Search

CN-121977323-A - Cold storage control system based on solar drive

CN121977323ACN 121977323 ACN121977323 ACN 121977323ACN-121977323-A

Abstract

The invention belongs to the fields of refrigeration technology and new energy application, and particularly relates to a refrigeration house control system based on solar drive. The method comprises the steps of adopting a three-level control mode to control the compressor in different time periods, firstly taking factors directly related to the working state of the refrigeration house compressor as input prediction to obtain a second target frequency as a real-time control instruction to adjust the working state of the compressor, then taking a third target frequency as a real-time control instruction to adjust the working state of the compressor based on the prospective scheduling cold quantity of a thermal load prediction model, taking the control instruction with higher priority than the control instruction of the second target frequency, and finally taking the frequency of the compressor output by a multi-target collaborative optimization algorithm as the highest control instruction, comprehensively balancing a plurality of targets such as energy efficiency, temperature precision, battery health, compressor service life and the like, and ensuring the overall performance of the system to be optimal.

Inventors

  • JIN NI
  • YU YONG
  • CHEN LULU

Assignees

  • 杭州索乐光电有限公司

Dates

Publication Date
20260505
Application Date
20260403

Claims (10)

  1. 1. A freezer control system based on solar drive, the system comprising: The control unit is used for controlling each module in the refrigeration house control system; the photovoltaic power optimization module models a photovoltaic system power optimization process into a Markov decision process by deep reinforcement learning, and performs maximum power point tracking under a complex environment by trial and error learning to obtain first output power; The adaptive temperature control module obtains a second target frequency of the variable-frequency compressor by using the T1 as a time period and utilizing deep neural network DNN prediction Target opening degree of electronic expansion valve Will second target frequency Target opening degree of electronic expansion valve The working state of the compressor is adjusted as a real-time control instruction; the energy efficiency self-adaptive optimization module uses T2 as a time period, calculates the refrigeration demand in a future time window by using a thermal load prediction model, and calculates and obtains the frequency adjustment quantity of the third compressor according to the refrigeration demand Adjusting the frequency of the third compressor For a second target frequency Compensating to obtain a third target frequency Will third target frequency Adjusting compressor operating conditions as real-time control commands having a priority above a second target frequency Control instructions of (2); The multi-target collaborative optimization module takes T3 as a time period to output the first output power and the second target frequency Frequency adjustment of third compressor The method comprises the steps of taking battery SOC, ambient temperature, library temperature deviation, compressor running time and current as inputs, and optimizing by using a multi-objective function to obtain a target control instruction, wherein the target control instruction comprises MPPT power output setting, battery charging and discharging strategies and optimal compressor frequency; wherein T1< T2< T3.
  2. 2. The solar drive-based freezer control system of claim 1, the adaptive temperature control module comprising: the system prepares the sub-module, load the DNN controller model trained in advance; The state data acquisition sub-module takes T1 as a time period, acquires multidimensional state information and constructs an input vector; the DNN forward reasoning submodule sends the constructed input vector to a DNN controller, and the network directly outputs the control vector through multi-layer nonlinear transformation The output control vector u (k) includes a second target frequency of the inverter compressor Target opening degree of electronic expansion valve ; A first real-time controller for controlling the second target frequency Target opening degree of electronic expansion valve And adjusting the working state of the compressor as a real-time control instruction.
  3. 3. The solar-driven refrigeration storage control system according to claim 2, wherein the input vector comprises a current return air temperature Target set temperature Error in temperature Rate of change of temperature error Ambient temperature The evaporator air outlet temperature, the battery SOC and the current time.
  4. 4. The solar-driven freezer control system of claim 1, the energy efficiency adaptive optimization module comprising: the parameter loading module loads the heat load prediction model parameters; The data acquisition sub-module takes T2 as a time period and acquires the data of the environment temperature, the opening times and the duration of the garage door, the photovoltaic power generation power, the battery SOC and the temperature in the garage; a thermal load prediction sub-module that calculates refrigeration demand within a future time window using a thermal load prediction model; The control quantity prediction sub-module is used for judging whether the predicted thermal load exceeds a threshold value, executing prospective control if the high-temperature period or the frequent door opening operation is predicted, commanding the compressor to cool in advance when the photovoltaic power is abundant, constructing cold quantity reserve, and controlling the output to be the frequency adjustment quantity of the third compressor 。
  5. 5. The solar-driven freezer control system of claim 4, the energy efficiency adaptive optimization module further comprising: A second real-time controller for adjusting the frequency of the third compressor For a second target frequency Compensating to obtain a third target frequency Will third target frequency Adjusting compressor operating conditions as real-time control commands having a priority above a second target frequency Control instructions of (2).
  6. 6. The solar-driven freezer control system of claim 4, the thermal load prediction model being: Wherein, the The thermal load is predicted and the thermal load is predicted, As a result of the temperature of the environment, The frequency of the opening of the garage door, Is the power of the photovoltaic power generation, In order to determine the state of charge of the battery, 、 、 、 As coefficients of the model, the coefficients of the model, Is an error term.
  7. 7. The solar-driven freezer control system of claim 4, said third compressor frequency adjustment The calculation formula is as follows: Wherein, the As the gain factor of the gain factor, Is a thermal load threshold.
  8. 8. The solar-driven freezer control system of claim 1, the multi-objective function being Wherein, the The energy consumption cost is related to photovoltaic power and power grid electricity supplementing cost; is a library Wen Piancha; estimating the battery loss rate based on the SOC cycle times and the charge and discharge rate; For the fatigue degree of the compressor, based on the start-stop times and the load rate, 、 、 、 Is a weight.
  9. 9. The solar drive-based freezer control system of claim 1, the photovoltaic power optimization module specifically comprising: the system initialization submodule is used for initializing an intelligent agent for deep reinforcement learning, and comprises a strategy network, a target network and an experience playback buffer zone, and loads pre-trained network model weights; the state sensing submodule samples the output voltage of the photovoltaic array in each control period Output current And calculate the current power To N times of the current and the past Data as a state vector ; The action decision sub-module inputs the state vector s (k) into a strategy network of the DRL agent, and the network outputs an action a (k) which represents an adjustment instruction for a switching signal of the power converter; executing action and environment feedback sub-module for executing action Changing system operation parameters to adjust power output state of the photovoltaic array, sampling to obtain new power in next period The improvement degree of the power output characteristic is used as the instant rewarding ; Experience storage and learning sub-module for storing the experience tuple Periodically randomly sampling a batch of empirical data from the buffer for updating the weights of the policy network with the goal of maximizing the long-term jackpot; the output control sub-module is used for obtaining first output power through conversion of actions output by the strategy network, generating an actual adjusting instruction according to the first output power, driving an executing device in the power controller, and waiting for the next control period to repeatedly execute the photovoltaic power optimizing process.
  10. 10. A refrigeration storage realized based on the refrigeration storage control system of any one of claims 1-9.

Description

Cold storage control system based on solar drive Technical Field The invention belongs to the fields of refrigeration technology and new energy application, and particularly relates to a refrigeration house control system based on solar drive. Background The traditional refrigeration house system mainly depends on a power grid for power supply, so that the energy consumption is high, and the refrigeration house system is difficult to apply in a power-free or power-deficient area. Although the dependence on a power grid is relieved to a certain extent, the development of the solar-driven refrigeration house system faces two major core challenges, namely, how to efficiently capture unstable solar energy and how to realize accurate and stable control of the temperature in the refrigeration house under the condition of limited energy. At present, most solar refrigeration house systems adopt a traditional disturbance observation method to track the Maximum Power Point (MPPT) in the aspect of energy management, and the problems of continuous oscillation near the maximum power point, tracking misalignment when illumination changes rapidly and the like exist, so that the energy capturing efficiency is low. In the aspect of temperature control, the conventional system mostly adopts simple switch control or PID (proportion integration differentiation) regulation of fixed parameters, and is difficult to effectively cope with load change of a refrigeration house, environmental temperature fluctuation and thermal disturbance generated in the defrosting process, so that the temperature fluctuation in the refrigeration house is obvious, the storage quality of articles is influenced, and energy waste is caused. In addition, the existing system generally lacks sensing and self-adaption capability to user habit, and also lacks system health state prediction and maintenance reminding functions based on data driving, so that the overall intelligent level is highly needed to be improved. Therefore, there is a need to develop a solution for a refrigeration house system that has high efficiency and strong adaptability in both energy capture and temperature control, and that incorporates intelligent management and autonomous decision making capabilities. Disclosure of Invention The invention aims to overcome the defects of the prior art and provide a solar-driven refrigeration house system which is high in photovoltaic energy capturing efficiency, accurate in temperature control and intelligent and stable in operation, and the compressors are controlled in a three-stage control mode in different time periods. The invention provides a refrigeration house control system based on solar drive, which comprises: and the control unit is used for controlling each module in the refrigeration house control system. And the photovoltaic power optimization module models the photovoltaic system power optimization process into a Markov decision process by deep reinforcement learning, and performs maximum power point tracking under a complex environment by trial and error learning to obtain the first output power. The adaptive temperature control module obtains a second target frequency of the variable-frequency compressor by using the T1 as a time period and utilizing deep neural network DNN predictionTarget opening degree of electronic expansion valveWill second target frequencyTarget opening degree of electronic expansion valveAnd adjusting the working state of the compressor as a real-time control instruction. The energy efficiency self-adaptive optimization module uses T2 as a time period, calculates the refrigeration demand in a future time window by using a thermal load prediction model, and calculates and obtains the frequency adjustment quantity of the third compressor according to the refrigeration demandAdjusting the frequency of the third compressorFor a second target frequencyCompensating to obtain a third target frequencyWill third target frequencyAdjusting compressor operating conditions as real-time control commands having a priority above a second target frequencyControl instructions of (2); The multi-target collaborative optimization module takes T3 as a time period to output the first output power and the second target frequency Frequency adjustment of third compressorBattery SOC, ambient temperatureLibrary Wen PianchaCompressor run timeAnd current flowAs input, the core of the module is to define a comprehensive optimization objective function (i.e. a multi-objective function) and map the input parameters to quantized values of the optimization terms in the objective function. And optimizing by using a multi-objective function to obtain a target control instruction, wherein the target control instruction comprises MPPT power output setting, a battery charging and discharging strategy and an optimal compressor frequency, and the optimal compressor frequency is used as the highest-level control instruction. Wherein T1< T2< T3. T