Instant freezer operation parameter optimization method and system based on artificial intelligence

Abstract

The embodiment of the invention provides an instant freezer operation parameter optimization method and system based on artificial intelligence, and the method comprises the steps: firstly integrating instant freezer operation parameters, environmental factors and product feature information to generate a comprehensive database, extracting operation records and fault report text key information in a production process, and supplementing the operation records and fault report text key information to the database; then, core elements of the reinforcement learning optimization model are configured based on the comprehensive database, an instant freezer system is set as an intelligent agent, operation parameters are combined into an action space, environment and product information are combined into a state space, quick freezing effect indexes are reward function input parameters, and the intelligent agent executes parameter adjustment operation in the state space; updating strategy parameters through the reward function to generate an optimization strategy, collecting current environment and product information in real time to input the optimization strategy when the instant freezer operates, outputting a target operation parameter combination and applying the target operation parameter combination to a parameter control system, and achieving intelligent optimization of the operation parameters of the instant freezer.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and more specifically, to an artificial intelligence-based quick-freezing machine operating parameter optimization method and system. Background Art

[0002] In the production of frozen foods, quick-freezing machines are critical equipment. The proper setting of their operating parameters directly impacts quick-freezing results, product quality, and production efficiency. Currently, adjustment of quick-freezing machine operating parameters relies primarily on manual experience. Operators manually adjust parameters such as air speed, refrigerant flow rate, and conveyor speed based on past production data and a rough assessment of the current materials and environment. However, this method has numerous drawbacks. Firstly, manual experience is limited and cannot fully account for numerous complex factors, such as subtle changes in ambient temperature and humidity, and subtle differences between batches of materials. This results in inaccurate parameter adjustments and prevents the quick-freezing machine from consistently operating at optimal conditions. Secondly, manual adjustments often fail to respond promptly to sudden changes in the production environment or material characteristics, leading to unstable quick-freezing results, substandard product core temperatures, and degraded post-thaw quality, impacting production efficiency and product quality. Furthermore, manual parameter adjustment lacks systematicity and scientificity, making it difficult to analyze and summarize large amounts of production data, thus failing to provide an effective basis for subsequent parameter optimization. Summary of the Invention

[0003] In view of this, the purpose of this application is to provide a method and system for optimizing the operating parameters of a quick-freezing machine based on artificial intelligence.

[0004] According to a first aspect of the present application, a method for optimizing operating parameters of a quick-freezing machine based on artificial intelligence is provided, the method comprising: Integrate quick-freezing machine operating parameter information, environmental factor information, and product characteristic information to generate a comprehensive database, wherein the comprehensive database includes associated stored operating parameter records, environmental status records, and product characteristic records; Acquiring operation record text information and fault report text information during the production process, performing key information extraction processing on the operation record text information and the fault report text information to obtain text key information related to the quick-freezing machine operating parameters, and adding the text key information to the comprehensive database; Based on the comprehensive database, configuring core elements of a reinforcement learning optimization model, wherein the core elements include setting the quick-freezing machine system as an intelligent agent, setting the quick-freezing machine operating parameter combination as an action space, setting the combination of environmental factor information and product feature information as a state space, and setting the quick-freezing effect index as an input parameter of a reward function; Controlling the intelligent agent to perform parameter adjustment operations in the state space according to the action space, calculating a reward value corresponding to each parameter adjustment operation through the reward function, updating the policy parameters of the reinforcement learning optimization model according to the reward value, and generating a quick-freezer operating parameter optimization strategy; During the operation of the quick-freezer, the current environmental factor information and the current product characteristic information are collected in real time, the current environmental factor information and the current product characteristic information are input into the quick-freezer operation parameter optimization strategy, the target operation parameter combination is output, and the target operation parameter combination is applied to the parameter control system of the quick-freezer.

[0005] According to the second aspect of the present application, a quick-freeze machine operating parameter optimization system based on artificial intelligence is provided, and the quick-freeze machine operating parameter optimization system based on artificial intelligence includes a processor and a readable storage medium, and the readable storage medium stores a program, which, when executed by the processor, implements the aforementioned quick-freeze machine operating parameter optimization method based on artificial intelligence.

[0006] Based on any of the above aspects, the present application constructs a comprehensive database by integrating the quick-freezing machine operating parameter information, environmental factor information and product feature information, comprehensively covering all kinds of key data affecting the quick-freezing machine operation, extracting key information from the operation record text information and fault report text information in the production process and supplementing it to the comprehensive database, so that the database can reflect the operating status and problems of the quick-freezing machine under different circumstances, configuring the core elements of the reinforcement learning optimization model based on the comprehensive database, setting the quick-freezing machine system as an intelligent agent, setting the operating parameter combination as the action space, setting the combination of environmental factor information and product feature information as the state space, and setting the quick-freezing effect index as the input parameter of the reward function. A scientific and reasonable intelligent optimization framework has been constructed. The intelligent agent performs parameter adjustment operations in the state space, calculates the reward value through the reward function and updates the strategy parameters. It can autonomously learn and optimize the operation parameter adjustment strategy, so that the quick-freezing machine can automatically adjust the operation parameters according to different environmental conditions and product characteristics, and realize dynamic and accurate parameter optimization. During the operation of the quick-freezing machine, the current environmental factor information and product feature information are collected in real time and input into the optimization strategy. The target operation parameter combination is output and applied to the parameter control system to ensure that the quick-freezing machine always operates with the optimal parameters, effectively improving the quick-freezing effect and product quality stability, reducing energy consumption and production costs, and improving the company's production efficiency and competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 A schematic diagram of a process for optimizing the operating parameters of a quick-freezing machine based on artificial intelligence provided in an embodiment of the present application is shown; Figure 2A schematic diagram of the component structure of the quick-freezing machine operating parameter optimization system based on artificial intelligence provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0008] Figure 1 The following is a flow chart of an AI-based quick-freezer operating parameter optimization method provided in an embodiment of the present application. It should be understood that in other embodiments, the order of some steps in the AI-based quick-freezer operating parameter optimization method can be interchanged according to actual needs, or some steps can be omitted or deleted. The detailed steps of the AI-based quick-freezer operating parameter optimization method are described below.

[0009] Step S110: Integrate the quick-freezing machine operating parameter information, environmental factor information and product characteristic information to generate a comprehensive database, which includes associated stored operating parameter records, environmental status records and product characteristic records.

[0010] In the production of frozen foods, to effectively optimize the operating parameters of the freezer, the first step is to build a comprehensive database. This database integrates the freezer's operating parameters, environmental factors, and product characteristics, and stores records of operating parameters, environmental conditions, and product characteristics in an associative manner.

[0011] Step S111: collecting operating parameter information generated by the quick-freezing machine during historical operation, wherein the operating parameter information includes quick-freezing temperature adjustment parameters, conveyor belt operating speed parameters, cold air circulation frequency parameters and insulation layer pressure parameters.

[0012] During the long-term operation of the quick-freezing machine, various operation-related parameter information will be generated. These parameter information reflects the operating status of the quick-freezing machine under different production conditions.

[0013] Freezing temperature control parameters are a key factor influencing quick-freezing results. Different frozen products have different characteristics and therefore require different freezing temperatures. For example, products with high water content and fragile cell structures, such as fruits like strawberries, require a lower and more stable freezing temperature to prevent cell damage from ice crystal formation during the freezing process, thereby maintaining the product's taste and quality. On the other hand, for processed products with relatively low water content, such as quick-frozen dumplings, freezing temperature requirements may be more relaxed. During production, operators adjust the freezing temperature using the freezer's temperature control system based on the product's characteristics and production process requirements. Temperature sensors monitor the freezing chamber's temperature in real time and feed this data back to the control system for timely temperature adjustments. Data acquisition equipment records information such as the time of each temperature adjustment, the target temperature set, and the actual temperature achieved. This data can be used to analyze the specific freezing temperature requirements of different products and to evaluate the freezer's temperature control accuracy and stability.

[0014] The conveyor speed parameter directly affects the product's residence time in the freezer, which in turn affects the freezing quality and production efficiency. If the conveyor speed is too fast, the product's residence time in the freezer will be too short, potentially preventing adequate freezing. This can lead to uneven internal temperatures and poor product quality. Conversely, if the conveyor speed is too slow, while sufficient freezing can be achieved, it can reduce production efficiency and increase costs. In actual production, production managers will adjust the conveyor speed appropriately based on factors such as product type and size, freezing process, and production volume requirements. For example, larger products require longer freezing times due to their slower heat transfer rates, so the conveyor speed may be slower. Smaller products, on the other hand, can benefit from faster heat transfer rates and can be accelerated. Data acquisition equipment continuously records the conveyor speed for each production batch, as well as the timing and magnitude of speed adjustments. This data helps analyze the balance between freezing time and production volume during the production process.

[0015] The cold air circulation frequency parameter plays a key role in the distribution and circulation of cold air within the freezer. A reasonable cold air circulation frequency ensures even exposure to cold air across all product parts, resulting in uniform quick freezing. If the cold air circulation frequency is too low, the cold air distribution within the freezer may be uneven, resulting in poor quick freezing of some areas of the product, with some areas not fully frozen or thawed. On the other hand, if the cold air circulation frequency is too high, while ensuring even cold air distribution, it will increase energy consumption and wear on the equipment, and may also cause unnecessary impact on the product. During operation, the cold air circulation device operates at a preset frequency. The data acquisition system records the operating frequency of the cold air circulation device over different time periods in real time, as well as information related to frequency adjustments. By analyzing this data, the performance and operational effectiveness of the cold air circulation device can be evaluated, and the optimal cold air circulation frequency can be determined to improve quick freezing results and reduce energy consumption.

[0016] The insulation layer pressure parameters are closely related to the freezer's thermal insulation performance. Appropriate insulation layer pressure can reduce heat loss and improve energy efficiency. During freezer operation, the pressure within the insulation layer is affected by various factors, such as the freezer's startup and shutdown, and changes in ambient temperature and humidity. If the insulation layer pressure is unstable or does not meet requirements, it can lead to increased heat loss, requiring the freezer to consume more energy to maintain the set freezing temperature. Data acquisition equipment monitors pressure changes within the insulation layer in real time and records the relevant data. By analyzing this data, abnormal insulation layer pressure can be promptly detected and appropriate adjustments can be made, such as checking the insulation layer's sealing properties and adjusting the pressure regulating device, to ensure the freezer's thermal insulation performance and energy efficiency.

[0017] Step S112: collecting environmental factor information related to the operation of the quick-freezing machine, wherein the environmental factor information includes environmental temperature information, environmental humidity information, environmental pressure information and environmental dust concentration information.

[0018] In this embodiment, the ambient temperature directly affects the refrigeration load of the quick-freezing machine. In hot summer weather, the ambient temperature is high, and the quick-freezing machine needs to consume more energy to overcome external thermal interference and reduce the temperature in the quick-freezing chamber to the set quick-freezing temperature. This not only increases energy consumption, but may also affect the refrigeration efficiency and stability of the quick-freezing machine. In contrast, in cold winter weather, the ambient temperature is low, the refrigeration load of the quick-freezing machine is relatively small, and energy consumption is also reduced accordingly. By installing temperature sensors in the quick-freezing machine workshop, ambient temperature information can be collected in real time. The data acquisition equipment will record ambient temperature data for different time periods, including the highest temperature, lowest temperature, average temperature, etc. This data can be used to analyze the impact of ambient temperature on the operation of the quick-freezing machine. For example, in hot seasons, the operating time of the quick-freezing machine can be appropriately adjusted or the production load can be reduced to reduce energy consumption and equipment loss.

[0019] Furthermore, excessively high humidity can cause frost to form inside the freezer, impacting its proper operation and freezing performance. When humidity is high, moisture in the air condenses onto the cold freezer components, forming frost. Frost accumulation can affect the equipment's heat transfer performance, reducing cooling efficiency and potentially damaging components such as fans and piping. Furthermore, frost can impede air circulation within the freezer, hindering the circulation of cold air. To monitor humidity, humidity sensors are typically installed in the workshop. Data acquisition equipment records real-time humidity data and humidity trends. Analysis of this data can promptly identify abnormal humidity levels and enable appropriate measures, such as activating dehumidification equipment and increasing ventilation, to ensure the proper operation of the freezer.

[0020] Furthermore, ambient air pressure varies in different regions and under different weather conditions. These changes can affect the pressure balance within the freezer, impacting its performance. For example, at high altitudes, where ambient pressure is lower, the pressure within the freezer is relatively high. If the freezer's pressure control system fails to adjust in a timely manner, equipment failure or unstable operation may occur. To monitor ambient air pressure, air pressure sensors can be installed in the workshop. Data acquisition equipment records real-time ambient air pressure data and pressure fluctuations. By analyzing this data, the impact of ambient air pressure on the freezer's operation can be assessed, and the freezer's pressure control system can be adjusted as needed to ensure proper operation.

[0021] Regarding ambient dust concentration, if the dust concentration in the workshop is too high, it may clog the quick-freezing machine's vents and filters, reducing the efficiency of cold air circulation and affecting the quick-freezing effect. Furthermore, dust may enter the equipment, causing wear and damage to components and shortening its service life. To monitor ambient dust concentration, dust concentration sensors can be installed in the workshop. Data acquisition equipment regularly detects ambient dust concentration and records the relevant data. By analyzing this data, abnormal dust concentrations can be promptly identified and appropriate measures can be taken, such as strengthening workshop cleanliness and sanitation and installing air purification equipment, to ensure the normal operation of the quick-freezing machine and product quality.

[0022] Step S113: collecting product characteristic information of the product to be quick-frozen, wherein the product characteristic information includes product initial temperature information, product shape characteristic information, product water content information and product packaging material information.

[0023] In this embodiment, different product characteristics require different quick-freezing processes and parameter settings to ensure the quick-freezing effect and quality of the product.

[0024] The product's initial temperature directly impacts the quick-freezing time and energy consumption. Different products to be quick-frozen have different initial temperatures before entering the freezer. For example, the initial temperatures of products freshly transported from a room temperature environment and those that have undergone initial pre-cooling can differ significantly. If the product's initial temperature is higher, the temperature must be lowered more significantly during the quick-freezing process, which will increase the freezing time and energy consumption. Conversely, if the product has undergone initial pre-cooling and has a lower initial temperature, the freezing time and energy consumption can be reduced. Temperature detection equipment measures and records the product's initial temperature before it enters the freezer. This data can help determine appropriate quick-freezing parameters, such as freezing temperature and time. For example, if the product's initial temperature is higher, the freezing temperature can be appropriately lowered or the freezing time extended to ensure optimal quick-freezing.

[0025] Product shape characteristics significantly influence quick-freezing performance. Heat transfer during the quick-freezing process varies depending on the product's shape. Regularly shaped products, such as cubes and cylinders, transfer heat relatively evenly because their contact area with the cold air and the heat transfer path are relatively consistent across all parts. However, complexly shaped products, such as those with protrusions, depressions, or irregular shapes, may have heat transfer blind spots, resulting in poor quick-freezing performance in some areas. For example, a product with protrusions may experience slower heat transfer than other parts, making it prone to partial under-freezing. Image recognition technology or 3D scanning equipment can be used to capture product shape characteristics. These technologies can accurately capture product shape data and convert it into digital information for recording. By analyzing this data, appropriate measures can be taken during the subsequent quick-freezing process to improve heat transfer, such as adjusting the cold air circulation direction or increasing the quick-freezing time.

[0026] Product moisture content is directly related to the speed and quality of quick freezing. Products with high moisture content are more likely to form ice crystals during the quick freezing process, and the size and distribution of these ice crystals can affect the taste and quality of the product. Excessively large ice crystals can damage the product's cellular structure, leading to problems such as juice loss and a poor taste after thawing. To ensure product quality, quick freezing parameters need to be adjusted based on the product's moisture content. For example, for products with a high moisture content, the quick freezing temperature can be appropriately lowered or the quick freezing time increased to reduce the size of the ice crystals and improve the product's quick freezing quality. Moisture testing equipment can measure and record the product's moisture content.

[0027] Product packaging material information also affects the quick-freezing process. Different packaging materials have different thermal insulation and air permeability properties. For example, plastic packaging has good thermal insulation properties, which can reduce heat transfer and help maintain product temperature. Paper packaging, on the other hand, is relatively breathable but has poor thermal insulation properties. The choice of product packaging material affects the speed of heat exchange between the product and the external environment, which in turn affects the quick-freezing effect. When collecting product packaging material information, it is necessary to record parameters such as packaging material type, thickness, and density. By analyzing this data, we can understand the impact of different packaging materials on the quick-freezing process and adjust the quick-freezing parameters accordingly. For example, for packaging materials with good thermal insulation properties, the quick-freezing temperature can be appropriately increased or the conveyor speed can be increased to improve production efficiency.

[0028] Step S114: adding the same timestamp mark to the operating parameter information, the environmental factor information and the product feature information to establish an association relationship in the time dimension.

[0029] After collecting freezer operating parameters, environmental factors, and product characteristics, to accurately analyze the relationships between these information, it's necessary to add a timestamp to each set of information. A timestamp precisely records the time of data collection, allowing different types of information to be linked together at the same point in time or within a specific timeframe.

[0030] For example, when recording the quick-freezing temperature adjustment parameters at a certain moment, the ambient temperature at that moment, the initial product temperature, and other information are also associated. Therefore, in the subsequent data analysis and model building process, the interaction and influence of these factors can be studied based on the time dimension. For example, the impact of changes in quick-freezing temperature adjustment parameters on the quick-freezing effect of the product under different ambient temperatures can be analyzed; or the relationship between the initial product temperature and the conveyor belt speed can be studied to optimize the parameter settings in the production process. By establishing a correlation relationship in the time dimension, the dynamic changes of various factors during the operation of the quick-freezing machine can be more comprehensive and accurate.

[0031] Step S115: Based on the association relationship of the timestamp marks, the operating parameter information is stored as an operating parameter record, the environmental factor information is stored as an environmental status record, and the product feature information is stored as a product characteristic record. The operating parameter records, the environmental status records and the product characteristic records are integrated to generate a comprehensive database.

[0032] The collected information is classified, stored and integrated according to the association relationship established by the timestamp mark.

[0033] Operating parameter information is organized and stored as operating parameter records. Each operating parameter record contains information such as quick-freeze temperature adjustment parameters, conveyor belt speed parameters, cold air circulation frequency parameters, and insulation layer pressure parameters, and is associated with a corresponding timestamp. Thus, each record represents the operating status of the quick-freeze machine at a specific moment. By analyzing operating parameter records, we can understand the quick-freeze machine's operating patterns and performance changes over different time periods.

[0034] Similarly, environmental factor information is organized into environmental status records. These records include information such as ambient temperature, humidity, air pressure, and dust concentration. Each record is also accurately timestamped. These records reflect changes in the freezer's operating environment over time, helping to analyze the impact of environmental factors on freezer operation.

[0035] Product characteristic information is stored as a product characteristic record. This record includes information such as the product's initial temperature, shape, moisture content, and packaging material, all of which are mapped to corresponding timestamps. These records can help us understand the characteristics of different products as they enter the freezer.

[0036] Finally, records of operating parameters, environmental conditions, and product characteristics are integrated to construct a comprehensive database. This database stores key information about the freezer's operation using time as a chronological link. Within the database, relevant information can be quickly retrieved and queried using timestamps, facilitating data analysis and model building. For example, all records within a specific time period can be filtered based on the time range to analyze the freezer's operating conditions, changes in environmental factors, and the distribution of product characteristics within that period, providing more targeted recommendations for optimizing freezer operating parameters.

[0037] Step S120: Obtain the operation record text information and fault report text information in the production process, perform key information extraction processing on the operation record text information and the fault report text information, obtain text key information related to the quick-freezing machine operating parameters, and add the text key information to the comprehensive database.

[0038] In the frozen food production process, operation logs and fault reports contain a wealth of knowledge and experience about freezer operation. By extracting key information from these texts, we can obtain important information related to freezer operating parameters and add this information to the comprehensive database, further improving its content.

[0039] Step S121: Obtain the operation record text information and fault report text information in the production process, perform text standardization on the operation record text information and the fault report text information, perform paragraph division on the standardized operation record text information and the fault report text information, divide the operation record text information into multiple operation description paragraphs, and divide the fault report text information into multiple fault description paragraphs.

[0040] During the production process, operators will record daily operations in detail, forming textual records. When a quick-freezing machine malfunctions, a textual report will be generated. The format and content of these textual records may vary, so to facilitate subsequent information extraction, text standardization is required.

[0041] Text standardization involves unifying text formats, encodings, and language standards. For example, all text is converted to the same character encoding, and special symbols, irregular spaces, and line breaks are removed to make the text more standardized and consistent. Furthermore, the language used in the text is standardized, and the use of professional terminology is standardized to avoid the mixing of synonyms and near-synonyms. Text standardization improves the quality and readability of text information, laying the foundation for subsequent processing.

[0042] After standardization, the operation log text and fault report text are segmented. The operation log text is divided into multiple sections based on the different stages and content of the operation. For example, separate sections are created for startup preparation, parameter adjustment during operation, and shutdown. Each section details a complete operation flow, including the time, steps, equipment involved, and parameters. This segmentation helps clearly demonstrate the operation process and facilitates subsequent information extraction.

[0043] The fault report text is divided into sections based on the fault occurrence time, description of the phenomenon, and preliminary investigation results, forming multiple fault description sections. Each section records detailed information about the fault, such as the specific time of occurrence, symptoms (such as abnormal equipment sounds and temperature), preliminary investigation results, and emergency measures taken. This sectioning makes the fault report more organized and facilitates analysis and processing of the fault information.

[0044] Step S122: performing keyword recognition processing on the operation description paragraph and the fault description paragraph, and extracting a set of candidate keywords containing the names of quick-freezing machine operating parameters.

[0045] After obtaining the operation description paragraphs and the fault description paragraphs, it is necessary to perform keyword recognition processing on these paragraphs to extract words related to the names of the quick-freezing machine operating parameters. These words will be used as candidate keywords.

[0046] The keyword extraction algorithm in natural language processing technology is used to analyze the vocabulary in each paragraph. The keyword extraction algorithm will screen out representative vocabulary based on factors such as the frequency, importance and context of the vocabulary. In this process, focus on those words related to the names of the quick-freezing machine operating parameters, such as "quick-freezing temperature", "conveyor belt speed", "cold air circulation frequency", "insulation layer pressure", etc. These words may appear in different forms in the text, such as full names, abbreviations or related descriptive words. For example, "quick-freezing temperature" may be described as "quick-freezing chamber temperature", "set quick-freezing temperature", etc. Through keyword recognition processing, these words related to the names of the quick-freezing machine operating parameters are extracted to form a set of candidate keywords.

[0047] Each word in the candidate keyword set may contain important information related to the quick-freezing machine operating parameters. Through further analysis and screening of these candidate keywords, the target keywords that are truly related to the quick-freezing machine operating parameter adjustment behavior can be found.

[0048] Step S123: performing contextual semantic association analysis on the candidate keyword set, determining the semantic pointing object of each candidate keyword in the corresponding paragraph, and screening out target keywords pointing to the quick-freezing machine operating parameter adjustment behavior.

[0049] Step S1231: Sentence segmentation is performed on the operation description paragraph or fault description paragraph where each candidate keyword is located to obtain a plurality of semantically complete sentence units.

[0050] To accurately analyze the semantics of candidate keywords, we first need to segment the paragraph containing each candidate keyword into sentences. Based on the grammatical structure and semantic integrity of the sentences, we break the paragraph into multiple independent sentence units. Each sentence unit expresses relatively complete semantic information, potentially including an operation step, a description of the fault phenomenon, or a related explanation.

[0051] For example, a paragraph describing an operation may contain multiple sentences, such as "At 10:00 AM, the quick-freeze temperature was adjusted to -30°C." "After adjustment, the temperature was observed for 10 minutes and stabilized at -29°C." Sentence segmentation allows these two sentences to be processed as independent units. This segmentation helps to more carefully analyze the role and semantics of each candidate keyword in a specific sentence, avoiding semantic confusion caused by overly long paragraphs.

[0052] Step S1232: performing dependency syntactic analysis on each sentence unit, identifying the subject-verb-object structure and the adverbial-predicate structure in the sentence unit, and determining the grammatical role of the candidate keyword in the sentence unit.

[0053] After obtaining sentence units, we conduct an in-depth analysis of each sentence using dependency parsing techniques. Dependency parsing can identify grammatical relationships within a sentence, such as subject-verb-object structure and adverbial-verb structure. The subject-verb-object structure clearly defines the main action performer (subject), the action (predicate), and the action object (object); the adverbial-verb structure describes adverbial information such as the time, place, and manner of the action.

[0054] Dependency analysis is used to determine the grammatical role of candidate keywords within a sentence. For example, in the sentence "adjust the quick-freeze temperature to -30°C," "quick-freeze temperature" is the object of the action (the verb), and "adjust" is the action (the verb). Understanding the grammatical role of candidate keywords helps us understand their logical relationships with other words.

[0055] Step S1233: determining whether the candidate keyword is directly associated with an action verb based on the grammatical role, where the action verbs include adjust, modify, set, increase, and decrease.

[0056] Based on the candidate keyword's grammatical role in the sentence, determine whether it has a direct relationship with a specific action verb. These action verbs, such as "adjust," "modify," "set," "increase," and "lower," usually indicate the act of manipulating the operating parameters of the quick-freezing machine.

[0057] If a candidate keyword is directly related to these action verbs, it suggests it may be related to adjusting the operating parameters of a quick-freezer. For example, in the sentence "Adjust the conveyor belt speed to 2 meters per minute," the candidate keyword "conveyor belt speed" is directly associated with the action verb "adjust," indicating that this keyword may involve adjusting the operating parameters of a quick-freezer. This analysis allows us to initially identify candidate keywords that may be related to the act of adjusting operating parameters.

[0058] Step S1234: For candidate keywords with direct association, extract their corresponding action verbs and action objects, and determine whether the action object is a quick-freezing machine operating parameter.

[0059] For candidate keywords that are directly related to action verbs, we further extract their corresponding action verbs and action objects. For example, in the sentence "Set the insulation layer pressure to 1.2 MPa", "set" is the action verb and "insulation layer pressure" is the action object.

[0060] Next, we determine whether the action target is a freezer operating parameter. If so, such as freezing temperature, conveyor speed, cold air circulation frequency, insulation layer pressure, etc., then the candidate keyword refers to the behavior of adjusting freezer operating parameters. This determination can more accurately filter out keywords truly related to freezer operating parameter adjustments.

[0061] Step S1235: determining the candidate keyword whose action object is the quick-freezing machine operating parameter as the target keyword pointing to the quick-freezing machine operating parameter adjustment behavior.

[0062] After the above analysis and judgment, the candidate keywords whose action objects are the operating parameters of the quick-freezing machine are determined as target keywords. These target keywords accurately reflect the relevant information on adjusting the operating parameters of the quick-freezing machine during the production process.

[0063] For example, among a series of candidate keywords, we screened and identified keywords such as "quick freezing temperature" and "conveyor belt speed" that are directly associated with action verbs and whose action targets are quick-freezing machine operating parameters as target keywords. These target keywords will serve as the focus of subsequent information extraction to obtain detailed information related to quick-freezing machine operating parameter adjustments.

[0064] Step S124: Based on the target keyword and its contextual semantic pointing object, extract the adjustment condition information, adjustment range information and adjustment result information related to the quick-freezing machine operating parameters, and combine the adjustment condition information, the adjustment range information and the adjustment result information into text key information.

[0065] After determining the target keyword, combined with its contextual semantic pointing object, the adjustment condition information, adjustment range information and adjustment result information related to the quick-freezing machine operating parameters are further extracted.

[0066] Adjustment condition information typically describes why the quick-freezing machine's operating parameters need to be adjusted. For example, this could be due to a change in product type, environmental factors, or specific equipment requirements. The operation log text or fault report text will include a description explaining the reason for the parameter adjustment. For example, "Since the production product has changed from strawberries to blueberries, the quick-freezing temperature has been adjusted from -25°C to -30°C." Here, "the production product has changed from strawberries to blueberries" is the condition information for adjusting the quick-freezing temperature.

[0067] Adjustment amplitude information describes the specific extent of the adjustment to the operating parameters. For example, "Reducing the conveyor belt speed from 2 m / min to 1.5 m / min" or "Reducing from 2 m / min to 1.5 m / min" represents the adjustment amplitude for the conveyor belt speed. By extracting this adjustment amplitude information, you can understand the specific details of the parameter adjustment.

[0068] Adjustment result information reflects the effect of parameter adjustments. For example, "After adjusting the quick-freezing temperature, the product qualification rate increased from 80% to 90%" or "The product qualification rate increased from 80% to 90%" are the result information of adjusting the quick-freezing temperature. Adjustment result information can help evaluate the effectiveness of parameter adjustments.

[0069] The extracted adjustment condition information, adjustment range information, and adjustment result information are combined to form text key information. This text key information supplements the detailed information on quick-freezer operating parameter adjustments in the comprehensive database. Finally, this text key information is added to the previously constructed comprehensive database to further improve the database.

[0070] Step S130: Based on the comprehensive database, the core elements of the reinforcement learning optimization model are configured, wherein the core elements include setting the quick-freezing machine system as an intelligent agent, setting the quick-freezing machine operating parameter combination as an action space, setting the combination of environmental factor information and product feature information as a state space, and setting the quick-freezing effect index as an input parameter of the reward function.

[0071] To optimize the freezing machine's operating parameters, a reinforcement learning optimization model was used. This model continuously learns and adjusts its strategy through the interaction between the agent and the environment to achieve optimal performance. When configuring a reinforcement learning optimization model, its core elements must be determined, including the agent, action space, state space, and reward function.

[0072] Step S131: Based on the operating parameter records in the comprehensive database, all historical quick-freezing machine operating parameter combinations are extracted, and the set of quick-freezing machine operating parameter combinations is defined as the action space of the reinforcement learning optimization model.

[0073] From the operating parameter records in the comprehensive database, we screened all historical quick-freeze operating parameter combinations. Each combination includes specific values ​​for multiple parameters, including quick-freeze temperature adjustment, conveyor speed, cold air circulation frequency, and insulation layer pressure. These combinations are set during actual production based on different products and environmental conditions, reflecting various possible operating conditions.

[0074] These different operating parameter combinations are organized and collected into a set. Each element in this set represents a possible configuration of the quick-freezing machine's operating parameters. In the reinforcement learning optimization model, this set is defined as an action space. The agent (i.e., the quick-freezing machine system) can select the appropriate operating parameter combination from this action space to apply to the quick-freezing process. The definition of the action space provides the agent with multiple options, allowing it to try different operating parameter configurations under different circumstances to find the optimal solution.

[0075] Step S132: Based on the environmental status records and product characteristic records in the comprehensive database, the combined data of environmental factor information and product feature information is defined as the state space of the reinforcement learning optimization model, and each state in the state space contains a set of environmental factor information and a corresponding set of product feature information.

[0076] Based on the environmental status records and product characteristic records in the comprehensive database, environmental factor information and product characteristic information are combined. Environmental factor information includes ambient temperature, humidity, air pressure, and dust concentration; product characteristic information includes initial product temperature, shape, moisture content, and packaging material information.

[0077] Each set of environmental factor information is combined with the corresponding set of product feature information to form a state. Each state represents the environmental and product conditions faced by the freezer at a specific moment. The collection of all these states constitutes the state space of the reinforcement learning optimization model.

[0078] In different production scenarios, quick-freezing machines face different environmental and product characteristics. Each state in the state space represents a specific production situation. The agent can select an appropriate action (i.e., a combination of operating parameters) based on the current state. For example, when the ambient temperature is high and the initial product temperature is also high, the agent needs to select an operating parameter combination from the action space that effectively addresses these conditions to ensure quick-freezing results and production efficiency.

[0079] Step S133: The quick-freezing machine system is defined as an intelligent agent of the reinforcement learning optimization model, and the intelligent agent is capable of selecting a quick-freezing machine operating parameter combination from the action space and acting on the quick-freezing environment.

[0080] In the reinforcement learning optimization model, the quick-freeze system is defined as an intelligent agent. The agent has the ability to make autonomous decisions. It can perceive its current state (i.e., a combination of environmental factors and product characteristics) and select the appropriate quick-freeze operating parameter combination from the action space based on the set strategy.

[0081] Once the agent selects a set of operating parameters, it can apply them to the freezing environment. This means the freezer system will adjust its operating state based on the selected parameter combination, such as the freezing temperature, conveyor speed, cold air circulation frequency, and insulation pressure. In this way, the agent can influence the freezing process and, in turn, the freezing effect.

[0082] For example, based on information such as the current ambient temperature and the product's initial temperature, the agent determines that the freezing temperature needs to be lowered and the conveyor speed needs to be increased. It then selects a corresponding combination of operating parameters from the action space and applies them to the freezer. The freezer system then adjusts based on these parameters to achieve optimal freezing results.

[0083] Step S134: extracting evaluation data related to quick-freezing effect from the comprehensive database, and determining the evaluation data as quick-freezing effect indicators, wherein the quick-freezing effect indicators include product qualification rate indicators, production rhythm indicators and energy consumption indicators.

[0084] Evaluation data related to quick-freezing performance was extracted from a comprehensive database. This data reflects the actual operating performance of the quick-freezing machine under different operating parameter configurations and is of great significance for evaluating the decision-making quality of the intelligent agent and optimizing operating parameters.

[0085] The product qualification rate reflects the proportion of qualified products after quick freezing. Qualified products are those that meet quality standards, such as intact appearance, good taste, and well-preserved nutrients. The product qualification rate is a key indicator of quick freezing quality. A high qualification rate indicates that the quick freezer's operating parameters are effectively ensuring product quality; otherwise, adjustments may be necessary.

[0086] The cycle time indicator represents the number of products that can be quickly frozen in a given timeframe, reflecting production efficiency. In practice, companies typically aim to maximize cycle time while ensuring product quality, thereby increasing output and reducing costs. The cycle time indicator reflects the production capacity of a quick-freezing machine under different operating parameters.

[0087] Energy consumption indicators record the amount of energy consumed by a freezer during operation. Energy consumption is a significant component of a company's production costs, and reducing it can improve its economic benefits. By analyzing energy consumption indicators, we can understand the freezer's energy efficiency under different operating parameters and identify ways to reduce energy consumption.

[0088] These evaluation data are determined as quick-freezing effect indicators, which will be used as input parameters of the reward function to evaluate the pros and cons of the operating parameter combination selected by the intelligent agent.

[0089] Step S135: constructing a reward function based on the quick-freezing effect index, wherein the input parameter of the reward function is the actual measurement value of the quick-freezing effect index, and the output is a reward value corresponding to the actual measurement value.

[0090] Step S1351: setting weight coefficients for the product qualification rate index, production rhythm index and energy consumption index in the quick-freezing effect index respectively, and the weight coefficients are determined based on the priority requirements of quick-freezing production.

[0091] Based on the specific priorities of quick-frozen production, weight coefficients are set for product qualification rate, production cycle time, and energy consumption. Different companies or production scenarios may place different emphasis on these indicators.

[0092] If a company prioritizes product quality, the weighting factor for the product qualification rate indicator may be relatively high. This is because high-quality products enhance a company's market competitiveness and earn consumer trust. For example, for a company producing high-end frozen foods, product qualification rate is a primary consideration, and the weighting factor for this indicator may be set higher.

[0093] Companies pursuing production efficiency will assign a higher weight to the production cycle indicator. Improving the production cycle can increase output, meet market demand, and reduce unit production costs. For example, in some large-scale frozen food companies, production efficiency is key, so they place greater emphasis on the production cycle indicator.

[0094] If a company is considering reducing production costs, the weighting of energy consumption indicators will increase accordingly. Reducing energy consumption can reduce energy costs and improve the company's economic benefits. For example, in regions with high energy prices or when companies are facing significant cost pressures, they will pay more attention to energy consumption indicators.

[0095] These weight coefficients reflect the importance of different indicators in the overall evaluation. By reasonably setting the weight coefficients, the reward function can be made more in line with the actual needs of the enterprise.

[0096] Step S1352: normalizing the actual measured value of the product qualification rate indicator to obtain a standardized value of the product qualification rate, wherein the standardized value of the product qualification rate is positively correlated with the actual measured value of the product qualification rate indicator.

[0097] In order to make different quick-freezing effect indicators comparable, it is necessary to standardize the actual measured values ​​of the product qualification rate indicator. The purpose of standardization is to map the actual measured values ​​of the product qualification rate indicator to a unified range and eliminate the dimensional differences between different indicators.

[0098] Through standardization, a standardized value for the product pass rate is obtained. This value is positively correlated with the actual measured value of the product pass rate indicator; that is, the higher the actual measured value, the higher the standardized value. Standardization can be performed using a variety of methods, such as linear transformation and normalization. For example, the actual measured value of the product pass rate can be mapped to a range of 0 to 1, where 0 represents the lowest pass rate and 1 represents the highest pass rate. This allows the pass rates of products from different batches or under different operating parameters to be compared on the same scale.

[0099] Step S1353: normalizing the actual measured value of the production tact index to obtain a normalized production tact value, wherein the normalized production tact value is positively correlated with the actual measured value of the production tact index.

[0100] Similarly, the actual measured values ​​of the production cycle indicators are standardized. The actual measured values ​​of the production cycle indicators may vary due to factors such as production scale and equipment performance. In order to facilitate comparison and analysis, they need to be standardized.

[0101] After normalization, the normalized takt value is obtained. This value is positively correlated with the actual measured value of the takt indicator, meaning that the faster the takt, the higher the normalized takt value. Normalization makes takt data under different production conditions comparable, helping to accurately evaluate the agent's performance in improving production efficiency.

[0102] Step S1354: normalizing the actual measured value of the energy consumption index to obtain a normalized energy consumption value, wherein the normalized energy consumption value is negatively correlated with the actual measured value of the energy consumption index.

[0103] Energy consumption indicators also require standardization. Since lower energy consumption is better, the standardized energy consumption value is negatively correlated with the actual measured value of the energy consumption indicator. That is, the lower the actual measured value of the energy consumption indicator, the higher the standardized energy consumption value.

[0104] Through standardization, energy consumption data under different operating parameters can be uniformly compared. For example, the actual measured value of the energy consumption indicator can be mapped to a range of 0 to 1, where 0 represents the highest energy consumption and 1 represents the lowest energy consumption. This allows the intelligent agent to more intuitively understand the energy consumption under different operating parameter configurations and choose the most energy-efficient solution.

[0105] Step S1355: Multiply the standardized value of the product qualification rate, the standardized value of the production rhythm, and the standardized value of the energy consumption by the corresponding weight coefficients respectively and then sum them up to obtain a reward value corresponding to the actual measured value of the quick-freezing effect index.

[0106] Multiply the standardized values ​​for product pass rate, production cycle time, and energy consumption by the previously set weight coefficients. This step takes into account the importance of different indicators in the overall evaluation. For example, if the weight coefficient for product pass rate is 0.5 and the standardized value is 0.8, the result after multiplication is 0.4, which represents the contribution of product pass rate to the reward value.

[0107] These products are then summed to produce a reward corresponding to the actual measured value of the quick-freezing performance indicator. This reward, which takes into account factors such as product quality, production efficiency, and energy consumption, is used to evaluate the agent's performance in selecting operating parameter combinations. The agent's goal is to maximize this reward through continuous learning and selecting appropriate actions.

[0108] Step S140: Control the intelligent agent to perform parameter adjustment operations in the state space according to the action space, calculate the reward value corresponding to each parameter adjustment operation through the reward function, update the strategy parameters of the reinforcement learning optimization model according to the reward value, and generate a quick freezer operating parameter optimization strategy.

[0109] After configuring the core elements of the reinforcement learning optimization model, the agent begins learning and making decisions in the state space. By continuously performing parameter adjustments, calculating rewards, and updating policy parameters, a policy for optimizing the freezer's operating parameters is gradually generated.

[0110] Step S141: randomly selecting a set of environmental factor information and a corresponding set of product feature information from the comprehensive database as an initial state, and inputting the initial state into the intelligent agent.

[0111] At the beginning of reinforcement learning, the agent needs to be given an initial state. A set of environmental factors and a corresponding set of product characteristics are randomly selected from a comprehensive database and combined into a state. This state represents the environmental and product conditions facing the freezer at a specific moment.

[0112] This initial state is input into the freezer system, which serves as an intelligent agent. After receiving this initial state, the agent can select an appropriate combination of operating parameters from the action space based on its current policy parameters. The random selection of the initial state helps the agent begin learning in different scenarios, increasing the diversity and comprehensiveness of its learning.

[0113] Step S142: Control the intelligent agent to select an initial quick-freezing machine operating parameter combination from the action space as an initial action, and execute a parameter adjustment operation corresponding to the initial action.

[0114] After receiving the initial state, the agent can select an initial combination of quick-freeze operating parameters from the action space based on its own policy parameters as its initial action. Policy parameters are parameters that the agent continuously adjusts and optimizes during the learning process, and they determine how the agent makes decisions in different states.

[0115] Once the initial action is selected, the freezer system executes the parameter adjustment operations corresponding to that action. This means that the freezer system adjusts its operating state based on the selected parameter combination, such as the freezing temperature, conveyor speed, cold air circulation frequency, and insulation pressure. By executing these parameter adjustment operations, the agent begins to influence the freezing process.

[0116] Step S143: collecting actual measurement values ​​of the quick-freezing effect index of the quick-freezing machine after executing the initial action, inputting the actual measurement values ​​into the reward function, and calculating the initial reward value.

[0117] After a period of initial operation, the actual measured values ​​of the quick-freezing machine's quick-freezing performance indicators need to be collected. These indicators include product qualification rate, production cycle time, and energy consumption. Through actual measurement, the specific values ​​of these indicators under the current operating parameter configuration can be obtained.

[0118] These actual measurements are fed into the previously constructed reward function. The reward function calculates the reward value corresponding to these measurements based on the input actual measurements, combined with the previously set weight coefficients and normalization results. This reward value reflects the agent's performance under the initial action.

[0119] Step S144: Based on the initial state, the initial action, and the initial reward value, update the strategy parameters of the agent according to the reinforcement learning algorithm.

[0120] Step S1441: Based on the initial state and the initial action, the next state is determined by the state transition function of the reinforcement learning algorithm, and the next state is a combination of the environmental factor information and product feature information of the quick freezer after the initial action is performed.

[0121] Based on the initial state and initial action, the state transition function in the reinforcement learning algorithm is used to determine the next state. The state transition function describes the next state the system will transition to after performing a certain action in the current state.

[0122] In the freezer scenario, after the initial action is executed, the freezer's operating state will change, and environmental factors and product characteristics may also be affected. The state transition function calculates the next state based on these changes. The next state is also a combination of environmental factor information and corresponding product characteristics, representing the new environmental and product conditions of the freezer after the initial action is executed.

[0123] Step S1442: The initial state, the initial action, the initial reward value, and the next state are combined into an experience sample, and stored in an experience replay buffer.

[0124] The initial state, initial action, initial reward value, and next state are combined to form an experience sample. The experience sample records the complete information of the agent during a decision-making process, including the state it is in, the action taken, the reward obtained, and the next state it moves to.

[0125] The experience sample is stored in the experience replay buffer. The experience replay buffer is a container for storing experience samples, which can store multiple experience samples. The experience replay buffer can be used to randomly sample experience samples, avoiding correlation between samples and improving learning stability and efficiency.

[0126] Step S1443: When the number of experience samples in the experience replay buffer reaches a preset threshold, batches of experience samples are randomly extracted from the experience replay buffer.

[0127] The experience replay buffer continuously accumulates experience samples. When the number of experience samples reaches a preset threshold, it indicates that there is enough data for effective policy updates. At this point, a batch of experience samples is randomly drawn from the experience replay buffer.

[0128] The purpose of random sampling is to avoid correlation between samples. Consecutive learning with similar experience samples can cause the agent to become stuck in a local optimum and be unable to find the global optimal strategy. Random sampling allows the agent to learn from different experience samples, increasing learning diversity and generalization capabilities.

[0129] Step S1444: For each experience sample in the batch of experience samples, a target Q value is calculated based on the current policy parameters, where the target Q value is a weighted sum of the current reward value and the future reward value.

[0130] For each experience sample in the batch of extracted experience samples, the target Q value is calculated based on the agent's current policy parameters. The target Q value is an important indicator that represents the long-term cumulative reward that can be obtained by taking a certain action in a certain state.

[0131] The target Q-value is composed of a weighted sum of the current reward and the future reward. The current reward is the reward received immediately after performing an action, reflecting the immediate effect of the current action. The future reward takes into account the potential rewards from subsequent states and actions, reflecting the impact of the current action on the future. By weighting the future reward, we can balance the impact of short-term and long-term rewards.

[0132] For example, in an experience sample, the current reward value is 10, the future reward value is predicted to be 20 based on the current policy parameters, and the weighting coefficient is 0.8, then the target Q value is 26.

[0133] Step S1445: Calculate the difference between the target Q value and the current Q value using a loss function, where the current Q value is a Q value calculated based on the current strategy parameters for the experience sample.

[0134] After calculating the target Q value, the loss function is used to calculate the difference between the target Q value and the current Q value. The current Q value is the Q value calculated based on the agent's current policy parameters and the experience sample.

[0135] The loss function measures the difference between the target Q-value and the current Q-value. This difference reflects the gap between the agent's current strategy and the optimal strategy. Common loss functions include the mean squared error (MSE) loss function. For example, the MSE loss function calculates the difference between the target Q-value and the current Q-value. The smaller the difference, the closer the agent's current strategy is to the optimal strategy.

[0136] Step S1446: Update the strategy parameters of the agent through the back propagation algorithm according to the difference, and reduce the difference between the target Q value and the current Q value.

[0137] Based on the calculated difference, the backpropagation algorithm is used to update the agent's policy parameters. The backpropagation algorithm is a commonly used optimization algorithm that gradually reduces the value of the loss function by adjusting the weight parameters in the neural network.

[0138] During this process, the weight parameters in the neural network can be adjusted based on the size and direction of the difference. The goal of this adjustment is to gradually bring the current Q value closer to the target Q value, thereby optimizing the agent's decision-making strategy. By continuously updating the strategy parameters, the agent can learn more optimal decision-making methods and improve its ability to select the optimal action in different states.

[0139] Step S145: Repeat the process of selecting a new state from the state space, the agent selecting a new action, calculating a new reward value, and updating the strategy parameters until the strategy parameters converge, and generate a quick-freezer operating parameter optimization strategy.

[0140] After completing a policy parameter update, the process of selecting a new state from the state space, selecting a new action, calculating a new reward, and updating the policy parameters is repeated. Each cycle allows the agent to make decisions in a different state and continuously optimize the policy parameters based on the reward value obtained.

[0141] As the number of cycles increases, the policy parameters gradually converge. Convergence means that the agent's decision-making strategy becomes increasingly stable and optimized, choosing similar actions in the same state and achieving stable rewards. When the policy parameters converge to a certain level, a relatively optimal decision-making strategy has been found, which is defined as the freezer operating parameter optimization strategy.

[0142] This optimization strategy can guide the quick-freezing machine to select the most appropriate operating parameter combination in different production scenarios, thereby improving the quick-freezing effect, production efficiency and reducing energy consumption.

[0143] Step S150: During the operation of the quick freezer, current environmental factor information and current product feature information are collected in real time, the current environmental factor information and the current product feature information are input into the quick freezer operation parameter optimization strategy, the target operation parameter combination is output, and the target operation parameter combination is applied to the parameter control system of the quick freezer.

[0144] After generating the quick-freezing machine operating parameter optimization strategy, it is necessary to apply the strategy during the actual quick-freezing machine operation process to achieve real-time optimization of the operating parameters.

[0145] Step S151: During the operation of the quick-freezing machine, the current environmental temperature information, the current environmental humidity information, the current environmental air pressure information and the current environmental dust concentration information are collected in real time through environmental sensors, and combined to form the current environmental factor information.

[0146] During the actual operation of the quick-freezer, environmental sensors installed in the workshop collect current environmental information in real time. The ambient temperature sensor continuously monitors changes in ambient temperature and records the current ambient temperature information. The ambient humidity sensor measures the ambient humidity level and obtains the current ambient humidity information. The ambient pressure sensor collects ambient pressure data in real time and generates the current ambient pressure information. The ambient dust concentration sensor detects the dust concentration in the workshop and obtains the current ambient dust concentration information.

[0147] The collected information is combined to form the current environmental factor information. The current environmental factor information reflects the current environmental status of the quick-freezing machine and is of great significance for adjusting the operating parameters of the quick-freezing machine.

[0148] Step S152: The product detection device collects the current initial temperature information, current shape characteristic information, current water content information and current packaging material information of the product to be quick-frozen in real time, and combines them to form current product characteristic information.

[0149] Product inspection equipment performs real-time inspections on frozen products. Temperature detection equipment measures the product's current initial temperature and obtains this information. Shape detection equipment (such as a 3D scanner) can obtain the product's current shape characteristics. Moisture content detection equipment analyzes the product's current moisture content. Packaging material identification equipment can determine the product's current packaging material.

[0150] These product characteristic information are combined to form the current product characteristic information. The current product characteristic information reflects the characteristics of the product to be quick-frozen when it enters the quick-freezing machine and is an important basis for adjusting the operating parameters of the quick-freezing machine.

[0151] Step S153: performing feature normalization processing on the current environmental factor information and the current product feature information, so that the feature dimensions of the current environmental factor information and the current product feature information are consistent with the state feature dimensions in the state space.

[0152] To ensure that the current environmental factors and product features match the state space of the reinforcement learning optimization model, they need to be normalized. This involves normalizing and scaling the data so that data with different features are on the same scale.

[0153] At the same time, ensure that the feature dimensions of the current environmental factor information and the current product feature information are consistent with the state feature dimensions in the state space. This means that the processed information should contain the same number and types of features as in the state space. For example, if the state in the state space contains features such as ambient temperature, ambient humidity, initial product temperature, and product moisture content, then the current environmental factor information and the current product feature information should also contain these features after normalization, and the feature dimensions and order should be consistent.

[0154] By standardizing the features, the compatibility and validity of the data can be guaranteed when this information is input into the optimization strategy of the quick-freezing machine operating parameters.

[0155] Step S154: Input the standardized current environmental factor information and the current product feature information into the quick-freeze machine operating parameter optimization strategy, and the quick-freeze machine operating parameter optimization strategy selects the optimal quick-freeze machine operating parameter combination from the action space based on the current environmental factor information and the current product feature information.

[0156] The standardized current environmental factors and product characteristics are input into a freezer operating parameter optimization strategy. This freezer operating parameter optimization strategy is trained through reinforcement learning and can select the optimal freezer operating parameter combination from the action space based on the input environmental and product characteristics.

[0157] Optimization strategies take into account various factors, such as product quality, production efficiency, and energy consumption, to achieve the best quick-freezing results. For example, based on the current ambient temperature and the initial product temperature, the strategy determines whether to lower the quick-freezing temperature or increase the conveyor speed. Depending on the product's shape and moisture content, the strategy can adjust the frequency of cold air circulation and quick-freezing time.

[0158] Step S155: The selected optimal quick-freezing machine operating parameter combination is determined as the target operating parameter combination and outputted.

[0159] After the quick-freezing machine operating parameter optimization strategy is determined, the optimal quick-freezing machine operating parameter combination is determined as the target operating parameter combination. The target operating parameter combination is the parameter configuration that can enable the quick-freezing machine to achieve the best operating effect under the current environmental and product characteristics.

[0160] Step S156: applying the target operating parameter combination to the parameter control system of the quick-freezing machine.

[0161] For example, step S1561: parsing the target operating parameter combination into quick-freezing temperature adjustment parameter instructions, conveyor belt operating speed parameter instructions, cold air circulation frequency parameter instructions and insulation layer pressure parameter instructions.

[0162] The target operating parameter combination is parsed and split into specific parameter instructions. The target operating parameter combination includes information such as quick freezing temperature adjustment parameters, conveyor belt operating speed parameters, cold air circulation frequency parameters, and insulation layer pressure parameters. These parameter information is converted into corresponding instructions.

[0163] For example, converting the quick-freeze temperature adjustment parameter into a quick-freeze temperature adjustment parameter instruction explicitly instructs the quick-freeze machine to adjust the temperature to the specified value. Converting the conveyor belt speed parameter into a conveyor belt speed parameter instruction instructs the conveyor belt control module to adjust the conveyor belt speed. Converting the cold air circulation frequency parameter into a cold air circulation frequency parameter instruction instructs the air circulation control module to adjust the operating frequency of the cold air circulation device. Converting the insulation layer pressure parameter into an insulation layer pressure parameter instruction instructs the pressure control module to adjust the insulation layer pressure setting.

[0164] Step S1562: Send the quick-freezing temperature adjustment parameter instruction to the temperature control module of the quick-freezing machine, and control the temperature control module to adjust the temperature setting in the quick-freezing chamber according to the quick-freezing temperature adjustment parameter instruction.

[0165] The quick freezing temperature adjustment parameter instruction is sent to the temperature control module of the quick freezing machine. After receiving the instruction, the temperature control module can adjust the temperature setting in the quick freezing chamber according to the instruction requirements.

[0166] The temperature control module controls the operation of the refrigeration equipment, such as starting and stopping the compressor and adjusting the cooling capacity, so that the temperature in the quick-freeze chamber gradually reaches the specified quick-freeze temperature. During this adjustment process, the temperature sensor monitors temperature changes in real time and feeds this data back to the temperature control module, allowing timely adjustments to ensure the temperature remains stable within the specified range.

[0167] Step S1563: Send the conveyor belt running speed parameter instruction to the conveyor belt control module of the quick-freezing machine, and control the conveyor belt control module to adjust the running speed of the conveyor belt according to the conveyor belt running speed parameter instruction.

[0168] The conveyor belt speed parameter command is sent to the conveyor belt control module of the quick freezer. The conveyor belt control module will adjust the conveyor belt speed according to the command.

[0169] The conveyor belt control module adjusts the motor speed and transmission ratio to keep the conveyor belt running at a specified speed. During this adjustment process, the speed sensor monitors the conveyor belt's speed in real time and feeds this data back to the conveyor belt control module, ensuring that the conveyor belt's speed remains stable within the specified range. This ensures that the product's residence time in the freezer meets production requirements, achieving optimal quick-freezing results.

[0170] Step S1564: Send the cold air circulation frequency parameter instruction to the air circulation control module of the quick-freezing machine, and control the air circulation control module to adjust the operating frequency of the cold air circulation device according to the cold air circulation frequency parameter instruction.

[0171] The cold air circulation frequency parameter instruction is sent to the air circulation control module of the quick freezer. The air circulation control module will adjust the operating frequency of the cold air circulation device according to the instruction.

[0172] The air circulation control module controls the fan speed and operating mode, ensuring that the cold air circulates within the freezer at a specified frequency. By adjusting the cold air circulation frequency, we ensure that all parts of the product are evenly exposed to the cold air, achieving uniform freezing. During this adjustment process, the wind speed sensor monitors the cold air circulation speed in real time and feeds this data back to the air circulation control module, allowing it to promptly adjust the fan's operating status to ensure that the cold air circulation frequency remains stable within the specified range.

[0173] Step S1565: sending the insulation layer pressure parameter instruction to the pressure control module of the quick-freezing machine, and controlling the pressure control module to adjust the pressure setting of the insulation layer according to the insulation layer pressure parameter instruction.

[0174] The insulation layer pressure parameter command is sent to the pressure control module of the quick freezer. The pressure control module will adjust the insulation layer pressure setting according to the command.

[0175] The pressure control module adjusts pressure regulating devices, such as valve opening and pump operating status, to maintain the specified pressure within the insulation layer. During this adjustment process, pressure sensors monitor pressure changes within the insulation layer in real time and provide feedback to the pressure control module for timely adjustments, ensuring that the insulation layer pressure remains stable within the specified range. Appropriate insulation layer pressure reduces heat loss and improves the freezer's energy efficiency.

[0176] Further, Figure 2FIG. 1 shows a hardware structure diagram of an artificial intelligence-based quick-freezing machine operating parameter optimization system 100 for implementing the method provided in an embodiment of the present application. Figure 2 As shown, the artificial intelligence-based quick-freezing machine operating parameter optimization system 100 may include at least one processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, a transmission device 106 for communication functions, and a controller 108. It will be understood by those skilled in the art that Figure 2 The structure shown is only for illustration and does not limit the structure of the quick-freezing machine operating parameter optimization system 100 based on artificial intelligence. For example, the quick-freezing machine operating parameter optimization system 100 based on artificial intelligence may also include Figure 2 More or fewer components than shown, or with Figure 2 Different configurations shown.

[0177] The memory 104 can be used to store software programs and modules of application software, such as program instructions corresponding to the above-mentioned method embodiments in the embodiments of the present application. The processor 102 executes the software programs and modules stored in the memory 104 to perform various functional applications and data processing, thereby implementing the above-mentioned artificial intelligence-based quick-freezer operating parameter optimization method. The transmission device 106 is used to obtain or send data via a network.

[0178] Those skilled in the art will understand that all or part of the steps for implementing the above embodiments may be accomplished by hardware, or may be accomplished by instructing the relevant hardware through a program, and the above program may be stored in a computer-readable storage medium, which may be a read-only memory, a disk, or an optical disk, etc.

Claims

1. A method for optimizing operating parameters of a quick-freezing machine based on artificial intelligence, characterized in that: The method comprises: Integrate quick-freezing machine operating parameter information, environmental factor information, and product characteristic information to generate a comprehensive database, wherein the comprehensive database includes associated stored operating parameter records, environmental status records, and product characteristic records; Acquiring operation record text information and fault report text information during the production process, performing key information extraction processing on the operation record text information and the fault report text information to obtain text key information related to the quick-freezing machine operating parameters, and adding the text key information to the comprehensive database; Based on the comprehensive database, configuring core elements of a reinforcement learning optimization model, wherein the core elements include setting the quick-freezing machine system as an intelligent agent, setting the quick-freezing machine operating parameter combination as an action space, setting the combination of environmental factor information and product feature information as a state space, and setting the quick-freezing effect index as an input parameter of a reward function; Controlling the intelligent agent to perform parameter adjustment operations in the state space according to the action space, calculating a reward value corresponding to each parameter adjustment operation through the reward function, updating the policy parameters of the reinforcement learning optimization model according to the reward value, and generating a quick-freezer operating parameter optimization strategy; During the operation of the quick-freezer, the current environmental factor information and the current product characteristic information are collected in real time, the current environmental factor information and the current product characteristic information are input into the quick-freezer operation parameter optimization strategy, the target operation parameter combination is output, and the target operation parameter combination is applied to the parameter control system of the quick-freezer.

2. The method for optimizing operating parameters of a quick-freezing machine based on artificial intelligence according to claim 1, characterized in that: The integration of quick-freezing machine operating parameter information, environmental factor information and product feature information to generate a comprehensive database includes: Collecting operating parameter information generated by the quick-freezing machine during historical operation, the operating parameter information includes quick-freezing temperature adjustment parameters, conveyor belt operating speed parameters, cold air circulation frequency parameters and insulation layer pressure parameters; Collecting environmental factor information related to the operation of the quick-freezing machine, the environmental factor information includes ambient temperature information, ambient humidity information, ambient air pressure information and ambient dust concentration information; Collecting product characteristic information of the product to be quick-frozen, the product characteristic information including product initial temperature information, product shape characteristic information, product water content information and product packaging material information; Adding the same timestamp to the operating parameter information, the environmental factor information, and the product feature information to establish an association relationship in the time dimension; Based on the association relationship of the timestamp marks, the operating parameter information is stored as an operating parameter record, the environmental factor information is stored as an environmental status record, and the product feature information is stored as a product characteristic record. The operating parameter records, the environmental status records and the product characteristic records are integrated to generate a comprehensive database.

3. The method for optimizing operating parameters of a quick-freezing machine based on artificial intelligence according to claim 1, characterized in that: The obtaining of the operation record text information and the fault report text information during the production process includes: Acquiring operation record text information and fault report text information during a production process, performing text standardization processing on the operation record text information and the fault report text information, and performing paragraph division processing on the standardized operation record text information and the fault report text information, dividing the operation record text information into a plurality of operation description paragraphs, and dividing the fault report text information into a plurality of fault description paragraphs; Performing keyword recognition processing on the operation description paragraph and the fault description paragraph to extract a candidate keyword set containing names of quick-freezing machine operating parameters; Performing contextual semantic association analysis on the candidate keyword set to determine the semantically directed object of each candidate keyword in the corresponding paragraph, and screening out target keywords directed to the quick-freezing machine operating parameter adjustment behavior; Based on the target keyword and its contextual semantic pointing object, the adjustment condition information, adjustment range information and adjustment result information related to the quick-freezing machine operating parameters are extracted, and the adjustment condition information, the adjustment range information and the adjustment result information are combined into text key information.

4. The method for optimizing operating parameters of a quick-freezing machine based on artificial intelligence according to claim 3, characterized in that: The process of performing contextual semantic association analysis on the candidate keyword set, determining the semantically directed object of each candidate keyword in the corresponding paragraph, and screening out target keywords directed to the quick-freezing machine operating parameter adjustment behavior, includes: Sentence segmentation is performed on the operation description paragraph or fault description paragraph where each candidate keyword is located to obtain multiple semantically complete sentence units; Perform dependency syntax analysis on each sentence unit to identify the subject-verb-object structure and adverbial-verb structure in the sentence unit, and determine the grammatical role of the candidate keyword in the sentence unit; Determining whether the candidate keyword has a direct association relationship with an action verb based on the grammatical role, the action verbs including adjust, modify, set, increase, and decrease; For candidate keywords with direct correlation, extract their corresponding action verbs and action objects, and determine whether the action object is a quick-freezing machine operating parameter; The candidate keywords whose action objects are the quick-freezing machine operating parameters are determined as target keywords pointing to the quick-freezing machine operating parameter adjustment behavior.

5. The method for optimizing operating parameters of a quick freezer based on artificial intelligence according to claim 1, characterized in that: The core elements of the reinforcement learning optimization model are configured based on the comprehensive database. The core elements include setting the quick-freezing machine system as an intelligent agent, setting the quick-freezing machine operating parameter combination as an action space, setting the combination of environmental factor information and product feature information as a state space, and setting the quick-freezing effect index as an input parameter of the reward function, including: Based on the operating parameter records in the comprehensive database, all historically used quick-freezing machine operating parameter combinations are extracted, and the set of quick-freezing machine operating parameter combinations is defined as the action space of the reinforcement learning optimization model; Based on the environmental state records and product characteristic records in the comprehensive database, the combined data of the environmental factor information and the product characteristic information is defined as a state space of the reinforcement learning optimization model, where each state in the state space includes a set of environmental factor information and a corresponding set of product characteristic information; The quick-freezing machine system is defined as an intelligent agent of the reinforcement learning optimization model, wherein the intelligent agent is capable of selecting a quick-freezing machine operating parameter combination from the action space and acting on the quick-freezing environment; Extracting evaluation data related to quick-freezing effect from the comprehensive database, and determining the evaluation data as quick-freezing effect indicators, wherein the quick-freezing effect indicators include a product qualification rate indicator, a production cycle indicator, and an energy consumption indicator; A reward function is constructed based on the quick-freezing effect index, wherein the input parameter of the reward function is the actual measurement value of the quick-freezing effect index, and the output is a reward value corresponding to the actual measurement value.

6. The method for optimizing operating parameters of a quick freezer based on artificial intelligence according to claim 5, characterized in that: The constructing of a reward function based on the quick-freezing effect indicator includes: Setting weight coefficients for the product qualification rate index, production rhythm index, and energy consumption index in the quick-freezing effect index, respectively, wherein the weight coefficients are determined based on the priority requirements of quick-freezing production; Standardizing the actual measured value of the product qualification rate indicator to obtain a standardized value of the product qualification rate, wherein the standardized value of the product qualification rate is positively correlated with the actual measured value of the product qualification rate indicator; Normalizing the actual measured value of the production tact index to obtain a normalized production tact value, wherein the normalized production tact value is positively correlated with the actual measured value of the production tact index; Normalizing the actual measured value of the energy consumption indicator to obtain a normalized energy consumption value, wherein the normalized energy consumption value is negatively correlated with the actual measured value of the energy consumption indicator; The standardized value of the product qualification rate, the standardized value of the production rhythm and the standardized value of the energy consumption are multiplied by the corresponding weight coefficients respectively and then summed to obtain a reward value corresponding to the actual measured value of the quick-freezing effect index.

7. The method for optimizing operating parameters of a quick freezer based on artificial intelligence according to claim 1, characterized in that: The controlling the intelligent agent to perform parameter adjustment operations in the state space according to the action space, calculating a reward value corresponding to each parameter adjustment operation through the reward function, updating the strategy parameters of the reinforcement learning optimization model according to the reward value, and generating a quick-freezing machine operating parameter optimization strategy, including: Randomly selecting a set of environmental factor information and a corresponding set of product feature information from the comprehensive database as an initial state, and inputting the initial state into the intelligent agent; Controlling the intelligent agent to select an initial quick-freezing machine operating parameter combination from the action space as an initial action, and executing a parameter adjustment operation corresponding to the initial action; Collecting actual measurement values ​​of the quick-freezing effect index of the quick-freezing machine after executing the initial action, inputting the actual measurement values ​​into the reward function, and calculating an initial reward value; Based on the initial state, the initial action, and the initial reward value, updating the strategy parameters of the agent according to a reinforcement learning algorithm; The process of selecting a new state from the state space, selecting a new action for the agent, calculating a new reward value, and updating the strategy parameters is repeated until the strategy parameters converge, thereby generating a strategy for optimizing the operating parameters of the quick-freezing machine.

8. The method for optimizing operating parameters of a quick freezer based on artificial intelligence according to claim 7, characterized in that: The updating of the agent's strategy parameters according to a reinforcement learning algorithm based on the initial state, the initial action, and the initial reward value includes: Based on the initial state and the initial action, determining a next state through a state transition function of a reinforcement learning algorithm, wherein the next state is a combination of environmental factor information and product feature information of the quick-freezing machine after the initial action is performed; The initial state, the initial action, the initial reward value and the next state form an experience sample, and store the sample in an experience replay buffer; When the number of experience samples in the experience replay buffer reaches a preset threshold, randomly extracting batches of experience samples from the experience replay buffer; For each experience sample in the batch of experience samples, calculate a target Q value based on current policy parameters, where the target Q value is a weighted sum of a current reward value and a future reward value; Calculate the difference between the target Q value and the current Q value using a loss function, where the current Q value is a Q value calculated based on the current strategy parameters for the experience sample; The strategy parameters of the agent are updated according to the difference through the back propagation algorithm to reduce the difference between the target Q value and the current Q value.

9. The method for optimizing operating parameters of a quick freezer based on artificial intelligence according to claim 1, characterized in that: During the operation of the quick-freezing machine, current environmental factor information and current product feature information are collected in real time, the current environmental factor information and current product feature information are input into the quick-freezing machine operation parameter optimization strategy, and a target operation parameter combination is output, including: During the operation of the quick-freezing machine, the current ambient temperature information, current ambient humidity information, current ambient air pressure information and current ambient dust concentration information are collected in real time through environmental sensors and combined to form the current environmental factor information; The product detection device collects the current initial temperature information, current shape characteristic information, current moisture content information and current packaging material information of the product to be quick-frozen in real time, and combines them to form the current product characteristic information; performing feature normalization processing on the current environmental factor information and the current product feature information so that feature dimensions of the current environmental factor information and the current product feature information are consistent with state feature dimensions in the state space; Inputting the standardized current environmental factor information and the current product feature information into the quick-freezing machine operating parameter optimization strategy, wherein the quick-freezing machine operating parameter optimization strategy selects the optimal quick-freezing machine operating parameter combination from the action space based on the current environmental factor information and the current product feature information; The selected optimal quick-freezing machine operating parameter combination is determined as the target operating parameter combination and output.

10. An artificial intelligence-based quick-freezing machine operating parameter optimization system, characterized in that: The method comprises a processor and a readable storage medium, wherein the readable storage medium stores a program, and when the program is executed by the processor, the method for optimizing the operating parameters of a quick-freezing machine based on artificial intelligence as described in any one of claims 1 to 9 is implemented.

Images