Refrigerator multi-objective energy consumption optimization method and system based on fusion of genetic algorithm and xLSTM

Abstract

The application discloses a refrigerator multi-objective energy consumption optimization method and system based on genetic algorithm and xLSTM fusion, comprising obtaining the temperature, ambient temperature, door opening and closing state, compressor state and power of the current refrigerator, and preprocessing the obtained data; inputting the running state and temperature history information of the current refrigerator into the xLSTM model, determining how much historical information to retain by the forgetting gate, determining the number of new information to retain by the input gate and the candidate memory, processing the forgetting gate and the input gate after being indexed through the memory unit, and generating the final prediction output by the output gate in combination with the memory unit; fusing the energy consumption target, temperature control target and equipment life target into a comprehensive target function, iteratively outputting the optimal parameters by the genetic algorithm based on the prediction result, and executing the equipment according to the optimal parameters; and repeating the above process in each optimization cycle. The application can effectively reduce the energy consumption of the refrigerator and prolong the service life of the compressor under the premise of ensuring the safety of goods.

Description

Technical Field

[0001] This invention relates to the field of energy-saving and consumption-reducing technology for commercial freezers, and in particular to a multi-objective energy consumption optimization method and system for freezers based on the fusion of genetic algorithm and xLSTM. Background Technology

[0002] In the catering and retail industries, commercial refrigerated display cases, as core equipment in the cold chain, are widely used in scenarios such as food storage, refrigerated display, and beverage preservation. Their energy consumption typically accounts for more than 40% of a store's total electricity consumption, and even higher in some high-density usage scenarios. This high energy consumption not only directly increases the operating costs of stores but also becomes one of the main sources of carbon emissions in the industry. Therefore, improving the energy efficiency of commercial refrigerated display cases and reducing their operating energy consumption is not only an important means for enterprises to control operating costs and improve profitability, but also a key link in implementing energy conservation and emission reduction policies and promoting the green transformation of the industry, possessing significant economic value and environmental significance.

[0003] In recent years, with the rapid development of automatic control theory and artificial intelligence technology, intelligent control technology has been gradually introduced into the commercial refrigerator field to replace traditional temperature control methods based on switch control or PID regulation. Current applications include methods such as fuzzy control, neural network control, and expert systems, which achieve a certain degree of energy efficiency optimization by dynamically modeling and adaptively adjusting the refrigerator's operating status.

[0004] However, existing intelligent control technologies still have significant shortcomings in the commercial refrigeration equipment sector. First, most methods rely on offline training or fixed rule bases, lacking the ability to adapt to real-time environmental changes (such as ambient temperature, door opening / closing frequency, and food loading volume), resulting in limited energy efficiency improvements under actual operating conditions. Second, while traditional LSTM (Long Short-Term Memory) networks and their variants have been attempted for temperature prediction, their memory update mechanism is limited by a linear gating structure, making it difficult to efficiently capture multi-scale temporal features in temperature changes. Especially under non-stationary conditions such as door opening / closing disturbances and sudden changes in ambient temperature, the model's response to load changes is significantly lagging, easily leading to the risk of temperature exceeding limits and affecting the safety of stored goods.

[0005] Furthermore, most existing intelligent control strategies are geared towards a single optimization objective, such as focusing solely on reducing energy consumption or pursuing temperature stability, neglecting the inherent conflict between energy efficiency, temperature control accuracy, and compressor lifespan. In actual operation, frequent compressor start-ups and shutdowns, while beneficial for temperature control, shorten equipment lifespan and increase energy consumption; conversely, excessively extending the compressor operating cycle, while reducing the number of start-ups and shutdowns, may lead to excessive temperature fluctuations. Existing methods lack effective multi-objective collaborative optimization mechanisms, making it difficult to automatically search for the optimal control parameters while meeting temperature constraints, thus failing to achieve a global balance between energy efficiency, temperature control, and equipment lifespan. Summary of the Invention

[0006] To overcome the aforementioned problems in the existing technology, this invention proposes a multi-objective energy consumption optimization method and system for freezers based on the fusion of genetic algorithm and xLSTM.

[0007] The technical solution adopted by this invention to solve its technical problem is: a multi-objective energy consumption optimization method for freezers based on the fusion of genetic algorithm and xLSTM, comprising the following steps: Step 1: Obtain the current temperature inside the freezer, ambient temperature, door open / close status, compressor status, and power, and preprocess the acquired data; Step 2: Input the current operating status and historical temperature information of the freezer into the xLSTM model. The forget gate determines how much historical memory to retain, the input gate determines the importance of new information, the candidate memory extracts the new information at the current moment, the forget gate and the input gate are exponentialized and then processed through the memory unit, and the output gate is combined with the memory unit to generate the final prediction output. Step 3: The energy consumption target, temperature control target, and equipment life target are integrated into a comprehensive objective function. Based on the prediction results of Step 2, under the conditions of satisfying the upper and lower temperature limits and the compressor start-stop count constraints, the optimal parameters are iterated through a genetic algorithm, and the equipment executes according to the optimal parameters. Step 4: Repeat steps 1-3 within each optimization cycle. If an abnormal situation of a sharp rise in temperature occurs between optimization cycles, the system will immediately interrupt the optimization process and forcibly start the compressor to cool down quickly.

[0008] The above-mentioned multi-objective energy consumption optimization method for freezers based on the fusion of genetic algorithm and xLSTM, specifically step 2 is as follows: Step 2.1, passing through the forgetting gate To determine how much of the past memory to retain, among which It is the input at the current moment. It is a summary of memories from the previous moment. It's the Sigmoid function, which outputs a value between 0 and 1; input gate. and candidate memories They jointly decide how much new information to store in memory; Step 2.2: After obtaining the forget gate and input gate, perform indexing on both the forget gate and input gate. ; Step 2.3, the memory unit passes through Update, among which This indicates element-wise multiplication. This represents the system's long-term memory of temperature change patterns. Represents the Gate of Oblivion Indicates candidate memories, Indicates the input gate; Step 2.4, Output Gate Combined with memory units, through Generate the final prediction output. That is, the temperature prediction value several steps in the future. .

[0009] The above-mentioned multi-objective energy consumption optimization method for freezers based on the fusion of genetic algorithm and xLSTM, specifically the comprehensive objective function in step 3 is as follows:

[0010] in , , It is an adjustable weighting coefficient. Energy consumption target; For temperature control objectives, For equipment lifespan targets.

[0011] The above-mentioned multi-objective energy consumption optimization method for freezers based on the fusion of genetic algorithm and xLSTM, the specific calculation formulas for the energy consumption objective, temperature control objective, and equipment lifespan objective are as follows:

[0012]

[0013]

[0014] in, It is the number of predicted steps. The power is calculated based on the compressor's predicted operating status. It is a time interval; It is the highest temperature permissible for food safety. Ensure that penalties are only incurred when the temperature exceeds the upper limit; This refers to the number of times the compressor will start and stop within a certain period of time. It is the penalty coefficient for each start-stop cycle. It is the single continuous run time. This is the maximum allowed runtime. This is the penalty coefficient for exceeding the time limit.

[0015] The above-mentioned multi-objective energy consumption optimization method for refrigerated display cases based on the fusion of genetic algorithm and xLSTM, in step 3, the genetic algorithm specifically involves: generating an initial population containing multiple random individuals, with each control scheme representing an individual; evaluating the quality of each individual using a fitness function; randomly selecting two individuals from the population for comparison each time, with the individual with higher fitness advancing to the next generation, and the two individuals with the highest fitness in each generation being directly retained for the next generation; using a simulated binary crossover method, independently calculating the expansion factor for each decision variable, so that the offspring can inherit the characteristics of the parents; employing a multinomial mutation method, randomly perturbing the parameters of each individual with a fixed probability, with the variable length controlled by the distribution exponent; and finally outputting the optimal control parameters after multiple generations of iteration.

[0016] The fitness function of the above-mentioned multi-objective energy consumption optimization method for freezers based on the fusion of genetic algorithm and xLSTM is as follows:

[0017] Where J represents the comprehensive objective function, To prevent extremely small positive numbers with a denominator of zero.

[0018] The aforementioned multi-objective energy consumption optimization system for freezers based on the fusion of genetic algorithm and xLSTM is used to execute the multi-objective energy consumption optimization method described above. It includes a data acquisition layer, an edge computing layer, an execution control layer, and a cloud platform layer. The data acquisition layer is used to collect freezer operation data and environmental data. The edge computing layer outputs the temperature change trend and load status of the future prediction time domain through an xLSTM model based on the data collected by the data acquisition layer. Based on the prediction data, a genetic algorithm is executed to iteratively find the optimal control parameters that balance the energy consumption target, temperature control target, and equipment lifespan target. The energy consumption target is the total energy consumption over a future period, the temperature control target is the cumulative penalty for exceeding the temperature limit, and the equipment lifespan target is the number of compressor start-stop cycles and the penalty for single operation time. The execution control layer executes specific instructions based on the optimal control parameters solved by the edge computing layer; The cloud platform layer is deployed in the data center and is used for remote monitoring, model management, parameter configuration, and data analysis.

[0019] The beneficial effects of this invention are that it introduces an xLSTM model, which adds an exponential gating mechanism to the traditional LSTM model, expanding the dynamic range of memory updates and enabling it to more accurately learn the multi-scale characteristics of temperature changes. Based on this, xLSTM can accurately grasp the dynamic patterns of temperature changes with door opening and closing and ambient temperature variations from historical data, predicting upcoming load changes in advance, effectively avoiding the risk of exceeding limits due to temperature response lag, and significantly improving the safety of commodity storage.

[0020] This invention constructs a multi-objective optimization function encompassing energy consumption, temperature control, and equipment lifespan. By combining this with the global optimization capability of a genetic algorithm, it effectively balances the inherent conflicts among these three objectives. Under the premise of satisfying temperature constraints, it searches for the control parameters with the optimal overall performance, achieving synergistic optimization of energy efficiency, temperature control, and equipment lifespan. Furthermore, this invention operates using a rolling time-domain control method, enabling the control strategy to dynamically adapt to environmental changes and load fluctuations. It also activates a safety protection mechanism in the event of temperature anomalies, ensuring product safety. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the process of this invention; Figure 2 This is a flowchart of the xLSTM model processing of the present invention; Figure 3 This is a flowchart of the genetic algorithm of the present invention. Detailed Implementation

[0022] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] This invention proposes an intelligent control method for commercial refrigerators that integrates genetic algorithms and xLSTM. It operates in a rolling optimization manner: periodically, the system collects the current state, uses xLSTM to predict the temperature change trend over a future period, then uses a genetic algorithm to solve for a set of optimal control parameters, and finally sends control commands to the compressor for execution. This process is repeated continuously to achieve synergistic optimization of three objectives: energy consumption, temperature stability, and compressor lifespan. The specific implementation process is as follows: Figure 1 As shown.

[0024] The temperature inside a freezer is not constant but is affected by various heat exchanges. This invention simplifies this process into a single heat balance equation: The left side of this equation represents the rate of change of the cooling capacity inside the cabinet over time, where... This refers to the heat capacity of the freezer, which can be understood as its ability to store cold air. It's the temperature inside the cabinet. This represents the rate of temperature change. The right side of the equation represents the three forces affecting temperature: It is the food load, which is the amount of calories brought in by food; This refers to heat intrusion caused by opening and closing the door; the more frequently the door is opened, the higher this value becomes. This refers to the cooling capacity produced by the compressor. When the incoming heat is greater than the cooling capacity, the temperature rises; conversely, it falls. The cooling capacity itself is closely related to the compressor's operating state and can be expressed as... ,in It is the compressor operating frequency. and These are equipment characteristic parameters determined by the refrigeration system of the freezer. The frequency is adjusted... This allows you to control the cooling capacity, which in turn affects the temperature.

[0025] However, adjusting the compressor solely based on the current temperature is insufficient because temperature changes have a lag. Waiting until the temperature has already exceeded the limit before activating the compressor often poses a safety hazard to the product. Therefore, this invention introduces an xLSTM model to predict future temperature trends. xLSTM is an improved recurrent neural network that learns the temporal patterns of temperature changes through a series of gating mechanisms. The xLSTM model processing flow is as follows: Figure 2 As shown. Specifically, the model first uses a forgetting gate. To determine how much of the past memory to retain, among which It is the input at the current moment (including current temperature, ambient temperature, door open / closed status, etc.). It is a summary of memories from the previous moment. It's the Sigmoid function, which outputs a value between 0 and 1. Also, the input gate... and candidate memories Together, they decide how much new information to store in memory. Unlike standard LSTM, xLSTM obtains the forget gate. and input gate Then, it is further indexed, that is... and This expands the gating value range from (0,1) to (1, e), thereby enhancing the dynamic range of the gating. This exponential gating mechanism allows the model to strengthen the retention of historical memory when facing long-term patterns in the operation of the freezer (such as diurnal temperature variation cycles), while responding quickly to new information when facing sudden changes (such as frequent door opening and closing). Subsequently, the memory unit... Update, among which This indicates element-wise multiplication. This is equivalent to the system's long-term memory of temperature change patterns. Finally, the output gate... Combined with memory units, through Generate the final prediction output. That is, the temperature prediction value several steps in the future. Through this mechanism, xLSTM can learn from historical data how temperature changes with door opening and closing and ambient temperature, thus predicting upcoming load changes in advance.

[0026] Having predicted future temperatures, the next step is to determine how to control the compressor to minimize energy consumption and wear and tear while still meeting temperature requirements. This is a typical multi-objective optimization problem because there are conflicts between the three objectives: energy saving, temperature control, and extending compressor life. This invention integrates these three objectives into a single comprehensive objective function. ,in , , These are adjustable weighting coefficients, set according to actual needs. For example, the weight of temperature targets can be appropriately increased in summer. Energy consumption targets. Defined as the total energy consumption over a future period of time, i.e. ,here It is the number of predicted steps. The power is calculated based on the compressor's predicted operating status. It's a time interval; the entire formula sums up the electrical energy consumed at each future step. Temperature control target. Only cases where the temperature exceeds the limit are punished, defined as

[0027] ,in It is the highest temperature permissible for food safety. The penalty is only applied when the temperature exceeds the upper limit; the greater the exceedance and the longer the duration, the greater the penalty. Equipment lifespan target. This focuses on compressor start-up and shutdown wear and long-term operational damage, defined as... ,in This refers to the number of times the compressor will start and stop within a certain period of time. It is the penalty coefficient for each start-stop cycle. It is the single continuous run time. This is the maximum allowed runtime. This is the penalty coefficient for exceeding the timeout period. This formula means that the more frequent the start-stop cycles and the longer the single run time, the greater the wear and tear on the compressor's lifespan.

[0028] Faced with such a complex optimization problem, traditional analytical methods are insufficient to solve it. Therefore, this invention employs a genetic algorithm to search for the optimal control parameters. The genetic algorithm process is as follows: Figure 3 As shown. In genetic algorithms, each control scheme is encoded as an individual, for example, using a vector. To indicate, among which It's setting the temperature. It is the minimum running time. It is the minimum downtime. This is the highest allowed frequency. The algorithm first generates an initial population containing multiple random individuals, each representing a possible control policy, and then uses a fitness function... To assess the quality of each individual, among which It is the previously calculated objective function value. It is a very small positive number to prevent the denominator from being zero, fitness and The fitness of offspring is inversely proportional to the energy consumption, temperature control, and lifespan loss. This invention employs a tournament selection method to screen superior individuals. Specifically, two individuals are randomly selected from the population for comparison each time, and the individual with higher fitness is selected for the next generation. This selection method is computationally efficient and effectively avoids premature entrapment in local optima. Simultaneously, the two individuals with the highest fitness in each generation are directly retained for the next generation, ensuring that the optimal solution is not lost. For the crossover operation, since the control parameters in this invention are continuous real numbers, a simulated binary crossover method is used. This method independently calculates the expansion factor for each decision variable, allowing offspring to naturally inherit parental characteristics. The crossover probability is set to 0.8, and the distribution index is set to 20 to ensure that offspring maintain appropriate diversity while inheriting superior genes. For the mutation operation, a multinomial mutation method is used, randomly perturbing the parameters of each individual with a probability of 0.1. The mutation length is controlled by the distribution index (also set to 20), resulting in a small mutation amplitude and fine-grained searching within a local range. This avoids destroying superior genes while maintaining the population's exploratory ability. After multiple generations of iteration, the population typically converges after 50 generations of evolution. At this point, the optimal set of control parameters obtained represents the best trade-off between energy consumption, temperature control, and equipment lifespan under the current environmental conditions.

[0029] The entire system operates using rolling time-domain control, forming a closed loop. In each optimization cycle, for example every 15 minutes, the system first collects real-time data such as the cabinet temperature, ambient temperature, door open / closed status, and compressor operating status. This data is then input into a pre-trained xLSTM model, which outputs the temperature trend and load conditions for a predicted time domain (e.g., one hour in the future). Next, a genetic algorithm, based on this prediction, searches for the optimal combination of control parameters while satisfying upper and lower temperature limits and compressor start / stop frequency constraints. Finally, the controller executes according to these optimal parameters, including adjusting the temperature setpoint and setting compressor start / stop thresholds or operating frequencies. At the start of the next optimization cycle, the system collects the latest data again, performs prediction and optimization again, and so on, allowing the control strategy to dynamically adapt to environmental changes and load fluctuations. If an abnormal situation of a sharp temperature rise occurs between optimization cycles, the system immediately interrupts the optimization process, forcibly starts the compressor for rapid cooling, and prioritizes food safety.

[0030] By integrating the predictive capabilities of xLSTM and the multi-objective optimization capabilities of genetic algorithms into a control architecture, this invention can effectively reduce the energy consumption of commercial freezers while ensuring food storage temperature safety. It also reduces the number of compressor start-ups and shutdowns and the mechanical wear caused by long-term continuous operation, thereby achieving synergistic optimization of energy efficiency, temperature control, and lifespan.

[0031] This invention provides an intelligent control system for commercial refrigerators that integrates genetic algorithms and xLSTM. Its core idea is to combine the accurate time-series prediction capability of xLSTM networks with the global optimization capability of genetic algorithms to construct an intelligent control architecture that operates in a rolling optimization manner, thereby achieving coordinated optimization of three objectives: energy consumption, temperature stability, and compressor lifespan.

[0032] The system mainly consists of four layers: a data acquisition layer, an edge computing layer, an execution control layer, and a cloud platform layer. The data acquisition layer is responsible for collecting the freezer's operating data and environmental data; the edge computing layer implements load prediction and optimized control functions; the execution control layer is responsible for issuing control commands to the refrigeration system for execution; and the cloud platform layer provides remote monitoring, model updates, and data analysis functions.

[0033] The data acquisition layer is used to collect real-time data such as the cabinet temperature, ambient temperature, door opening / closing status, and compressor operating status, providing basic information for subsequent prediction and optimization.

[0034] The edge computing layer is the core device of this system, comprising three core service modules. The data acquisition service writes preprocessed data to shared memory; the prediction service periodically reads current data and calls a trained xLSTM model to output the temperature change trend and load conditions for the next predicted time domain. This model learns the temporal patterns of temperature changes through forget gates, input gates, output gates, and exponential gating mechanisms, understanding the patterns of temperature changes with door opening / closing and ambient temperature from historical data; the optimization service executes a genetic algorithm based on the prediction results, encoding the control scheme into individuals containing set temperature, minimum running time, minimum downtime, and maximum allowable frequency. Through fitness function evaluation, selection, crossover, and mutation operations, iteratively searches for a set of optimal control parameters that strike the best balance between energy consumption, temperature control, and equipment lifespan. The energy consumption target is the total energy consumption over a future period, the temperature control target is the cumulative penalty for exceeding temperature limits, and the equipment lifespan target is the number of compressor start / stop cycles and the penalty for each running cycle.

[0035] The cloud platform is deployed in the data center, providing functions such as remote monitoring, model management, parameter configuration, and data analysis. The cloud platform and the edge controller are connected via TLS-encrypted MQTT protocol to ensure the confidentiality and integrity of data transmission.

[0036] The execution control layer includes a variable frequency compressor, which can adjust its operating power in real time according to control commands. The system operates in a rolling time-domain control manner. In each optimization cycle, after the edge computing layer obtains the optimal control parameters, the execution control layer executes according to these parameters, including adjusting the temperature setpoint, setting the compressor's start / stop threshold or operating frequency. When the next optimization cycle arrives, prediction and optimization are performed again, so that the control strategy dynamically adapts to environmental changes and load fluctuations. If an abnormal situation of a sharp temperature rise occurs between optimization cycles, the system immediately interrupts the optimization process, forcibly starts the compressor for rapid cooling, and prioritizes food safety.

[0037] Table 1 shows a comparison of energy consumption between the proposed solution and the traditional solution, comparing three control methods: traditional switching control, PID control, and the xLSTM+genetic algorithm proposed in this invention.

[0038] Table 1

[0039] As shown in Table 1, taking a 620L upright refrigerated display case as an example, under continuous power-on operation for 24 hours, the door opening and closing frequency is set according to business patterns: 30 times per hour during the three peak periods (morning, noon, and evening), and 5 times per hour during the remaining off-peak periods. The weights of the three optimization objectives are 0.5, 0.3, and 0.2, respectively, reflecting an emphasis on energy saving while also considering temperature control and equipment reliability. The comparison shows that the method of this invention is superior to other solutions in terms of energy consumption, temperature control, and compressor lifespan.

[0040] The above embodiments are merely exemplary embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art can make various modifications or equivalent substitutions to the present invention within its scope and spirit, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of the present invention.

Claims

1. A multi-objective energy consumption optimization method for refrigerated display cases based on the fusion of genetic algorithm and xLSTM, characterized in that, Includes the following steps: Step 1: Obtain the current temperature inside the freezer, ambient temperature, door open / close status, compressor status, and power, and preprocess the acquired data; Step 2: Input the current operating status and historical temperature information of the freezer into the xLSTM model. The forget gate determines how much historical information to retain, the input gate determines the importance of new information, the candidate memory extracts the new information at the current moment, the forget gate and the input gate are exponentialized and processed through the memory unit, and the output gate is combined with the memory unit to generate the final prediction output. Step 3: The energy consumption target, temperature control target, and equipment life target are integrated into a comprehensive objective function. Based on the prediction results of Step 2, under the conditions of satisfying the upper and lower temperature limits and the compressor start-stop count constraints, the optimal parameters are iterated through a genetic algorithm, and the equipment executes according to the optimal parameters. Step 4: Repeat steps 1-3 within each optimization cycle. If an abnormal situation of a sharp rise in temperature occurs between optimization cycles, the system will immediately interrupt the optimization process and forcibly start the compressor to cool down quickly.

2. The multi-objective energy consumption optimization method for refrigerated display cases based on the fusion of genetic algorithm and xLSTM as described in claim 1, characterized in that, Step 2 specifically involves: Step 2.1, passing through the forgetting gate To determine how much of the past memory to retain, among which It is the input at the current moment. It is a summary of memories from the previous moment. It is the Sigmoid function, which outputs a value between 0 and 1; Input gate and candidate memories They jointly decide how much new information to store in memory; Step 2.2: After obtaining the forget gate and input gate, perform indexing on both the forget gate and input gate. ; Step 2.3, the memory unit passes through Update, among which This indicates element-wise multiplication. This represents the system's long-term memory of temperature change patterns. Represents the Gate of Oblivion Indicates candidate memories, Indicates the input gate; Step 2.4, Output Gate Combined with memory units, through Generate the final prediction output. That is, the temperature prediction value several steps in the future. .

3. The multi-objective energy consumption optimization method for refrigerated display cases based on the fusion of genetic algorithm and xLSTM as described in claim 1, characterized in that, The specific objective function in step 3 is as follows: in , , It is an adjustable weighting coefficient. Energy consumption target; For temperature control objectives, For equipment lifespan targets.

4. The multi-objective energy consumption optimization method for refrigerated display cases based on the fusion of genetic algorithm and xLSTM as described in claim 3, characterized in that, The specific calculation formulas for the energy consumption target, temperature control target, and equipment lifespan target are as follows: in, It is the number of predicted steps. The power is calculated based on the compressor's predicted operating status. It is a time interval; It is the highest temperature permissible for food safety. Ensure that penalties are only incurred when the temperature exceeds the upper limit; This refers to the number of times the compressor will start and stop within a certain period of time. It is the penalty coefficient for each start-stop cycle. It is the single continuous run time. This is the maximum allowed runtime. This is the penalty coefficient for exceeding the time limit.

5. The multi-objective energy consumption optimization method for refrigerated display cases based on the fusion of genetic algorithm and xLSTM as described in claim 3, characterized in that, The genetic algorithm in step 3 specifically involves: generating an initial population containing multiple random individuals, with each control scheme representing one individual; evaluating the quality of each individual using a fitness function; randomly selecting two individuals from the population for comparison each time, with the individual with higher fitness advancing to the next generation, and the two individuals with the highest fitness in each generation being directly retained for the next generation; employing a simulated binary crossover method, independently calculating the expansion factor for each decision variable, enabling offspring to inherit features from their parents; using a multinomial mutation method, randomly perturbing the parameters of each individual with a fixed probability, with the variable length controlled by a distribution exponent; and finally outputting the optimal control parameters after multiple generations of iteration.

6. The multi-objective energy consumption optimization method for refrigerated display cases based on the fusion of genetic algorithm and xLSTM as described in claim 5, characterized in that, The fitness function is specifically: Where J represents the comprehensive objective function, To prevent extremely small positive numbers with a denominator of zero.

7. The multi-objective energy consumption optimization system for refrigerated display cases based on the fusion of genetic algorithm and xLSTM as described in claim 1, characterized in that, The method for executing the multi-objective energy consumption optimization method as described in any one of claims 1-6 includes a data acquisition layer, an edge computing layer, an execution control layer, and a cloud platform layer. The data acquisition layer is used to collect freezer operation data and environmental data. The edge computing layer outputs the temperature change trend and load status of a future prediction time domain through an xLSTM model based on the data collected by the data acquisition layer. Based on the prediction data, a genetic algorithm is executed to iteratively find the optimal trade-off control parameters between the energy consumption target, the temperature control target, and the equipment lifespan target. The energy consumption target is the total energy consumption over a future period, the temperature control target is the cumulative penalty for exceeding the temperature limit, and the equipment lifespan target is the number of compressor start-stop cycles and the penalty for single operation time. The execution control layer executes specific instructions based on the optimal control parameters solved by the edge computing layer; The cloud platform layer is deployed in the data center and is used for remote monitoring, model management, parameter configuration, and data analysis.

Images