behavior is as follows: , in, Indicates the distance between an individual and its prey; Indicates the location of the prey. Indicates the location of the gray wolf. For the number of iterations, This is the coefficient vector.
[0027] The gray wolf's location is updated as follows: , in, Indicates the final location of an individual; This is the coefficient vector.
[0028] The system vector is calculated as follows: , , in, The convergence factor decreases linearly from 2 to 0 with the number of iterations, as shown in the formula: , This represents the maximum number of iterations. It is a random vector in the interval [0,1].
[0029] The mathematical model for the attack process is as follows: Calculate Grey Wolf Tracking Wolf, wolves and The process of the wolf: , in, They represent Wolf, wolves and The wolf's current position vector. Representing other individuals respectively and Wolf, wolves and The distance of the wolves It is a random vector.
[0030] Calculate the individual orientation of a wolf pack Wolf, wolves and The process of the wolf approaching: , in, It is a random vector; These represent the individual wolves in the pack moving towards... Wolf, wolves and The wolf's approaching stride length.
[0031] Final update on the final location of the individual gray wolf:
[0032] Please see Figures 2 to 5 This invention provides a method for cogeneration load allocation in cogeneration plants based on an improved gray wolf algorithm, comprising the following steps: S1: Set parameters, including the number of gray wolf individuals, maximum number of iterations, problem dimension, percentage of elite wolves, and percentage of potential wolves.
[0033] Specifically, the population size of gray wolves is set as follows: N The maximum number of iterations is max_iter The problem dimension is dim Elite Wolf Ratio elite_rate (In this embodiment, the default percentage is 20%), Potential Wolf Ratio potential_ratio(In this embodiment, the default value is 20%). These parameters are set based on a comprehensive consideration of the complexity of the cogeneration system and the efficiency of algorithm optimization. For example, the number of individuals in the population. N This determines the coverage of the algorithm in the search space. N If the value is too small, it may not be able to fully explore the solution space, leading to missing the globally optimal solution; if... N If the value is too high, it will increase the computational load and time cost. Through extensive experiments and practical application experience, it has been determined that this parameter setting can ensure the accuracy of the algorithm while also taking into account computational efficiency.
[0034] S2: Use Logistic mapping to generate chaotic sequences, map the chaotic variables in the chaotic sequences to the solution space of the cogeneration load allocation problem, and obtain the initial population individuals.
[0035] Specifically, the process of step S2 is as follows: Using the Logistic mapping, its mathematical model is as follows: , in, As a control parameter, its value range is usually between (3.57, 4]. It is a chaotic variable, and its initial value is randomly selected in the interval (0, 1); Assume the solution space of the cogeneration load allocation problem in a cogeneration power plant is ,in, This represents the lower bound of the decision variable; This represents the upper limit of the decision variable; The chaotic variables in the chaotic sequence are mapped to the solution space of the cogeneration load allocation problem in a combined heat and power plant through the following linear transformation. Thus, the initial population of individuals is obtained; , in, It is a chaotic variable; This represents the value taken by an individual in the corresponding solution space.
[0036] This initialization method utilizes the characteristics of chaotic systems to avoid the initial population from clustering in local areas, thus providing a rich starting point for subsequent searches.
[0037] S3: Calculate the fitness value based on the optimization objective function of the cogeneration power plant and select the best individuals.
[0038] Specifically, the fitness value of each individual in the population is calculated, and the three individuals with the best fitness values are selected as the top three. Wolf, wolves and Wolves, the rest of the individuals in the population are classified as The fitness value is calculated based on the optimization objective function of the cogeneration power plant, which can be either minimizing total coal consumption or maximizing economic benefits. In this embodiment, the optimization objective function of the cogeneration power plant is based on total coal consumption. The minimum target is considered, taking into account the content of cogeneration power plants. The objective function for the combined heat and power (CHP) unit is: , in, Indicates the first The coal consumption function of the unit is obtained based on the unit's energy consumption curve; Indicates the first Power generation of the unit; Indicates the first Heating supply of the unit; The value of the objective function calculated using the objective function is the fitness value.
[0039] Specifically, when calculating fitness, constraints need to be considered simultaneously: the internal constraint is that the energy supply of a single unit has upper and lower limits, that is:
[0040] in, Indicates the first The lower limit of power generation capacity of the unit; Indicates the first The maximum power generation capacity of the unit; Indicates the first The lower limit of the heating capacity of the unit; Indicates the first The maximum heat supply capacity of the unit; The global constraint is that the total energy supply of the combined heat and power plant must meet the electrical and thermal load demands of the power grid and the heating network, that is: , in, This represents the electrical load demand of the power grid and heating network; This represents the heat load demand of the power grid and heating network.
[0041] Only individuals that meet these constraints have valid fitness values.
[0042] S4: Based on the fitness calculated in step S3, the population is stratified into different levels, and different position update methods are adopted according to the characteristics of different levels.
[0043] Specifically, step S4 involves the following process: the entire population is ranked according to the fitness calculated in step S3. Using a preset elite wolf percentage (preset at 20%) as the selection criterion, individuals ranking in the top 20% of the population in terms of fitness are designated as elite wolves. Using a preset potential wolf percentage (preset at 20%) as the selection criterion, individuals whose fitness has decreased twice consecutively are selected from the remaining individuals in the population as potential wolves. The number of potential wolves is at most the preset upper limit (i.e., the total number of gray wolf individuals). N (20%), while the rest of the wolves in the population are ordinary wolves.
[0044] Specifically, in Elite Wolves, the mathematical model for the Elite Wolf's attack process is as follows: Computational Elite Wolf Tracking Wolf, wolves and The process of the wolf: , in, Indicates the position of the elite wolf; This represents the number of iterations. They represent Wolf, wolves and The wolf's current position vector; These represent other elite wolf individuals. and Wolf, wolves and The distance of the wolves; It is a random vector; Calculate the individual elite wolf Wolf, wolves and The process of the wolf approaching: , in, This indicates that individual wolves in the elite group are... Wolf, wolves and The step size for moving closer; It is a random vector; Final update on the final location of the elite wolf individual: , in, Represented as the individual's historical best position; The weighted value represents the individual's historical best position, and its value ranges from (0, 1). Indicates the final location of an individual.
[0045] Elite wolves use a weighted average of individual historical best and group best, leveraging the high-quality information they have accumulated throughout history, to guide the population to gather in the currently perceived optimal area and conduct localized, refined searches.
[0046] Specifically, in the Potential Wolf attack, the Exponentially Weighted Moving Average (EWMA) algorithm is used. The mathematical model of the Potential Wolf attack process is as follows: Calculate potential wolf tracking Wolf, wolves and The process of the wolf: , in, Indicates the potential wolf's position; This represents the number of iterations. They represent Wolf, wolves and The wolf's current position vector; These represent other potential wolf individuals. and Wolf, wolves and The distance of the wolves; It is a random vector; Calculating the potential of individual wolves Wolf, wolves and The process of the wolf approaching: , in, This indicates the potential of individual wolves. Wolf, wolves and The wolf's approaching stride length; It is a random vector; Final update on the final position of the potential wolf individual: , in, This is a trend factor with a value range of (0, 1), representing the degree to which an individual maintains its self-movement inertia; It is an exponentially weighted shift vector; This is the movement vector for the current iteration. The attenuation coefficient has a value range of (0, 1) and represents the weight of the historical movement vector.
[0047] The potential wolf employs the EWMA algorithm, which allows individuals to retain some historical movement trend information while placing greater emphasis on recent positional changes. This enables them to actively explore new trend directions while developing existing ones, avoiding the loss of new, better solutions due to over-reliance on historical trends. While the potential wolf's exploration depth is not as deep as the elite wolf's, it enriches the population's diversity and uncovers new potential search areas in a relatively wider space.
[0048] Specifically, in the ordinary wolf algorithm, the wolf retains the original attack logic of the gray wolf algorithm while adding a random perturbation term to increase the probability of the population escaping local optima. The mathematical model of the ordinary wolf's attack process is as follows: Calculate ordinary wolf tracking Wolf, wolves and The process of the wolf: , in, Indicates the position of a regular wolf; This represents the number of iterations. They represent Wolf, wolves and The wolf's current position vector; These represent other ordinary wolf individuals. and Wolf, wolves and The distance of the wolves; It is a random vector; Calculate the individual characteristics of a common wolf Wolf, wolves and The process of the wolf approaching: , in, This indicates that an individual wolf is... Wolf, wolves and The wolf's approaching stride length; It is a random vector; Final update of the final location of the ordinary wolf individual: , in, This is a random disturbance term.
[0049] S5: Iteration and Output Results.
[0050] Specifically, the process of step S5 is as follows: if the maximum number of iterations has not been reached... max_iterIf the fitness value is not found, repeat steps S3 to S4 (i.e., repeatedly calculate the fitness value, population stratification, and update the individual position). Otherwise, output the best individual in the population, which is the result of the algorithm iteration. The best individual corresponds to the heat and power load allocation strategy that minimizes the total coal consumption of the entire plant under various constraints of the cogeneration power plant.
[0051] Specifically, this method employs chaotic mapping to initialize the population. Leveraging the randomness, ergodicity, and regularity of chaotic systems, a seemingly random sequence is generated within a specific range according to certain rules using a Logistic mapping. This sequence is then mapped onto the problem search space to obtain the initial population. The generated initial population is evenly distributed throughout the search space. Compared to traditional random initialization, this effectively overcomes the drawback of population clustering in localized areas, which is common in traditional random initialization. This allows the algorithm to have a rich variety of search starting points from the initial stage, greatly expanding the search range and enabling the algorithm to better adapt to real-time system changes, improving real-time performance and significantly increasing the probability of finding the global optimum in complex nonlinear and strongly coupled problems such as cogeneration systems.
[0052] Specifically, this method cleverly divides the gray wolf population into elite wolves, potential wolves, and ordinary wolves. Elite wolves consist of the gray wolves with the best fitness values. Utilizing a weighted average of individual historical optimal positions and the group's optimal position, the population is precisely guided towards the currently perceived optimal region, focusing on executing refined local search tasks. In the optimization of cogeneration systems, this mechanism can precisely adjust the unit's power generation and heating parameters, thereby effectively reducing coal consumption. Potential wolves employ an exponentially weighted moving average algorithm, which retains some historical movement trend information while focusing on recent position changes, thereby uncovering entirely new potential search areas. This significantly enhances the algorithm's global exploration capability in high-dimensional solution spaces and multi-objective conflict scenarios, effectively preventing the algorithm from getting trapped in local optima. Ordinary wolves, by retaining the original attack logic and adding random perturbation terms, further increase the probability of the population escaping local optima, ensuring the algorithm's stable operation under various complex conditions. These three different levels of wolves—elite, potential, and ordinary—achieve a comprehensive trade-off across multiple objectives through different strategies.
[0053] Specifically, this method balances these two capabilities through the collaborative operation of elite wolves and potential wolves. Elite wolves focus on refined local searches, improving local search accuracy; potential wolves explore new potential search areas in a relatively broader space, enhancing global exploration capabilities. By dynamically adjusting the algorithm's global and local search capabilities, the algorithm can quickly converge to near-optimal solutions while maintaining sufficient exploration ability to escape local optima when solving complex optimization problems. This improves the overall performance of the algorithm, effectively solving optimization problems in complex scenarios of combined heat and power (CHP) plants. It can accurately formulate more reasonable CHP load allocation strategies, effectively reducing unnecessary energy losses, thereby significantly reducing coal consumption and improving energy efficiency, bringing tangible energy-saving benefits to CHP systems.
[0054] The above embodiments merely illustrate the basic principles and characteristics of the present invention, but are not limited to the above implementation schemes. It should be understood that those skilled in the art can make various changes and modifications to the present invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined freely by the appended claims and their equivalents.
Claims
1. A method for allocating cogeneration load in a combined heat and power plant based on an improved gray wolf algorithm, characterized in that, Includes the following steps: S1: Set parameters; S2: Use Logistic mapping to generate chaotic sequences, map the chaotic variables in the chaotic sequences to the solution space of the cogeneration load allocation problem, and obtain the initial population individuals; S3: Calculate the fitness value based on the optimization objective function of the cogeneration power plant and select the best individuals; S4: Based on the fitness calculated in step S3, the population is stratified into different levels, and different position update methods are adopted according to the characteristics of different levels. S5: Iteration and Output Results.
2. The cogeneration load allocation method for cogeneration power plants based on the improved gray wolf algorithm according to claim 1, characterized in that, The parameters in step S1 include the number of gray wolf individuals, the maximum number of iterations, the problem dimension, the proportion of elite wolves, and the proportion of potential wolves.
3. The cogeneration load allocation method for cogeneration power plants based on the improved gray wolf algorithm according to claim 2, characterized in that, The specific process of step S2 is as follows: Using the Logistic mapping, the mathematical model is as follows: ， in, For control parameters; It is a chaotic variable, and its initial value is randomly selected in the interval (0, 1); The solution space of the cogeneration load allocation problem in a cogeneration power plant is: ,in, This represents the lower bound of the decision variable; This represents the upper limit of the decision variable; Mapping chaotic variables in chaotic sequences to the solution space of the combined heat and power (CHP) load allocation problem in cogeneration power plants. Thus, the initial population of individuals is obtained; ， in, It is a chaotic variable; This represents the value taken by an individual in the corresponding solution space.
4. The cogeneration load allocation method for cogeneration power plants based on the improved gray wolf algorithm according to claim 3, characterized in that, In step S3, the optimization objective function of the cogeneration power plant is based on the total coal consumption of the entire plant. The minimum target is considered, taking into account the content of cogeneration power plants. The objective function for the Taiwan combined heat and power unit is: ， in, Indicates the first The coal consumption function of the unit is obtained based on the unit's energy consumption curve; Indicates the first Power generation of the unit; Indicates the first Heating supply of the unit; The objective function value calculated using the objective function is the fitness value. The three individuals with the best fitness values are selected as the top three individuals. Wolf, wolves and Wolves, the rest of the individuals in the population are classified as Wolf.
5. The cogeneration load allocation method for cogeneration power plants based on the improved gray wolf algorithm according to claim 4, characterized in that, In step S3, when calculating fitness, constraints need to be considered simultaneously: the internal constraint is that the energy supply of a single unit has upper and lower limits, that is: in, Indicates the first The lower limit of power generation capacity of the unit; Indicates the first The maximum power generation capacity of the unit; Indicates the first The lower limit of the heating capacity of the unit; Indicates the first The maximum heat supply capacity of the unit; The global constraint is that the total energy supply of the combined heat and power plant must meet the electrical and thermal load demands of the power grid and the heating network, that is: ， in, This represents the electrical load demand of the power grid and heating network; This represents the heat load demand of the power grid and heating network.
6. The cogeneration load allocation method for cogeneration power plants based on the improved gray wolf algorithm according to claim 5, characterized in that, The specific process of step S4 is as follows: the entire population is arranged according to the fitness calculated in step S3. The individuals in the population whose fitness ranking is within the specified percentage are selected as elite wolves, and the individuals whose fitness has dropped twice consecutively are selected as potential wolves, based on the specified percentage of potential wolves. The number of potential wolves is at most the specified upper limit. The remaining wolves in the population are ordinary wolves.
7. The cogeneration load allocation method for cogeneration power plants based on the improved gray wolf algorithm according to claim 6, characterized in that, In step S4, the mathematical model for the attack process of the elite wolf is as follows: Computational Elite Wolf Tracking Wolf, wolves and The process of the wolf: ， in, Indicates the position of the elite wolf; This represents the number of iterations. They represent Wolf, wolves and The wolf's current position vector; These represent other elite wolf individuals. and Wolf, wolves and The distance of the wolves; It is a random vector; Calculate the individual characteristics of elite wolves Wolf, wolves and The process of the wolf approaching: ， in, This indicates that individual wolves in the elite group are... Wolf, wolves and The step size for moving closer; It is a random vector; Final update on the final location of the elite wolf individual: ， in, Represented as the individual's historical best position; The weighted value represents the individual's historical best position, and its value ranges from (0, 1). Indicates the final location of an individual.
8. The cogeneration load allocation method for cogeneration power plants based on the improved gray wolf algorithm according to claim 7, characterized in that, In step S4, the mathematical model for the attack process of the potential wolf is as follows: Calculate potential wolf tracking Wolf, wolves and The process of the wolf: ， in, Indicates the potential wolf's position; This represents the number of iterations. They represent Wolf, wolves and The wolf's current position vector; These represent other potential wolf individuals. and Wolf, wolves and The distance of the wolves; It is a random vector; Calculating the potential of individual wolves Wolf, wolves and The process of the wolf approaching: ， in, This indicates the potential of individual wolves. Wolf, wolves and The wolf's approaching stride length; It is a random vector; Final update on the final position of the potential wolf individual: ， in, This is a trend factor with a value range of (0, 1), representing the degree to which an individual maintains its self-movement inertia; It is an exponentially weighted shift vector; This is the movement vector for the current iteration. The attenuation coefficient has a value range of (0, 1) and represents the weight of the historical movement vector.
9. The cogeneration load allocation method for cogeneration power plants based on the improved gray wolf algorithm according to claim 8, characterized in that, In step S4, the mathematical model for the attack process of a common wolf is as follows: Calculate ordinary wolf tracking Wolf, wolves and The process of the wolf: ， in, Indicates the position of a regular wolf; This represents the number of iterations. They represent Wolf, wolves and The wolf's current position vector; These represent other ordinary wolf individuals. and Wolf, wolves and The distance of the wolves; It is a random vector; Calculate the individual characteristics of a common wolf Wolf, wolves and The process of the wolf approaching: ， in, This indicates that an individual wolf is... Wolf, wolves and The step length the wolf takes to approach; It is a random vector; Final update of the final position of the ordinary wolf individual: ， in, This is a random disturbance term.
10. The cogeneration load allocation method for cogeneration power plants based on the improved gray wolf algorithm according to claim 9, characterized in that, The specific process of step S5 is as follows: if the maximum number of iterations has not been reached, repeat steps S3 to S4; otherwise, output the optimal individual in the population, which is the algorithm iteration result. The optimal individual is the heat and power load allocation strategy that minimizes the total coal consumption of the entire plant under various constraints of the cogeneration power plant.