Compressor rotating speed dynamic optimization system based on deep learning

Abstract

The invention relates to the technical field of industrial automation and intelligent control, in particular to a compressor rotating speed dynamic optimization system based on deep learning, which constructs a closed-loop intelligent optimization architecture through cooperative work of storage, training, testing, acquisition, analysis, execution, recording and verification units. The training unit learns and generates optimization strategy parameters and evaluation criteria based on historical data; the analysis unit compares the target with the actual parameters in real time, and generates an optimization strategy; and the verification unit comprehensively evaluates the optimization effect, calculates an optimization characterization value reflecting the temperature tracking precision and the response speed, and compares the optimization characterization value with an evaluation reference. And when the judgment result is unqualified, the system can automatically diagnose reasons and execute online parameter correction or feed back to the training unit to update the model. Through deep fusion of deep learning and a classical control theory, adaptive and multi-target dynamic optimization of the rotating speed of the compressor is realized, and the temperature control precision and the system response speed are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of industrial automation and intelligent control technology, and in particular to a compressor speed dynamic optimization system based on deep learning. Background Technology

[0002] In industrial refrigeration and HVAC systems, compressors are critical energy-consuming devices, and their operating efficiency directly affects the overall system energy efficiency. Currently, compressor speed control mainly relies on traditional PID (Proportional-Integral-Derivative) control or variable frequency regulation based on fixed rules. These methods have the following limitations: Poor adaptability: Traditional PID parameters are usually fixed values ​​or require manual tuning, making it difficult to adapt to complex operating conditions such as load fluctuations, environmental changes, and equipment aging, leading to decreased control accuracy and slow response. Single optimization objective: Most control strategies only focus on the steady-state tracking of single parameters such as temperature and pressure, lacking multi-objective collaborative optimization of the overall system performance (such as response speed, energy efficiency ratio, and equipment safety). Lack of self-learning and self-evolution capabilities: The system cannot autonomously extract experience from historical operating data, discover optimization patterns, or perform intelligent diagnosis and parameter self-correction when the control effect is poor, resulting in gradual deterioration of energy efficiency over long-term operation.

[0003] Chinese Patent Application No. CN107917558A discloses an intelligent control technology for variable-load evaporative refrigeration cycles in fighter jets. This technology features an optimized and improved novel refrigeration cycle. The invention includes a comprehensive control scheme for the refrigeration system and control strategies for the compressor and expansion valve. This invention employs a method combining fuzzy algorithms, expert control, and PID controllers to control the compressor and expansion valve, achieving effective control of the fighter jet's evaporative refrigeration system. This allows the fighter jet's electronic equipment to reach normal operating temperatures faster and more efficiently. The technology exhibits high reliability and stability, meeting the normal operating requirements of fighter jet electronic equipment.

[0004] However, existing technologies still have the following problems: Lacking the ability to learn and evolve autonomously from actual system operating data, it is unable to adapt to complex and ever-changing working conditions or equipment performance degradation during long-term operation. Summary of the Invention

[0005] To address this, the present invention provides a compressor speed dynamic optimization system based on deep learning, which overcomes the problem in the prior art that it lacks the ability to autonomously and continuously learn and evolve from actual system operating data, and cannot adapt to complex and changing working conditions or equipment performance degradation during long-term operation.

[0006] To achieve the above objectives, this invention provides a compressor speed dynamic optimization system based on deep learning. It includes: Storage unit, used to store historical operating data of the compressor system; The training unit, connected to the storage unit, is used to determine the generation parameters and evaluation benchmarks for the speed optimization process based on the historical operating data. The test unit is used to generate target parameters and perform optimization tests before the actual optimization tests. The data acquisition unit is used to collect the actual operating parameters of the compressor system in real time during the optimization test. An analysis unit, connected to the training unit, the testing unit, and the acquisition unit respectively, is used to compare the actual operating parameters with the target parameters and generate an optimization strategy based on the comparison results. An execution unit, connected to the analysis unit, is used to execute the optimization strategy; A recording unit, connected to the execution unit, is used to record the speed adjustment process data during the execution of the optimization strategy by the execution unit; The verification unit is connected to the recording unit, training unit, and execution unit respectively. It is used to calculate the optimization characterization value based on the speed adjustment process data and the system state before and after optimization, compare the optimization characterization value with the evaluation benchmark, and determine whether the optimization is qualified. When the optimization is determined to be unqualified, a correction instruction is generated and executed.

[0007] Specifically, there are no restrictions on the specific structure of the analyzer; it can be composed of logic components, including field-programmable processors, computers, and microprocessors within computers.

[0008] Furthermore, the verification unit calculates the optimized characterization value based on the speed regulation process data and the system state before and after optimization, including: The verification unit calculates the temperature change value, strategy response time, and execution optimization time based on the speed adjustment process data. The basic optimization characterization value is obtained by weighted summing of the temperature change value, the strategy response time, and the execution optimization time. The additional state parameters of the system during the optimization test are obtained, including the entropy difference between the required temperature and the ambient temperature, the real-time load rate of the compressor, and the system energy efficiency ratio. The basic optimization characterization value is corrected based on the additional state parameters to obtain the corrected optimization characterization value. Wherein, the temperature change value is the absolute difference between the actual change in system temperature before and after optimization and the preset target change value; the strategy response time is the time interval from when the analysis unit generates the optimization strategy to when the execution unit starts executing the optimization strategy; and the optimization execution time is the time consumed by the execution unit to complete the optimization strategy.

[0009] Further, the verification unit compares the optimized characterization value with the evaluation benchmark to determine whether the optimization is qualified, including: If the optimized characterization value is greater than or equal to the preset optimized characterization value, the verification unit determines that the optimization is qualified; If the optimized characterization value is less than the preset optimized characterization value, the verification unit determines that the optimization is unqualified and adjusts the frequency of the inverter.

[0010] Furthermore, the verification unit adjusts the frequency of the inverter, including: Obtain the temperature change difference, which is the difference between the optimized actual temperature and the expected temperature; The frequency of the inverter is adjusted in the same direction based on the temperature change difference, wherein the larger the difference, the greater the adjustment range of the inverter frequency.

[0011] Furthermore, during the process of adjusting the inverter frequency based on the temperature change difference, the verification unit performs inverter frequency correction, including: Real-time acquisition of the current value of the compressor bearing; Based on the current value of the bearing, the frequency of the inverter is negatively corrected; The larger the current value of the bearing, the greater the frequency correction range of the inverter.

[0012] Furthermore, when the verification unit adjusts the inverter frequency based on the temperature change difference, it also coordinates the compressor displacement, including: Obtain the real-time speed of the compressor before and after the frequency of the inverter is adjusted, and calculate the change in speed; Based on the change in rotational speed, a displacement adjustment command is generated; The smaller the change in rotational speed, the greater the displacement adjustment range indicated by the displacement adjustment command.

[0013] Furthermore, after determining that the optimization is unqualified and completing the adjustment of the inverter frequency, if the recalculated optimization characteristic value is still less than the preset optimization characteristic value, the verification unit determines whether to repeatedly adjust the inverter frequency based on the motor efficiency, including: Obtain the current operating efficiency of the compressor motor; Compare the current operating efficiency with a preset efficiency threshold; If the current operating efficiency is greater than the preset efficiency threshold, the verification unit repeatedly adjusts the frequency of the inverter. If the current operating efficiency is less than or equal to the preset efficiency threshold, the verification unit determines that the reason for the failure is the system response performance and adjusts the proportional coefficient of the controller in the execution unit.

[0014] Further, the verification unit adjusts the proportional coefficient of the controller in the execution unit, including: Calculate the difference in optimized characterization values, where the difference in optimized characterization values ​​is the difference between the preset optimized characterization value and the currently recalculated optimized characterization value; The proportional coefficient of the controller is adjusted in the same direction based on the difference in the optimized characterization values; wherein, the larger the difference in the optimized characterization values, the greater the increase in the proportional coefficient.

[0015] Furthermore, during the process of adjusting the proportional coefficient of the controller based on the difference in the optimized characterization value, the verification unit performs proportional coefficient correction, including: After adjusting the proportional coefficient based on the difference in the optimized characterization value, the operating current of the compressor motor is acquired in real time, and the fluctuation range of the operating current is calculated. Based on the fluctuation range of the operating current, the adjusted proportional coefficient is negatively corrected; The greater the fluctuation range of the operating current, the greater the correction range of the proportional coefficient.

[0016] Furthermore, during the process of adjusting the proportional coefficient of the controller based on the difference in the optimized characterization value, the verification unit performs correlated adjustment on the integral time of the controller, including: The integral time of the controller is adjusted according to the increase in the proportional coefficient; wherein, the greater the increase in the proportional coefficient, the greater the increase in the integral time.

[0017] Compared with existing technologies, the beneficial effects of this invention lie in its weighted fusion of three core dimensions: control precision (temperature change value), decision agility (strategy response time), and execution efficiency (execution optimization time). Furthermore, it incorporates operating condition difficulty (entropy difference, load rate) and economic indicators (energy efficiency ratio) for dynamic correction, thereby achieving a comprehensive, fair, and intelligent integrated score for each optimization process. All calculation weights, reference values, and correction rules are autonomously learned by the training unit from historical high-performing operating data, rather than being manually fixed and preset. This ensures that the evaluation criteria always use the system's own best historical performance as a benchmark and can be dynamically adjusted according to changes in equipment status and operating conditions, thus enabling the system to possess self-learning and continuous performance optimization capabilities. The calculated optimization characterization value is a quantifiable and comparable comprehensive performance score. It is not only used to simply determine whether something is qualified or unqualified, but more importantly, its constituent components (such as which indicator has a severe score loss) and absolute values ​​provide precise, data-driven decision inputs for subsequent diagnostic and correction strategies (such as adjusting frequency and PID parameters), significantly improving the efficiency and accuracy of the optimization closed loop.

[0018] Furthermore, by selecting statistical values ​​from historical high-performing samples (such as the top 30% quantile) as a benchmark, the system essentially sets its previously achieved high performance level as the target to be maintained or surpassed. This ensures both the advanced nature and achievability of the standard, avoiding the problems of setting targets that are too high to be achieved or too low to be effective in motivating the system.

[0019] Furthermore, this invention directly uses the real-time deviation of the core control target (temperature) as the adjustment input, ensuring that the control command is purposeful and responsive. It can quickly and effectively compensate for any remaining temperature deviations after optimization, effectively shortening the system's steady-state time and improving control accuracy and dynamic response speed. The adjustment range is not a rigid fixed ratio, but rather a dynamic decision made by the data-driven model provided by the training unit based on the current comprehensive operating conditions (temperature difference, initial state, environment, load) to prevent overshoot and oscillation. This adaptive nonlinear adjustment capability significantly improves control quality and robustness under complex operating conditions.

[0020] Furthermore, this invention proactively intervenes before control commands are executed by real-time monitoring of bearing current, a key early parameter that directly reflects the risk of bearing electrolytic corrosion damage. This elevates fault protection from traditional reactive maintenance or periodic upkeep to predictive maintenance and online proactive protection based on real-time status, effectively preventing unplanned downtime and major equipment damage caused by sudden bearing failure. When a performance enhancement command (increasing frequency) may jeopardize equipment safety (increasing bearing current), an optimal compromise point within a safety boundary is calculated using a risk assessment-based negative correction model. This allows the system to dynamically adjust the aggressiveness of performance targets based on the real-time risk level (current magnitude), achieving a refined and intelligent operation strategy that maximizes performance while ensuring safety.

[0021] Furthermore, this invention dynamically couples and coordinates speed regulation and displacement regulation. Through the organic cooperation of these two physical quantities (n and V), it achieves faster, more precise, and more stable control of the cooling capacity (Q) output, significantly improving the overall regulation capability of complex thermodynamic systems. Addressing the inherent mechanical and electrical inertia of variable frequency speed control systems that causes speed response delays, this invention uses the rapid and significant response of displacement (V) to actively compensate for the sluggish changes in speed (n). When a gradual increase in speed (small change) is detected, the system automatically strengthens the displacement regulation, ensuring that the total cooling capacity output is not constrained by a single bottleneck, effectively shortening the dynamic response time and improving the speedliness of the control system.

[0022] Furthermore, this invention introduces motor operating efficiency as a key state indicator to intelligently diagnose performance bottlenecks. High efficiency indicates an execution problem, and adjustments are continued in the original direction (frequency regulation); low efficiency indicates a control loop response problem, and the system switches to another optimization dimension (PID parameter adjustment). Using the motor efficiency threshold as a decision switch essentially incorporates the motor's overall thermal, electrical, and magnetic state into the core considerations of optimization decisions. When motor efficiency is low, the system can identify that insufficient driving force is not due to control commands, but rather that the motor itself is in a sub-healthy or overloaded state. In this case, the decision to "no longer repeatedly adjust the frequency" is a proactive protection for the motor, preventing accelerated failure caused by applying greater stress to weak points, demonstrating the system's ability to perceive and protect the health of core power equipment.

[0023] Furthermore, this invention establishes a direct mapping relationship between the comprehensive performance gap (optimized characteristic value difference) and the core control parameter (proportional coefficient P), significantly improving the intelligence level and efficiency of parameter tuning. A unique dual adjustment and constraint mechanism is introduced. First, the proportional coefficient is increased based on the performance gap to improve response speed; then, by immediately monitoring the key real-time stability indicator of motor current fluctuations, an excessively large proportional coefficient is negatively fine-tuned (online damping injection). This allows the system to maximize response speed while automatically suppressing potential oscillations and overshoot, ensuring that the control process is both fast and stable, effectively resolving the contradiction between speed and stability in classic PID control. Not only are the P and I parameters adjusted independently, but the I parameter (integral time) is intelligently adjusted in conjunction with the adjustment range of the P parameter. Through this coordinated adjustment, it is ensured that when the proportional action is enhanced to accelerate the response, the integral action can be matched and constrained accordingly, preventing integral saturation, increased overshoot, or decreased steady-state accuracy caused by mismatch between the two, ensuring that the PID controller as a whole always maintains a coordinated and efficient working state. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the compressor speed dynamic optimization system based on deep learning according to the present invention. Figure 2 This is a flowchart of the verification unit in the deep learning-based compressor speed dynamic optimization system of the present invention. Detailed Implementation

[0025] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0026] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0027] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0028] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0029] Please see Figure 1 - Figure 2 As shown, Figure 1 This is a schematic diagram of the compressor speed dynamic optimization system based on deep learning according to the present invention. Figure 2 This is a flowchart of the verification unit in the deep learning-based compressor speed dynamic optimization system of the present invention.

[0030] This invention relates to a deep learning-based compressor speed dynamic optimization system, comprising: Storage unit, used to store historical operating data of the compressor system; The training unit, connected to the storage unit, is used to determine the generation parameters and evaluation benchmarks for the speed optimization process based on the historical operating data. The test unit is used to generate target parameters and perform optimization tests before the actual optimization tests. The data acquisition unit is used to collect the actual operating parameters of the compressor system in real time during the optimization test. An analysis unit, connected to the training unit, the testing unit, and the acquisition unit respectively, is used to compare the actual operating parameters with the target parameters and generate an optimization strategy based on the comparison results. An execution unit, connected to the analysis unit, is used to execute the optimization strategy; A recording unit, connected to the execution unit, is used to record the speed adjustment process data during the execution of the optimization strategy by the execution unit; The verification unit is connected to the recording unit, training unit, and execution unit respectively. It is used to calculate the optimization characterization value based on the speed adjustment process data and the system state before and after optimization, compare the optimization characterization value with the evaluation benchmark, and determine whether the optimization is qualified. When the optimization is determined to be unqualified, a correction instruction is generated and executed.

[0031] Specifically, there are no restrictions on the specific structure of the analyzer; it can be composed of logic components, including field-programmable processors, computers, and microprocessors within computers.

[0032] In this embodiment of the invention, the historical operating data includes, but is not limited to, compressor speed, suction / discharge pressure and temperature, evaporator and condenser temperatures, system load rate, ambient temperature and humidity, inverter output frequency and current, motor operating voltage / current / power / efficiency, bearing temperature and vibration data, refrigerant flow rate, and the corresponding system operating energy efficiency ratio (EER / COP) and the set target temperature parameters.

[0033] Specifically, the verification unit calculates the optimization characterization value based on the speed regulation process data and the system state before and after optimization, including: The verification unit calculates the temperature change value, strategy response time, and execution optimization time based on the speed adjustment process data. The basic optimization characterization value is obtained by weighted summing of the temperature change value, the strategy response time, and the execution optimization time. The additional state parameters of the system during the optimization test are obtained, including the entropy difference between the required temperature and the ambient temperature, the real-time load rate of the compressor, and the system energy efficiency ratio. The basic optimization characterization value is corrected based on the additional state parameters to obtain the corrected optimization characterization value. Wherein, the temperature change value is the absolute difference between the actual change in system temperature before and after optimization and the preset target change value; the strategy response time is the time interval from when the analysis unit generates the optimization strategy to when the execution unit starts executing the optimization strategy; and the optimization execution time is the time consumed by the execution unit to complete the optimization strategy.

[0034] In this embodiment of the invention, the system records the initial temperature before optimization (e.g., room temperature 25℃) and the final temperature after optimization (e.g., 20℃). The preset target for this optimization is a temperature reduction of 10℃, i.e., an expected temperature of 15℃. Therefore, the actual change is from 25℃ to 20℃, a reduction of 5℃; the target change is a reduction of 10℃. The absolute difference between the two is |5℃ - 10℃| = 5℃. This 5℃ is the temperature change value. The smaller it is, the closer the actual cooling effect is to the expected target. The preset target change is determined by the testing unit automatically calculating it based on the current operating conditions (e.g., initial temperature, equipment status) and user needs (e.g., rapid cooling or energy-saving mode) or querying it from a preset rule base. The system records the time when the analysis unit completes the analysis and generates the optimization strategy (e.g., 12:00:00) and the time when the execution unit receives the instruction and starts driving the frequency converter (e.g., 12:00:02). The time difference between these two moments, i.e., 2 seconds, is the strategy response time. It reflects the delay from system decision to action; the shorter the better. The system records the moment the execution unit begins its operation (12:00:02) and the moment the compressor speed is fully adjusted to the new set value, at which the system considers the optimization operation complete (e.g., 12:01:30). This time difference, 1 minute and 28 seconds (88 seconds), is the optimization execution duration. It reflects the speed of the actual adjustment process. The verification unit assigns importance weights to the above three indicators. For example, temperature change value (accuracy) is considered the most important, with a weight of 5; strategy response time (agility) is the second most important, with a weight of 3; and execution optimization time (execution speed) is also important, with a weight of 2. The value of each indicator is multiplied by its weight and then summed. However, to make the results more intuitive, the indicator values ​​are usually standardized first (e.g., divided by a reference value to narrow the values ​​to a similar range) before being weighted, ultimately obtaining a basic optimization characterization value representing overall performance. The higher this value, the better the basic performance of this optimization. The method for determining the weights and standardized reference values: These weights and reference values ​​(such as the maximum acceptable temperature difference, typical response time, etc.) are statistically derived by the training unit in the early stages through analysis of a large number of historical excellent operating cases and are solidified in the system as part of the evaluation benchmark. Calculate the degree of difference between the user's required indoor temperature (e.g., 20℃) and the outdoor ambient temperature (e.g., 35℃). The larger the temperature difference, the heavier the burden on the system to overcome environmental influences, and the greater the optimization difficulty. Compressor real-time load rate: Read the average load of the compressor during the optimization process, for example, the current load is 80% of the rated capacity. System energy efficiency ratio: Calculated as the ratio of cooling capacity generated by the system to electrical energy consumed during this optimization process. If the system still achieves a good basic optimization performance value under unfavorable operating conditions such as high entropy difference (e.g., demand 20℃ but ambient temperature 35℃) and high load rate (e.g., 80%), then this optimization is particularly valuable.The verification unit will therefore increase the basic optimization characterization value. If the system energy efficiency ratio is very low, it means that although the temperature is adjusted quickly and controlled accurately, the energy consumption is high and the economy is poor. The verification unit will therefore decrease the basic optimization characterization value.

[0035] This invention weights and integrates three core dimensions: control precision (temperature change value), decision agility (strategy response time), and execution efficiency (execution optimization time). It also introduces operating condition difficulty (entropy difference, load rate) and economic indicators (energy efficiency ratio) for dynamic correction, thereby achieving a comprehensive, fair, and intelligent integrated score for each optimization process. All calculation weights, reference values, and correction rules are autonomously learned by the training unit from historical high-performing operating data, rather than being manually fixed and preset. This ensures that the evaluation criteria always use the system's own best historical performance as a benchmark and can be dynamically adjusted according to changes in equipment status and operating conditions, thus enabling the system to have self-learning and continuous performance optimization capabilities. The calculated optimization characterization value is a quantifiable and comparable comprehensive performance score. It is not only used to simply determine whether something is qualified or unqualified, but more importantly, its constituent components (such as which indicator has a serious score loss) and absolute values ​​provide precise, data-driven decision inputs for subsequent diagnostic and correction strategies (such as adjusting frequency and PID parameters), significantly improving the efficiency and accuracy of the optimization closed loop.

[0036] Specifically, the verification unit compares the optimized characterization value with the evaluation benchmark to determine whether the optimization is qualified, including: If the optimized characterization value is greater than or equal to the preset optimized characterization value, the verification unit determines that the optimization is qualified; If the optimized characterization value is less than the preset optimized characterization value, the verification unit determines that the optimization is unqualified and adjusts the frequency of the inverter.

[0037] In this embodiment of the invention, the preset optimization characterization value is determined based on statistical learning (offline benchmark) of historical excellent samples, including: the training unit selects optimization process records that have been verified to have excellent comprehensive performance (such as high energy efficiency, fast response, and stable temperature control) from the historical running data of the storage unit in the initial stage or periodically, as a sample set, calculates the optimization characterization value of each optimization process in the sample set, and then takes the statistical characteristics of these values ​​(e.g., the average value or the value in the top 30 percentile) as the preset optimization characterization value.

[0038] This invention uses the statistical values ​​of historical high-performing samples (such as the top 30% quantile) as a benchmark. Essentially, the system sets its previously achieved high performance level as the target to be maintained or surpassed. This ensures both the standard's advancement and its attainability, avoiding the problems of setting targets that are too high to achieve or too low to provide motivation.

[0039] Specifically, the verification unit adjusts the frequency of the inverter, including: Obtain the temperature change difference, which is the difference between the optimized actual temperature and the expected temperature; The frequency of the inverter is adjusted in the same direction based on the temperature change difference, wherein the larger the difference, the greater the adjustment range of the inverter frequency.

[0040] In this embodiment of the invention, assuming a refrigeration scenario, after an optimization process, the verification unit determines it to be unqualified and initiates a frequency adjustment program to obtain the difference: Expected temperature: The target temperature set for this optimization is 20℃. Actual temperature: After optimization, the actual system temperature measured by the acquisition unit is 22℃; Calculate the difference: Temperature change difference = Actual temperature - Expected temperature = 22℃ - 20℃ = +2℃. This positive difference indicates that the temperature did not meet the standard, and the actual temperature is higher than expected. Execute adjustment: Current frequency: In the execution unit, the current output frequency of the inverter is 45Hz (corresponding to a certain speed of the compressor).

[0041] Determine the adjustment direction: Since the difference is positive, the actual temperature is too high. To lower the temperature, the cooling capacity needs to be enhanced, which means increasing the compressor speed. Therefore, the adjustment direction is the same, adjusting the frequency in the direction of increase. Determine the adjustment range: The system has a preset correspondence between temperature difference and frequency adjustment. For example, the rule is set as follows: for every 1°C deviation in temperature difference, the basic frequency adjustment is 5% of the current frequency. Calculate the adjustment amount: The current temperature difference is +2°C, so the basic adjustment range is 2 × 5% = 10% of the current frequency. The current frequency is 45Hz, so 10% is 4.5Hz. Execute the adjustment command: The verification unit sends a command to the execution unit (inverter) to adjust the target frequency from 45Hz to 49.5Hz (45 + 4.5). In this embodiment, the frequency adjustment is not based on a simple fixed proportional relationship, but on a dynamic, nonlinear mapping relationship learned from historical data. Its core logic is as follows, based on the principle of temperature difference feedback control. Theoretical basis: This is a direct application of closed-loop negative feedback control in the temperature field. The detected temperature difference (deviation) is the input signal of the control system, and frequency adjustment is the output action. The larger the difference, the larger the adjustment amplitude, reflecting the idea of ​​proportional control, aiming to quickly eliminate steady-state error. Purpose: To quickly reduce the gap between the actual state and the target state. Core basis: Historical operating condition-performance mapping model (provided by the training unit). Dynamic mapping table / function: The training unit analyzes massive amounts of historical data to learn the most effective frequency adjustment amount under complex operating conditions such as different initial frequencies, temperature difference magnitudes, ambient temperatures, and system loads. It generates one or a set of intelligent "temperature difference-frequency adjustment amount" mapping rules as the core basis for verifying the unit's adjustment. Source of rules in the example: The rule of "adjusting by 5% for every 1℃ of temperature difference" in the aforementioned example is a simplified linear approximation of the output of this mapping model under the current specific operating conditions (such as medium ambient temperature and medium load). In more complex models, the adjustment amplitude may not be linear. For example, when the temperature difference is small (such as 0.5℃), fine-tuning (adjusting by 2%) is used; when the temperature difference is large (such as 3℃), a bolder adjustment (adjusting by 8%) is used to avoid overshoot. Supporting Basis: System safety and efficiency boundary constraints. Frequency Upper and Lower Limits: Any adjustments must be made within the safe operating frequency range allowed by the inverter (e.g., 30-60Hz). The adjustment algorithm will use this as a hard constraint. Optimal Energy Efficiency Range: The training unit will also learn the optimal energy efficiency operating range of the compressor under different loads. If the current frequency is already at the edge of this range, even if the temperature difference is large, the adjustment algorithm will tend to use a gentler adjustment range, or combine it with other means (such as adjusting the displacement) to keep the system operating in the high energy efficiency range as much as possible.

[0042] This invention directly uses the real-time deviation of the core control target (temperature) as the adjustment input, ensuring that the control command is clear in purpose and direct in response. It can quickly and effectively compensate for temperature deviations that still exist after optimization, effectively shortening the time for the system to reach steady state and improving control accuracy and dynamic response speed. The adjustment range is not a rigid fixed ratio, but rather a dynamic decision made by a data-driven model provided by the training unit based on the current comprehensive operating conditions (temperature difference, initial state, environment, load) to prevent overshoot and oscillation. This adaptive nonlinear adjustment capability significantly improves control quality and robustness under complex operating conditions.

[0043] Specifically, during the process of adjusting the inverter frequency based on the temperature change difference, the verification unit performs inverter frequency correction, including: Real-time acquisition of the current value of the compressor bearing; Based on the current value of the bearing, the frequency of the inverter is negatively corrected; The larger the current value of the bearing, the greater the frequency correction range of the inverter.

[0044] In this embodiment of the invention, the verification unit has calculated a preliminary instruction to increase the inverter frequency from 45Hz to 49.5Hz based on a temperature difference of +2℃. Before issuing this instruction, the system executes a frequency correction process. Bearing current is collected: While executing the adjustment instruction, the system monitors the bearing's conduction current (i.e., the current flowing through the bearing, a key parameter reflecting the risk of bearing erosion and lubrication status) in real time using a current sensor installed on the compressor's main shaft bearing. Assuming the real-time bearing current is 0.8 amperes (A) when the frequency is about to increase to 49.5Hz, the corrected current is 0.8 amperes (A). Negative correction is performed: Correction rules: The system has preset safety correction rules based on bearing current. For example: a bearing current below 0.5A is considered a safe range and no correction is performed; a current between 0.5A and 1.0A is a warning range, and the target frequency increment is reduced proportionally to the percentage of current exceeding 0.5A. Correction amount is calculated: the current bearing current is 0.8A, and the portion exceeding the safe threshold of 0.5A is 0.8 - 0.5 = 0.3A. The warning range width is 1.0 - 0.5 = 0.5A, therefore the percentage exceeding this range is 0.3 / 0.5 = 60%. The original planned frequency increment was 49.5Hz - 45Hz = 4.5Hz. The negative correction was 4.5Hz × 60% = 2.7Hz. The final frequency was determined by subtracting the correction from the original target frequency, resulting in the final execution frequency: 49.5Hz - 2.7Hz = 46.8Hz. The verification unit ultimately issued the following instruction to the inverter: adjust the frequency to 46.8Hz, not the initial 49.5Hz. The frequency increase was significantly suppressed. The adjustment process was based on the fundamental principle of bearing current and its safety threshold. Physical correlation: In inverter-driven motors, due to factors such as high-frequency switching and common-mode voltage, shaft voltage is generated across the motor bearings. When this voltage exceeds the insulation strength of the bearing grease, bearing current will be generated. This current causes electrolytic corrosion of the bearing raceways and balls, creating tiny melting pits, accelerating bearing wear, generating noise and vibration, and ultimately leading to premature bearing failure. It is one of the main failure modes of variable frequency drive systems. Safety Threshold: The magnitude of the bearing current directly reflects the level of electrolytic corrosion risk. The higher the current value, the exponentially increasing rate of electrolytic corrosion damage, and the drastically higher risk of equipment damage. Therefore, the bearing current must be controlled within safe limits. Core Basis: A causal relationship model between frequency, voltage, and bearing current. Influence Relationship: The output frequency of the frequency converter is closely related to the voltage / current waveform characteristics. Generally, under the same load, increasing the output frequency tends to exacerbate shaft voltage and bearing current. The training unit learns and establishes a predictive model for frequency adjustment commands to estimate bearing current changes using historical data. Correction Logic: When the verification unit issues a frequency increase command, the model predicts the potential bearing current increase. If the predicted value or real-time monitored value approaches or exceeds the safety threshold, a correction is triggered.The essence of this correction is to sacrifice some performance by reducing the target frequency increase in exchange for the safety and lifespan of critical equipment components (bearings). The rule is that the greater the current, the greater the correction (suppression) magnitude; this is a non-linear safety protection strategy that matches the intensity of electrolytic corrosion risk. Supporting factors include system reliability and total lifecycle cost considerations. Avoiding unplanned downtime: Bearing damage is the main cause of forced compressor shutdowns and major overhauls. Actively protecting bearings through current correction is primarily based on ensuring the continuity and reliability of system operation, avoiding significant losses due to minor issues. Reducing maintenance costs: The cost and downtime of preventative bearing replacement are far lower than the cost of repairing the entire compressor after a failure. This correction strategy is based on the economic principle of minimizing total lifecycle costs.

[0045] This invention proactively intervenes before control commands are executed by real-time monitoring of bearing current, a key early indicator of bearing electrolytic corrosion risk. This elevates fault protection from traditional reactive maintenance or periodic upkeep to predictive maintenance and online proactive protection based on real-time status, effectively preventing unplanned downtime and major equipment damage caused by sudden bearing failure. When performance-enhancing commands (increasing frequency) may jeopardize equipment safety (increasing bearing current), a risk-assessment-based negative correction model calculates an optimal compromise point within a safety boundary. This allows the system to dynamically adjust the aggressiveness of performance targets based on the real-time risk level, i.e., the current magnitude, achieving a refined and intelligent operation strategy that maximizes performance while ensuring safety.

[0046] Specifically, when the verification unit adjusts the inverter frequency based on the temperature change difference, it also coordinates the compressor displacement, including: Obtain the real-time speed of the compressor before and after the frequency of the inverter is adjusted, and calculate the change in speed; Based on the change in rotational speed, a displacement adjustment command is generated; The smaller the change in rotational speed, the greater the displacement adjustment range indicated by the displacement adjustment command.

[0047] In this embodiment of the invention, the verification unit, considering both temperature difference and bearing current, ultimately determined to adjust the inverter frequency from 45Hz to 46.8Hz. Speed ​​change was obtained: Before adjustment: At the instant before the frequency adjustment command was issued, the system measured the compressor's real-time speed as 2800 revolutions per minute (RPM) (corresponding to 45Hz) using an encoder. After adjustment: After the inverter completed frequency adjustment and the speed stabilized, the system again measured the compressor's real-time speed as 2920 RPM (corresponding to 46.8Hz).

[0048] Calculate the change: Speed ​​change = |2920 - 2800| = 120 RPM. This value reflects the actual change in speed brought about by this frequency adjustment. Generate and execute the displacement adjustment command: Adjustment rule: The system has a preset coordinated strategy of speed change and displacement adjustment. For example, the strategy is: set a reference speed change (e.g., 200 RPM). When the actual speed change is less than this reference, it indicates that the speed increase is gradual; at this time, the compressor displacement should be increased significantly (e.g., increase the suction valve opening or adjust the slide valve position) to fully utilize the increased speed capacity to increase the cooling capacity. The adjustment range is proportional to the difference between the reference change and the actual change. Calculate the displacement adjustment range: Reference change = 200 RPM. Actual change = 120 RPM.

[0049] Difference = 200 - 120 = 80 RPM. Assuming the rule is: for every 40 RPM difference, the opening of the displacement regulating valve will increase by 10% (from the current opening). Therefore, the displacement adjustment range this time is (80 / 40) × 10% = an increase of 20% in opening. Command generation and execution: The verification unit generates a displacement adjustment command and sends it to the displacement adjustment mechanism in the execution unit (such as the control unit of the electronic expansion valve, suction regulating valve, or variable displacement mechanism), instructing it to increase the displacement by 20%. The basis for the adjustment process, fundamentally based on the compressor's working principle and the cooling capacity formula. Theoretical relationship: The theoretical cooling capacity (Q) of the compressor is approximately proportional to the rotational speed (n) × displacement (V), i.e., Q ∝ n × V. The cooling capacity is determined by both rotational speed and displacement. Adjustment logic: When increasing the rotational speed (n) by increasing the frequency, theoretically, the displacement (V) should also increase simultaneously to achieve the expected linear increase in cooling capacity (Q) and realize the control target (lowering the temperature). If only the rotational speed is increased without increasing the displacement, the improvement in cooling capacity will be significantly reduced. Core Basis: The complementary model of rotational speed response characteristics and displacement adjustment. Problem Identification: Variable frequency drive systems have inertia; speed changes are not instantaneous. "Small speed change" may indicate two situations: first, the system has high inertia and heavy load, resulting in a slow speed response; second, the frequency adjustment command itself is conservative (possibly due to limitations such as bearing current), resulting in a small increase in the target rotational speed. Complementary Strategy: In either case, relying solely on increasing rotational speed to achieve a leap in cooling capacity faces bottlenecks or delays. At this point, actively and significantly increasing the displacement (V) becomes a key complementary means to rapidly increase cooling capacity (Q). This is the inherent logic that the smaller the speed change, the larger the displacement adjustment range. Rapid and significant displacement adjustment compensates for or coordinates with the insufficient speed change, ensuring that the overall cooling capacity output keeps pace with the optimization strategy. Model Source: This nonlinear collaborative mapping relationship is learned by the training unit through analyzing the relationship between rotational speed, displacement, and final cooling effect in historical data. It is essentially a dynamic feedforward compensator. Supporting Basis: Energy efficiency optimization and system stability. Preventing over-adjustment and oscillation: If the engine speed increases rapidly (large change), the system already has the capacity to quickly increase cooling capacity. Increasing the displacement significantly at this point can easily lead to overshoot in cooling capacity and drastic fluctuations in system pressure, thus reducing energy efficiency and stability. Therefore, when the engine speed changes significantly, displacement adjustment should be more cautious (smaller increments) or even temporarily suspended, relying primarily on the engine speed's own adjustment. Maintaining the optimal pressure ratio: The compressor needs a suitable compression ratio to operate in its high-efficiency range. One of the underlying goals of coordinated adjustment of engine speed and displacement is to enable the system to quickly transition and stabilize at the new high-efficiency operating point, rather than simply pursuing increased cooling capacity output.

[0050] This invention dynamically couples and coordinates speed regulation and displacement regulation. Through the organic cooperation of these two physical quantities (n and V), it achieves faster, more precise, and more stable control of the cooling capacity (Q) output, significantly improving the overall regulation capability of complex thermodynamic systems. Addressing the inherent mechanical and electrical inertia of variable frequency speed control systems that causes speed response delays, this invention uses the rapid and significant response of displacement (V) to actively compensate for the sluggish changes in speed (n). When a gradual increase in speed (small change) is detected, the system automatically strengthens the displacement regulation, ensuring that the total cooling capacity output is not constrained by a single bottleneck, effectively shortening the dynamic response time and improving the speedliness of the control system.

[0051] Specifically, after determining that the optimization is unqualified and completing the adjustment of the inverter frequency, if the recalculated optimization characteristic value is still less than the preset optimization characteristic value, the verification unit determines whether to repeatedly adjust the inverter frequency based on the motor efficiency, including: Obtain the current operating efficiency of the compressor motor; Compare the current operating efficiency with a preset efficiency threshold; If the current operating efficiency is greater than the preset efficiency threshold, the verification unit repeatedly adjusts the frequency of the inverter. If the current operating efficiency is less than or equal to the preset efficiency threshold, the verification unit determines that the reason for the failure is the system response performance and adjusts the proportional coefficient of the controller in the execution unit.

[0052] In this embodiment of the invention, the analysis unit generates a strategy, and the execution unit adjusts the frequency. After optimization, the verification unit calculates an optimized characterization value of 65. Initial judgment: The training unit issued a preset optimized characterization value of 75 for the current operating condition. 65 < 75, therefore it is deemed unqualified, and a first-level correction (frequency adjustment) is performed: The verification unit adjusts the frequency from 45Hz to 46.8Hz based on the temperature difference (e.g., +2℃) and bearing current. Second evaluation: After the frequency adjustment stabilizes, the system runs a short evaluation cycle again, and the verification unit recalculates the optimized characterization value, obtaining 68. Second judgment: The new value 68 is still less than the preset value 75, and the optimization is still unqualified. At this point, the system initiates an advanced diagnostic process. Obtaining the current motor efficiency: Under the operating state of 46.8Hz, the system collects the motor's input electrical power and estimated output mechanical power in real time, calculating the current motor operating efficiency to be 92%. Comparison with the threshold: The training unit issued a preset efficiency threshold of 90% for the current operating condition and motor model. Because the current efficiency of 92% is greater than the threshold of 90%, the verification unit determines that the motor's energy conversion capability is good and is not a performance bottleneck. The problem may lie in the insufficient strength or precision of frequency adjustment. Decision: Repeatedly adjust the inverter frequency. The system will initiate a new round of frequency adjustment, potentially with a larger amplitude, based on the current 46.8Hz (while incorporating safety corrections such as bearing current), attempting to further approach the target. If the measured motor efficiency in this state is 88% (less than or equal to the 90% threshold), the diagnosis is: the motor itself is operating in a low-efficiency state, with insufficient ability to convert electrical energy into mechanical energy. At this point, blindly increasing the frequency to drive the motor will not only have limited cooling effects but will also lead to further motor overload, increased heat generation, and further efficiency decline, creating a vicious cycle. Decision: The verification unit determines that the root cause of the failure is poor "system response performance," meaning insufficient speed and accuracy in the generation and execution of control commands (control loop), failing to fully utilize the potential of the existing equipment. Therefore, instead of repeatedly adjusting the frequency, the proportional gain of the controller in the execution unit will be adjusted to optimize the performance of the control loop itself. The preset optimization characteristic value is dynamically determined by the training unit. During the initial learning phase, the system will collect a large number of historical data segments marked as "operating well" under different operating conditions. For each typical operating condition (such as ambient temperature range or load range), the optimized characteristic values ​​of these excellent samples are calculated, and their average or median is taken as the initial preset optimized characteristic value for that condition. During operation, this value will be slowly and periodically adjusted upwards according to the long-term performance improvement target of the system to drive continuous optimization. The preset efficiency threshold is related to the characteristics of the motor itself. The determination basis includes: motor rated efficiency curve: referring to the efficiency-load characteristic curve provided by the motor manufacturer, the higher level of motor efficiency that can be achieved in the typical operating load range (such as 50%-80% load) (e.g., 95%-98% of rated efficiency) is selected as the benchmark.Historical High-Efficiency Operation Data: The training unit analyzes historical data to determine the statistical distribution of motor efficiency during periods when the optimized performance values ​​were satisfactory. The lower quantile of the efficiency value (e.g., the 25th percentile) is set as the "preset efficiency threshold." This means that when historical performance is good, the motor efficiency has at least a 75% probability of being higher than this value. If the current efficiency is lower than this historical lower limit, the motor is considered to be in an abnormal state or poorly controlled. Dynamic Adjustment: This threshold can also be fine-tuned based on motor operating time (aging) and ambient temperature (affecting heat dissipation).

[0053] This invention intelligently diagnoses performance bottlenecks by introducing motor operating efficiency as a key performance indicator. High efficiency indicates an execution problem, and adjustments are continued in the original direction (frequency regulation); low efficiency indicates a control loop response problem, and the system switches to another optimization dimension (PID parameter adjustment). Using the motor efficiency threshold as a decision switch essentially incorporates the motor's overall thermal, electrical, and magnetic state into the core considerations of optimization decisions. When motor efficiency is low, the system can identify that insufficient driving force is not due to control commands, but rather to the motor itself being in a sub-healthy or overloaded state. In this case, the decision to stop repeated frequency regulation is a proactive protection for the motor, preventing accelerated failure caused by applying greater stress to weak points, demonstrating the system's ability to perceive and protect the health of core power equipment.

[0054] Specifically, the verification unit adjusts the proportional coefficient of the controller in the execution unit, including: Calculate the difference in optimized characterization values, where the difference in optimized characterization values ​​is the difference between the preset optimized characterization value and the currently recalculated optimized characterization value; The proportional coefficient of the controller is adjusted in the same direction based on the difference in the optimized characterization values; wherein, the larger the difference in the optimized characterization values, the greater the increase in the proportional coefficient.

[0055] In this embodiment of the invention, the difference in optimized characterization values ​​is calculated as follows: Preset optimized characterization value: The qualified standard value issued by the training unit under the current operating condition is 75. Current optimized characterization value: The latest value recalculated after the first frequency correction is 68. Calculate the difference: Optimized characterization value difference = Preset value - Current value = 75 - 68 = 7. This difference represents the comprehensive performance gap between the current system's actual performance and the expected target. Adjust the proportional coefficient based on the difference: Current proportional coefficient: The current proportional coefficient (P) of the PID controller in the execution unit is 2.0. Adjustment rule: The system has a preset mapping relationship of "performance gap - proportional coefficient adjustment amount". For example, a simplified linear rule is: for every 1 point increase in the difference, the proportional coefficient increases by 5% of the current value. Calculate the adjustment amount: The difference is 7. Adjustment amplitude = 7 × 5% = 35% of the current coefficient value. Adjustment amount = 2.0 × 35% = 0.7. Execute the adjustment: The verification unit issues an instruction to the PID controller of the execution unit to adjust the proportional coefficient from 2.0 to 2.7 (2.0 + 0.7). This is an increased unidirectional adjustment. The fundamental basis for the adjustment process is the causal relationship between the proportional control principle and the performance gap. Control principle: In PID control, the proportional coefficient (P) directly determines the system's response strength to deviation. The larger the P value, the faster and more "powerful" the controller's response to the error, aiming to reduce the deviation more quickly. Problem mapping: Currently, a large difference in the optimized characteristic value (a comprehensive performance deviation) indicates that the overall system response is slow or the accuracy is insufficient. From a control perspective, this often means insufficient gain in the control action, i.e., the proportional coefficient is too small, causing the system to be unable to quickly approach the target. Core basis: A self-tuning model that maps high-level performance indicators to low-level parameters. Limitations of traditional tuning: Traditional PID parameter tuning (such as the critical proportional method) relies on the observation and formula calculation of the step response of a single controlled variable (such as temperature), which is cumbersome and unsuitable for changing operating conditions. Intelligent mapping of this invention: The training unit establishes an innovative mapping relationship by learning historical data: comprehensive performance gap, suggested proportional coefficient adjustment amount. It bypasses complex intermediate analysis and trial and error, directly transforming the achievement of the highest-level, composite optimization objective (optimization representation value) into a direct adjustment command for the lowest-level, core control parameter (proportional coefficient P). The adjustment logic: "The larger the difference, the greater the increase" is based on the fact that large performance gaps usually require the controller to take more decisive and forceful actions to reverse the situation, thus requiring a larger increase in the proportional coefficient. This relationship is non-linear and determined by the learned model. Auxiliary basis: The trade-off between response speed and stability. Performance orientation: The primary basis for this adjustment is to improve system response speed and control accuracy to narrow the performance gap. Increasing the proportional coefficient is the most direct way to achieve this goal. Stability constraints: The adjustment process is not unlimited. Stability boundary knowledge is embedded in the training unit when building the mapping model.It knows that under current operating conditions, exceeding a certain upper limit of the proportional gain will cause system oscillations. Therefore, its recommended "increase" is the maximum effective adjustment that ensures stability. If the gap remains large and the system remains stable after the initial adjustment, the model will recommend further increases in subsequent adjustments.

[0056] Specifically, during the process of adjusting the proportional coefficient of the controller based on the difference in the optimized characterization value, the verification unit performs proportional coefficient correction, including: After adjusting the proportional coefficient based on the difference in the optimized characterization value, the operating current of the compressor motor is acquired in real time, and the fluctuation range of the operating current is calculated. Based on the fluctuation range of the operating current, the adjusted proportional coefficient is negatively corrected; The greater the fluctuation range of the operating current, the greater the correction range of the proportional coefficient.

[0057] In this embodiment of the invention, after adjusting the proportional gain of the PID controller from 2.0 to 2.7, the system enters the proportional gain correction stage. Current fluctuation amplitude is acquired: During a short observation period (e.g., 10 seconds) after the coefficient is adjusted to 2.7, the system collects the three-phase current of the motor at a high frequency (e.g., 100 times per second). The standard deviation of the effective current value relative to its average value during this period is calculated, resulting in a current fluctuation amplitude of 3.5 amperes (A). Negative correction is performed: Correction rule: The preset rule is that for every 1 ampere exceeding the reference value (e.g., 1.0A) of the current fluctuation amplitude, the newly adjusted proportional gain is reduced to 2% of its value. Correction amount is calculated: The reference value is 1.0A, the current fluctuation is 3.5A, exceeding 3.5-1.0=2.5A. The reduction ratio is 2.5×2%=5%. Correction amount = 2.7×5%=0.135. Final coefficient is determined: The verification unit issues a final instruction to correct the proportional gain to 2.7-0.135=××2.565××, and locks this value. Adjustment Basis, Core Basis (Stability Assurance): The fluctuation amplitude of the motor current directly reflects the dynamic stability of the system. Larger fluctuations indicate that the control action may be too aggressive, causing oscillations in the controlled object or drastic changes in motor torque. Correction Logic (Overshoot Suppression): According to classical control theory, while an excessively large proportional gain can accelerate the response, it can also easily lead to overshoot and oscillations. Therefore, based on the real-time stability indicator of current fluctuation amplitude, a reverse fine-tuning (negative correction) is performed on the newly increased proportional gain, which is a form of online damping injection. Larger fluctuations mean a stronger oscillation trend, requiring a larger adjustment of the proportional gain to suppress it. Fundamental Purpose: To achieve the optimal balance between pursuing rapid response (increasing the proportional gain) and ensuring smooth operation (suppressing current oscillations), thus achieving fast and stable control.

[0058] Specifically, during the process of adjusting the proportional coefficient of the controller based on the difference in the optimized characterization value, the verification unit performs correlated adjustment on the integral time of the controller, including: The integral time of the controller is adjusted according to the increase in the proportional coefficient; wherein, the greater the increase in the proportional coefficient, the greater the increase in the integral time.

[0059] In this embodiment of the invention, after the system completes the final adjustment of the proportional coefficient (e.g., from 2.0 to 2.565), it immediately enters the correlation adjustment of the integral time. The increase in the proportional coefficient is calculated as follows: The original proportional coefficient was 2.0. The final proportional coefficient after adjustment is 2.565. The increase in the proportional coefficient is (2.565-2.0) / 2.0×100%=28.25%. The correlation adjustment of the integral time is executed: Current integral time: The current integral time (Ti) of the PID controller is 5.0 seconds. Correlation rule: The preset rule is: the increase in the proportional coefficient is directly used as the percentage increase in the integral time. The adjustment amount is calculated: The increase in the proportional coefficient is 28.25%. The adjustment amount of the integral time = 5.0 seconds × 28.25% = 1.4125 seconds. The final integral time is determined: The verification unit issues a command to the PID controller to adjust the integral time from 5.0 seconds to 6.4125 seconds. The fundamental basis for adjustment lies in the coupling relationship between PID parameters and classic tuning rules: In PID control, the proportional (P) and integral (I) actions need to work together. The proportional coefficient (P) is responsible for rapid response and eliminating most of the deviation, while the integral time (Ti) is responsible for ultimately eliminating the steady-state error. There is an inherent relationship between the two. According to classic engineering tuning rules such as the Ziegler-Nichols method, when the proportional coefficient increases, it is generally recommended to simultaneously increase the integral time (i.e., weaken the integral action) to prevent the integral component from accumulating too quickly due to the enhanced proportional action, leading to integral saturation or increased overshoot. The core basis is intelligent correlation mapping based on system response.

[0060] This invention establishes a direct mapping relationship between the overall performance gap (optimized characteristic value difference) and the core control parameter (proportional coefficient P), significantly improving the intelligence and efficiency of parameter setting. A unique dual adjustment and constraint mechanism is introduced. First, the proportional coefficient is increased based on the performance gap to improve response speed; then, by immediately monitoring the key real-time stability indicator of motor current fluctuations, an excessively large proportional coefficient is negatively fine-tuned (online damping injection). This allows the system to maximize response speed while automatically suppressing potential oscillations and overshoot, ensuring that the control process is both fast and stable, effectively resolving the contradiction between speed and stability in classic PID control. Not only are the P and I parameters adjusted independently, but the I parameter (integral time) is intelligently adjusted in conjunction with the adjustment range of the P parameter. Through this coordinated adjustment, it is ensured that when the proportional action is enhanced to accelerate the response, the integral action can be matched and constrained accordingly, preventing integral saturation, increased overshoot, or decreased steady-state accuracy caused by mismatch between the two, ensuring that the PID controller as a whole always maintains a coordinated and efficient working state.

[0061] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0062] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A deep learning-based compressor speed dynamic optimization system, characterized in that, The method comprises the following steps: a storage unit for storing historical operation data of the compressor system; a training unit connected to the storage unit, for determining the generation parameters of the speed optimization process and the evaluation benchmark based on the historical operation data; a test unit for generating target parameters before actual optimization test and conducting optimization test; a collection unit for collecting real-time actual operation parameters of the compressor system during the optimization test; an analysis unit connected to the training unit, the test unit and the collection unit, for comparing the actual operation parameters with the target parameters and generating an optimization strategy according to the comparison result; an execution unit connected to the analysis unit, for executing the optimization strategy; a recording unit connected to the execution unit, for recording the speed regulation process data during the execution of the optimization strategy by the execution unit; a verification unit connected to the recording unit, the training unit and the execution unit, for calculating an optimization characteristic value based on the speed regulation process data and the system state before and after optimization, comparing the optimization characteristic value with the evaluation benchmark, and determining whether the optimization is qualified; and generating and executing a correction instruction when the optimization is determined to be unqualified.

2. The deep learning-based compressor speed dynamic optimization system of claim 1, wherein, The verification unit calculates the optimization characteristic value based on the speed regulation process data and the system state before and after optimization, which comprises: The verification unit calculates the temperature change value, the strategy response time and the execution optimization time based on the speed regulation process data; The temperature change value, the strategy response time and the execution optimization time are weighted and summed to obtain a basic optimization characteristic value; obtaining additional state parameters of the system during the optimization test, the additional state parameters including the entropy difference of the demand temperature and the environment temperature, the real-time load rate of the compressor and the system energy efficiency ratio; based on the additional state parameters, the basic optimization characteristic value is corrected to obtain a corrected optimization characteristic value; wherein the temperature change value is the absolute difference between the actual change amount and the preset target change amount of the system temperature before and after optimization, the strategy response time is the time interval from the generation of the optimization strategy by the analysis unit to the start of the execution of the optimization strategy by the execution unit, and the execution optimization time is the time consumed by the execution unit to complete the optimization strategy.

3. The deep learning-based compressor speed dynamic optimization system of claim 1, wherein, The verification unit compares the optimization characteristic value with the evaluation benchmark to determine whether the optimization is qualified, which comprises: If the optimization characteristic value is greater than or equal to the preset optimization characteristic value, the verification unit determines that the optimization is qualified; If the optimization characteristic value is less than the preset optimization characteristic value, the verification unit determines that the optimization is unqualified, and adjusts the frequency of the frequency converter.

4. The deep learning-based compressor speed dynamic optimization system of claim 3, wherein, The verification unit adjusts the frequency of the frequency converter, which comprises: obtaining a temperature change difference value, which is the difference between the actual temperature after optimization and the expected temperature; adjusting the current frequency of the frequency converter in the same direction according to the temperature change difference value, wherein the larger the difference value is, the larger the adjustment range of the frequency of the frequency converter is.

5. The deep learning-based compressor speed dynamic optimization system of claim 4, wherein, The verification unit executes the frequency correction of the frequency converter during the adjustment of the frequency of the frequency converter according to the temperature change difference value, which comprises: real-time collection of the current value of the compressor bearing; The frequency of the frequency converter is negatively corrected based on the current value of the bearing. The greater the current value of the bearing, the greater the correction range of the frequency of the frequency converter.

6. The deep learning-based compressor speed dynamic optimization system of claim 4, wherein, The verification unit cooperatively adjusts the displacement of the compressor when adjusting the frequency of the frequency converter according to the temperature change difference, including: The real-time rotating speed of the compressor before and after the adjustment of the frequency of the frequency converter is obtained, and the rotating speed change is calculated. The displacement adjustment instruction is generated according to the rotating speed change. The greater the rotating speed change, the greater the displacement adjustment range indicated by the displacement adjustment instruction.

7. The deep learning-based compressor speed dynamic optimization system of claim 4, wherein, After determining that the optimization is unqualified and completing the adjustment of the frequency of the frequency converter, if the re-calculated optimization characteristic value is still less than the preset optimization characteristic value, whether to repeatedly adjust the frequency of the frequency converter is determined based on the motor efficiency, including: The current operating efficiency of the compressor motor is obtained. The current operating efficiency is compared with a preset efficiency threshold. If the current operating efficiency is greater than the preset efficiency threshold, the verification unit repeatedly adjusts the frequency of the frequency converter. If the current operating efficiency is less than or equal to the preset efficiency threshold, the verification unit determines that the unqualified reason is the system response performance, and adjusts the proportional coefficient of the controller in the execution unit.

8. The deep learning-based compressor speed dynamic optimization system of claim 1, wherein, The verification unit adjusts the proportional coefficient of the controller in the execution unit, including: The optimization characteristic value difference is calculated, which is the difference between the preset optimization characteristic value and the current re-calculated optimization characteristic value. The proportional coefficient of the controller is positively adjusted according to the optimization characteristic value difference. The greater the optimization characteristic value difference, the greater the increase range of the proportional coefficient.

9. The deep learning-based compressor speed dynamic optimization system of claim 8, wherein, The verification unit adjusts the proportional coefficient of the controller according to the optimization characteristic value difference, including: After adjusting the proportional coefficient based on the optimization characteristic value difference, the operating current of the compressor motor is obtained in real time, and the fluctuation range of the operating current is calculated. The adjusted proportional coefficient is negatively corrected based on the fluctuation range of the operating current. The greater the fluctuation range of the operating current, the greater the correction range of the proportional coefficient.

10. The deep learning-based compressor speed dynamic optimization system of claim 8, wherein, The verification unit adjusts the integral time of the controller in association with the adjustment of the proportional coefficient of the controller according to the optimization characteristic value difference, including: The integral time of the controller is adjusted according to the increase range of the proportional coefficient. The greater the increase range of the proportional coefficient, the greater the increase range of the integral time. The verification unit adjusts the integral time of the controller in association with the adjustment of the proportional coefficient of the controller according to the optimization characteristic value difference, including: The integral time of the controller is adjusted according to the increase range of the proportional coefficient. The greater the increase range of the proportional coefficient, the greater the increase range of the integral time.

Images