Efficient refrigerating machine room performance optimization method and device

Abstract

The invention discloses an efficient refrigerating machine room performance optimization method and device, and relates to the technical field of intelligent energy-saving optimizing.The efficient refrigerating machine room performance optimization method comprises the steps that operation data of refrigerating equipment are collected through various sensors and preprocessed; based on the preprocessed operation data, constructing an MT-LSTM model to predict a future cold load demand, and generating a cold load demand prediction value; on the basis of the cold load demand predicted value, generating an optimal scheduling strategy through an NSGA-II optimization algorithm; and according to the optimal scheduling strategy, equipment starting and closing and load proportion distribution are executed through IoT, operation data are monitored in real time, and the operation state of the refrigeration equipment is dynamically adjusted and optimized. According to the method, the future cold load demand is predicted by constructing the MT-LSTM model, and accurate demand prediction is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent energy-saving optimization, and particularly to a method and device for optimizing the performance of an efficient refrigeration machine room. Background Art

[0002] In recent years, with the rapid popularization of intelligent management and energy-saving technologies, refrigeration machine rooms, as an important part of the heating and cooling of building clusters, have received extensive attention. Existing technologies mainly rely on traditional refrigeration control strategies, using simple sensors to sense environmental variables such as temperature and humidity, and performing manual adjustments based on empirical rules; with the introduction of Internet of Things and artificial intelligence technologies, advanced intelligent control systems have gradually been applied to refrigeration machine rooms, improving the level of intelligence and automation; at the same time, LSTM and multi-objective optimization algorithms have shown significant potential in prediction and optimization, providing a feasible way to achieve more efficient refrigeration equipment scheduling;

[0003] Although traditional refrigeration management systems meet the cold load requirements to a certain extent, there are still many deficiencies in the real-time performance, accuracy of data collection and processing, and the level of intelligence in scheduling optimization; traditional systems often rely on manual experience for regulation, resulting in significant energy consumption, high failure rates, and difficulty in meeting the dynamically changing cold load requirements. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for optimizing the performance of an efficient refrigeration machine room to solve the energy consumption management problem.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for optimizing the performance of an efficient refrigeration machine room, which includes collecting the operation data of refrigeration equipment through various sensors and performing preprocessing; based on the preprocessed operation data, constructing an MT-LSTM model to predict future cold load requirements and generating cold load demand prediction values; based on the cold load demand prediction values, generating an optimized scheduling strategy through the NSGA-II optimization algorithm; according to the optimized scheduling strategy, turning on and off IoT execution devices and allocating load ratios, and real-time monitoring the operation data to dynamically adjust and optimize the operation state of refrigeration equipment.

[0008] As a preferred embodiment of the method for optimizing the performance of the efficient refrigeration machine room according to the present invention, wherein: the operation data of the equipment includes environmental temperature data, environmental humidity data, equipment status data, and historical load data.

[0009] As a preferred embodiment of the method for optimizing the performance of the high-efficiency refrigeration machine room according to the present invention, the preprocessing includes denoising, data standardization, and missing value filling.

[0010] As a preferred embodiment of the method for optimizing the performance of the high-efficiency refrigeration machine room according to the present invention, based on the preprocessed operation data, an MT-LSTM model is constructed to predict the future cooling load demand and generate a predicted value of the cooling load demand. The specific steps are as follows:

[0011] The preprocessed operation data is weighted and fused to generate a unified data set;

[0012] The unified data set is divided into a training set, a validation set, and a test set;

[0013] An MT-LSTM model is constructed based on LSTM, and the input of the MT-LSTM model is defined as the training set, and the output is the predicted value of the cooling load demand;

[0014] The time series data in the training set is extracted and split into multiple time periods according to the time steps of the time series data to form a time segment sequence, which is input to the multi-layer LSTM for processing;

[0015] The first layer of LSTM receives the time segment sequence, processes the time segment sequence through the gating mechanism, captures the short-term dependence information, and inputs it to the second layer of LSTM;

[0016] The second layer of LSTM receives the short-term dependence information, processes the short-term dependence information through the gating mechanism and filtering, captures the medium-term dependence information, and inputs it to the third layer of LSTM;

[0017] The third layer of LSTM receives the medium-term dependence information, combines the short-term dependence information of the first layer of LSTM and the medium-term dependence information of the second layer of LSTM through the gating mechanism to capture the long-term dependence information, and outputs it to the fully connected layer;

[0018] The fully connected layer receives and integrates the short-term dependence information, medium-term dependence information, and long-term dependence information, and calculates the predicted value of the cooling load demand. The expression is:

[0019]

[0020] Wherein, is the predicted value of the cooling load demand, D is the cooling load demand, pred is the predicted value, T e,t is the environmental temperature data at the t-th time point, H e,t is the environmental humidity data at the t-th time point, e is the environmental identifier, S t is the equipment operation status data at the t-th time point, D t-1 is the historical load data at the t-th time point, t is the time, and m is the total number of time;

[0021] Evaluate the performance of the MT-LSTM model using the validation set and adjust the parameters;

[0022] Use the test set to measure the generalization ability and prediction accuracy of the MT-LSTM model.

[0023] As a preferred embodiment of the high-efficiency refrigeration plant performance optimization method described in the present invention, wherein: based on the predicted value of the cooling load demand, an optimized scheduling strategy is generated through the NSGA-II optimization algorithm. The specific steps are as follows:

[0024] Select the NSGA-II optimization algorithm and define the optimization objective as the minimum energy consumption of the refrigeration equipment;

[0025] Set the start time and shutdown time of multiple refrigeration equipment according to the predicted value of the cooling load, and allocate the load ratio of multiple refrigeration equipment;

[0026] Convert the start time, shutdown time, and load ratio allocation of each refrigeration equipment into chromosome coding;

[0027] Use the NSGA-II optimization algorithm to generate an initial scheduling plan based on the chromosome coding;

[0028] Optimize the initial scheduling plan through crossover simulation and mutation simulation to generate a new scheduling plan;

[0029] For the new scheduling plan, calculate the total energy consumption of the refrigeration equipment. The expression is:

[0030]

[0031] Where, E is the total energy consumption of the refrigeration equipment, P i (t) is the power of the i-th equipment at time t, Q i (t) is the running duration of the i-th equipment, i is the equipment, and n is the total number of equipment;

[0032] Evaluate the total energy consumption of the refrigeration equipment in the new scheduling plan through non-dominated sorting to generate an optimized scheduling strategy.

[0033] As a preferred embodiment of the high-efficiency refrigeration plant performance optimization method described in the present invention, wherein: the step of using the NSGA-II optimization algorithm to generate an initial scheduling plan based on the chromosome coding is as follows:

[0034] Define the start time of the chromosome coding as a five-bit binary number to generate an initial chromosome;

[0035] Define the shutdown time of the chromosome coding as a five-bit binary number to generate an initial chromosome;

[0036] Define the load ratio allocation of the chromosome coding as a four-bit binary number to generate an initial chromosome;

[0037] Combine the initial chromosomes of the startup time, shutdown time, and load ratio allocation to generate encoded information;

[0038] For the encoded information, perform disassembly and decoding to obtain the specific startup time, shutdown time, and load ratio allocation of each refrigeration device, and generate an initial scheduling plan.

[0039] As a preferred solution of the method for optimizing the performance of the high-efficiency refrigeration machine room described in the present invention, wherein: according to the optimized scheduling strategy, the IoT execution device is used to perform startup and shutdown and load ratio allocation, and the operation data is monitored in real time, and the operation state of the refrigeration device is dynamically adjusted and optimized. The specific steps are as follows:

[0040] Extract the startup time, shutdown time, and load ratio allocation of the refrigeration device from the optimized scheduling strategy as control instructions;

[0041] Establish the connection between each refrigeration device and the Internet of Things through the IoT;

[0042] Send the control instructions for extracting the startup time, shutdown time, and load ratio allocation of the refrigeration device to the corresponding refrigeration device through the IoT for startup, shutdown, and load ratio allocation;

[0043] Use the ambient temperature sensor, ambient humidity sensor, and device operation state monitor to collect operation data in real time;

[0044] Transmit and store the collected real-time operation data to the central control center through the IoT;

[0045] The central control center monitors the operation data in real time through the support vector machine algorithm. When the operation data is abnormal, it indicates that the device has a fault;

[0046] The central control center automatically sends an alarm and starts adjustment according to the fault of the device;

[0047] If it is slightly abnormal, automatically adjust the load ratio allocation and temperature parameters of the refrigeration device;

[0048] If it is severely abnormal, start the standby refrigeration device.

[0049] Second aspect, the present invention provides a device for optimizing the performance of an efficient refrigeration machine room, including an acquisition module, a prediction module, an optimization module, and a monitoring module; the acquisition module is used to collect the operation data of refrigeration equipment through a variety of sensors and perform preprocessing; the prediction module is used to construct an MT-LSTM model based on the preprocessed operation data to predict future cooling load requirements and generate a predicted value of the cooling load requirements; the optimization module is used to generate an optimized scheduling strategy through the NSGA-II optimization algorithm based on the predicted value of the cooling load requirements; the monitoring module is used to turn on and off IoT execution devices and allocate the load ratio according to the optimized scheduling strategy, and monitor the operation data in real time to dynamically adjust and optimize the operation state of the refrigeration equipment.

[0050] Third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the method for optimizing the performance of an efficient refrigeration machine room as described in the first aspect of the present invention is implemented.

[0051] Fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the method for optimizing the performance of an efficient refrigeration machine room as described in the first aspect of the present invention is implemented.

[0052] The beneficial effects of the present invention are as follows: the present invention predicts future cooling load requirements by constructing an MT-LSTM model, achieving accurate demand prediction, which plays an important role in optimizing the scheduling of refrigeration equipment. The MT-LSTM model can capture short-term, medium-term, and long-term dependency information. By processing a variety of feature data, it generates reliable predicted values of the cooling load requirements. In this way, even in the face of complex environmental changes and equipment operation states, the MT-LSTM model can still provide accurate demand prediction, thereby avoiding over-cooling or under-cooling, improving the overall efficiency of the operation of the refrigeration equipment, reducing unnecessary energy consumption, and ensuring the accurate satisfaction of the cooling load requirements. Detailed implementation manners

[0053] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following detailed description of the specific implementation manners of the present invention is provided.

[0054] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0055] Second, the "one embodiment" or "embodiment" referred to herein means a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or selectively exclusive embodiments from other embodiments.

[0056] Embodiment 1 provides an efficient refrigeration machine room performance optimization method, including the following steps:

[0057] S1. Collect the operation data of the refrigeration equipment through a variety of sensors and perform preprocessing.

[0058] Furthermore, deploy a variety of sensors to monitor multiple key parameters of the refrigeration machine room in real time;

[0059] Collect ambient temperature data through an ambient temperature sensor;

[0060] Collect ambient humidity data through an ambient humidity sensor;

[0061] Collect ambient humidity data through a humidity sensor;

[0062] The equipment operation status monitor collects equipment status data. The equipment operation status monitor includes current / voltage sensors, vibration sensors, temperature sensors, and pressure sensors;

[0063] The current / voltage sensor is used to measure the current and voltage data during equipment operation to infer the load status of the equipment. By monitoring the current and voltage data in real time, the working load and efficiency of the equipment can be evaluated;

[0064] The vibration sensor is used to monitor the vibration data of equipment (such as compressors, pumps, etc.) to determine whether there is abnormal vibration and then judge the health status of the equipment;

[0065] The temperature sensor is used to monitor the temperature changes of equipment such as compressors and cooling towers to obtain temperature data and judge their operating conditions;

[0066] The pressure sensor is used to measure the pressure changes of the equipment to obtain pressure data. For example, the pressure in the cooling cycle can help judge the working state of the refrigeration;

[0067] Through these equipment operation status monitors, it is possible to collect data such as the operation status, load conditions, and health status of the equipment in real time, providing a reference for the change of the cooling load;

[0068] Collect historical load data from historical operations, which is the refrigeration demand data within a certain past time period;

[0069] Use a signal filter to remove the noise data in the collection process of a variety of sensors;

[0070] To ensure the uniformity of the input operation data, standardization is adopted to convert all operation data into the same dimension range;

[0071] Missing values in the operation data are filled by interpolation or other methods to ensure the integrity of the operation data;

[0072] It should be noted that by deploying a variety of sensors for real-time monitoring, comprehensively collecting the environmental temperature, environmental humidity and equipment status data of the refrigeration machine room, and combining with historical load data, a comprehensive data basis can be provided; using a signal filter to remove noise to ensure the quality of the operation data; standardization processing and missing value filling ensure the consistency and integrity of the input operation data; effectively improving the accuracy and integrity of the operation data, providing reliable data support for subsequent cold load prediction and scheduling optimization, and greatly improving the accuracy and efficiency of monitoring and optimized scheduling.

[0073] S2. Based on the preprocessed operation data, construct an MT-LSTM model to predict the future cold load demand and generate a cold load demand prediction value.

[0074] Furthermore, the preprocessed operation data is weighted and fused to generate a unified data set; the expression is:

[0075] X′ t =α×T e,t +β×H e,t +γ×S t +δ×D t-1 ;

[0076] Among them, X' t is the unified data set after preprocessing, α is the weight coefficient of the environmental temperature data, T e,t is the environmental temperature data at the t-th time point, β is the weight coefficient of the environmental humidity data, H e,t is the environmental humidity data at the t-th time point, e is the environment, S t is the equipment operation status data at the t-th time point, γ is the weight coefficient of the equipment operation status data, D t-1 is the historical load at the t-th time point, δ is the weight coefficient of the historical load data, and t is the time;

[0077] The unified data set is divided into a training set, a validation set and a test set. The common ratio is 70% for training, 15% for validation, and 15% for testing;

[0078] Based on LSTM, construct an MT-LSTM model, and define the input of the MT-LSTM model as the training set and the output as the cold load demand prediction value;

[0079] Extract the time series data in the training set, split it into multiple time periods according to the time steps of the time series data (such as every hour, every day, etc.) to form a time segment sequence, and input it to a multi-layer LSTM for processing. For example, the data of the past n time steps is used as an input time segment, and the number of input time segments is determined according to the length of the unified data set and the time step size;

[0080] The first layer of LSTM receives the time segment sequence, processes the time segment sequence through the gating mechanism (input gate, forget gate, output gate) to capture short-term dependence information, and inputs it to the second layer of LSTM;

[0081] Specifically, the first layer of LSTM receives the time segment sequence of the input ambient temperature data, humidity data, device status data, and historical load data;

[0082] The input gate of the gating mechanism determines how much new input information to receive. It combines the new data of the current time step (the time segment composed of ambient temperature, humidity, device status, and historical load data) and the short-term information generated in the previous time step, evaluates which new inputs should be memorized, and synthesizes these data (the time segment composed of ambient temperature, humidity, device status, and historical load data) to determine how much new input information to receive;

[0083] The forget gate of the gating mechanism processes the comprehensive analysis between the short-term dependence information and the historical information in the unified data set, and determines how much historical information to retain and discard;

[0084] The output gate of the gating mechanism combines the input data (ambient temperature data, ambient humidity data, device status data, and historical load data) and the memory cell state data (that is, the memory information accumulated from the past time steps) to generate a hidden state output, capture short-term dependence information, generate short-term dependence information after updating, and pass it to the second layer of LSTM;

[0085] The second layer of LSTM receives the short-term dependence information, processes the short-term dependence information through the gating mechanism and filtering to capture medium-term dependence information, and inputs it to the third layer of LSTM;

[0086] Specifically, the short-term dependence information received by the second layer of LSTM is the hidden state and memory cell state data transmitted from the first layer of LSTM;

[0087] The input data processed by the input gate of the gating mechanism includes both the short-term dependence information from the first layer and the new time segment data in the unified data set, which is used to evaluate the reception of new input information, combines the output of the first layer and the new time segment data (such as ambient temperature data, ambient humidity data, device status data, and historical load data), and captures medium-term information;

[0088] The forget gate of the gating mechanism combines the time segment data in the unified dataset and the output of the first-layer LSTM for processing, determining how much short-term information to retain and discard to update the memory cell state;

[0089] The output gate of the gating mechanism synthesizes the input, memory cell state, and short-term information to generate a new hidden state output, capturing medium-term dependency information and passing it as output to the third-layer LSTM;

[0090] The third-layer LSTM receives the medium-term dependency information, combines the short-term dependency information of the first-layer LSTM and the medium-term dependency information of the second-layer LSTM through the gating mechanism, captures long-term dependency information, and outputs it to the fully connected layer;

[0091] Specifically, the third-layer LSTM receives the medium-term dependency information of the second layer as a new input time segment sequence (such as environmental temperature data, environmental humidity data, device status data, and historical load data);

[0092] The new input data processed by the input gate of the gating mechanism includes the medium-term dependency information from the second layer and the time segment data in the unified dataset, comprehensively analyzing and capturing long-term information;

[0093] The forget gate of the gating mechanism combines the output information of the previous two layers (short-term and medium-term dependency information) and the latest input time segment sequence (environmental temperature, humidity, device status, and historical load data), comprehensively processing the time information and determining how much information to retain and discard to update the memory cell state;

[0094] The output gate of the gating mechanism synthesizes the input, memory, and medium-term information to generate a new hidden state output, capturing long-term dependency information and passing it as output to the fully connected layer;

[0095] The fully connected layer receives and integrates the short-term dependency information, medium-term dependency information, and long-term dependency information, calculating the predicted value of the cooling load demand. The expression is:

[0096]

[0097] Among them, is the predicted value of the cooling load demand, representing the predicted value of the cooling load demand at the future time point t. D is the cooling load demand, which is the variable to be predicted. pred is the predicted value. T e,t is the environmental temperature data at the t-th time point. H e,t is the environmental humidity data at the t-th time point. e is the environmental identifier, used to distinguish and identify that a specific variable is related to the environment. S t is the device operating status data at the t-th time point. D t-1 is the historical load data at the t-th time point. t is the time, and m is the total number of time;

[0098] Evaluate the performance of the MT-LSTM model using the validation set, and adjust the parameters to improve the generalization ability of the MT-LSTM model;

[0099] Specifically, during the training process of the MT-LSTM model, the validation set data is often used to evaluate the MT-LSTM model at regular intervals (such as every epoch);

[0100] Input the validation set data into the MT-LSTM model to generate cold load demand prediction values;

[0101] Calculate the loss value (such as mean square error, root mean square error, etc.) between the cold load demand prediction value and the true value of the validation set;

[0102] Calculate the accuracy rate (such as accuracy, recall rate, F1 score, etc. in classification tasks);

[0103] According to the evaluation results of the loss value and accuracy rate of the validation set, adjust the hyperparameters (such as learning rate, batch size, etc.) of the MT-LSTM model;

[0104] If the loss value on the validation set cannot continue to decrease, or overfitting occurs (low training loss but high validation loss), it may be necessary to adjust the MT-LSTM model structure (such as increasing or decreasing the number of LSTM layers, changing the number of neurons, etc.);

[0105] Set an early stopping strategy to abort the training in advance when the loss value on the validation set no longer decreases, to prevent the MT-LSTM model from overfitting;

[0106] Use the test set to measure the generalization ability and prediction accuracy of the MT-LSTM model. Specifically, after completing the training and validation of the MT-LSTM model, use the MT-LSTM model with the best performance for the final test set evaluation;

[0107] Input the test set data into the MT-LSTM model to generate cold load demand prediction values;

[0108] Calculate the loss value (such as mean square error) between the cold load demand prediction value and the true value of the test set;

[0109] Calculate the accuracy rate (such as accuracy, recall rate, F1 score, etc. in classification tasks);

[0110] Compare the loss values and accuracy rates on the training set, validation set, and test set to evaluate the generalization ability and prediction accuracy of the MT-LSTM model on unseen data;

[0111] Use the evaluation results on the test set as a reference index for the final performance of the MT-LSTM model to ensure the effect in real data applications;

[0112] It should be noted that by constructing an MT-LSTM model and accurately predicting future cold load demands, the scheduling strategy of refrigeration equipment can be effectively optimized; by capturing short-term, medium-term, and long-term dependency information through a multi-layer MT-LSTM model, a cold load demand prediction value is generated based on a comprehensive analysis of environmental temperature, humidity, equipment status, and historical load data, and reasonable weighted fusion and time series processing further improve the prediction accuracy; ultimately, an accurate cold load demand prediction value is provided, significantly reducing energy consumption, fully meeting dynamic cold load demands, and enhancing the operating efficiency and reliability of the entire refrigeration system.

[0113] S3. Based on the cold load demand prediction value, an optimal scheduling strategy is generated through the NSGA-II optimization algorithm.

[0114] Furthermore, the NSGA-II optimization algorithm is selected to handle multi-objective optimization problems. The optimization objectives are defined as minimizing the energy consumption of refrigeration equipment (under the premise of meeting the cold load demand value and other refrigeration objectives, the total energy consumption of refrigeration equipment is reduced as much as possible through the optimal scheduling strategy), while ensuring that the cold load demand prediction value is met.

[0115] Set the start time and shutdown time of multiple refrigeration equipment according to the cold load prediction value, and allocate the load ratio among multiple refrigeration equipment.

[0116] Specifically, collect the cold load demand prediction value and equipment status data, and analyze the load changes at different time points.

[0117] According to the peak load period, arrange the equipment to start in advance to ensure sufficient refrigeration before the peak load.

[0118] According to the load decline trend, arrange the equipment to shut down gradually and reasonably adjust the operating load of the equipment.

[0119] Divide the predicted cold load demand prediction value into several time periods to ensure appropriate load ratio allocation for each time period.

[0120] Calculate the load demand of each equipment in different time periods to ensure that the total load meets the cold load prediction value and each equipment operates within a reasonable load range.

[0121] Convert the start time, shutdown time, and load ratio allocation of each refrigeration equipment into chromosome encoding.

[0122] Specifically, it is agreed that each gene position of the chromosome represents equipment and time information. For example, the 1st - 5th positions represent the start time of equipment 1, and the 6th - 10th positions represent the load ratio allocation of equipment 1, etc.

[0123] According to the set encoding rules, convert the equipment start, shutdown, and load ratio allocation into independent chromosomes to prepare for the input of the subsequent optimization algorithm.

[0124] For example: Suppose there are three refrigeration devices now, and the on / off state and load ratio distribution of the i-th device need to be encoded in the following way during a specific time period;

[0125] The encoding details are that the start time and the stop time are represented by 5-bit binary numbers (for example, the maximum 23 is 5 bits, that is, 00000 to 10111 represent 24-hour time points);

[0126] The load ratio distribution is represented by 4-bit binary numbers (0000 to 1111, corresponding to 0.0 to 1.0);

[0127] If the start and stop times of the device are within the 24-hour time points, they are represented by 0 - 23;

[0128] If the load ratio distribution of the device is within 0 - 1, then 0.5 represents that the device load is 50%;

[0129] Suppose there are three devices (A, B, C), and their on / off times and load ratio distributions are encoded;

[0130] The start time of device A is: 2 (2 am), and the converted encoding is: 00010;

[0131] The stop time is: 10 (10 am), and the converted encoding is: 01010;

[0132] The load ratio distribution is: 0.6, and the converted encoding is: 1001;

[0133] The start time of device B is: 6 (6 am), and the converted encoding is: 00110;

[0134] The stop time is: 18 (6 pm), and the converted encoding is: 10010;

[0135] The load ratio distribution is: 0.8, and the converted encoding is: 1100;

[0136] The start time of device C is: 8 (8 am), and the converted encoding is: 01000;

[0137] The stop time is: 16 (4 pm), and the converted encoding is: 10000;

[0138] The load ratio distribution is: 0.4, and the converted encoding is: 0110;

[0139] Combining the encodings of devices A, B, and C into a chromosome encoding is: 000100101010011000100101100000010000100000110;

[0141] Generate an initial scheduling plan based on chromosome coding using the NSGA-II optimization algorithm;

[0142] Furthermore, define the start time of chromosome coding as a five-bit binary number to generate an initial chromosome;

[0143] Define the shutdown time of chromosome coding as a five-bit binary number to generate an initial chromosome;

[0144] Define the load ratio allocation of chromosome coding as a four-bit binary number to generate an initial chromosome;

[0145] Combine the initial chromosomes of start time, shutdown time, and load ratio allocation to generate coding information;

[0146] For the coding information, perform disassembling and decoding to obtain the specific start time, shutdown time, and load ratio allocation of each refrigeration device, and generate an initial scheduling plan.

[0147] Illustrate with an example:

[0148] Suppose there are three refrigerations, namely A, B, and C;

[0149] Among them, for device A: start time 2 (00010), shutdown time 10 (01010), load ratio 0.6 (1001);

[0150] For device B: start time 6 (00110), shutdown time 18 (10010), load ratio 0.8 (1100);

[0151] For device C: start time 8 (01000), shutdown time 16 (10000), load ratio 0.4 (0110);

[0152] Then combine the initial chromosomes of A, B, and C to generate coding information as: 000100101010011000100101100000010000100000110;

[0153] Based on the combined coding information, then perform disassembling and decoding to obtain the specific start time, shutdown time, and load ratio allocation of each refrigeration device;

[0154] Time 0 - 1: All devices are off;

[0155] Time 2 - 6: Device A starts, load 0.6;

[0156] Time 6 - 8: Devices A and B start, load A: 0.6, B: 0.8;

[0157] Time 8 - 10: Devices A, B, and C start, load A: 0.6, B: 0.8, C: 0.4;

[0158] At 10 - 16: Equipment B and C start, and the load is distributed according to the load ratio.

[0159] At 16 - 18: Equipment B runs and C stops.

[0160] At 18 - 24: All equipment shuts down.

[0161] By iteratively repeating the above encoding and decoding, obtain the specific start time, shutdown time, and load ratio distribution of each refrigeration equipment, generate multiple initial scheduling schemes, and ensure the diversity of the population size (such as 20 - 100 initial scheduling schemes).

[0162] Optimize the initial scheduling schemes through crossover simulation and mutation simulation to generate new scheduling schemes.

[0163] Specifically, select chromosomes with better genes in the initial scheduling schemes through crossover simulation, and generate new scheduling schemes through crossover operations (such as pairwise crossover recombination).

[0164] Randomly select some chromosomes through mutation simulation, and generate new scheduling schemes through mutation operations (such as randomly changing gene position information) to ensure diversity.

[0165] Repeatedly perform crossover and mutation simulations, evaluate the new scheduling schemes, and continuously optimize the scheduling strategy.

[0166] For the new scheduling schemes, calculate the total energy consumption of the refrigeration equipment, and the expression is:

[0167]

[0168] Among them, E is the total energy consumption of the refrigeration equipment, P i (t) is the power of the i - th equipment at time t, Q i (t) is the running duration of the i - th equipment, i is the equipment, and n is the total number of equipment.

[0169] Evaluate the total energy consumption of the refrigeration equipment in the new scheduling schemes through non - dominated sorting (Pareto sorting) to generate an optimized scheduling strategy.

[0170] Specifically, sort each optimized scheduling scheme according to the total energy consumption and load satisfaction of the refrigeration equipment, and select the optimal solution that is not "dominated" by other schemes.

[0171] For multiple new scheduling schemes, evaluate the total energy consumption and balanced load satisfaction of each scheme, and finally determine the optimized scheduling strategy.

[0172] The final optimized scheduling strategy includes the specific start time, shutdown time, load ratio distribution, and running power of each equipment.

[0173] Generate the final optimized scheduling strategy on the premise of minimizing the total energy consumption and meeting the load.

[0174] It should be noted that by using the NSGA-II optimization algorithm to generate the optimized scheduling strategy based on the predicted value of the cold load demand, the efficient scheduling of the refrigeration equipment is realized. On the premise of meeting the cold load demand, by optimizing the start time, shutdown time and load ratio allocation of the equipment, the total energy consumption of the refrigeration equipment is minimized to the greatest extent. Combining the crossover simulation and mutation simulation ensures the diversity and optimization degree of the scheduling scheme, significantly reduces the energy consumption, improves the operation efficiency and reliability of refrigeration, and ensures the long-term stable, economical and efficient operation of each equipment.

[0175] S4. According to the optimized scheduling strategy, use the IoT execution device to turn on and off and allocate the load ratio, and monitor the operation data in real time to dynamically adjust and optimize the operation state of the refrigeration equipment.

[0176] Furthermore, extract the start time, shutdown time and load ratio allocation of the refrigeration equipment from the optimized scheduling strategy as control instructions.

[0177] Specifically, if device A starts in a certain time period (0, 1), the optimized scheduling strategy will clearly indicate that device A starts at time t = 0.

[0178] If device A starts in a certain time period (0, 1), the shutdown time will indicate that device A shuts down at time t = 1.

[0179] If device A accounts for 50% of the load in the time period (0, 1), the optimized scheduling strategy will clearly indicate that the load ratio allocation of device A in this time period is 0.5.

[0180] Establish the connection between each refrigeration equipment and the Internet of Things through the IoT.

[0181] Specifically, install relevant sensors (such as temperature, humidity, pressure, flow sensors) and controllers (such as programmable logic controller (PLC), intelligent controller (IoT) and variable frequency controller (VFD)) on each refrigeration equipment (such as compressor, cooling tower, water pump, etc.).

[0182] Connect a variety of sensors and controllers to the refrigeration equipment by wired or wireless means to realize the collection of operation data and the transmission of control signals.

[0183] Select a suitable network connection method according to the specific situation of the refrigeration machine room, such as Wi-Fi, Ethernet, Zigbee, LoRa or 5G communication.

[0184] Configure a variety of sensors and controllers into the selected network to ensure that these devices can communicate with the Internet of Things platform stably.

[0185] Install one or more Internet of Things (IoT) gateways in the refrigeration machine room to collect the operation data from various sensors and controllers and transmit this operation data to the cloud or the central control center;

[0186] Select a suitable data transmission protocol, such as MQTT, HTTP, CoAP, etc., to ensure the efficient transmission and security of the operation data;

[0187] Send the control instructions for extracting the start time, shutdown time, and load ratio allocation of the refrigeration equipment to the corresponding refrigeration equipment via IoT for startup, shutdown, and load ratio allocation;

[0188] Use environmental temperature sensors, environmental humidity sensors, and equipment operation status monitors to collect operation data in real time;

[0189] Transmit and store the collected real-time operation data to the central control center via IoT;

[0190] The central control center monitors the operation data in real time through the support vector machine algorithm. When the operation data is abnormal, it indicates that the equipment has a fault;

[0191] The central control center automatically sends an alarm and initiates an adjustment according to the faults that occur in the equipment;

[0192] If it is slightly abnormal, automatically adjust the load ratio allocation and temperature parameters of the refrigeration equipment;

[0193] If it is severely abnormal, start the standby refrigeration equipment;

[0194] Specifically, collect the historical operation data of the equipment under normal operation conditions, including environmental temperature data, environmental humidity data, equipment status data, and historical load data;

[0195] Perform data cleaning, noise removal, and missing value handling on the historical operation data to ensure the quality of the operation data;

[0196] Extract key features from the operation data, such as extracting the temperature change rate from the environmental temperature data, extracting the pressure fluctuation from the pressure data in the equipment status data, and extracting the flow rate fluctuation from the current and voltage data in the equipment status data, etc.;

[0197] Build a support vector machine model based on the support vector machine algorithm;

[0198] Divide the extracted key features into a training set and a test set. The training set is used to train the support vector machine model, and the test set is used to verify the performance of the support vector machine model;

[0199] Train a support vector machine model using support vector machine algorithms (such as linear SVM, radial basis kernel SVM, etc.) to learn the characteristic patterns of normal operating states;

[0200] Collect the operating data of the device in real time, such as environmental temperature data, environmental humidity data, device status data, and historical load data;

[0201] Standardize and extract features from the real-time operating data, and convert it into a format consistent with the training set;

[0202] Input the real-time data into the trained support vector machine model for pattern matching and analysis;

[0203] The support vector machine model analyzes the input real-time operating data to determine whether it matches the normal operating mode;

[0204] When the support vector machine model detects that the input real-time data does not match the normal mode, it is determined as abnormal data;

[0205] Identify specific types such as temperature anomalies, pressure anomalies, or flow anomalies based on the characteristics of the abnormal data;

[0206] Once the support vector machine model detects abnormal data, the central control center will immediately mark and generate an abnormal event;

[0207] The central control center will immediately send an alarm notification to the designated management personnel or maintenance team via text message, email, Internet of Things applications (such as mobile apps), etc. The alarm content includes the abnormal device, abnormal type, and specific data indicators;

[0208] Automatically evaluate the severity of the abnormal situation based on the abnormal data and abnormal event type;

[0209] For minor anomalies, adjust parameters such as the load ratio distribution and temperature setting of the device automatically to restore the normal operating state. For example, adjust the load of the compressor and lower the temperature;

[0210] For minor anomalies, if the abnormal situation is relatively serious, automatically start the standby device to ensure uninterrupted cooling. For example, when device A fails, automatically start device B to replace it;

[0211] Record all abnormal events and adjustment operations to ensure that the subsequent maintenance and review can trace back to the specific process;

[0212] It should be noted that by using the Internet of Things (IoT) to execute equipment start / stop and load distribution, real-time monitoring of the operating data of the equipment is carried out to achieve dynamic adjustment. The support vector machine algorithm is used to monitor anomalies, and the equipment load and temperature settings are automatically adjusted, or standby equipment is started in case of severe anomalies, ensuring the efficient operation of the equipment, significantly reducing energy consumption and failures, while improving the overall stability and reliability, and providing comprehensive automated control and optimization solutions.

[0213] This embodiment also provides a device for optimizing the performance of an efficient refrigeration machine room, including: a collection module, a prediction module, an optimization module, and a monitoring module; the collection module collects the operating data of the refrigeration equipment through a variety of sensors and performs preprocessing; the prediction module constructs an MT-LSTM model based on the preprocessed operating data to predict the future cooling load demand and generates a predicted value of the cooling load demand; the optimization module generates an optimized scheduling strategy through the NSGA-II optimization algorithm based on the predicted value of the cooling load demand; the monitoring module, according to the optimized scheduling strategy, executes equipment start / stop and load ratio distribution through the IoT and real-time monitors the operating data to dynamically adjust and optimize the operating state of the refrigeration equipment.

[0214] This embodiment also provides a computer device applicable to the situation of the method for optimizing the performance of an efficient refrigeration machine room, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method for optimizing the performance of an efficient refrigeration machine room as proposed in the above embodiment.

[0215] This computer device can be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of this computer device is used to provide computing and control capabilities. The memory of this computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of this computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of this computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of this computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the shell of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0216] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for optimizing the performance of an efficient refrigeration machine room as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, abbreviated as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, abbreviated as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, abbreviated as EPROM), programmable read-only memory (Programmable Red-Only Memory, abbreviated as PROM), read-only memory (Read-Only Memory, abbreviated as ROM), magnetic memory, flash memory, magnetic disk or optical disc.

[0217] In summary, the present invention constructs an MT-LSTM model to predict future cooling load demands, achieving accurate demand prediction, which plays an important role in optimizing the scheduling of refrigeration equipment. The MT-LSTM model can capture short-term, medium-term, and long-term dependency information. By processing various feature data, it generates reliable cooling load demand prediction values. In this way, even in the face of complex environmental changes and equipment operating states, the MT-LSTM model can still provide accurate demand prediction, thereby avoiding over-cooling or under-cooling, improving the overall efficiency of the operation of refrigeration equipment, reducing unnecessary energy consumption, and at the same time ensuring the accurate satisfaction of cooling load demands.

[0218] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for optimizing the performance of a high-efficiency refrigeration room, characterized in that: include, Collect the operating data of refrigeration equipment through various sensors and perform pre-processing; Based on the preprocessed operation data, the MT-LSTM model is constructed to predict the future cooling load demand and generate the cooling load demand prediction value; Based on the predicted value of cooling load demand, the optimal scheduling strategy is generated through the NSGA-II optimization algorithm; According to the optimized scheduling strategy, IoT is used to execute equipment startup and shutdown and load ratio distribution, and to monitor operating data in real time to dynamically adjust and optimize the operating status of the refrigeration equipment.

2. The high-efficiency refrigeration room performance optimization method according to claim 1, characterized in that: The operation data of the equipment includes ambient temperature data, ambient humidity data, equipment status data and historical load data.

3. The high-efficiency refrigeration room performance optimization method according to claim 2, characterized in that: The preprocessing includes denoising, data standardization and missing value filling.

4. The high-efficiency refrigeration room performance optimization method according to claim 3, characterized in that: The MT-LSTM model is constructed based on the preprocessed operation data to predict the future cooling load demand and generate the cooling load demand prediction value. The specific steps are: Perform weighted fusion on the preprocessed running data to generate a unified data set; Divide the unified dataset into training set, validation set and test set; Build an MT-LSTM model based on LSTM, define the MT-LSTM model input as the training set, and the output as the cold load demand forecast value; Extract the time series data in the training set, split it into multiple time periods according to the time steps of the time series data, form a time segment sequence, and input it into the multi-layer LSTM for processing; The first layer LSTM receives the time segment sequence, processes the time segment sequence through the gating mechanism, captures the short-term dependency information, and inputs it to the second layer LSTM; The second layer LSTM receives short-term dependency information, processes it through gating mechanisms and filtering, captures medium-term dependency information, and inputs it to the third layer LSTM; The third LSTM layer receives medium-term dependency information, combines the short-term dependency information of the first LSTM layer and the medium-term dependency information of the second LSTM layer through a gating mechanism, captures long-term dependency information, and outputs it to the fully connected layer; The fully connected layer receives and integrates short-term dependency information, medium-term dependency information, and long-term dependency information to calculate the cold load demand forecast value, which is expressed as: in, is the predicted value of cooling load demand, D is the cooling load demand, pred is the predicted value, T e,t is the ambient temperature data at the tth time point, H e,t is the ambient humidity data at the tth time point, e is the environmental identifier, S t is the equipment operation status data at the tth time point, D t-1 is the historical load data at the t-th time point, t is the time, and m is the total time; Use the validation set to evaluate the performance of the MT-LSTM model and adjust the parameters; The test set is used to measure the generalization ability and prediction accuracy of the MT-LSTM model.

5. The high-efficiency refrigeration room performance optimization method according to claim 4, characterized in that: The optimization scheduling strategy is generated by the NSGA-II optimization algorithm based on the cold load demand forecast value. The specific steps are: Select the NSGA-II optimization algorithm and define the optimization goal as minimizing the energy consumption of the refrigeration equipment; Set the start-up and shutdown times of multiple refrigeration devices according to the cold load prediction value, and distribute the load proportions of multiple refrigeration devices; Convert the start-up time, shutdown time and load ratio distribution of each refrigeration equipment into chromosome code; The NSGA-II optimization algorithm is used to generate the initial scheduling plan based on the chromosome coding; Optimize the initial scheduling plan through crossover simulation and mutation simulation to generate a new scheduling plan; According to the new scheduling scheme, the total energy consumption of the refrigeration equipment is calculated, and the expression is: Where E is the total energy consumption of refrigeration equipment, P i (t) is the power of the ith device at time t, Q i (t) is the operating time of the i-th device, i is the device, and n is the total number of devices; The total energy consumption of refrigeration equipment in the new scheduling scheme is evaluated by the non-dominated sorting method to generate an optimal scheduling strategy.

6. The high-efficiency refrigeration room performance optimization method according to claim 5, characterized in that: The NSGA-II optimization algorithm is used to generate an initial scheduling plan based on chromosome coding, and the specific steps are: Define the start time of chromosome encoding as a five-bit binary number to generate the initial chromosome; Define the closing time of chromosome encoding as a five-bit binary number and generate the initial chromosome; Define the load ratio of chromosome encoding as a four-bit binary number to generate the initial chromosome; The initial chromosomes of the startup time, shutdown time and load ratio distribution are combined to generate encoding information; The coded information is disassembled and decoded to obtain the specific start-up time, shutdown time and load ratio distribution of each refrigeration equipment, and generate an initial scheduling plan.

7. The high-efficiency refrigeration room performance optimization method according to claim 6, characterized in that: According to the optimization scheduling strategy, the device startup and shutdown and load ratio distribution are executed through IoT, and the operation data is monitored in real time to dynamically adjust and optimize the operation status of the refrigeration equipment. The specific steps are as follows: Extracting the start-up time, shutdown time and load ratio distribution of the refrigeration equipment from the optimization scheduling strategy as control instructions; Establish connections between each refrigeration equipment and the Internet of Things through IoT; The control instructions for extracting the start-up time, shutdown time and load ratio distribution of the refrigeration equipment are sent to the corresponding refrigeration equipment through IoT to start, shut down and distribute the load ratio; Use ambient temperature sensors, ambient humidity sensors and equipment operation status monitors to collect operation data in real time; The collected real-time operation data is transmitted and stored in the central control center via IoT; The central control center monitors the operation data in real time through the support vector machine algorithm. When the operation data is abnormal, it means that the equipment is faulty; The central control center automatically sends alarms and initiates adjustments based on equipment failures; If there is a slight abnormality, the load ratio distribution and temperature parameters of the refrigeration equipment will be automatically adjusted; If there is a serious abnormality, start the backup refrigeration equipment.

8. A high-efficiency refrigeration room performance optimization device, based on the high-efficiency refrigeration room performance optimization method according to any one of claims 1 to 7, characterized in that: Including, acquisition module, prediction module, optimization module and monitoring module; The acquisition module is used to collect the operating data of the refrigeration equipment through various sensors and perform pre-processing; The prediction module is used to construct an MT-LSTM model to predict future cooling load demand based on the preprocessed operation data and generate a cooling load demand prediction value; The optimization module is used to generate an optimization scheduling strategy through the NSGA-II optimization algorithm based on the cold load demand forecast value; The monitoring module is used to execute equipment startup and shutdown and load ratio distribution through IoT according to the optimized scheduling strategy, and to monitor the operation data in real time, dynamically adjust and optimize the operation status of the refrigeration equipment.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the high-efficiency refrigeration room performance optimization method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the high-efficiency refrigeration room performance optimization method according to any one of claims 1 to 7 are implemented.

Images