An energy-saving intelligent temperature control system for cattle and sheep slaughterhouse cooling rooms

Abstract

This invention provides an energy-saving intelligent temperature control system for cooling rooms in cattle and sheep slaughterhouses, comprising the following functional modules: a data sensing and preprocessing module: collecting relevant characteristic data on temperature control safety in the cooling room and preprocessing the raw data; a comprehensive cooling safety assessment module: generating characteristic data scores from the preprocessed relevant characteristic data and generating a comprehensive cooling safety index; a model cognition confidence assessment module: collecting confidence assessment factors in the intelligent temperature control system and generating a model cognition confidence index; and a comprehensive decision-making and mode adaptive switching module: generating a comprehensive decision index based on the comprehensive cooling safety index and the model cognition confidence index. This invention constructs a comprehensive cooling safety index and a model cognition confidence index, enabling the system to provide cooling on demand, dynamically allocate control permissions, and proactively degrade operation when model cognition is insufficient, effectively improving the intelligence level and operational reliability of the temperature control system.

Description

Technical Field

[0001] This invention relates to the field of intelligent temperature control technology, specifically to an energy-saving intelligent temperature control system for cattle and sheep slaughtering and cooling rooms. Background Technology

[0002] In the slaughtering and processing of cattle and sheep, the cooling room is a crucial link in ensuring the hygiene, safety, and quality of meat products. The main goal of the cooling process is to reduce the core temperature of the carcass to below the microbial inhibition threshold within a specified time, while simultaneously meeting the requirements of the aging process for cooling rate and final temperature. However, when intelligent systems encounter abnormal conditions in the cooling room or have blind spots in their own cognition, they often blindly execute decisions due to a lack of self-assessment of decision reliability, leading to food safety risks or control failures. Therefore, by using the safety status of the cooling room and the reliability of the model to determine when human intervention is appropriate, the energy-saving optimization potential of the intelligent model can be fully utilized while ensuring the safety of the meat cooling process. This effectively solves the problems of delayed response, high energy consumption, and lack of decision reliability assessment in temperature control systems.

[0003] The prior art, disclosed in CN119915066B, discloses an intelligent temperature control system and method for seafood refrigeration. This technology includes: deploying multiple gas sensing units within the cold storage facility to construct a dynamic pollution map by detecting the types and concentration distribution of characteristic gases of corrosion; when any sensing unit detects that the concentration of a characteristic gas exceeds a primary threshold, a directional ventilation mode is activated, with the exhausted cold air flowing through a heat exchange zone to pre-cool the introduced fresh air; a humidity compensation system is simultaneously activated during the ventilation process; when three consecutive sensing units detect that the concentration exceeds a secondary threshold, a pollution source tracing mode is triggered, inferring the core area of ​​corrosion based on the gas diffusion path; an automated removal command is triggered according to the positioning coordinates, and the gas concentration change at the original cargo location is monitored in real time during removal; if the concentration does not decrease to a safe value after removal, a secondary tracing is triggered.

[0004] However, the aforementioned existing technologies rely on fixed process curves or manual experience for refrigeration regulation, making it difficult to perceive multi-dimensional dynamic information such as changes in the core temperature of meat products, disturbances from door opening, equipment operational health, and batch material differences in real time. Due to the lack of quantitative assessment of the overall safety margin of the cooling process, the system often experiences control lag or over-adjustment when facing complex operating conditions, resulting in high energy consumption and food safety risks. At the same time, traditional control systems cannot assess the reliability of their own decisions, rigidly executing preset logic when the model's understanding is insufficient or the operating conditions are abnormal, failing to achieve a smooth transition in human-machine collaboration and adaptive mode switching.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide an energy-saving intelligent temperature control system for cattle and sheep slaughtering cooling rooms to solve the problems mentioned in the background art. By constructing a comprehensive cooling safety index and a model cognitive confidence index, and integrating the two to generate a comprehensive decision index, the system enables adaptive switching between four modes of the cooling process: fully automatic, semi-automatic, manual monitoring, and emergency forced operation. This maximizes the operating efficiency, energy efficiency, and energy-saving effect of the refrigeration system while ensuring meat cooling safety and process compliance.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] An energy-saving intelligent temperature control system for cattle and sheep slaughterhouse cooling rooms includes the following functional modules:

[0009] Data sensing and preprocessing module: Collects relevant characteristic data on temperature control safety in the cooling room, including the core temperature of meat products in the cooling room, the opening and closing status and duration of the door, the refrigeration health, and the cooling load; and preprocesses the raw data of the relevant characteristic data, including outlier handling.

[0010] Cooling Comprehensive Safety Assessment Module: Scores the relevant characteristic data after preprocessing to generate characteristic data scores; the characteristic data scores include: meat core temperature safety score, warehouse door disturbance score, system health score, and load characteristic score; and generates a cooling comprehensive safety index based on the characteristic data scores, which is used to evaluate the overall safety margin and control effect of the cooling system in real time;

[0011] Model cognitive confidence assessment module: Collects confidence assessment factors from the intelligent temperature control system. These confidence assessment factors include experience coverage score, decision stability score, and physical consensus score. The module also preprocesses the confidence assessment factors, including outlier detection and correction. The preprocessed confidence assessment factors generate a model cognitive confidence index, which is used to quantify the system's familiarity with the current operating conditions and the reliability of its decisions.

[0012] The integrated decision-making and adaptive mode switching module generates an integrated decision index based on the cooling integrated safety index and the model cognitive confidence index. The integrated decision index is dynamically mapped to a preset decision range by integrating the cooling integrated safety index and the model cognitive confidence index, so as to realize the adaptive switching of four control modes of the cooling process: fully automatic, semi-automatic, manual monitoring and emergency forced. This ensures that when the model's cognition is insufficient, the system can actively degrade its operation and request manual intervention.

[0013] Furthermore, the raw data of the relevant feature data are preprocessed, including outlier handling; a dynamic confidence interval is set for continuously sampled time-series data using the statistical 3σ criterion, outliers exceeding the upper and lower limits are marked as outliers and removed, and the data is completed by linear interpolation or forward padding to ensure the continuity and integrity of the data sequence; logical verification is performed in conjunction with industrial field experience thresholds to further filter out invalid data.

[0014] Furthermore, the preprocessed relevant feature data are quantitatively evaluated to generate corresponding feature data scores. These feature data scores include: meat core temperature safety score, warehouse door disturbance score, system health score, and load characteristic score. The meat core temperature safety score is based on the degree of conformity between the real-time monitored meat core temperature and its expected cooling trajectory, used to characterize the current cooling process's compliance with food safety and process requirements. The warehouse door disturbance score quantifies the intensity of disturbance caused by external heat and moisture intrusion to the warehouse thermal environment based on statistical analysis of the frequency of warehouse door opening, single duration, and cumulative opening time. The system health score evaluates the operating efficiency and stability of the compressor, heat exchanger, and air cooler by integrating the changing trends of key operating parameters of the refrigeration unit, reflecting the equipment's ability to provide the required cooling capacity. The load characteristic score comprehensively evaluates the cooling difficulty of the current batch of meat based on its fat coverage thickness, initial temperature, and inherent carcass specifications, enabling the system to adapt to material differences between different batches.

[0015] Furthermore, a comprehensive cooling safety index is generated based on the feature data scores, and the comprehensive cooling safety index is obtained using the following formula:

[0016]

[0017] Where: T represents the safe center temperature of the meat product;

[0018] D represents the Kumen disturbance component;

[0019] H represents the system health score;

[0020] L represents the load characteristic segment;

[0021] These are the weighting coefficients for the meat product core temperature safety score, the warehouse door disturbance score, the system health score, and the load characteristic score, respectively; and the weighting coefficients satisfy the following relationships: The relationship between the weighting coefficients is as follows: .

[0022] Furthermore, the comprehensive cooling safety index integrates multi-dimensional temperature control safety-related characteristic data into a unified quantitative indicator, used to assess the overall safety status of meat cooling processes in cattle and sheep slaughtering cooling rooms in real time. By differentially weighting and synthesizing the meat core temperature safety score, system health score, warehouse door disturbance score, and load characteristic score, the monitoring information with different physical meanings is normalized to an intuitive scale of 0 to 100, thereby characterizing the degree to which the current cooling conditions meet food safety requirements and process objectives. The value of the comprehensive cooling safety index directly reflects the risk level of the cooling process: a high score indicates that the meat cooling rate is ideal, the equipment is operating stably, and external interference is controllable; a low score indicates the existence of food safety hazards, equipment abnormalities, or deteriorating operating conditions, requiring timely warnings and manual intervention.

[0023] Furthermore, confidence assessment factors are collected in the intelligent temperature control system. These confidence assessment factors are generated in the following ways: The experience coverage score is obtained by comparing the historical training dataset with the feature vector of the current working condition. Specifically, it is obtained by extracting the meat attributes, environmental parameters, and equipment status at the current moment to form a multi-dimensional feature vector. The nearest neighbor search algorithm is used to find the distribution density of its neighboring samples in the feature space of the training data, and then mapping it into a quantitative score representing the model's familiarity with the current scenario. The decision stability score is generated through the uncertainty estimation technology of the neural network. During the model inference stage, multiple forward propagations are performed on the same input, and the statistical variance of the multiple output results is collected and analyzed. The inverse mapping value of the output variance measures the degree of disagreement within the model. The smaller the variance, the higher the decision stability score. The physical consensus score is generated by running the neural network model and the built-in simplified physical reference model in parallel. The same working condition feature is simultaneously input into the two types of models to obtain the relative deviation between the neural network prediction value and the prediction value of the physical model based on thermodynamic mechanism. The deviation is mapped into a consensus score representing whether the intelligent decision of the system conforms to physical laws based on the comparison relationship between the deviation and the preset threshold.

[0024] Furthermore, the preprocessing includes: employing a detection strategy based on a combination of range constraints and rate of change constraints: firstly, a valid range is preset according to the physical meaning of each factor, and values ​​exceeding this range are directly determined as invalid; for continuous time series data, the rate of change between adjacent time points is calculated, and if the change exceeds an empirical threshold, it is considered an abnormal jump; for detected outliers, linear interpolation or previous value preservation is used for correction to ensure the temporal continuity and stability of the confidence assessment factors; at the same time, logical consistency verification is combined to further assist in identifying potential anomalies.

[0025] Furthermore, a model cognitive confidence index is generated based on the preprocessed confidence assessment factors. The model cognitive confidence index is calculated using the following formula:

[0026]

[0027] Where: E is the experience coverage score;

[0028] S represents the decision stability score;

[0029] P represents the physical consensus score;

[0030] These are the weighting coefficients for the experience coverage score, decision stability score, and physical consensus score, respectively, and the weighting coefficients satisfy the following relationship: The relationship between the weighting coefficients is as follows: .

[0031] Furthermore, a comprehensive decision index is generated based on the cooling comprehensive safety index and the model cognitive confidence index. The comprehensive decision index is obtained through the following formula:

[0032]

[0033] in: It serves as a comprehensive decision-making index.

[0034] Furthermore, the comprehensive decision index integrates the intelligent cooling comprehensive safety index, which reflects the safety of meat cooling and equipment operation, with the model cognitive confidence index, which characterizes the credibility of the intelligent model, through multiplication to form a comprehensive score from 0 to 100. The comprehensive decision index is directly mapped to a preset decision range, thereby enabling the system to adaptively and seamlessly switch between four control modes: fully automatic, semi-automatic, manual monitoring, and emergency forced control. When the CDI is high, it indicates that the site is safe and the model is highly confident, and the system authorizes the intelligent model to fully execute refined energy-saving control. When the CDI drops to a low level, it means that there is a safety risk or insufficient model cognition, and the system actively degrades its operation and requests manual intervention.

[0035] Compared with existing technologies, the beneficial effects of this invention are as follows: By constructing a multi-dimensional feature data fusion mechanism, this invention can collect real-time data on the core temperature of meat products, the status of the storage door, equipment health, and material load characteristics, thus solving the problems of single-dimensional perception and lagging control in traditional single-point temperature control modes; by introducing a model cognitive confidence index based on experience coverage, decision stability, and physical consensus, the system can quantify the reliability of its own decisions and proactively degrade operation when there is insufficient cognition, effectively avoiding the risk of model misjudgment; by generating a comprehensive decision index and dynamically mapping it to four control modes—fully automatic, semi-automatic, manual monitoring, and emergency forced—it achieves human-machine collaboration and seamless switching under the principle of safety priority; while ensuring the precise execution of meat cooling processes and food safety, it significantly reduces the energy consumption of the refrigeration system, improves the stability of equipment operation and the response capability to abnormal conditions, and provides an intelligent, adaptive, and highly reliable temperature control solution for the cattle and sheep slaughtering and cooling process. Attached Figure Description

[0036] Figure 1 A system block diagram of an energy-saving intelligent temperature control system for cattle and sheep slaughterhouse cooling rooms;

[0037] Figure 2 This is a schematic diagram of the process of an energy-saving intelligent temperature control system for cattle and sheep slaughtering and cooling rooms according to the present invention. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0039] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0040] Example:

[0041] Please see Figures 1-2 The present invention provides a technical solution:

[0042] An energy-saving intelligent temperature control system for cattle and sheep slaughterhouse cooling rooms includes the following functional modules:

[0043] S1: Collect relevant characteristic data on temperature control safety in the cooling room. Specifically, this data includes: real-time monitoring of the meat's core temperature, obtained through implanted temperature sensors to capture core temperature changes in representative carcasses, characterizing the real-time effect of the cooling process; door opening and closing status and duration, recorded via door magnetic switches, with single opening duration and cumulative opening frequency counted to quantify the disturbance intensity of external hot and humid air on the internal environment; refrigeration health, comprehensively evaluating the refrigeration system's operating efficiency and stability by collecting key operating parameters such as compressor suction and discharge pressure, temperature, and superheat; and cooling load, quantifying the cooling difficulty by acquiring the fat thickness, initial core temperature, and inherent carcass specifications of the current batch of meat, enabling the system to adapt to material differences between different batches.

[0044] The raw data of the relevant feature data is preprocessed, including outlier handling. Due to the complex factors in the cooling room environment, such as low temperature, high humidity, equipment vibration, and electromagnetic interference, the raw time-series data collected by the sensors inevitably contains gross errors, transient spikes, or abnormal jumps caused by communication packet loss. If such outliers are not identified and corrected, they will directly affect the accuracy of subsequent feature extraction and index calculation. Anomaly detection is performed on the continuously sampled time-series data using the statistical 3σ criterion: for each measurement point, the mean and standard deviation are calculated based on its historical data within a sliding window, and a dynamic confidence interval is set accordingly. Isolated data points outside the specified range are identified as outliers and removed. For the data gaps left after removal, linear interpolation is used to smoothly fill the gaps based on the valid data points before and after the gaps. Alternatively, if forward filling is not possible, valid values ​​from nearby times are used as substitutes to ensure the continuity and integrity of the data sequence. In addition, relying solely on statistical criteria is insufficient to identify invalid data caused by sensor physical limitations or logical errors. Therefore, it is also necessary to combine industrial field experience thresholds for logical verification to remove values ​​exceeding the sensor's upper limit, false trigger records where the duration of the gate opening / closing state is less than the set minimum duration, and abrupt changes where the temperature change rate exceeds the physical limit.

[0045] S2: Quantitatively evaluate the relevant feature data after preprocessing to generate corresponding feature data scores; the feature data scores include: meat core temperature safety score, warehouse door disturbance score, system health score, and load characteristic score; wherein, the meat core temperature safety score is based on the degree of conformity between the real-time monitored meat core temperature and its expected cooling trajectory, by comparing the current measured temperature value with the standard cooling curve preset according to food safety standards and aging process requirements point by point, comprehensively considering factors such as absolute temperature deviation, deviation duration, and the current cooling stage, and dynamically calculating the compliance margin of the current temperature relative to the target value; the meat core temperature safety score is used to characterize the real-time compliance status of the current cooling process with food safety and process requirements, and is the core indicator for evaluating the cooling effect;

[0046] The chiller door disturbance score is based on statistical analysis of chiller door opening frequency, single duration and cumulative opening time. By setting a weighted cumulative algorithm within different time windows, it quantifies the comprehensive disturbance intensity caused by the intrusion of external hot and humid air to the thermal environment inside the chiller. This score can effectively identify abnormal heat load impacts caused by operational factors such as frequent entry and exit from the chiller or chiller doors not being closed tightly.

[0047] The system health score is assessed by integrating the changing trends of key operating parameters of the refrigeration unit. This score comprehensively considers real-time data from multiple dimensions, including the fluctuation range of compressor suction and discharge pressure and temperature, the degree of superheat deviating from the ideal range, the stability of the air cooler's operating current, and the evaporator frosting cycle. Through multi-parameter coupled analysis, it comprehensively evaluates the operating efficiency and stability of core components such as the compressor, heat exchanger, and air cooler. This score directly reflects the refrigeration system's ability to provide the required cooling capacity under current operating conditions and is an important basis for judging whether there are potential faults or performance degradation in the equipment.

[0048] The load characteristic score comprehensively evaluates the inherent properties of the current batch of meat. This score integrates characteristic parameters such as fat coverage thickness, initial core temperature, carcass grade, and breed type obtained from manual input or automated detection equipment. Based on a cooling difficulty benchmark model built from historical big data, it compares the heat dissipation characteristics of the current batch of material with standard operating conditions, quantifying its relative cooling difficulty. This score enables the system to adaptively identify and respond to material differences between different batches, avoiding overcooling or undercooling problems caused by using a uniform control strategy.

[0049] A comprehensive cooling safety index is generated based on the feature data scores, and the comprehensive cooling safety index is obtained using the following formula:

[0050]

[0051] Where: T represents the safe center temperature of the meat product;

[0052] D represents the Kumen disturbance component;

[0053] H represents the system health score;

[0054] L represents the load characteristic segment;

[0055] These are the weighting coefficients for the meat product core temperature safety score, the warehouse door disturbance score, the system health score, and the load characteristic score, respectively; and the weighting coefficients satisfy the following relationships: The relationship between the weighting coefficients is as follows: The meat core temperature safety score directly reflects whether the meat is within a safe temperature range and whether the cooling rate meets the process requirements. It is a core criterion for food safety and the ultimate control target of the cooling process, therefore it is given the highest weight. The operating status of the refrigeration equipment is the material basis for ensuring cooling capacity. Equipment malfunctions will directly lead to cooling failure. Its importance is second only to the meat temperature, so it is given the second highest weight for system health. The opening of the storage door is the most important external source of heat and humidity interference in the cooling room. Although it has a significant impact on safety, it is an adjustable operating condition disturbance, so the weight of the storage door disturbance score is lower than that of the equipment health score. The inherent properties of meat determine the difficulty of batch cooling, but this factor is relatively static and can be partially compensated by adjusting process parameters, so it is given the lowest weight for load characteristic score.

[0056] The intelligent cooling comprehensive safety index is a comprehensive evaluation tool that integrates multi-dimensional temperature control safety-related characteristic data into a single quantitative indicator. Its design purpose is to quantitatively evaluate the overall safety status of meat cooling process in cattle and sheep slaughtering and cooling rooms in real time and intuitively. This index uses differentiated weighted calculations of four sub-items with different physical meanings: meat core temperature safety score, system health score, warehouse door disturbance score, and load characteristic score. It maps monitoring information from different sources and with different dimensions to a standardized scale of 0 to 100, thus accurately depicting the degree to which cooling conditions meet food safety requirements and process objectives. The ICSI value directly reflects the real-time risk level of the cooling process: when the index is in the high range, it indicates that the meat cooling rate is as expected, the refrigeration equipment is operating stably, external thermal and humidity disturbances are controllable, and the system is in a safe and efficient state; when the index drops to the middle range, it indicates a slight deviation or disturbance, requiring closer attention; when the index enters the low range, it means there may be risks such as excessive meat core temperature, abnormal key equipment parameters, or serious deterioration of operating conditions, requiring timely triggering of the early warning mechanism or requesting manual intervention; if the index remains at an extremely low level, it indicates that the system faces serious food safety hazards or equipment failure, and emergency measures must be taken immediately. This index can be used not only for real-time risk monitoring and alarms, but also as an optimization objective function for automatic control systems to guide the dynamic adjustment of refrigeration strategies. At the same time, its historical change curves can provide important quantitative basis for system operation trend analysis, control parameter optimization, and fault backtracking diagnosis, thereby achieving intelligent and refined management and control of the cooling process while ensuring meat safety.

[0057] S3: Collect confidence assessment factors from the intelligent temperature control system; the experience coverage score is calculated based on the comparison between the historical training dataset and the current operating condition feature vector; firstly, extract key information such as meat attributes, environmental parameters, and equipment status at the current moment from the preprocessed real-time data stream, and construct a multi-dimensional feature vector representing the current operating condition after feature standardization. Subsequently, a nearest neighbor search algorithm suitable for fast retrieval in high-dimensional space is used to traverse and search for a preset number of samples that are closest to the current feature vector in the feature space of the historical training data stored in the system database; by statistically analyzing the distribution density of neighboring samples in the feature space, the sample density is converted into a quantitative score between 0 and 100 according to a preset density-score mapping function; this score is the experience coverage, and the higher the value, the higher the frequency of the current operating condition during the model training stage, the more sufficient the model's experience reserve for this type of scenario, and the more reliable the decision basis.

[0058] The decision stability score is generated using uncertainty estimation techniques in neural networks. To quantify the model's internal confidence in the current input, multiple forward propagations are performed on the same input feature during the model inference phase. In each propagation, different sub-network structures of the model are simulated by randomly discarding some neurons or injecting small noise, thereby obtaining multiple output results with reasonable differences. The key predicted values ​​of all forward propagation outputs are collected, and their statistical variance or standard deviation is calculated as an uncertainty measure. The smaller the variance, the lower the dispersion between multiple prediction results, the smaller the internal disagreement of the model, and the more confident the judgment on the current input. Conversely, the larger the variance, the more hesitant the model is, and the lower the credibility of the output results. After normalization, the variance value is mapped to an inverse score, that is, a decision stability score in the range of 0 to 100 is obtained, with the smaller the variance, the higher the corresponding score.

[0059] The physical consensus score is generated by running a neural network model and a built-in simplified physical reference model in parallel. The simplified physical reference model is constructed based on the first law of thermodynamics, the fundamental equations of heat transfer, and the refrigeration cycle mechanism. It employs the lumped parameter method to ensure real-time computation and can output theoretical predictions that conform to physical laws based on current operating conditions. For input features at the same time, both the neural network model and the physical reference model are driven to perform inference, obtaining the prediction outputs of the two models respectively, and then calculating the relative deviation between them. This deviation reflects the degree of deviation of the intelligent decision from objective physical laws. The relative deviation is compared with a preset physical consensus threshold. The smaller the deviation, the higher the mapped physical consensus score, indicating that the intelligent model's decision is highly consistent with the thermodynamic mechanism. If the deviation exceeds the threshold, it indicates that the model may produce abnormal outputs that violate basic physical laws, and its credibility assessment needs to be reduced accordingly.

[0060] The confidence assessment factors are preprocessed, including: preprocessing the generated confidence assessment factors to ensure their accuracy and reliability as inputs for subsequent model cognitive confidence index calculations; the preprocessing includes employing a composite detection strategy based on a combination of range constraints and rate of change constraints; and setting reasonable value ranges for the experience coverage score, decision stability score, and physical consensus score according to the physical meaning and preset domain of each confidence assessment factor, and directly determining and removing values ​​that exceed the range due to calculation anomalies or data transmission errors as invalid data.

[0061] For factor data with continuous time-series characteristics, the rate of change or absolute difference between adjacent sampling times is calculated, and the magnitude of the change is compared with an empirical threshold set based on historical statistical distribution. If the rate of change of a value at a certain time is far beyond the normal fluctuation range, it is determined that there is an abnormal jump at that point, which may be due to model inference jitter or instantaneous interference. For the various outliers detected above, a linear interpolation method is used to smooth the filling with the effective data adjacent to the outlier. When linear interpolation cannot be implemented, the effective value of the previous time is continued by preserving the previous value to ensure the temporal continuity and dynamic stability of the confidence assessment factor sequence.

[0062] To address more subtle logical anomalies, it is also necessary to verify logical consistency by combining the inherent relationships between various factors. The physical consensus score should not fluctuate frequently and significantly during the stable operation of the refrigeration system. The experience coverage score and the decision stability score should maintain a certain synergistic trend under specific operating conditions. If isolated abnormal combinations that violate conventional logic occur, further assistance can be provided to identify potential data quality problems.

[0063] The model cognitive confidence index is generated based on the preprocessed confidence assessment factors. The model cognitive confidence index is calculated using the following formula:

[0064]

[0065] Where: E is the experience coverage score;

[0066] S represents the decision stability score;

[0067] P represents the physical consensus score;

[0068] These are the weighting coefficients for the experience coverage score, decision stability score, and physical consensus score, respectively, and the weighting coefficients satisfy the following relationship: The relationship between the weighting coefficients is as follows: The experience coverage score is given the highest weight because the reliability of the model's decision depends primarily on the completeness of its training data, i.e., whether the current working condition is covered by the historical sample space. This is the basic premise for the credibility of the model's output. The decision stability score is given the second highest weight to quantify the degree of consensus within the model for the same input, reflecting the consistency of its output under different random subnetwork structures. It is a key indicator for measuring the accuracy of the model. The physical consensus score is given a relatively low weight. Its role is to impose objective constraints of physical laws on intelligent decision-making, serving as a fallback check to prevent the model from producing abnormal outputs that violate the basic principles of thermodynamics.

[0069] The core function of the model cognitive confidence index is to quantitatively assess the decision-making reliability of the model itself in an intelligent temperature control system, providing an objective basis for the allocation of control authority and human-machine collaboration. This index integrates three cognitive dimensions—experience coverage, decision stability, and physical consensus—to convert the model's familiarity with the current operating conditions, internal consensus, and the physical rationality of its output into a quantitative score of 0 to 100. This allows for real-time dynamic evaluation of the model's output credibility. A high score indicates that the model has sufficient experience reserves, high internal confidence, and conforms to physical laws, making its decisions trustworthy and allowing for fully automated control to fully leverage the intelligent model's optimization and energy-saving potential. A low score indicates that the model has encountered unfamiliar operating conditions, internal disagreements exist, or its output violates common sense physics.

[0070] S4: Generate a comprehensive decision index based on the comprehensive cooling safety index and the model cognitive confidence index. The comprehensive decision index is obtained through the following formula:

[0071]

[0072] in: As a comprehensive decision-making index;

[0073] The core function of the comprehensive decision index is to organically fuse two indices with different evaluation dimensions—namely, the intelligent cooling comprehensive safety index, which characterizes the safety status of meat cooling and equipment operation, and the model cognitive confidence index, which characterizes the reliability of the model's own decision-making—through multiplication, generating a single comprehensive score between 0 and 100. This multiplication fusion mechanism reflects the decision-making logic of prioritizing safety: a low score for either index will significantly lower the overall CDI score, ensuring that the system can respond sensitively to both on-site safety and model confidence risks.

[0074] The CDI value is directly mapped to the corresponding control mode range based on preset multi-level decision thresholds, thereby driving the system to achieve adaptive and seamless switching between four control modes: fully automatic, semi-automatic, manual monitoring, and emergency forced switching. Specifically, when the CDI is in the high range, it indicates that the current cooling process is safe and reliable and the model is highly confident in its own decision-making. Based on this, the system authorizes the model to fully execute refined energy-saving control, giving full play to its optimization and regulation capabilities. When the CDI drops to the middle range, it indicates that there is a certain degree of uncertainty or risk. The system switches to semi-automatic mode, where the intelligent model provides control suggestions, which are then executed after manual confirmation.

[0075] When the CDI (Consumer Digital Indicator) enters a low range, it indicates a significant safety risk on-site or insufficient model understanding. The system automatically switches to manual monitoring mode, where the intelligent model only monitors data while manual control remains in place. When the CDI remains below a preset danger threshold, the system forcibly switches to an emergency conservative operation mode and issues a high-level warning requesting immediate manual intervention, prioritizing meat safety. Through this mechanism, the CDI enables dynamic and adaptive allocation of system control authority, maximizing the energy-saving optimization potential of the intelligent model while ensuring food safety.

[0076] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0077] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0078] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0079] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. An energy-saving intelligent temperature control system for cattle and sheep slaughterhouse cooling rooms, characterized in that, Includes the following functional modules: Data sensing and preprocessing module: Collects relevant characteristic data on temperature control safety in the cooling room, including the core temperature of meat products in the cooling room, the opening and closing status and duration of the door, the refrigeration health, and the cooling load; and preprocesses the raw data of the relevant characteristic data, including outlier handling. Cooling Comprehensive Safety Assessment Module: Scores the preprocessed relevant feature data and generates feature data scores; The feature data scores include: meat core temperature safety score, warehouse door disturbance score, system health score, and load characteristic score; and a comprehensive cooling safety index is generated based on the feature data scores. The comprehensive cooling safety index is used to evaluate the overall safety margin and control effect of the cooling system in real time. Model cognitive confidence assessment module: Collects confidence assessment factors from the intelligent temperature control system. These confidence assessment factors include experience coverage score, decision stability score, and physical consensus score. The module also preprocesses the confidence assessment factors, including outlier detection and correction. The preprocessed confidence assessment factors generate a model cognitive confidence index, which is used to quantify the system's familiarity with the current operating conditions and the reliability of its decisions. The integrated decision-making and adaptive mode switching module generates an integrated decision index based on the cooling integrated safety index and the model cognitive confidence index. The integrated decision index is dynamically mapped to a preset decision range by integrating the cooling integrated safety index and the model cognitive confidence index, so as to realize the adaptive switching of four control modes of the cooling process: fully automatic, semi-automatic, manual monitoring and emergency forced. This ensures that when the model's cognition is insufficient, the system can actively degrade its operation and request manual intervention.

2. The energy-saving intelligent temperature control system for cattle and sheep slaughtering cooling rooms according to claim 1, characterized in that: The raw data of the relevant feature data are preprocessed, including outlier handling; a dynamic confidence interval is set for continuously sampled time series data using the statistical 3σ criterion, outliers exceeding the upper and lower limits are marked as outliers and removed, and the data is completed by linear interpolation or forward padding to ensure the continuity and integrity of the data sequence; logical verification is performed in combination with industrial field experience thresholds to further filter out invalid data.

3. The energy-saving intelligent temperature control system for cattle and sheep slaughtering cooling rooms according to claim 2, characterized in that: The preprocessed relevant feature data are quantitatively evaluated to generate corresponding feature data scores; The characteristic data scores include: meat core temperature safety score, warehouse door disturbance score, system health score, and load characteristic score. The meat core temperature safety score is based on the degree of conformity between the real-time monitored meat core temperature and its expected cooling trajectory, used to characterize the current cooling process's compliance with food safety and process requirements. The warehouse door disturbance score quantifies the intensity of disturbance caused by external heat and moisture intrusion to the warehouse thermal environment based on statistical analysis of the frequency of warehouse door opening, the duration of each opening, and the cumulative opening time. The system health score evaluates the operating efficiency and stability of the compressor, heat exchanger, and air cooler by integrating the changing trends of key operating parameters of the refrigeration unit, reflecting the equipment's ability to provide the required cooling capacity. The load characteristic score comprehensively assesses the cooling difficulty of the current batch of meat based on its fat coverage thickness, initial temperature, and inherent carcass specifications, enabling the system to adapt to material differences between different batches.

4. The energy-saving intelligent temperature control system for cattle and sheep slaughtering cooling rooms according to claim 3, characterized in that: A comprehensive cooling safety index is generated based on the feature data scores, and the comprehensive cooling safety index is obtained using the following formula: Where: T represents the safe center temperature of the meat product; D represents the Kumen disturbance component; H represents the system health score; L represents the load characteristic segment; These are the weighting coefficients for the meat product core temperature safety score, the warehouse door disturbance score, the system health score, and the load characteristic score, respectively; and the weighting coefficients satisfy the following relationships: The relationship between the weighting coefficients is as follows: .

5. The energy-saving intelligent temperature control system for cattle and sheep slaughtering cooling rooms according to claim 4, characterized in that: The cooling comprehensive safety index integrates multi-dimensional temperature control safety-related characteristic data into a unified quantitative index, which is used to evaluate the overall safety status of meat cooling process in cattle and sheep slaughtering cooling rooms in real time. By differentially weighting and synthesizing the meat core temperature safety score, system health score, warehouse door disturbance score and load characteristic score, the monitoring information with different physical meanings is normalized to an intuitive scale of 0 to 100, thereby characterizing the degree of compliance of the current cooling conditions with food safety requirements and process objectives. The overall safety index of cooling directly reflects the risk level of the cooling process: a high score indicates that the meat cooling rate is ideal, the equipment is operating stably, and external interference is controllable; a low score indicates that there are potential food safety hazards, equipment abnormalities, or deteriorating operating conditions, requiring timely warnings and human intervention.

6. The energy-saving intelligent temperature control system for cattle and sheep slaughtering cooling rooms according to claim 1, characterized in that: The confidence assessment factors in the intelligent temperature control system are generated in the following ways: The experience coverage score is obtained by comparing the historical training dataset with the feature vector of the current working condition. Specifically, it is obtained by extracting the meat attributes, environmental parameters, and equipment status at the current moment to form a multi-dimensional feature vector. The nearest neighbor search algorithm is used to find the distribution density of its neighboring samples in the feature space of the training data, and then mapping it into a quantitative score representing the model's familiarity with the current scenario. The decision stability score is generated by the uncertainty estimation technology of the neural network. During the model inference stage, multiple forward propagations are performed on the same input, and the statistical variance of the multiple output results is collected and analyzed. The inverse mapping value of the output variance measures the degree of disagreement within the model. The smaller the variance, the higher the decision stability score. The physical consensus score is generated by running the neural network model and the built-in simplified physical reference model in parallel. The same working condition feature is simultaneously input into the two models to obtain the relative deviation between the neural network prediction value and the prediction value of the physical model based on the thermodynamic mechanism. The deviation is mapped into a consensus score representing whether the intelligent decision of the system conforms to the physical laws based on the comparison relationship between the deviation and the preset threshold.

7. The energy-saving intelligent temperature control system for cattle and sheep slaughtering cooling rooms according to claim 6, characterized in that: The preprocessing includes: employing a detection strategy based on a combination of range constraints and rate of change constraints: firstly, a valid range is preset according to the physical meaning of each factor, and values ​​exceeding this range are directly determined as invalid; for continuous time series data, the rate of change between adjacent time points is calculated, and if the change exceeds an empirical threshold, it is considered an abnormal jump; for detected outliers, linear interpolation or previous value preservation is used for correction to ensure the temporal continuity and stability of the confidence assessment factors; at the same time, logical consistency verification is combined to further assist in identifying potential anomalies.

8. The energy-saving intelligent temperature control system for cattle and sheep slaughtering cooling rooms according to claim 7, characterized in that: The model cognitive confidence index is generated based on the preprocessed confidence assessment factors. The model cognitive confidence index is calculated using the following formula: Where: E is the experience coverage score; S represents the decision stability score; P represents the physical consensus score; These are the weighting coefficients for the experience coverage score, decision stability score, and physical consensus score, respectively, and the weighting coefficients satisfy the following relationship: The relationship between the weighting coefficients is as follows: .

9. The energy-saving intelligent temperature control system for cattle and sheep slaughtering cooling rooms according to claim 1, characterized in that: A comprehensive decision index is generated based on the cooling comprehensive safety index and the model cognitive confidence index. The comprehensive decision index is obtained through the following formula: in: It serves as a comprehensive decision-making index.

10. The energy-saving intelligent temperature control system for cattle and sheep slaughtering cooling rooms according to claim 9, characterized in that: The comprehensive decision index integrates the intelligent cooling comprehensive safety index, which reflects the safety of meat cooling and equipment operation, and the model cognitive confidence index, which characterizes the credibility of the intelligent model, through multiplication to form a comprehensive score from 0 to 100. The comprehensive decision index is directly mapped to a preset decision range. When the CDI is high, it indicates that the site is safe and the model is highly confident, and the system authorizes the intelligent model to fully execute refined energy-saving control. When the CDI drops to a low level, it means that there is a safety risk or insufficient model cognition, and the system actively degrades its operation and requests manual intervention.

Images