CN-116101496-B - Control method for energy efficiency optimization of electric environmental control system of airplane

Abstract

The invention discloses a control method for energy efficiency optimization of an aircraft Electric Environmental Control System (EECS), which changes the traditional regional control strategy into a combination of three-variable optimization and double-target control, takes coefficient of performance (COP) of the system as an index, and maximizes the COP by solving an optimal solution set of optimization variables under different working conditions. On the basis, the controller calculates the values of the two control quantities to meet the performance requirements of the system. Firstly, a GRNN model is established, and through a large amount of offline data training under each working condition, the steady-state COP of the system under different working conditions can be accurately predicted. Subsequently, the DE algorithm searches the solution space for the optimal variable solution set corresponding to the optimal COP that satisfies the component performance constraints. And finally, solving corresponding control quantity under the optimal operation point by the controller, thereby realizing energy efficiency optimization control.

Inventors

• ZHENG FENGYING
• HE ZHONGZE
• ZHANG JINGYANG
• HU WENCHAO
• FU JIECHENG
• JIN XINGJIAN
• ZHU WENJIE

Assignees

• 南京航空航天大学

Dates

Publication Date

20260512

Application Date

20221114

Claims (4)

1. 1. The control method of energy efficiency optimization of an aircraft electric environmental control system comprises the steps that an electric environmental control system EECS is additionally provided with a motor-driven air compressor to compress ambient air as an air source, air is directly introduced from the ambient air, the motor drives the air compressor to compress ram air, the ram air is pressurized by the air compressor after being cooled by a primary radiator and then enters a secondary heat exchanger to be continuously cooled, the cooled air is cooled and dehumidified by a condenser and then sequentially enters a primary turbine cooler and a secondary turbine cooler to be cooled, and finally the obtained cooled air is supplied to a cabin to balance heat load; The method is characterized in that a traditional regional control strategy is changed into a combination of three-variable optimization and double-target control, the COP of the refrigeration coefficient of EECS is used as an index, the COP is maximized by solving an optimal solution set of the optimization variable under different working conditions, and on the basis, the two control quantity values are solved by a controller to meet the performance requirement of a system; The method comprises the following steps: S1, designing EECS control scheme and building EECS model On the basis of meeting the refrigerating capacity and air supply capacity demand of a cabin, according to the component configuration, control function and framework demand analysis of EECS, designing an optimized variable, a control object and a control capacity, wherein the control object of EECS is the air supply capacity meeting EECS performance and the air supply temperature calculated through the refrigerating capacity and the air supply capacity, an actuating mechanism comprises an electric compressor, a cold air channel valve, a bypass valve, an economic refrigerating valve and a low-limit valve, wherein the bypass valve, the economic refrigerating valve and the low-limit valve are changed to be heat flow, the corresponding node temperatures can be adjusted to finally influence the air supply temperature, the cold air channel valve is changed to be cold flow, the bypass valve, the economic refrigerating valve and the low-limit valve are selected as optimized variables for maximizing the COP of EECS, the cold air channel valve is used as the control capacity to meet the air supply temperature demand, and the electric compressor speed is used as the control capacity to meet the air supply capacity demand; s2, acquiring a learning sample of the COP prediction model The method comprises the steps of predicting steady-state COP (coefficient of performance) of EECS under different environment parameters and operation parameters through a generalized regression neural network GRNN, wherein the input of the GRNN comprises three operation parameters including a cruising height, a flight Mach number, a cabin heat load, a bypass valve opening, an economic refrigeration valve opening and a low-limit valve opening, the output of the GRNN is the outlet temperature of an electric compressor and the air supply temperature of the cabin, a large number of simulation data are used for collecting learning samples, the number of the samples is not less than 2400 groups, the three environment parameters including the cruising height, the flight Mach number and the cabin heat load are set according to gradients, the three operation parameters including the bypass valve opening, the economic refrigeration valve opening and the low-limit valve opening are set according to gradients under one group of environment parameters, the air supply quantity requirement and the refrigeration quantity requirement of the cabin are met by respectively controlling the rotation speed of the electric compressor and the opening of the cold air channel valve through two PID controllers, and the outlet temperature of the electric compressor and the air supply temperature of the cabin under the steady state are recorded as learning samples; s3, constructing a topological structure of a generalized regression neural network GRNN GRNN is four-layer network, including input layer, mode layer, summation layer and output layer, the input layer sets up six neurons, corresponds current cruising height, flight Mach number, cabin heat load, bypass valve aperture, economic refrigeration valve aperture and low limit valve aperture respectively, and the input layer neuron is direct to be transmitted the input variable for the mode layer, and the neuron number of mode layer is the study sample number, and the output of single neuron is: (1) Wherein, the Is the six input vectors of the network, Is the first A learning sample corresponding to each neuron, Is the width of the neuron; The summation layer performs summation calculation by adopting two kinds of neurons, wherein the total number is one added to the dimension of the output vector, and all the outputs of the first mode layer are directly added, and the outputs of the first mode layer are output Described as (2) The second type performs weighted summation on the outputs of the mode layers, the outputs of which Can be described as (3) Wherein, the Is the first The first output sample An element; the output layer is provided with two neurons which respectively correspond to the outlet temperature of the electric compressor And cabin air supply temperature The output layer divides the output of the summing layer, namely: (4) let the cabin air recirculation flow be half the supply, take the temperature rise of the recirculation air through the fan to 4 ℃, then the COP of EECS is obtained by: (5) Wherein, the For the refrigerating capacity of the cabin, The power consumption of the system is EECS, Is the constant pressure specific heat capacity of the air, For the air supply temperature, In order to be at the temperature of the environment, Is the air supply flow; S4, training a generalized regression neural network 3-network GRNN model; equally dividing the learning sample obtained in the step S2 into two groups, wherein one group is used as a training set for training the generalized regression neural network, and the other group is used as a test set for verifying the accuracy of GRNN (generic regression neural network) on the prediction of COP (coefficient of performance) of systems under different working conditions; s5, setting individual and fitness functions of a differential evolution algorithm; Each individual in the differential evolution algorithm represents a group of valve opening degrees, the individual value after each iteration is substituted into a prediction model to obtain a corresponding predicted value, and the corresponding predicted value is converted to obtain a COP value as an adaptability value of the differential evolution algorithm; s6, setting differential evolution algorithm parameters The population number of the differential evolution algorithm is 30-100, the maximum iteration number is 100-200, and the values of the variation factor and the crossover factor are selected according to the advantages and disadvantages of the convergence result of the algorithm; s7, constraint is applied to the differential evolution algorithm according to EECS performance requirements and the working characteristics of the components; The solution set obtained by the differential evolution algorithm needs to meet the following conditions, namely, each group of solutions needs to be generated under the same refrigerating capacity and air supply capacity, 2) each group of solutions needs to consider the running state EECS, the opening degree of a valve needs to be within 0-90 ℃, and the outlet temperature of a gas compressor and the inlet temperature of a condenser need to be within a normal working range; For the first type of conditions, since the learning samples of the GRNN are processed by the PID controller, only constraint is needed to be applied to the second type of conditions, the outlet temperature of the air compressor is lower than 210 ℃, the inlet temperature of the condenser is higher than the dew point, and the opening of the valve is within 0-90 ℃; s8, performing iterative computation of a differential evolution algorithm, and solving operation parameters corresponding to the optimal COP (1) Initialization of Each individual in the differential evolution algorithm is a three-dimensional vector representing a set of valve openings, described as: The initial population may be generated randomly from the feasible domain, (6) Wherein, the The lower bound of the value is taken for the individual, The upper bound of the value is taken for the individual, Is the population number; (2) Variation of The variant vector is generated by weighting two independent individuals onto one another, which lays the foundation for the offspring population, the variant individual descriptions are: (7) Wherein, the Is a variation factor; (3) Crossover To enhance diversity in a population, partially mutated individuals are placed into the population: (8) Wherein, the As a result of the crossover factor, Is a random integer of 1 to 3, ensuring that at least one dimensional variable is derived from the mutation operation; (4) Selection of After each iteration is completed, selecting a better individual to perform the next iteration by calculating and comparing the fitness value of the individual, repeating the steps (2) to (4) until the preset maximum iteration times are reached, and finally obtaining the valve opening corresponding to the optimal COP: (9) Wherein, the For the number of iterations, Is the fitness function, i.e., COP.
2. 2. The method for energy efficiency optimization control of an electronic control system of an aircraft according to claim 1, wherein in S2, 4200 groups are collected by simulating data to collect the number of learning samples.
3. 3. The method for optimizing and controlling the energy efficiency of the aircraft electric environmental control system according to claim 1 or 2, wherein in the step S6, the values of the variation factor and the crossover factor are selected according to the advantages and disadvantages of the algorithm convergence result, namely ten mutually independent random initialization tests are carried out on the algorithm, when the algorithm can converge within the maximum iteration number of each test result, and the deviation between the maximum value and the minimum value of COP in all convergence results is smaller than 5%, the values of the variation factor and the crossover factor are considered to be suitable, otherwise, the variation factor and the crossover factor are scaled, and ten mutually independent random initialization tests are carried out again until the convergence result can meet the evaluation criterion.
4. 4. The method for optimizing and controlling the energy efficiency of the electric environmental control system of the aircraft according to claim 3, wherein in S6, the differential evolution algorithm takes the population number of 50, the maximum iteration number of 200, the variation factor of 0.7 and the crossover factor of 0.5.

Description

Control method for energy efficiency optimization of electric environmental control system of airplane Technical Field The invention relates to aircraft cabin environment control, in particular to a control method for energy efficiency optimization of an aircraft Electric Environmental Control System (EECS), and belongs to the technical field of energy efficiency management and control of aircraft thermal management systems. Background The aircraft environmental control system performs ventilation and temperature control functions by providing air to the cabin to maintain a comfortable cabin environment for humans. Currently, most environmental control systems employ an engine bleed air solution, which is regulated to the appropriate temperature and pressure to supply the nacelle by implementing a series of control and power optimization techniques on the air circulation components. However, the air discharged from the engine has a negative effect on the fuel consumption of the engine. To ensure that the engine is operated in an economy mode, the temperature and pressure of the bleed air are typically set to constant values, which are often higher than the actual needs of the environmental control system, resulting in a significant amount of power wastage. With the increasing complexity of flight tasks and the increasing performance requirements of the complete machine, this approach is increasingly limited. The electrical control system (EECS) eliminates the conventional pneumatic system and converts the air supply from the engine bleed air to ambient air compressed by the electrical compressor. The system adopts a refrigerating mode of two-stage compression and two-stage expansion, the internal circulation state of the system is adjusted by a plurality of groups of valves in a coordinated manner, and the rotating speed of the electric compressor and the electric fan can be freely adjusted according to different working conditions. For the same flight mission, the same control objective can be realized through different flow distribution among the flow passages, so that the possibility of energy efficiency optimization is generated. However, the whole system has strong nonlinearity, multiple execution mechanisms and strong coupling effect among control targets, the traditional control strategy cannot give consideration to the synergistic effect among all control channels, and the energy efficiency optimization of the system is difficult to realize. From the current research situation at home and abroad, in terms of control strategies, a region control strategy is commonly used, namely, the aim of reducing the order is achieved by dividing EECS into a plurality of regions with independent functions, and then a control algorithm is independently designed for each region. Optimization technology research is mainly focused on component-level parameter optimization and system architecture optimization. For example, part of scholars respectively model the refrigerating system by a heat flow method and a parameter matching method, and give out the optimal structure and size parameters of the system from the angles of minimum total take-off weight and fuel loss. However, for specific system and specific task requirements, how to combine the optimization method with the control method and achieve the purpose of optimization by adjusting the working state of the system, and the mature theoretical system and evaluation method are still lacking at present. In recent years, the rise of artificial neural networks has attracted extensive attention from students. A particular advantage of neural networks in the control and optimization field is their ability to approximate nonlinear functions with high precision. In the optimization problem of insufficient object information, neural networks are often used as object substitutes. By building a neural network with a suitable structure according to a specific problem, an object can be approximated with a high accuracy with limited data. And solving the trained neural network model by combining a numerical optimization algorithm to obtain an approximate optimal solution of the original problem. At present, in the field of environmental control systems, the design of corresponding controllers and optimizers by using a neural network is still rarely reported, but in the problems of optimization and control of complex systems and complex processes, the neural network is widely applied, and has certain reference significance. Disclosure of Invention The technical problem to be solved by the invention is to overcome the defects of the prior art, and provide a control method for optimizing the energy efficiency of an aircraft Electric Environmental Control System (EECS), which adopts the following technical scheme that the electric environmental control system EECS directly introduces air from the environment atmosphere by additionally configuring air compressor compressed environm