CN106371318A

CN106371318A - Facility environment multi-objective optimization control method based on cooperative game

Info

Publication number: CN106371318A
Application number: CN201610965089.4A
Authority: CN
Inventors: 王浩云; 徐焕良; 任守纲; 王珂; 翟肇裕
Original assignee: Nanjing Agricultural University
Current assignee: Nanjing Agricultural University
Priority date: 2016-10-31
Filing date: 2016-10-31
Publication date: 2017-02-01
Anticipated expiration: 2036-10-31
Also published as: CN106371318B

Abstract

一种基于合作博弈的设施环境多目标优化控制方法，它包括以下步骤：S1、建立设施环境的三个控制目标：环境参数‑温湿度控制目标、能耗控制目标以及控制设备物理极限性，即设备损耗控制目标；S2、建立合作博弈模型，采用基于合作博弈模型的多目标模型预测控制方法对前述设施环境的三个控制目标的进行处理，得到每个控制目标的优化函数；S3、求解所得的目标的优化函数，获取各目标函数的最优控制量。本发明借鉴博弈论思想，将各个控制目标看作各个博弈方，通过建立合作竞争模型，使得每个博弈方在考虑个人利益的同时也兼顾他人利益，从而解决多目标冲突问题。A kind of facility environment multi-objective optimal control method based on cooperative game, it comprises the following steps: S1, establish three control objectives of facility environment: environmental parameter - temperature and humidity control target, energy consumption control target and control equipment physical limitation, namely Equipment loss control objectives; S2, establish a cooperative game model, and use the multi-objective model predictive control method based on the cooperative game model to process the three control objectives of the aforementioned facility environment, and obtain the optimization function of each control objective; S3, the obtained solution The optimization function of the objective of each objective function is obtained to obtain the optimal control amount of each objective function. The present invention draws on the ideas of game theory, regards each control target as each player, and establishes a cooperation and competition model to make each player take into account the interests of others while considering the interests of others, thereby solving the multi-objective conflict problem.

Description

Multi-objective Optimal Control Method of Facility Environment Based on Cooperative Game

技术领域technical field

本发明涉及智能农业领域，尤其是一种基于合作博弈的设施环境多目标优化控制方法。The invention relates to the field of intelligent agriculture, in particular to a cooperative game-based multi-objective optimal control method for facilities and environments.

背景技术Background technique

目前，以现代化农业生产方式、集成应用感知识别技术、监控技术、通信网络技术、云计算与大数据技术等泛在网(互联网、物联网)技术为显著特征，实现农业生产过程的智能控制、智慧生产、农产品全程质量控制和可追溯，是农业4.0所体现的智能化、产业化、标准化的高效集约、绿色低碳现代农业业态。设施农业是现代农业发展的象征，是当今世界最具活力的产业之一，也是世界各国用以提供新鲜农产品的主要技术措施，而农业4.0所倡导的智能(环境)控制是其关键技术。At present, ubiquitous network (Internet, Internet of Things) technologies such as modern agricultural production methods, integrated application perception and identification technology, monitoring technology, communication network technology, cloud computing and big data technology are prominent features to realize intelligent control of agricultural production process, Smart production, whole-process quality control and traceability of agricultural products are intelligent, industrialized, standardized, efficient, intensive, green and low-carbon modern agricultural formats embodied in Agriculture 4.0. Facility agriculture is a symbol of modern agricultural development and one of the most dynamic industries in the world today. It is also the main technical measure used by countries around the world to provide fresh agricultural products, and the intelligent (environmental) control advocated by Agriculture 4.0 is its key technology.

一般地，设施环境中的主要控制方法为PID(Proportion IntegrationDifferentiation)控制和模糊控制，近年来，由于模型预测控制(Model PredictiveControl,MPC)技术的发展及其在工业过程中显现出的处理复杂约束优化控制问题的巨大潜力，使得MPC的应用从传统的工业过程，例如炼油、石化、化工行业，延伸到电力、钢铁、船舶、环境、医疗、农业等领域。传统的温室控制仅考虑单目标的控制，即要求设施环境的温度、湿度、光照及营养等因子达到一个合适的值，而对能耗等其他因素以及设备的物理极限性等其他控制目标缺乏考虑，导致在理论上的优化控制算法由于运行成本高而不能有效应用于实际设施环境中。因此，在设施环境控制中，考虑生产能耗以及设备损耗显得尤为重要。而既能保证作物生长所需的最佳环境，从而使得作物产量最高，又能够有效降低温室生产的能耗和生产成本，从而获得更大的经济效益，这样的控制策略正受到越来越多的关注。由此可见，设施环境的控制实际上是一个多目标控制问题。Generally, the main control methods in the facility environment are PID (Proportion Integration Differentiation) control and fuzzy control. The huge potential of control problems makes the application of MPC extend from traditional industrial processes, such as oil refining, petrochemical, chemical industry, to electric power, steel, shipbuilding, environment, medical treatment, agriculture and other fields. Traditional greenhouse control only considers single-objective control, which requires factors such as temperature, humidity, light, and nutrition in the facility environment to reach an appropriate value, but lacks consideration of other factors such as energy consumption and physical limitations of equipment. , leading to the theoretical optimal control algorithm cannot be effectively applied to the actual facility environment due to high operating costs. Therefore, it is particularly important to consider production energy consumption and equipment loss in facility environment control. And it can not only ensure the best environment for crop growth, so as to make the highest crop yield, but also effectively reduce the energy consumption and production cost of greenhouse production, so as to obtain greater economic benefits. Such control strategies are being increasingly accepted. s concern. It can be seen that the control of facility environment is actually a multi-objective control problem.

在多目标优化设计中，各目标相互之间一般是冲突的，例如，在设施环境控制中，精确控制与控制能耗之间。为了调和这些冲突，就需要以某种方式来解决各目标之间的矛盾。而博弈论是研究决策主体的行为在直接相互作用时，如何进行决策以及这种决策如何达到均衡的问题，目前博弈论在工程设计领域的应用已经越来越多。鉴于多目标优化问题和博弈问题的相似性，可以将博弈论思想和方法引入到多目标优化设计问题的求解之中，以克服传统多目标优化设计问题求解方法的不足。In multi-objective optimization design, the objectives are generally in conflict with each other, for example, between precise control and control of energy consumption in facility environment control. In order to reconcile these conflicts, it is necessary to resolve the contradictions between the various goals in some way. Game theory is the study of how to make decisions and how to achieve equilibrium when the behavior of decision-making subjects interacts directly. At present, game theory has been applied more and more in the field of engineering design. In view of the similarity between multi-objective optimization problems and game problems, game theory ideas and methods can be introduced into the solution of multi-objective optimization design problems to overcome the shortcomings of traditional multi-objective optimization design problem solving methods.

发明内容Contents of the invention

本发明的目的是针对设施环境中的多目标控制问题，提出一种基于合作博弈的设施环境多目标优化控制方法。根据控制目标，将整体控制系统分解为若干个子系统，每个子系统通过优化各自的目标函数求得系统整体最优。为了实现各个控制目标的均衡求解，借鉴博弈论思想，将各个控制目标看作各个博弈方，通过建立合作竞争模型，使得每个博弈方在考虑个人利益的同时也兼顾他人利益，从而解决多目标冲突问题。The purpose of the present invention is to propose a multi-objective optimal control method for the facility environment based on cooperative game aiming at the multi-objective control problem in the facility environment. According to the control objective, the overall control system is decomposed into several subsystems, and each subsystem obtains the overall optimum of the system by optimizing its own objective function. In order to realize the balanced solution of each control target, we refer to game theory and regard each control target as each game party. By establishing a cooperative competition model, each game party can consider the interests of others while considering the interests of others, so as to solve the multi-objective problem. Conflict issues.

本发明的技术方案是：Technical scheme of the present invention is:

一种基于合作博弈的设施环境多目标优化控制方法，它包括以下步骤：A multi-objective optimal control method for facility environment based on cooperative game, which comprises the following steps:

S1、建立设施环境的三个控制目标：环境参数-温湿度控制目标、能耗控制目标以及控制设备物理极限性，即设备损耗控制目标；S1. Establish three control objectives of the facility environment: environmental parameters - temperature and humidity control objectives, energy consumption control objectives, and physical limit of control equipment, that is, equipment loss control objectives;

S2、建立合作博弈模型，采用基于合作博弈模型的多目标模型预测控制方法对前述设施环境的三个控制目标的进行处理，得到每个控制目标的优化函数；S2. Establishing a cooperative game model, using a multi-objective model predictive control method based on the cooperative game model to process the three control objectives of the aforementioned facility environment, and obtaining an optimization function for each control objective;

S3、求解所得的目标的优化函数，获取各目标函数的最优控制量。S3. Solve the obtained optimization function of the objective, and obtain the optimal control quantity of each objective function.

本发明的步骤S1具体为：Step S1 of the present invention is specifically:

S1-1、建立将设施的环境参数-温湿度控制目标函数f₁，描述为：S1-1. Establish the environmental parameter of the facility - the temperature and humidity control objective function f ₁ , which is described as:

${f f}_{11} = = m m i i n no [[{Σ Σ}_{j j = = 11}^{p p} {(({T T}_{i i n no} ((k k + + j j | | k k)) - - {T T}_{s the s e e t t}))}^{22} + + {Σ Σ}_{j j = = 11}^{p p} {(({H h}_{i i n no} ((k k + + j j | | k k)) - - {H h}_{s the s e e t t}))}^{22}]]$

其中：P表示预测时域，j表示预测步长，k表示当前时刻；T_in(k+j|k)表示当前时刻k获取的未来时刻k+j的室内温度(根据设施温室小气候预测模型获得)，H_in(k+j|k)表示当前时刻k获取的未来时刻k+j的室内湿度(根据设施温室小气候预测模型获得)，T_set和H_set分别为作物生长所适宜的温度值和湿度值；Among them: P represents the forecast time domain, j represents the forecast step size, k represents the current moment; T _in (k+j|k) represents the indoor temperature at the future time k+j obtained at the current moment k (obtained according to the facility greenhouse microclimate prediction model ), H _in (k+j|k) represents the indoor humidity at the future time k+j obtained at the current time k (according to the greenhouse microclimate prediction model of the facility), T _set and H _set are the temperature values suitable for crop growth and Humidity value;

S1-2、建立设施控制的能耗控制目标函数f₂，描述为：S1-2. Establish an energy consumption control objective function f ₂ for facility control, described as:

${f f}_{22} = = m m i i n no [[{Σ Σ}_{i i = = 11}^{N N} {Σ Σ}_{{j j}^{' '} = = 00}^{M m - - 11} (({u u}_{i i} ((k k + + {j j}^{' '} | | k k)) / / {u u}_{i i__m m a a x x})) / / M m]]$

其中，N表示控制设备的总数量，i表示当前控制设备的编号；M表示控制时域，j′表示控制步长；u_i(k+j′|k)表示当前时刻k获取的未来k+j时刻控制设备i的最优控制量(根据控制子系统优目标函数得到)，u_{i_max}为控制设备i的最大控制量(控制设备例如设施内加热器的最大功率、天窗的最大开启角度、风机的开启个数等)；Among them, N represents the total number of control devices, i represents the serial number of the current control device; M represents the control time domain, j′ represents the control step size; u _i (k+j′|k) represents the future k+ Control the optimal control quantity of equipment i at time j (according to the optimal objective function of the control subsystem), u _{i_max} is the maximum control quantity of control equipment i (control equipment such as the maximum power of the heater in the facility, the maximum opening angle of the skylight, the maximum opening angle of the fan number of openings, etc.);

S1-3、建立设施中设备损耗控制目标函数f₃，描述为：S1-3. Establish an objective function f ₃ for equipment loss control in the facility, described as:

${f f}_{33} = = m m i i n no {Σ Σ}_{i i = = 11}^{N N} (({u u}_{i i} ((k k + + {j j}^{' '} + + 11 | | k k)) - - {u u}_{i i} ((k k + + {j j}^{' '} | | k k)))) / / {u u}_{i i__max max}$

其中：u_i(k+j′+1|k)表示当前时刻k获取的未来k+j+1时刻控制设备i的最优控制量(根据控制子系统优目标函数得到)。Among them: u _i (k+j′+1|k) represents the optimal control quantity of the control device i obtained at the current moment k at the future k+j+1 moment (according to the optimal objective function of the control subsystem).

本发明的步骤S2具体为：Step S2 of the present invention is specifically:

S2-1、将三个控制目标即三个设施控制子系统分别作为博弈方，采用下述公式建立合作博弈模型：S2-1. Taking the three control objectives, that is, the three facility control subsystems as game parties respectively, the following formula is used to establish a cooperative game model:

${s the s}_{l l} (({u u}_{l l}^{* *})) = = \underset{{u u}_{l l} &Element; &Element; {U u}_{l l},, {u u}_{{l l}^{' '}} &Element; &Element; {U u}_{{l l}^{' '}}}{m m a a x x} [[{w w}_{l l l l} \times \times {\overset{&OverBar; &OverBar;}{s the s}}_{l l} (({u u}_{l l},, {u u}_{{l l}^{' '}})) + + {Σ Σ}_{{l l}^{' '} = = 11 (({l l}^{' '} &NotEqual; &NotEqual; l l))}^{m m} {w w}_{{ll ll}^{' '}} \times \times {\overset{^^}{s the s}}_{{ll ll}^{' '}} (({u u}_{l l},, {u u}_{{l l}^{' '}}))]]$

式中，l和l′表示博弈方的编号，m表示博弈方的总数量；U_l和U_l′表示博弈方l和l′的策略集，u_l和u_l′表示博弈方l和l′采用的策略，表示博弈方l能够采用的最优策略；表示博弈方l和l′分别采用策略u_l和u_l′时博弈方l的绝对收益，表示博弈方l和l′分别采用策略u_l和u_l′时博弈方l′的绝对收益，表示博弈方l采取最优策略时所有博弈方能获得的最大绝对收益总和；w_ll和w_ll′为比例系数，w_ll表示博弈方l的不合作程度，w_ll′表示博弈方l的合作程度，且 In the formula, l and l' represent the numbers of the players, m represents the total number of players; U _l and U _l' represent the strategy sets of the players l and l', u _l and u _l' represent the players l and l ' the strategy adopted, Indicates the optimal strategy that player l can adopt; Indicates the absolute revenue of player l when players l and l′ adopt strategies u _l and u _l ′ respectively, Indicates the absolute income of the player l' when the players l and l' respectively adopt strategies u _l and u _l' , Indicates that player l adopts the optimal strategy When is the maximum sum of absolute benefits that all players can obtain; w _ll and w _ll′ are proportional coefficients, w _ll represents the degree of non-cooperation of player l, w _ll′ represents the degree of cooperation of player l, and

S2-2、基于前述合作博弈模型建立多目标模型预测控制方法，得到设施环境中各子系统的优化目标函数为：S2-2. Establish a multi-objective model predictive control method based on the aforementioned cooperative game model, and obtain the optimization objective function of each subsystem in the facility environment as:

${S S}_{l l} (({u u}_{l l__i i}^{* *})) = = max max [[{w w}_{l l l l} \times \times \frac{{f f}_{l l} (({u u}_{l l__i i}^{* *},, {u u}_{{l l}^{' '}__{i i}^{' '}}))}{{f f}_{l l} (({u u}_{l l__i i},, {u u}_{{l l}^{' '}__{i i}^{' '}}))} + + {Σ Σ}_{{l l}^{' '} = = 11 (({l l}^{' '} &NotEqual; &NotEqual; l l))}^{33} {w w}_{{ll ll}^{' '}} \times \times \frac{{f f}_{{l l}^{' '}} (({u u}_{l l__i i}^{* *},, {u u}_{{l l}^{' '}__{i i}^{' '}}))}{{f f}_{{l l}^{' '}} (({u u}_{l l__i i},, {u u}_{{l l}^{' '}__{i i}^{' '}}))}]]$

其中：l和l′表示设施控制子系统的编号，即博弈方；u_{l_i}和u_{l′_i′}表示控制子系统l和l′内控制设备i和i′的控制量，即博弈方采用的策略；表示控制子系统l内控制设备i的最优控制量，即最优策略；f_l和f_l′表示控制子系统l和l′的控制目标函数，即绝对收益；f_l(u_{l_i},u_{l′_i′})和f_l′(u_{l_i},u_{l′_i′})表示控制子系统l和l′中控制设备i和i′的控制量为u_l，i和u_l′，i′时各子系统的绝对收益；表示当控制子系统l内控制设备i采用最优控制量时整个控制系统的最大绝对收益。Among them: l and l' represent the number of the facility control subsystem, that is, the player; u _{l_i} and u _{l'_i'} represent the control amount of the control equipment i and i' in the control subsystem l and l', that is, the game party adopts Strategy; Indicates the optimal control amount of the control device i in the control subsystem l, that is, the optimal strategy; f _l and f _l' represent the control objective functions of the control subsystem l and l', that is, the absolute profit; f _l (u _{l_i} ,u _{l′_i′} ) and f _l′ (u _{l_i} ,u _{l′_i′} ) represent the control variables of control devices i and i′ in control subsystems l and l′ when u _l,i and u _l′,i′ Absolute benefits of each subsystem; Indicates the maximum absolute gain of the entire control system when the control device i in the control subsystem l adopts the optimal control quantity.

本发明的步骤S3具体为：Step S3 of the present invention is specifically:

S3-1、各控制子系统分别随机初始化各自控制设备的控制序列u_{l_i}(0)；S3-1. Each control subsystem randomly initializes the control sequence u _{l_i} (0) of each control device;

u_{l_i}(0)＝[u_{l_i}(0|0),u_{l_i}(1|0),...,u_{l_i}(M-1|0)],l＝1,2,3u _{l_i} (0)=[u _{l_i} (0|0),u _{l_i} (1|0),...,u _{l_i} (M-1|0)], l=1,2,3

其中：M表示控制时域，l表示控制子系统编号，i表示控制设备的编号，u_{l_i}(0|0)表示控制子系统l中控制设备i第0时刻控制量的随机预设值，u_{l_i}(M-1|0)表示控制子系统l中控制设备i在第0时刻对其在M-1时刻控制量的随机预设值；Among them: M represents the control time domain, l represents the number of the control subsystem, i represents the number of the control device, u _{l_i} (0|0) represents the random preset value of the control quantity of the control device i in the control subsystem l at the 0th moment, u _{l_i} (M-1|0) represents the random preset value of control device i in the control subsystem l at time 0 for its control quantity at time M-1;

S3-2、各控制子系统之间相互通信，分别将自己时刻k的控制序列u_{l_i,q}(k)发送给其他子系统；S3-2. Each control subsystem communicates with each other, and sends its own control sequence u _{l_i,q} (k) at time k to other subsystems;

u_{l_i,q}(k)＝[u_{l_i}(k|k),u_{l_i}(k+1|k),...,u_{l_i}(k+M-1|k)]^q u _{l_i,q} (k)＝[u _{l_i} (k|k),u _{l_i} (k+1|k),...,u _{l_i} (k+M-1|k)] ^q

其中：u_{l_i}(k+M-1|k)表示控制子系统l中控制设备i在时刻k对其在时刻k+M-1控制量的预设值，q表示时刻k控制子系统l与其它子系统交互控制量的次数；Among them: u _{l_i} (k+M-1|k) represents the preset value of control device i in control subsystem l at time k to its control quantity at time k+M-1, and q represents the control subsystem l and The number of times other subsystems interact with control quantities;

S3-3、在获得其他控制子系统在时刻k第q次发送的控制序列后，控制子系统l根据自己的优化目标函数利用粒子群算法求出最优控制序列 S3-3. After obtaining the control sequence sent by other control subsystems for the qth time at time k, the control subsystem l uses the particle swarm optimization algorithm to find the optimal control sequence according to its own optimization objective function

${u u}_{l l__i i,, q q}^{* *} ((k k)) = = {[[{u u}_{l l__i i}^{* *} ((k k | | k k)),, {u u}_{l l__i i}^{* *} ((k k + + 11 | | k k)),, ... ...,, {u u}_{l l__i i}^{* *} ((k k + + M m - - 11 | | k k))]]}^{q q}$

其中，表示控制子系统l中控制设备i在时刻k对时刻k+M-1控制量的最优预设值；in, Indicates the optimal preset value of the control quantity of the control device i in the control subsystem l at time k to time k+M-1;

S3-4、对于任意一个博弈方，在时刻k如果满足q＝q_max或者则停止发送控制序列，并执行最优控制序列的第1个控制量之后，返回步骤S4-2进行时刻k+1的最优控制序列计算。S3-4. For any player, at time k if q=q _max or Then stop sending the control sequence and execute the optimal control sequence The first control amount of After that, return to step S4-2 to calculate the optimal control sequence at time k+1.

其中，q_max表示控制子系统在某一时刻发送控制序列次数的最大值，ε表示控制序列改进的门限值。Among them, q _max represents the maximum number of times the control subsystem sends the control sequence at a certain moment, and ε represents the threshold value of the control sequence improvement.

本发明的有益效果：Beneficial effects of the present invention:

本发明针对设施环境中的多目标控制问题，采用分布式模型预测控制方法，根据控制目标，将整体控制系统分解为若干个子系统，每个子系统通过优化各自的目标函数求得系统整体最优。为了实现各个控制目标的均衡求解，借鉴博弈论思想，将各个控制目标看作各个博弈方，通过建立合作竞争模型，使得每个博弈方在考虑个人利益的同时也兼顾他人利益，从而解决多目标冲突问题。为说明问题的方便起见，以设施环境的温、湿度参数及相应控制设备组成的多目标控制系统为例(可拓展到多参数形成的多目标控制系统)，在MATLAB仿真平台上验证了算法的有效性。Aiming at the multi-objective control problem in the facility environment, the present invention adopts a distributed model predictive control method, and decomposes the overall control system into several subsystems according to the control objectives, and each subsystem obtains the overall optimum of the system by optimizing its own objective function. In order to realize the balanced solution of each control target, we refer to game theory and regard each control target as each game party. By establishing a cooperative competition model, each game party can consider the interests of others while considering the interests of others, so as to solve the multi-objective problem. Conflict issues. For the convenience of explaining the problem, taking the multi-objective control system composed of the temperature and humidity parameters of the facility environment and the corresponding control equipment as an example (which can be extended to a multi-objective control system formed by multiple parameters), the algorithm is verified on the MATLAB simulation platform. effectiveness.

本发明对设施环境中多目标控制问题进行研究，针对设施环境模型的非线性，目标函数的复杂性以及约束性，提出了分布式模型预测控制在设施环境中的应用框架，并借鉴博弈论，采用合作博弈方法解决设施环境中多目标，即精确控制、能耗最小、设备开启频繁度最低三个目标的冲突问题。通过在MATLAB平台上以2015年春季、夏季、秋季和冬季各一天的气象数据为例进行了仿真与验证，并与单目标控制、传统的线性加权多目标控制进行对比。实验结果表明，本发明提出的算法在解决多目标冲突问题上具有明显的优势，不仅能够使得室内温湿度能够满足作物生长所需的适宜范围，同时大大地降低了控制设备的能耗以及设备损耗。The present invention studies the problem of multi-objective control in the facility environment. Aiming at the non-linearity of the facility environment model, the complexity and constraints of the objective function, the application framework of the distributed model predictive control in the facility environment is proposed, and game theory is used for reference. The cooperative game method is used to solve the multi-objective conflict problem in the facility environment, that is, precise control, minimum energy consumption, and minimum equipment opening frequency. The simulation and verification were carried out on the MATLAB platform by taking the meteorological data of each day in the spring, summer, autumn and winter of 2015 as an example, and compared with single-objective control and traditional linear weighted multi-objective control. The experimental results show that the algorithm proposed by the present invention has obvious advantages in solving the problem of multi-objective conflicts. It can not only make the indoor temperature and humidity meet the appropriate range required for crop growth, but also greatly reduce the energy consumption and equipment loss of the control equipment. .

具体实施方式detailed description

下面结合实施例对本发明作进一步的说明。The present invention will be further described below in conjunction with embodiment.

${f f}_{22} = = m m i i n no [[{Σ Σ}_{i i = = 11}^{N N} {Σ Σ}_{j j = = 00}^{M m - - 11} (({u u}_{i i} ((k k + + {j j}^{' '} | | k k)) / / {u u}_{i i__m m a a x x})) / / M m]]$

${f f}_{33} = = m m i i n no {Σ Σ}_{i i = = 11}^{N N} (({u u}_{i i} ((k k + + {j j}^{' '} + + 11 | | k k)) - - {u u}_{i i} ((k k + + {j j}^{' '} | | k k)))) / / {u u}_{i i__m m a a x x}$

其中：l和l′表示设施控制子系统的编号，即博弈方；u_{l_i}和u_{l′_i′}表示控制子系统l和l′内控制设备i和i′的控制量，即博弈方采用的策略；表示控制子系统l内控制设备i的最优控制量，即最优策略；f_l和f_l′表示控制子系统l和l′的控制目标函数，即绝对收益；f_l(u_{l_i},u_{l′_i′})和f_l′(u_{l_i},u_{l′_i′})表示控制子系统l和l′中控制设备i和i′的控制量为u_l,i和u_l′,i′时各子系统的绝对收益；表示当控制子系统l内控制设备i采用最优控制量时整个控制系统的最大绝对收益。Among them: l and l' represent the number of the facility control subsystem, that is, the player; u _{l_i} and u _{l'_i'} represent the control amount of the control equipment i and i' in the control subsystem l and l', that is, the game party adopts Strategy; Indicates the optimal control quantity of control device i in control subsystem l, that is, the optimal strategy; f _l and f _l′ represent the control objective functions of control subsystem l and l’, that is, the absolute profit; f _l (u _{l_i} ,u _{l′_i′} ) and f _l′ (u _{l_i} ,u _{l′_i′} ) represent the control variables of control devices i and i′ in control subsystems l and l′ when u _l,i and u _l′,i′ Absolute benefits of each subsystem; Indicates the maximum absolute gain of the entire control system when the control device i in the control subsystem l adopts the optimal control quantity.

具体实施时：When implementing it:

本发明的试验温室以下述条件为例：位于苏州御亭现代农业产业园，四连栋塑料温室，主体采用轻型钢结构，单栋跨度为8.0m，肩高3.0m，顶高5.0m，栋长为44.0m。以该设施环境为例，选择2015年1月下旬连续5天的实测室外温湿度以及室内温湿度值作为样本数据，采用粒子群算法对设施环境机理模型的参数进行辨识，得到模型参数如表1所示：The test greenhouse of the present invention takes the following conditions as an example: it is located in Suzhou Yuting Modern Agricultural Industrial Park, a four-span plastic greenhouse, the main body adopts a light steel structure, the span of a single building is 8.0m, the shoulder height is 3.0m, and the top height is 5.0m. The length is 44.0m. Taking the facility environment as an example, the measured outdoor temperature and humidity and indoor temperature and humidity values for 5 consecutive days in late January 2015 were selected as sample data, and the parameters of the facility environment mechanism model were identified using the particle swarm optimization algorithm, and the model parameters were obtained as shown in Table 1 Shown:

表1机理模型的参数辨识结果Table 1 The parameter identification results of the mechanism model

注：τ为太阳辐射透过率；C_d为自然通风的流量系数；C_w为自然通风的的综合风压系数；h_c为室内空气与空气通过覆盖材料的热交换系数；C_p为空气的定压比热。然后在该实际模型的基础上，结合测量的外界环境气象数据对设施环境的温湿度环境因子进行仿真控制。为了验证控制算法的有效性与可靠性，选取了每天不同的外界气象数据，对控制算法进行了8次仿真试验。试验涉及的其他参数设置如下：在模型预测控制中，预测时域为5；控制时域为2。采用粒子群算法进行优化问题的计算，其涉及的参数取值为：粒子个数为50，学习因子c1、c2均为1.4962，权重w为0.7298，迭代次数为100。Note: τ is the solar radiation transmittance; C _d is the flow coefficient of natural ventilation; C _w is the comprehensive wind pressure coefficient of natural ventilation; h _c is the heat exchange coefficient between indoor air and air passing through the covering material; C _p is the air specific heat at constant pressure. Then, on the basis of the actual model, combined with the measured external environmental meteorological data, the temperature and humidity environmental factors of the facility environment are simulated and controlled. In order to verify the effectiveness and reliability of the control algorithm, different external meteorological data are selected every day, and 8 simulation tests are carried out on the control algorithm. The other parameters involved in the experiment are set as follows: In the model predictive control, the prediction time domain is 5; the control time domain is 2. The particle swarm optimization algorithm is used to calculate the optimization problem. The parameters involved are: the number of particles is 50, the learning factors c1 and c2 are both 1.4962, the weight w is 0.7298, and the number of iterations is 100.

为了说明试验效果，选取某次仿真试验进行分析，以春季2015年4月21日(24小时)为例。当天室外环境采用美国SPECTRUM公司的自动气象站Watchdog2900ET每隔30min对温室外的温度、湿度和太阳辐射、风速进行测量得到。设施环境控制的初始温湿度值根据要求设定，为了使得试验更接近真实环境，本发明将室外环境0:00时刻采集的温湿度值作为设施环境控制的初始值，即初始室内温度为11℃，初始室内绝对湿度为5.76g/m³，即相对湿度为62％。控制目标为使得室内温度达到27℃，室内绝对湿度达到18.1g/m³，即相对湿度为80％，同时能够降低设备能耗以及损耗。In order to illustrate the test effect, a simulation test is selected for analysis, taking April 21, 2015 (24 hours) as an example. The outdoor environment of the day was obtained by measuring the temperature, humidity, solar radiation, and wind speed outside the greenhouse every 30 minutes with the automatic weather station Watchdog2900ET of SPECTRUM Company of the United States. The initial temperature and humidity value of the facility environment control is set according to requirements. In order to make the test closer to the real environment, the present invention uses the temperature and humidity value collected at 0:00 in the outdoor environment as the initial value of the facility environment control, that is, the initial indoor temperature is 11°C , the initial indoor absolute humidity is 5.76g/m ³ , that is, the relative humidity is 62%. The control target is to make the indoor temperature reach 27°C, and the indoor absolute humidity reach 18.1g/m ³ , that is, the relative humidity is 80%, while reducing equipment energy consumption and loss.

综上所述，本发明对设施环境中多目标控制问题进行研究，针对设施环境模型的非线性，目标函数的复杂性以及约束性，提出的分布式模型预测控制在设施环境中的应用框架，并借鉴博弈论，采用合作博弈方法解决设施环境中多目标，即精确控制、能耗最小、设备开启频繁度最低三个目标的冲突问题。通过在MATLAB平台上以冬季一天的气象数据为例进行了仿真与验证，与单目标控制、传统的线性加权多目标控制进行对比。实验结果表明，本发明提出的算法在解决多目标冲突问题上具有明显的优势，不仅能够使得室内温湿度能够满足作物生长所需的适宜范围，同时大大地降低了控制设备的能耗以及设备损耗。To sum up, the present invention studies the multi-objective control problem in the facility environment. Aiming at the non-linearity of the facility environment model, the complexity and constraints of the objective function, the application framework of the distributed model predictive control in the facility environment is proposed, With reference to game theory, the cooperative game method is used to solve the conflict problem of multiple objectives in the facility environment, namely, precise control, minimum energy consumption, and minimum equipment opening frequency. The simulation and verification were carried out on the MATLAB platform by taking the meteorological data of one day in winter as an example, and compared with single-objective control and traditional linear weighted multi-objective control. The experimental results show that the algorithm proposed by the present invention has obvious advantages in solving the problem of multi-objective conflicts. It can not only make the indoor temperature and humidity meet the appropriate range required for crop growth, but also greatly reduce the energy consumption and equipment loss of the control equipment. .

本发明未涉及部分均与现有技术相同或可采用现有技术加以实现。The parts not involved in the present invention are the same as the prior art or can be realized by adopting the prior art.

Claims

1. A facility environment multi-objective optimization control method based on cooperative game, it is characterized in that it comprises the following steps:

S1. Establish three control objectives of the facility environment: environmental parameters - temperature and humidity control objectives, energy consumption control objectives, and physical limit of control equipment, that is, equipment loss control objectives;

S2. Establishing a cooperative game model, using a multi-objective model predictive control method based on the cooperative game model to process the three control objectives of the aforementioned facility environment, and obtaining an optimization function for each control objective;

S3. Solve the obtained optimization function of the objective, and obtain the optimal control quantity of each objective function.

2. The facility environment multi-objective optimization control method based on cooperative game according to claim 1, characterized in that step S1 is specifically:

S1-1. Establish the environmental parameter of the facility - the temperature and humidity control objective function f ₁ , which is described as:

{f f}_{11} = = m m i i n no [[{Σ Σ}_{j j = = 11}^{P P} {(({T T}_{i i n no} ((k k + + j j | | k k)) - - {T T}_{s the s e e t t}))}^{22} + + {Σ Σ}_{j j = = 11}^{P P} {(({H h}_{i i n no} ((k k + + j j | | k k)) - - {H h}_{s the s e e t t}))}^{22}]]

Among them: P represents the forecast time domain, j represents the forecast step size, k represents the current moment; T _in (k+j|k) represents the indoor temperature at the future time k+j acquired at the current moment k, H _in (k+j| k) Indicates the indoor humidity at the future time k+j obtained at the current time k, and T _set and H _set are the temperature and humidity values suitable for crop growth, respectively;

S1-2. Establish an energy consumption control objective function f ₂ for facility control, described as:

{f f}_{22} = = m m i i n no [[{Σ Σ}_{i i = = 11}^{N N} {Σ Σ}_{{j j}^{' '} = = 00}^{M m - - 11} (({u u}_{i i} ((k k + + {j j}^{' '} | | k k)) / / {u u}_{i i__m m a a x x})) / / M m]]

Among them, N represents the total number of control devices, i represents the serial number of the current control device; M represents the control time domain, j′ represents the control step size; u _i (k+j′|k) represents the future k+ j controls the optimal control quantity of device i at all times, u _{i_max} is the maximum control quantity of control device i;

S1-3. Establish an objective function f ₃ for equipment loss control in the facility, described as:

{f f}_{33} = = m m i i n no {Σ Σ}_{i i = = 11}^{N N} (({u u}_{i i} ((k k + + {j j}^{' '} + + 11 | | k k)) - - {u u}_{i i} ((k k + + {j j}^{' '} | | k k)))) / / {u u}_{i i__max max}

Where: u _i (k+j′+1|k) represents the optimal control amount of the control device i at the future k+j+1 time obtained at the current time k.

3. The facility environment multi-objective optimization control method based on cooperative game according to claim 1, characterized in that step S2 is specifically:

S2-1. Taking the three control objectives, that is, the three facility control subsystems as game parties, the following formula is used to establish a cooperative game model:

{s the s}_{l l} (({u u}_{l l}^{* *})) = = \underset{{u u}_{l l} &Element; &Element; {U u}_{l l},, {u u}_{{l l}^{' '}} &Element; &Element; {U u}_{{l l}^{' '}}}{m m a a x x} [[{w w}_{l l l l} \times \times {\overset{&OverBar; &OverBar;}{s the s}}_{l l} (({u u}_{l l},, {u u}_{{l l}^{' '}})) + + {Σ Σ}_{{l l}^{' '} = = 11 (({l l}^{' '} &NotEqual; &NotEqual; l l))}^{m m} {w w}_{{ll ll}^{' '}} \times \times {\overset{^^}{s the s}}_{{ll ll}^{' '}} (({u u}_{l l},, {u u}_{{l l}^{' '}}))]]

In the formula, l and l' represent the numbers of the players, m represents the total number of players; U _l and U _l' represent the strategy sets of the players l and l', u _l and u _l' represent the players l and l ' the strategy adopted, Indicates the optimal strategy that player l can adopt; Indicates the absolute revenue of player l when players l and l′ adopt strategies u _l and u _l ′ respectively, Indicates the absolute income of the player l' when the players l and l' respectively adopt strategies u _l and u _l' , Indicates that player l adopts the optimal strategy When is the maximum sum of absolute benefits that all players can obtain; w _ll and w _ll′ are proportional coefficients, w _ll represents the degree of non-cooperation of player l, w _ll′ represents the degree of cooperation of player l, and

S2-2. Establish a multi-objective model predictive control method based on the aforementioned cooperative game model, and obtain the optimization objective function of each subsystem in the facility environment as:

{S S}_{l l} (({u u}_{l l__i i}^{* *})) = = max max [[{w w}_{l l l l} \times \times \frac{{f f}_{l l} (({u u}_{l l__i i}^{* *},, {u u}_{{l l}^{' '}__{i i}^{' '}}))}{{f f}_{l l} (({u u}_{l l__i i},, {u u}_{{l l}^{' '}__{i i}^{' '}}))} + + {Σ Σ}_{{l l}^{' '} = = 11 (({l l}^{' '} &NotEqual; &NotEqual; l l))}^{33} {w w}_{{ll ll}^{' '}} \times \times \frac{{f f}_{{l l}^{' '}} (({u u}_{l l__i i}^{* *},, {u u}_{{l l}^{' '}__{i i}^{' '}}))}{{f f}_{{l l}^{' '}} (({u u}_{l l__i i},, {u u}_{{l l}^{' '}__{i i}^{' '}}))}]]

Among them: l and l' represent the number of the facility control subsystem, that is, the player; u _{l_i} and u _{l'_i'} represent the control amount of the control equipment i and i' in the control subsystem l and l', that is, the game party adopts Strategy; Indicates the optimal control quantity of control device i in control subsystem l, that is, the optimal strategy; f _l and f _l′ represent the control objective functions of control subsystem l and l’, that is, the absolute profit; f _l (u _{l_i} ,u _{l′_i′} ) and f _l′ (u _{l_i} , u _{l′_i′} ) represent the control quantities of control devices i and i’ in control subsystems l and l′ when u _l,i and u _l′,i′ Absolute benefits of each subsystem; Indicates the maximum absolute gain of the entire control system when the control device i in the control subsystem l adopts the optimal control quantity.

4. The facility environment multi-objective optimization control method based on cooperative game according to claim 1, characterized in that step S3 is specifically:

S3-1. Each control subsystem randomly initializes the control sequence u _{l_i} (0) of each control device;

u _{l_i} (0)=[u _{l_i} (0|0),u _{l_i} (1|0),...,u _{l_i} (M-1|0)], l=1,2,3

Among them: M represents the control time domain, l represents the number of the control subsystem, i represents the number of the control device, u _{l_i} (0|0) represents the random preset value of the control quantity of the control device i in the control subsystem l at the 0th moment, u _{l_i} (M-1|0) represents the random preset value of control device i in the control subsystem l at time 0 for its control quantity at time M-1;

S3-2. Each control subsystem communicates with each other, and sends its own control sequence u _{l_i, q} (k) at time k to other subsystems;

u _{l_i, q} (k)＝[u _{l_i} (k|k), u _{l_i} (k+1|k),...,u _{l_i} (k+M-1|k)] ^q

Among them: u _{l_i} (k+M-1|k) represents the preset value of control device i in control subsystem l at time k to its control quantity at time k+M-1, and q represents the control subsystem l and The number of times other subsystems interact with control quantities;

S3-3. After obtaining the control sequence sent by other control subsystems for the qth time at time k, the control subsystem l uses the particle swarm optimization algorithm to find the optimal control sequence according to its own optimization objective function

{u u}_{l l__i i,, q q}^{* *} ((k k)) = = {[[{u u}_{l l__i i}^{* *} ((k k | | k k)),, {u u}_{l l__i i}^{* *} ((k k + + 11 | | k k)),, ... ...,, {u u}_{l l__i i}^{* *} ((k k + + M m - - 11 | | k k))]]}^{q q}

in, Indicates the optimal preset value of the control quantity of the control device i in the control subsystem l at time k to time k+M-1;

S3-4. For any player, at time k if q=q _max or Then stop sending the control sequence and execute the optimal control sequence The first control amount of After that, return to step S4-2 to calculate the optimal control sequence at time k+1,

Among them, q _max represents the maximum number of times the control subsystem sends the control sequence at a certain moment, and ε represents the threshold value of the control sequence improvement.