CN-121977313-A - Energy-saving control system and method for single-stage compression refrigerator parallel refrigeration house system

CN121977313ACN 121977313 ACN121977313 ACN 121977313ACN-121977313-A

Abstract

The invention discloses an energy-saving control system and method for a parallel refrigeration house system of a single-stage compression refrigerator, wherein the energy-saving control system comprises a hardware layer and an algorithm layer, the hardware layer comprises a compressor unit, an evaporator, a condenser and other actuating mechanisms, a sensor module and a PLC (programmable logic controller), the algorithm layer comprises an upper computer algorithm module, the energy-saving control method collects system operation data through the sensor module, inputs the system operation data into the upper computer algorithm module after preprocessing, adopts multi-agent reinforcement learning MARL to realize local agent decision, combines a collaborative genetic algorithm CGA to complete global collaborative optimization, ensures decision consistency through a conflict detection and coordination mechanism, and finally executes control instructions by the PLC and realizes closed loop feedback update. The invention comprehensively considers the temperature stability, the energy efficiency ratio and the equipment reliability, realizes the global optimal control, effectively reduces the energy consumption, improves the overall operation efficiency of the refrigeration house group, and is suitable for the fields of food, medicine, industrial cold chain storage and transportation and the like.

Inventors

JIANG SHAN
Liang Haoda
SUN MINGXUE
SUN DAOPENG
LI ZHANYING
TANG YU
WANG LIANG

Assignees

大连工业大学
大连冰山嘉德自动化有限公司

Dates

Publication Date: 20260505
Application Date: 20260206

Claims (9)

1. An energy-saving control system of a single-stage compression refrigerator parallel refrigeration house system is characterized by comprising a hardware layer and an algorithm layer; The hardware layer comprises a compressor unit, an evaporator, a condenser, a fan, a liquid supply valve, a sensor module and a PLC controller; The algorithm layer comprises an upper computer algorithm module, wherein the upper computer algorithm module comprises an environment sensing module, a local agent decision module, a global optimization module, a conflict detection module and a control execution module; The exhaust end of the compressor unit is connected to the inlet of the condenser through an exhaust pipeline, the condenser condenses high-pressure gaseous refrigerant into high-pressure liquid, liquid is branched to each liquid supply valve through a liquid distribution device connected with the outlet of the condenser, the outlets of the liquid supply valves are respectively connected to the inlets of evaporators of corresponding refrigeration houses, the evaporators are arranged in the refrigeration houses and used for absorbing heat load in the refrigeration houses and gasifying the liquid refrigerant, and the gas outlet of the evaporators and a suction header pipe are connected to the suction end of the compressor unit in parallel to form a complete refrigerant cycle; The sensor module acquires the temperature of the refrigeration house, the suction and exhaust pressure of the compressor unit and the electric energy consumption data in real time and transmits the data to the upper computer algorithm module; The upper computer algorithm module outputs control instructions through multi-agent reinforcement learning MARL and a collaborative genetic algorithm CGA, and drives the compressor contactor, the fan relay and the liquid supply valve to act through the PLC.
2. The energy-saving control system of the parallel refrigeration house system of the single-stage compression refrigerator of claim 1, wherein the specific functions of each functional module of the upper computer algorithm module are as follows: the environment sensing module is used for collecting the temperature, the load demand and the running state of the compressor of each cold storage area and carrying out denoising, anomaly detection and normalization on the data; the local intelligent agent decision module is used for configuring independent intelligent agents for each refrigeration house, generating local control instructions based on MARL, and determining an optimization target by a reward function; The global optimization module is used for performing global optimization on the running state of the compressor group and the local intelligent agent instruction based on the CGA, wherein an optimization target is determined by a fitness function, and temperature out-of-range and equipment life protection constraint are added; the conflict detection module is used for judging the conflict between the local agent instruction and the global optimization result, and feeding back the corrected instruction to the reward function and the fitness function for parameter updating; and the control execution module outputs the optimized control instruction to the PLC in a digital quantity form, so that the temperature stability of the refrigeration house group and the energy-saving operation of the system are realized.
3. A control method of an energy-saving control system based on the parallel refrigeration house system of the single-stage compression refrigerator as claimed in claim 2 is characterized by comprising the following steps: s1, environmental perception, namely collecting the temperature of each refrigeration house through a sensor module Load demand, compressor suction and discharge pressure \ The running state and the energy consumption data are preprocessed to form a state vector; S2, local agent decision, namely, configuring independent agents for each refrigeration house based on MARL, and deciding on-off actions of a fan and a liquid supply valve by the agents according to state vectors, wherein a reward function is used as an optimization target; s3, global collaborative optimization, namely performing global optimization on the running state of the compressor and a local agent instruction by adopting CGA, and generating a global control instruction by taking an fitness function as a target; s4, conflict detection and coordination, namely coordinating local decisions and global optimization results through a conflict and feedback mechanism to ensure the consistency of the decisions; And S5, controlling execution and feedback updating, namely, executing a global control instruction by a PLC (programmable logic controller) to drive the compressor, the fan and the liquid supply valve to act, and feeding back a control effect to the reward function and the fitness function after the period is finished, and updating parameter weights to realize closed-loop optimization.
4. The method of claim 3, wherein S2 the bonus function is defined as: Wherein, the For the purpose of temperature error punishment, , The temperature of the cold storage at the current moment, Setting a target temperature; an estimate of compressor power to be apportioned to the pool; taking 1 when the compressor is turned from a stop state to a start state due to the action of the compressor, or taking 0 when the compressor is turned from the stop state to the start state; The method comprises the steps that an indication function is frequently switched for equipment, 1 is taken when switching times in unit time exceed a threshold value, and otherwise 0 is taken; Taking 1 when the temperature is within a set error range as a temperature stability indication function, otherwise taking 0; is a weight coefficient.
5. The method according to claim 4, wherein the step of The calculation mode of (2) comprises the following two modes: when reflecting total power split: wherein Is the total power of the current compressor, Is the rated power of the system; when reflecting load balance among refrigerators: wherein Is the average power.
6. The method of claim 3, wherein S3 the fitness function is: Wherein, the Normalized total energy consumption predicted for the candidate; normalizing the candidate start-stop times; accumulating for temperature violations; is an index of load imbalance; Is a security constraint item; is a weight coefficient.
7. The method of claim 3, wherein S4 the collision and feedback mechanism follows the following rules: (1) If the temperature of the refrigeration house exceeds the set safety boundary due to local actions, a global optimization scheme is forcedly adopted; (2) If the global scheme causes the high-frequency start-stop of the compressor or violates the minimum switching period constraint of the equipment, preferentially executing local actions, and applying punishment to the global fitness function; (3) And feeding back the action result after conflict resolution to the reward function and the fitness function, updating the parameter weight, and ensuring the gradual convergence of the system.
8. The method of claim 7, wherein the temperature safety margin constraint includes a target temperature and an allowable fluctuation range, wherein the bonus function applies a strong penalty when exceeded, wherein the equipment life protection constraint includes a minimum compressor downtime, a maximum number of starts, a minimum switching cycle of the fan and the supply valve, and wherein the safety protection logic is triggered when violated and included in the fitness function by Penalty items.
9. The method of claim 3, wherein S2 the temperature deviation term in the bonus function is calculated by the formula Computing recommendations =1 ℃, And triggers a penalty when the temperature deviation exceeds the stability dead zone. Wherein the method comprises the steps of The measured temperature of the ith refrigeration house at the current moment is shown and is the temperature data acquired by the sensor in real time. The target temperature preset by the ith refrigeration house is indicated, and the reference point is used for temperature control. The reference scale representing the temperature normalization is a fixed value. The temperature deviation of the ith cold storage is indicated and used for controlling the error of the reaction temperature. Then a temperature error is represented for the state vector of MARL.

Description

Energy-saving control system and method for single-stage compression refrigerator parallel refrigeration house system Technical Field The invention relates to the technical field of refrigeration houses and intelligent control, in particular to an energy-saving control system and method for a single-stage compression refrigerator parallel refrigeration house system. Background In the existing refrigeration system of the refrigeration house, the refrigeration house mostly adopts a single compressor refrigeration, manual management and constant control strategy, so that the automatic control of the refrigeration house is realized. However, the traditional single compressor is used for refrigeration, so that the energy consumption is high, the response is slow, and the traditional single compressor is difficult to adapt to the actual requirements of multi-area and differential temperature control. Particularly in a multi-refrigeration house system, the problems of uneven cold energy distribution, frequent start and stop of a compressor, insufficient temperature control precision and the like often exist. Disclosure of Invention The invention provides an energy-saving control system and method for a parallel refrigeration house system of a single-stage compressor, which aims to solve the problems that the traditional single-stage compressor is high in refrigeration energy consumption and slow in response, is difficult to adapt to the actual demands of multi-region and differential temperature control, and particularly has uneven cold quantity distribution, frequent start and stop of compressors, insufficient temperature control precision and the like in a multi-refrigeration house system. The technical scheme adopted by the invention for realizing the purpose comprises a hardware layer and an algorithm layer; The hardware layer comprises a compressor unit, an evaporator, a condenser, a fan, a liquid supply valve, a sensor module and a PLC controller; The algorithm layer comprises an upper computer algorithm module, wherein the upper computer algorithm module comprises an environment sensing module, a local agent decision module, a global optimization module, a conflict detection module and a control execution module; The exhaust end of the compressor unit is connected to the inlet of the condenser through an exhaust pipeline, the condenser condenses high-pressure gaseous refrigerant into high-pressure liquid, liquid is branched to each liquid supply valve through a liquid distribution device connected with the outlet of the condenser, the outlets of the liquid supply valves are respectively connected to the inlets of evaporators of corresponding refrigeration houses, the evaporators are arranged in the refrigeration houses and used for absorbing heat load in the refrigeration houses and gasifying the liquid refrigerant, and the gas outlet of the evaporators and a suction header pipe are connected to the suction end of the compressor unit in parallel to form a complete refrigerant cycle; The sensor module acquires the temperature of the refrigeration house, the suction and exhaust pressure of the compressor unit and the electric energy consumption data in real time and transmits the data to the upper computer algorithm module; The upper computer algorithm module outputs control instructions through multi-agent reinforcement learning MARL and a collaborative genetic algorithm CGA, and drives the compressor contactor, the fan relay and the liquid supply valve to act through the PLC. Preferably, the specific functions of each functional module of the upper computer algorithm module are as follows: The environment sensing module is used for collecting the temperature, load demand and compressor running state (suction and exhaust pressure, running time length and start and stop times) of each cold storage area and carrying out denoising, abnormality detection and normalization treatment on the data; the local intelligent agent decision module is used for configuring independent intelligent agents for each refrigeration house, generating local control instructions (a fan and a liquid supply valve are started and stopped) based on MARL, and determining an optimization target by a reward function; The global optimization module is used for performing global optimization on the running state of the compressor group and the local intelligent agent instruction based on the CGA, wherein an optimization target is determined by a fitness function, and temperature out-of-range and equipment life protection constraint are added; the conflict detection module is used for judging the conflict between the local agent instruction and the global optimization result, and feeding back the corrected instruction to the reward function and the fitness function for parameter updating; and the control execution module outputs the optimized control instruction to the PLC in a digital quantity form, so that the temperature stability of the r