Method and device for optimizing the efficiency of a furnace fan based on magnetic transmission

Abstract

The application discloses a furnace fan efficiency logical control optimization method and device based on magnetic transmission, and relates to the field of equipment automatic control. A magnetic transmission efficiency factor is calculated based on a digital twin model. When the magnetic transmission efficiency factor is lower than a set threshold, a magnetic field compensation mechanism is activated to adjust a magnetic field coupling state to maintain a set transmission efficiency. Based on a furnace temperature distribution, a computational fluid dynamics simplified model is used to real-time inverse a target air flow mode, a minimum necessary air volume and an optimal air flow incident angle to generate a fan performance demand curve. An optimization function is established with the minimum energy consumption, the maximum temperature uniformity and the minimum equipment fatigue loss as targets. In combination with constraint conditions, an improved multi-objective genetic algorithm is used to solve optimal control parameters online. The application realizes real-time evaluation and active compensation of the magnetic transmission state, accurately quantifies the influence of multiple factors on the transmission efficiency, significantly improves the stability and efficiency of the magnetic transmission, and reduces the loss caused by the magnetic performance attenuation.

Description

Technical Field

[0001] This invention relates to the field of equipment automation control technology, and in particular to a method and apparatus for optimizing the efficiency logic control of furnace fans based on magnetic drive. Background Technology

[0002] Magnetic drive technology, due to its contactless transmission and excellent sealing performance, is widely used in industrial furnace fan drive scenarios, effectively avoiding the wear and leakage problems of traditional mechanical transmission. However, existing magnetic drive control schemes for furnace fans mostly adopt a single speed regulation mode, lacking the ability to dynamically adapt to the core state of magnetic drive and the flow field requirements inside the furnace, resulting in many technical limitations.

[0003] In terms of magnetic drive, high-temperature environments easily lead to the decay of the magnetic properties of permanent magnets. Changes in air gap, radial offset, and magnetic field waveform distortion can significantly reduce transmission efficiency. Existing technologies cannot assess the transmission state under multi-parameter coupling in real time and can only passively adjust after a significant drop in efficiency, failing to compensate for performance losses in advance. In terms of flow field control, traditional solutions set the fan operating state based on fixed process parameters, which cannot infer airflow demand in real time based on the temperature distribution inside the furnace, easily leading to problems such as excessive energy consumption or uneven temperature distribution.

[0004] Meanwhile, existing control strategies fail to achieve synergistic optimization of energy consumption, process accuracy, and equipment lifespan, focusing primarily on single-objective control, leading to accelerated equipment fatigue and insufficient operational stability. Furthermore, typical disturbances such as furnace door opening and closing, and workpiece entry and exit easily disrupt the furnace's internal operating balance; traditional control responses are lag-dependent, further impacting process stability and transmission efficiency. These issues constrain the energy efficiency improvement and long-term reliable operation of the magnetic drive furnace fan system. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method and apparatus for optimizing the efficiency logic control of furnace fans based on magnetic drive. The technical solution adopted is as follows:

[0006] The method for optimizing the efficiency logic control of furnace fans based on magnetic drive includes the following steps:

[0007] Step 1: Establish a high-fidelity digital twin model of the magnetic drive system, deploy a multi-dimensional sensor array, and collect multi-dimensional operating parameters of the magnetic drive system in real time.

[0008] Step 2: Calculate the magnetic transmission efficiency factor based on the digital twin model. When the magnetic transmission efficiency factor is lower than the set threshold, activate the magnetic field compensation mechanism to adjust the magnetic field coupling state to maintain the set transmission efficiency.

[0009] Step 3: Based on the temperature distribution inside the furnace, a simplified computational fluid dynamics model is used to invert the target airflow pattern, minimum required air volume, and optimal airflow injection angle in real time, generating a fan performance requirement curve.

[0010] Step 4: Establish an optimization function with the objectives of minimizing energy consumption, maximizing temperature uniformity, and minimizing equipment fatigue loss. Combined with the constraints, use an improved multi-objective genetic algorithm to solve for the optimal control parameters online.

[0011] Step 5: Implement decoupled control of speed and torque, combine predictive disturbance rejection strategy to offset disturbances under typical operating conditions, and execute life balance mode to achieve dynamic adaptation of fan operating status.

[0012] Optionally, step 6 is also included: building a running case database, performing deep optimization training at set time intervals, and correcting the digital twin model parameters and optimizing the weight coefficients.

[0013] Optionally, the multi-dimensional operating parameters of the magnetic drive system include magnet temperature, actual air gap size, radial offset, and magnetic field waveform data.

[0014] Optionally, step 2, calculating the magnetic transmission efficiency factor, includes the following sub-steps:

[0015] Step 21: Calculate the temperature decay coefficient based on the multi-dimensional operating parameters. Air gap influence function Radial offset influence function Harmonic distortion rate influence function ;

[0016] This reflects the decrease in magnetic flux density and the risk of irreversible demagnetization caused by high temperatures;

[0017] Based on the inverse relationship between magnetic field strength and the square of the distance, edge effect correction is considered;

[0018] The influence of the offset direction is taken into account, reflecting the attenuation effect of radial misalignment on torque transmission;

[0019] This reflects the increased eddy current losses and torque pulsation caused by magnetic field waveform distortion;

[0020] Step 22, set the reference transmission efficiency Multiplying the combined magnetic transmission efficiency factor by the influence functions calculated in step 21 yields the overall magnetic transmission efficiency factor. .

[0021] Magnetic transmission efficiency factor Calculated using the following formula:

[0022] ;

[0023] The calculation formula is:

[0024] ;

[0025] Where β is the magnet temperature coefficient, a typical value for NdFeB permanent magnets: 0.001-0.00121 / ℃. It is the real-time average temperature of the magnet, measured by an embedded thermocouple. This is the reference operating temperature of the magnet, typically taken as 80℃. It is a temperature gradient correction factor, obtained by subtracting the minimum temperature from the maximum temperature and then dividing the result by the average temperature. It is the thermal time constant of the magnet, which is determined by the material and cooling conditions, with a typical value of 300-600s;

[0026] Calculated using the following formula:

[0027] ,when ;

[0028] ,when ;

[0029] in This is the nominal air gap, the design value for the air gap, typically 3-8mm. It is the actual air gap, measured in real time by an eddy current sensor. This is the air gap sensitivity coefficient, with a typical value of 1.2-2.0 mm. It is a level one warning air gap, with a value of 1.3 times. , It is a level 2 alarm air gap, with a value of 1.3 times. ;

[0030] The calculation formula is:

[0031] ;

[0032] ; ;

[0033] Where Δx and Δy are the radial offsets in the X and Y directions, respectively;

[0034] r is the effective radius of the magnetic ring. and These are the components of the magnetic coupling force in the X and Y directions, which are calculated by inversion using a magnetic sensor array.

[0035] The calculation formula is:

[0036] ;

[0037] THD is the total harmonic distortion rate of the magnetic field, obtained by FFT analysis of the magnetic flux density waveform. It is the main harmonic frequency. It is the fundamental frequency. It is the harmonic sensitivity coefficient, with a typical value of 0.015-0.025.

[0038] Optionally, the simplified computational fluid dynamics model in step 3 is obtained through offline high-precision simulation training, and the fan performance requirement curve is a mapping relationship curve of speed, torque and flow rate.

[0039] Optionally, the multi-objective optimization function in step 4 is:

[0040] ;

[0041] in It is a minimization objective function used to balance three major objectives: energy consumption, temperature uniformity, and equipment fatigue loss. P is the real-time power, and Prated is the rated power. The standard deviation of furnace temperature. The fatigue coefficient is based on the cumulative damage theory, and α, β, and γ are dynamic weighting coefficients.

[0042] The improved multi-objective genetic algorithm is the improved NSGA-II algorithm, with an online optimization iteration cycle of once every 5 minutes.

[0043] Optionally, the dynamic weighting coefficients are adaptively adjusted according to the process stage: α=0.4, β=0.5, γ=0.1 for the heating stage; α=0.6, β=0.3, γ=0.1 for the heat preservation stage; and α=0.3, β=0.3, γ=0.4 for the cooling stage.

[0044] Optionally, the predictive disturbance rejection strategy in step 5 involves establishing a typical disturbance pattern library, which includes at least furnace door opening and closing and workpiece entry and exit, and pre-adjusting the fan speed 300ms-500ms in advance to counteract the impact of airflow disturbances on the furnace working conditions.

[0045] The life balancing mode monitors the workload of each magnetic pole unit, switches the dominant magnetic pole region at set intervals, and achieves uniform magnetic pole wear through a circumferential offset of ±3°.

[0046] A furnace fan efficiency logic control optimization device based on magnetic drive is used to realize the furnace fan efficiency logic control optimization method based on magnetic drive, including an intelligent magnetic drive unit, a multi-sensor fusion acquisition module and an edge computing control box.

[0047] The intelligent magnetic drive unit includes a partitioned adjustable permanent magnet array, multiple independently controlled electromagnetic compensation coils, and an active air gap adjustment mechanism, which are used to respond to optimized control commands and adjust the magnetic field coupling state and fan operating parameters.

[0048] The multi-sensor fusion acquisition module includes an array of Hall sensors, an infrared temperature measurement matrix, and an ultrasonic anemometer, which are used to collect multi-dimensional parameters such as magnetic field, temperature, and airflow in real time and transmit them to the edge computing control box.

[0049] The edge computing control box has a built-in processor, a real-time optimization algorithm acceleration card, and a redundant dual-controller architecture. It pre-stores digital twin models and multi-objective optimization algorithms for performing data processing, optimization calculations, and control command output.

[0050] Optionally, the air gap active adjustment mechanism is driven by piezoelectric ceramics, and an array of Hall sensors are arranged at equal intervals in the circumferential direction to simultaneously monitor the magnitude and direction of the magnetic induction intensity.

[0051] In summary, the present invention has at least one of the following beneficial technical effects:

[0052] This invention provides a method and device for optimizing the efficiency logic control of furnace fans based on magnetic drive, enabling real-time evaluation and active compensation of the magnetic drive state, accurately quantifying the impact of multiple factors on the drive efficiency, significantly improving the stability and efficiency of magnetic drive, and reducing losses caused by magnetic performance decay.

[0053] By using flow field demand prediction feedforward control, the furnace temperature distribution demand is dynamically adapted, the airflow pattern and fan operating parameters are optimized, the furnace temperature uniformity is significantly improved, and ineffective energy consumption is reduced.

[0054] A multi-objective collaborative optimization strategy balances energy consumption, process accuracy, and equipment lifespan, and combines dynamic weight adjustment to adapt to different process stages, thereby improving the overall system performance.

[0055] Predictive disturbance rejection control and life balance mode work together to reduce the impact of typical operating condition disturbances on furnace conditions, uniform magnetic pole wear, and extend the service life of key components.

[0056] The self-learning iterative mechanism continuously optimizes model parameters and control strategies to adapt to different furnace types and process requirements, enhances system versatility and control accuracy, and improves overall equipment operating efficiency and reliability. Attached Figure Description

[0057] Figure 1This is a flowchart illustrating the furnace fan efficiency logic control optimization method based on magnetic drive according to the present invention.

[0058] Figure 2 This is a comparison curve of energy consumption at different process stages in a specific embodiment of the present invention;

[0059] Figure 3 This is a comparison diagram of the uniformity of furnace temperature distribution in specific embodiments of the present invention;

[0060] Figure 4 This is a graph showing the change in magnetic transmission efficiency over time in a specific embodiment of the present invention. Detailed Implementation

[0061] The present invention will be further described in detail below with reference to the accompanying drawings.

[0062] This invention discloses a method and apparatus for optimizing the efficiency logic control of furnace fans based on magnetic drive.

[0063] Reference Figures 1-4 Example 1, a method for optimizing the efficiency logic control of an in-furnace fan based on magnetic drive, includes the following steps:

[0064] Step 1: Establish a high-fidelity digital twin model of the magnetic drive system, deploy a multi-dimensional sensor array, and collect multi-dimensional operating parameters of the magnetic drive system in real time.

[0065] Step 2: Calculate the magnetic transmission efficiency factor based on the digital twin model. When the magnetic transmission efficiency factor is lower than the set threshold, activate the magnetic field compensation mechanism to adjust the magnetic field coupling state to maintain the set transmission efficiency.

[0066] Step 3: Based on the temperature distribution inside the furnace, a simplified computational fluid dynamics model is used to invert the target airflow pattern, minimum required air volume, and optimal airflow injection angle in real time, generating a fan performance requirement curve.

[0067] Step 4: Establish an optimization function with the objectives of minimizing energy consumption, maximizing temperature uniformity, and minimizing equipment fatigue loss. Combined with the constraints, use an improved multi-objective genetic algorithm to solve for the optimal control parameters online.

[0068] Step 5: Implement decoupled control of speed and torque, combine predictive disturbance rejection strategy to offset disturbances under typical operating conditions, and execute life balance mode to achieve dynamic adaptation of fan operating status.

[0069] Example 2 also includes step 6, which involves building a running case database, performing deep optimization training at set time intervals, and correcting the digital twin model parameters and optimizing the weight coefficients.

[0070] Example 3: The multi-dimensional operating parameters of the magnetic drive system include magnet temperature, actual air gap size, radial offset, and magnetic field waveform data.

[0071] By adopting the above technical solution, the principle of Example 1 lies in constructing a closed-loop control logic from sensing to regulation, forming a complete efficiency optimization link. Step 1 replicates the physical characteristics of the magnetic drive system using a high-fidelity digital twin model, while relying on a multi-dimensional sensor array to achieve comprehensive capture of core operating parameters, providing a data foundation and model support for subsequent control, and solving the problems of incomplete parameter sensing and large deviations between the model and reality in traditional control. Step 2 calculates the magnetic drive efficiency factor based on the model, which is essentially quantifying the comprehensive impact of multi-parameter coupling on transmission performance. Active compensation is triggered by threshold judgment, breaking the traditional passive adjustment mode, and maintaining stable transmission efficiency by dynamically correcting the magnetic field coupling state.

[0072] Step 3 uses the furnace temperature distribution as input and leverages computational fluid dynamics to simplify the model and invert the airflow demand. The core is to transform the temperature field information into executable operating parameters for the fan, achieving a precise match between flow field requirements and fan operation, avoiding energy waste or temperature unevenness caused by fixed parameter control. Step 4 establishes a multi-objective optimization function to balance the three core requirements of energy consumption, temperature uniformity, and equipment fatigue wear. Combined with the online optimization capability of an improved multi-objective genetic algorithm, it quickly solves for the optimal control parameters under different operating conditions, overcoming the limitations of single-objective control. Step 5 employs speed-torque decoupling control to achieve flexible adaptation of fan operating states. Simultaneously, predictive disturbance rejection strategies proactively offset typical operating condition disturbances, and a lifespan balancing mode reduces wear on key components, ensuring system stability and continuity.

[0073] The principle of Example 2 is to introduce a self-learning iterative mechanism to compensate for the lack of adaptability of fixed control strategies. By building a database of operational cases, control parameters and operational effect data under different operating conditions are accumulated, providing sample support for model optimization. The deep optimization training performed at set time intervals is essentially based on historical operational data and offline verification to correct the parameters of the digital twin model to improve modeling accuracy, adjust the optimization weight coefficients to adapt to changes in operating conditions, and enable the control strategy to have continuous iterative upgrade capabilities, gradually improving the system's adaptability and control accuracy under different furnace types and processes.

[0074] The principle of Example 3 is to clarify the selection logic of core sensing parameters to ensure the accuracy of magnetic drive state assessment and control decisions. Magnet temperature directly affects the magnetic properties of the permanent magnet and is a key factor leading to transmission efficiency decay; the actual air gap size determines the magnetic field coupling strength, directly related to transmission torque and efficiency; radial offset causes magnetic field misalignment, exacerbating torque pulsation and losses; magnetic field waveform data reflects the degree of magnetic field distortion, affecting eddy current losses and transmission stability. Selecting these four types of parameters as multi-dimensional operating parameters comprehensively covers the core factors affecting magnetic drive performance, providing accurate parameter inputs for magnetic drive efficiency factor calculation, magnetic field compensation, and subsequent optimized control, ensuring the reliability and effectiveness of the entire control logic.

[0075] Example 4, step 2, calculating the magnetic transmission efficiency factor includes the following sub-steps:

[0076] Step 21: Calculate the temperature decay coefficient based on the multi-dimensional operating parameters. Air gap influence function Radial offset influence function Harmonic distortion rate influence function ;

[0077] This reflects the decrease in magnetic flux density and the risk of irreversible demagnetization caused by high temperatures;

[0078] Based on the inverse relationship between magnetic field strength and the square of the distance, edge effect correction is considered;

[0079] The influence of the offset direction is taken into account, reflecting the attenuation effect of radial misalignment on torque transmission;

[0080] This reflects the increased eddy current losses and torque pulsation caused by magnetic field waveform distortion;

[0081] Step 22, set the reference transmission efficiency Multiplying the combined magnetic transmission efficiency factor by the influence functions calculated in step 21 yields the overall magnetic transmission efficiency factor. .

[0082] Magnetic transmission efficiency factor Calculated using the following formula:

[0083] ;

[0084] The calculation formula is:

[0085] ;

[0086] Where β is the magnet temperature coefficient, a typical value for NdFeB permanent magnets: 0.001-0.00121 / °C. It is the real-time average temperature of the magnet, measured by an embedded thermocouple. This is the reference operating temperature of the magnet, typically taken as 80℃. It is a temperature gradient correction factor, obtained by subtracting the minimum temperature from the maximum temperature and then dividing the result by the average temperature. It is the thermal time constant of the magnet, which is determined by the material and cooling conditions, with a typical value of 300-600s;

[0087] Calculated using the following formula:

[0088] ,when ;

[0089] ,when ;

[0090] in This is the nominal air gap, the design value for the air gap, typically 3-8mm. It is the actual air gap, measured in real time by an eddy current sensor. This is the air gap sensitivity coefficient, with a typical value of 1.2-2.0 mm. It is a level one warning air gap, with a value of 1.3 times. , It is a level 2 alarm air gap, with a value of 1.3 times. ;

[0091] The calculation formula is:

[0092] ;

[0093] ; ;

[0094] Where Δx and Δy are the radial offsets in the X and Y directions, respectively;

[0095] r is the effective radius of the magnetic ring. and These are the components of the magnetic coupling force in the X and Y directions, which are calculated by inversion using a magnetic sensor array.

[0096] The calculation formula is:

[0097] ;

[0098] THD is the total harmonic distortion rate of the magnetic field, obtained by FFT analysis of the magnetic flux density waveform. It is the main harmonic frequency. It is the fundamental frequency. It is the harmonic sensitivity coefficient, with a typical value of 0.015-0.025.

[0099] By adopting the above technical solution, a multi-factor coupled magnetic transmission efficiency quantification system is established, enabling accurate calculation of the transmission efficiency factor and providing a scientific basis for the active compensation mechanism. The core principle lies in transforming the four key factors affecting magnetic transmission efficiency—temperature, air gap, radial offset, and harmonic distortion—into quantifiable functional indicators, and then integrating them through coupling to obtain a comprehensive efficiency factor, thus overcoming the fuzziness and empirical limitations of traditional efficiency assessment.

[0100] Step 21 constructs a dedicated quantification function to address the mechanisms of action of each influencing factor. The temperature decay coefficient is calculated based on the characteristics of permanent magnet magnetic properties changing with temperature, combined with temperature gradient correction and thermal time constant, to accurately reflect the weakening of magnetic flux density and the risk of irreversible demagnetization caused by high temperatures, while taking into account real-time temperature and the hysteresis effect of heat conduction. The air gap influence function is based on the inverse square relationship between magnetic field strength and distance, introducing edge effect correction, and setting different calculation logics according to the air gap size to adapt to the characteristic changes of the air gap from the normal range to the warning and alarm states, accurately quantifying the degree of impact of air gap fluctuations on transmission efficiency.

[0101] The radial offset influence function combines geometric calculations and mechanical analysis. First, the degree of offset is determined by the radial offset amount. Then, the offset direction is calculated using the magnetic coupling force component. Finally, trigonometric functions and coefficient corrections quantify the attenuation effect of radial misalignment on torque transmission, taking into account both the magnitude and direction of the offset. The harmonic distortion rate influence function, based on magnetic field waveform analysis, reflects the degree of waveform distortion through the total harmonic distortion rate. Combined with corrections based on the ratio of harmonic frequency to fundamental frequency, it accurately quantifies the increase in eddy current losses and torque pulsation caused by magnetic field distortion, closely reflecting the actual operating characteristics of the magnetic field.

[0102] Step 22 uses a product integration method to obtain the comprehensive magnetic transmission efficiency factor. Since the effects of temperature, air gap, radial offset, and harmonic distortion on transmission efficiency are independent yet cumulative, the product form accurately reflects the combined effect of multiple factors. Based on the benchmark transmission efficiency, multiplying by each influencing function yields the actual efficiency factor, upgrading efficiency assessment from qualitative judgment to quantitative calculation. This provides precise numerical support for subsequent threshold judgment, ensuring that the triggering timing and adjustment range of the magnetic field compensation mechanism are scientifically reasonable, and effectively maintaining stable transmission efficiency.

[0103] In Example 5, the simplified computational fluid dynamics model in step 3 is obtained through offline high-precision simulation training, and the fan performance requirement curve is a mapping relationship curve of speed, torque and flow rate.

[0104] By adopting the above technical solution, the accuracy and real-time performance of flow field inversion are balanced, and a precise mapping relationship between flow field requirements and fan operating parameters is established, providing a reliable basis for subsequent optimized control. Its core logic is to optimize the computational fluid dynamics model through offline high-precision simulation training, while simultaneously transforming flow field requirements into performance curves that the fan can directly execute, solving the problems of time-consuming online flow field calculations and unintuitive parameter transformations.

[0105] The simplified computational fluid dynamics model employs the principle of offline high-precision simulation training to balance model complexity and computational efficiency. In offline scenarios, a complete high-precision computational fluid dynamics model can be used to simulate flow field characteristics under different process parameters and furnace operating conditions, accumulating massive flow field data samples. By fitting and reducing the order of the sample data, redundant computational steps are eliminated, core flow field influencing factors are retained, and a simplified model is trained. This design inherits the predictive accuracy of high-precision simulation while significantly reducing online computation time, meeting the control requirements for real-time inversion of the furnace flow field and ensuring rapid output of target flow field parameters.

[0106] The fan performance requirement curve is designed based on the principle of mapping relationship between speed, torque, and flow rate, serving as a bridge between flow field requirements and fan control commands. The target airflow pattern, minimum required airflow, and optimal airflow injection angle obtained from flow field inversion need to be converted into executable operating parameters for the fan. Speed ​​directly determines airflow rate and velocity, while torque is related to fan drive capability and energy consumption. The mapping relationship among these three can accurately reflect the optimal operating state of the fan under different flow field requirements. Through this curve, abstract flow field targets can be quickly transformed into specific speed and torque control commands, providing a clear execution basis for subsequent multi-objective optimization and dynamic regulation, ensuring precise matching between flow field requirements and fan operation.

[0107] Example 6, the multi-objective optimization function in step 4 is:

[0108] ;

[0109] in It is a minimization objective function used to balance three major objectives: energy consumption, temperature uniformity, and equipment fatigue loss. P is the real-time power, and Prated is the rated power. The standard deviation of furnace temperature. The fatigue coefficient is based on the cumulative damage theory, and α, β, and γ are dynamic weighting coefficients.

[0110] The improved multi-objective genetic algorithm is the improved NSGA-II algorithm, with an online optimization iteration cycle of once every 5 minutes.

[0111] Example 7: The dynamic weighting coefficients are adaptively adjusted according to the process stage: heating stage α=0.4, β=0.5, γ=0.1; heat preservation stage α=0.6, β=0.3, γ=0.1; cooling stage α=0.3, β=0.3, γ=0.4.

[0112] By adopting the above technical solutions, a scientific multi-objective optimization system is constructed, relying on suitable algorithms to efficiently solve for optimal control parameters, balancing control accuracy and real-time performance. The core logic is to construct the objective function through a weighted summation method, quantifying and integrating the three core objectives, while selecting an improved algorithm and setting a reasonable iteration cycle to solve the problems of multi-objective conflict and online optimization efficiency.

[0113] The multi-objective optimization function is designed using a weighted sum principle to achieve a dynamic balance between three major objectives: energy consumption, temperature uniformity, and equipment fatigue loss. The ratio of real-time power to rated power directly reflects the energy consumption level, the standard deviation of furnace temperature quantifies temperature uniformity, and the fatigue coefficient derived from cumulative damage theory characterizes the equipment's wear state. By dynamically weighting and integrating these three indicators, the multi-objective optimization is transformed into a single-objective optimization problem. This approach retains the core influence of each objective while allowing for weight adjustments to adapt to different operating conditions, preventing the optimization of a single objective from leading to the deterioration of other performance aspects.

[0114] The principle behind selecting the improved NSGA-II algorithm and setting an online optimization iteration cycle of once every five minutes is to balance optimization accuracy and system response speed. The improved NSGA-II algorithm has good multi-objective optimization capabilities and can quickly converge to the Pareto optimal solution set, meeting the parameter solution requirements under multiple constraints. The iteration cycle of once every five minutes allows sufficient time to complete algorithm optimization and parameter calculation, while also responding promptly to changes in furnace operating conditions, avoiding control fluctuations caused by excessive iteration frequency, and preventing parameter lag caused by excessively long iteration intervals.

[0115] By adapting weighting coefficients to the characteristics of different process stages, multi-objective optimization becomes more aligned with actual production needs, enhancing the targeting and effectiveness of control strategies. The core logic is based on the differences in core requirements across different process stages, pre-setting corresponding weighting schemes to dynamically adjust optimization objectives and ensure optimal overall system performance at each stage.

[0116] The heating stage prioritizes temperature uniformity because it requires rapidly establishing a uniform furnace temperature field to lay the foundation for subsequent processes, while also balancing energy consumption control and equipment wear. The holding stage increases the energy consumption weight because the temperature field is relatively stable at this stage, shifting the core requirement to reducing ineffective energy consumption while maintaining temperature uniformity and protecting basic equipment. The cooling stage increases the weight for equipment fatigue wear because temperature gradients and airflow fluctuations during cooling can exacerbate component wear; therefore, prioritizing equipment safety and lifespan while balancing energy consumption and temperature stability is crucial. By adaptively adjusting the weights in stages, the optimization function consistently aligns with the core processes of each stage, achieving optimal overall performance across the entire process.

[0117] Example 8: The predictive disturbance rejection strategy in step 5 establishes a typical disturbance pattern library, which includes at least furnace door opening and closing and workpiece entry and exit. The fan speed is pre-adjusted 300ms-500ms in advance to counteract the impact of airflow disturbance on the furnace working conditions.

[0118] The life balancing mode monitors the workload of each magnetic pole unit, switches the dominant magnetic pole region at set intervals, and achieves uniform magnetic pole wear through a circumferential offset of ±3°.

[0119] By adopting the above technical solutions, the dynamic control execution capability of step 5 is optimized. Through advance disturbance rejection and load balancing design, the stability of furnace operating conditions and the service life of key components are taken into account, further improving the practicality and reliability of closed-loop control. The core logic is to address the problems of typical operating condition disturbances and uneven magnetic pole wear by adopting predictive control and intelligent load distribution strategies respectively, to make up for the shortcomings of traditional control response lag and excessive component wear.

[0120] The principle of predictive disturbance mitigation strategies lies in transforming passive response into proactive prediction, proactively offsetting the impact of disturbances on furnace conditions. Furnace door opening and closing, and workpiece entry and exit are typical high-frequency disturbances within the furnace. The resulting airflow fluctuations can be accurately captured, and establishing a pattern library enables rapid matching of disturbance types, intensities, and impact ranges. Pre-adjusting fan speed 300 to 500 milliseconds in advance allows for matching the disturbance propagation speed with fan response delay. By specifically adjusting the speed to compensate for airflow loss or impact, the damage to the furnace's temperature and flow fields is weakened at the source, avoiding passive correction after significant fluctuations in operating conditions and ensuring process stability.

[0121] The principle of the lifespan balancing mode is to achieve uniform wear of magnetic pole units through intelligent load scheduling, thereby extending the overall service life of the magnetic drive unit. During magnetic drive, the working load of each magnetic pole unit varies, and long-term operation can easily lead to excessive localized wear, affecting transmission stability. By monitoring the load status of each magnetic pole unit in real time and periodically switching the dominant magnetic pole region, high loads can be dynamically transferred to different areas, avoiding a single magnetic pole from bearing high-intensity forces for extended periods. The ±3° circumferential offset design further fine-tunes the force position of the magnetic poles, ensuring uniform wear distribution. This maintains magnetic field coupling strength and transmission efficiency while balancing fatigue wear of each magnetic pole unit, significantly extending the service life of key components and reducing equipment maintenance costs.

[0122] Example 9: A furnace fan efficiency logic control optimization device based on magnetic drive, used to implement a furnace fan efficiency logic control optimization method based on magnetic drive, including an intelligent magnetic drive unit, a multi-sensor fusion acquisition module, and an edge computing control box;

[0123] The intelligent magnetic drive unit includes a partitioned adjustable permanent magnet array, multiple independently controlled electromagnetic compensation coils, and an active air gap adjustment mechanism, which are used to respond to optimized control commands and adjust the magnetic field coupling state and fan operating parameters.

[0124] The multi-sensor fusion acquisition module includes an array of Hall sensors, an infrared temperature measurement matrix, and an ultrasonic anemometer, which are used to collect multi-dimensional parameters such as magnetic field, temperature, and airflow in real time and transmit them to the edge computing control box.

[0125] The edge computing control box has a built-in processor, a real-time optimization algorithm acceleration card, and a redundant dual-controller architecture. It pre-stores digital twin models and multi-objective optimization algorithms for performing data processing, optimization calculations, and control command output.

[0126] In Example 10, the air gap active adjustment mechanism is driven by piezoelectric ceramics, and an array of Hall sensors are arranged at equal intervals in the circumferential direction to simultaneously monitor the magnitude and direction of magnetic induction intensity.

[0127] The following specific embodiments illustrate the implementation principle of the present invention:

[0128] This paper describes a magnetic drive furnace fan system used in a large industrial heat treatment furnace. It employs a magnetic drive-based furnace fan efficiency logic control optimization method and device to achieve dynamic optimization control of fan efficiency. The detailed implementation details are as follows.

[0129] Device configuration:

[0130] The optimized device used in this embodiment includes an intelligent magnetic drive unit, a multi-sensor fusion acquisition module, and an edge computing control box.

[0131] The intelligent magnetic drive unit is equipped with a zoned adjustable permanent magnet array, multiple independently controlled electromagnetic compensation coils, and an active air gap adjustment mechanism. The active air gap adjustment mechanism uses piezoelectric ceramic drive to respond to optimized control commands and adjust the magnetic field coupling state and fan operating parameters.

[0132] The multi-sensor fusion acquisition module includes an array of Hall effect sensors, an infrared temperature measurement matrix, and an ultrasonic anemometer. The array of Hall effect sensors, with multiple measurement points evenly spaced in a circular direction, can simultaneously monitor the magnitude and direction of magnetic field strength. The infrared temperature measurement matrix is ​​used to collect temperature distribution data within the furnace, and the ultrasonic anemometer is used to collect airflow parameters. The core function of the multi-sensor fusion acquisition module is to acquire multi-dimensional parameters such as magnetic field, temperature, and airflow in real time and transmit them to the edge computing control box.

[0133] The edge computing control box incorporates an industrial-grade processor, a real-time optimization algorithm acceleration card, and a redundant dual-controller architecture. It also includes a pre-stored high-fidelity digital twin model of the magnetic drive system and multi-objective optimization algorithms. Its core functions are data processing, optimization calculations, and control command output.

[0134] Control method execution steps:

[0135] Step 1: System Modeling and Parameter Awareness

[0136] The edge computing control box establishes a high-fidelity digital twin model of the magnetic drive system based on pre-stored model data. The multi-sensor fusion acquisition module collects multi-dimensional operating parameters of the magnetic drive system in real time according to deployment requirements. These parameters include magnet temperature, actual air gap size, radial offset, and magnetic field waveform data. The collected parameters are transmitted to the edge computing control box in real time, providing a data foundation for subsequent control steps.

[0137] Step 2, Calculation of magnetic transmission efficiency factor and magnetic field compensation:

[0138] The edge computing control box calculates the magnetic transmission efficiency factor based on a digital twin model. The calculation process includes two sub-steps. First, based on multi-dimensional operating parameters, the temperature attenuation coefficient, air gap influence function, radial offset influence function, and harmonic distortion rate influence function are calculated. The temperature attenuation coefficient reflects the decrease in magnetic flux density and the risk of irreversible demagnetization caused by high temperatures. The air gap influence function considers edge effect correction based on the inverse square relationship between magnetic field strength and distance. The radial offset influence function considers the influence of the offset direction, reflecting the attenuation effect of radial misalignment on torque transmission. The harmonic distortion rate influence function reflects the increase in eddy current losses and torque pulsation caused by magnetic field waveform distortion. Then, the reference transmission efficiency is multiplied by the above influence functions to obtain the comprehensive magnetic transmission efficiency factor.

[0139] When the magnetic transmission efficiency factor falls below a set threshold, the edge computing control box activates the magnetic field compensation mechanism. The intelligent magnetic transmission unit adjusts the magnetic field coupling state to maintain the set transmission efficiency.

[0140] Step 3: Flow field demand inversion and performance curve generation:

[0141] The edge computing control box uses a simplified computational fluid dynamics (CFD) model to invert the minimum required airflow and optimal airflow injection angle of the target airflow pattern in real time, based on the collected furnace temperature distribution. This simplified CFD model is obtained through offline high-precision simulation training. Based on the inversion results, a fan performance requirement curve is generated, which is a mapping curve between speed, torque, and flow rate.

[0142] Step 4: Multi-objective optimization and optimal parameter solution:

[0143] The edge computing control box establishes an optimization function with the objectives of minimizing energy consumption, maximizing temperature uniformity, and minimizing equipment fatigue loss. This multi-objective optimization function is in the form of a weighted sum, balancing the three objectives through dynamic weight coefficients. An improved multi-objective genetic algorithm, the improved NSGA-II algorithm, is used for online optimization, with an iteration cycle set to once every five minutes.

[0144] The dynamic weighting coefficients are adaptively adjusted according to the process stage. During the heating stage, the energy consumption weighting coefficient is 0.4, the temperature uniformity weighting coefficient is 0.5, and the equipment fatigue loss weighting coefficient is 0.1. During the heat preservation stage, the energy consumption weighting coefficient is 0.6, the temperature uniformity weighting coefficient is 0.3, and the equipment fatigue loss weighting coefficient is 0.1. During the cooling stage, the energy consumption weighting coefficient is 0.3, the temperature uniformity weighting coefficient is 0.3, and the equipment fatigue loss weighting coefficient is 0.4. The edge computing control box, combined with constraints, uses an improved NSGA-II algorithm to solve for the optimal control parameters online.

[0145] Step 5: Dynamic Adjustment and Execution:

[0146] The edge computing control box outputs optimal control parameters to implement decoupled speed and torque control. Simultaneously, it executes predictive disturbance rejection strategies and a lifespan balancing mode to achieve dynamic adaptation of the fan's operating state.

[0147] The predictive disturbance rejection strategy is implemented by establishing a typical disturbance pattern library. This library includes typical disturbance types such as furnace door opening and closing, and workpiece entry and exit. When a typical disturbance is detected, the system pre-adjusts the fan speed 300 to 500 milliseconds in advance to counteract the impact of airflow disturbances on the furnace's operating conditions.

[0148] The lifespan balancing mode is achieved by monitoring the workload of each magnetic pole unit. The system switches the dominant magnetic pole region every set period and achieves uniform magnetic pole wear through a circumferential offset of ±3 degrees.

[0149] Step 6: Self-learning optimization iteration:

[0150] The system builds and runs a case database to accumulate control parameters and operational performance data under different operating conditions. At set time intervals, the edge computing control box performs deep optimization training to correct the digital twin model parameters and optimize weight coefficients, continuously improving control accuracy and adaptability.

[0151] Figure 2 The layout of the multi-sensor fusion acquisition module within the furnace body and intelligent magnetic drive unit is demonstrated. It primarily includes an array of Hall effect sensors, an infrared temperature measurement matrix, an ultrasonic anemometer, embedded thermocouples, and eddy current sensors. The infrared temperature measurement matrix consists of multiple infrared temperature probes, evenly distributed within the furnace walls to form a zoned temperature monitoring network. The ultrasonic anemometer is positioned at the air inlet and outlet of the furnace fan to collect airflow parameters. Embedded thermocouples are embedded within the permanent magnets of the intelligent magnetic drive unit to collect magnet temperature. Eddy current sensors are positioned outside the outer magnetic ring assembly of the intelligent magnetic drive unit to collect actual air gap size and radial offset. The array of Hall effect sensors is positioned between the inner and outer magnetic ring assemblies of the intelligent magnetic drive unit to collect magnetic field waveform data.

[0152] Figure 2 The graph compares the energy consumption of the optimized method and the traditional single-speed control method in the three stages of heating, holding, and cooling of the same industrial heat treatment furnace. The horizontal axis represents time, and the vertical axis represents real-time power. The graph contains two curves: the solid line represents the energy consumption change of the optimized method, and the dashed line represents the energy consumption change of the traditional control method.

[0153] The test conditions were the same furnace load, the same process temperature requirements, and the same operating cycle. During the heating phase, the solid power peak was lower than the dashed line and fluctuated more gently, reflecting precise airflow control under the weight of temperature uniformity. During the heat preservation phase, the solid power remained stable at a lower level, significantly lower than the dashed line, reflecting the energy-saving effect after the energy consumption weight was increased. During the cooling phase, the solid power curve transitioned smoothly without significant fluctuations, corresponding to stable control under the weight of equipment fatigue wear. By comparing the curves, the optimization effect of this solution's multi-objective optimization and dynamic weight adjustment on the overall process energy consumption can be intuitively verified.

[0154] Figure 3 The temperature distribution cloud map of the same cross-section inside the furnace is divided into two parts: the left side shows the temperature distribution of the traditional control method, and the right side shows the temperature distribution of the optimized method. Both graphs use the same temperature scale range; darker colors represent higher temperatures, and lighter colors represent lower temperatures.

[0155] The test conditions involved setting the temperature during the heat preservation stage and collecting temperature data from 12 monitoring points using an infrared thermography matrix to generate a cloud map. The cloud map generated by the traditional control method shows obvious high-temperature clusters and low-temperature blind spots, with significant temperature differences. The cloud map generated by the control method of this invention shows a uniform temperature distribution with no obvious temperature difference areas, and the temperature standard deviation is significantly lower than that on the left. This figure verifies that the proposed solution, by simplifying the model through computational fluid dynamics to invert the flow field requirements, can effectively improve the temperature uniformity within the furnace, meeting the process accuracy requirements.

[0156] Figure 4 The graph shows the change in magnetic transmission efficiency over time, with the horizontal axis representing the running time and the vertical axis representing the magnetic transmission efficiency factor. The graph includes the efficiency threshold line, the efficiency curve before optimization, the efficiency curve after optimization, and the magnetic field compensation trigger marker.

[0157] The test conditions involved continuous high temperatures inside the furnace, simulating a real-world scenario of rising magnet temperature and slight fluctuations in the air gap. Before optimization, the curve fluctuated wildly, repeatedly falling below the efficiency threshold and failing to recover autonomously after the drop. After optimization, the curve remained consistently above the efficiency threshold, only rapidly recovering and stabilizing when an efficiency decay trend occurred, triggered by the magnetic field compensation mechanism. This figure verifies the effectiveness of the quantitative calculation of the magnetic transmission efficiency factor and the active magnetic field compensation mechanism in this scheme, ensuring stable transmission efficiency.

[0158] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for optimizing the logic control of the efficiency of the in-furnace fan based on magnetic transmission, characterized by, The method comprises the following steps: Step 1: Establish a high-fidelity digital twin model of the magnetic transmission system, deploy a multi-dimensional sensor array, and collect multi-dimensional operating parameters of the magnetic transmission system in real time; Step 2: Calculate the magnetic transmission efficiency factor based on the digital twin model. When the magnetic transmission efficiency factor is lower than the set threshold, activate the magnetic field compensation mechanism to adjust the magnetic field coupling state to maintain the set transmission efficiency; Step 3: Based on the temperature distribution in the furnace, use a simplified computational fluid dynamics model to real-time inverse the target air flow pattern, the minimum required air volume and the optimal air flow incidence angle, and generate the fan performance demand curve; Step 4: Establish an optimization function with the goal of minimizing energy consumption, maximizing temperature uniformity, and minimizing equipment fatigue loss, combine the constraint conditions, and use an improved multi-objective genetic algorithm to solve the optimal control parameters online; Step 5: Implement speed and torque decoupling control, combine predictive disturbance rejection strategy to offset typical operating condition disturbances, and simultaneously execute life balance mode to achieve dynamic adaptation of fan operating state.

2. The method for optimizing the logic control of the in-furnace fan efficiency based on magnetic transmission according to claim 1, characterized in that, It also includes step 6: Construct a running case database and perform deep optimization training every set time period to correct the digital twin model parameters and optimization weight coefficients.

3. The method for optimizing the logic control of the in-furnace fan efficiency based on magnetic transmission according to claim 2, characterized in that, The multi-dimensional operating parameters of the magnetic transmission system include magnet temperature, actual air gap size, radial offset, and magnetic field waveform data.

4. The method for optimizing the logic control of the in-furnace fan efficiency based on magnetic transmission according to claim 3, characterized in that, In step 2, calculating the magnetic transmission efficiency factor includes the following sub-steps: Step 21, respectively calculate temperature attenuation coefficient based on multi-dimension operating parameters , air gap influence function , radial offset influence function , and harmonic distortion rate influence function ; reflects the risk of irreversible demagnetization due to the decrease in magnetic flux density at high temperatures; Based on the magnetic field intensity and the distance square inverse relationship, considering the edge effect correction; The effect of the misalignment direction is considered, reflecting the attenuation effect of the radial misalignment on the torque transmission; reflects the increase in eddy current losses and torque pulsations due to the distortion of the magnetic field waveform; Step 22, calculate the reference drive efficiency Step 23, multiply the reference drive efficiency with the comprehensive magnetic drive efficiency factor .

5. The method for optimizing the logic control of the in-furnace fan efficiency based on magnetic transmission according to claim 4, characterized in that, The computational fluid dynamics simplified model in step 3 is obtained through offline high-precision simulation training, and the fan performance demand curve is a mapping relationship curve of speed, torque, and flow.

6. The method for optimizing the logic control of the in-furnace fan efficiency based on magnetic transmission according to claim 5, characterized in that, The multi-objective optimization function in step 4 is: ； wherein is a minimization objective function for balancing the three targets of energy consumption, temperature uniformity and equipment fatigue loss, P is real-time power, Prated is rated power, is the standard deviation of the furnace temperature, is a fatigue coefficient based on the cumulative damage theory, and α, β, γ are dynamic weight coefficients. The improved multi-objective genetic algorithm is an improved NSGA-II algorithm, and the online optimization iteration period is every 5 minutes.

7. The method for optimizing the logic control of the in-furnace fan efficiency based on magnetic transmission according to claim 6, characterized in that, The dynamic weight coefficients are adaptively adjusted according to the process stage: α=0.4, β=0.5, γ=0.1 in the heating stage; α=0.6, β=0.3, γ=0.1 in the holding stage; α=0.3, β=0.3, γ=0.4 in the cooling stage.

8. The method for optimizing the logic control of the in-furnace fan efficiency based on magnetic drive transmission according to claim 7, characterized in that, The predictive disturbance rejection strategy in step 5 adjusts the fan speed in advance by 300ms-500ms to offset the influence of air flow disturbance on the furnace operating conditions by establishing a typical disturbance mode library, which at least includes furnace door opening and closing and workpiece entering and leaving; The life balance mode realizes uniform wear of the magnetic poles by monitoring the work load of each magnetic pole unit, switching the dominant magnetic pole area every set period, and realizing the uniformization of magnetic pole wear through a circumferential offset of ±3°.

9. A device for optimizing the logical control of the efficiency of the in-furnace fan based on magnetic transmission, characterized by, For implementing the method for optimizing the logical control of the fan efficiency in the furnace based on the magnetic transmission of claim 8, an intelligent magnetic transmission unit, a multi-sensor fusion acquisition module, and an edge computing control box are used; The intelligent magnetic transmission unit includes a partition adjustable permanent magnet array, multiple groups of independently controlled electromagnetic compensation coils, and an air gap active adjustment mechanism, which are used to respond to the optimization control instructions to adjust the magnetic field coupling state and the fan operating parameters; The multi-sensor fusion acquisition module includes an array type Hall sensor, an infrared temperature measurement matrix, and an ultrasonic anemometer, which are used to collect multi-dimensional operating parameters in real time and transmit them to the edge computing control box; The edge computing control box is internally provided with a processor, a real-time optimization algorithm acceleration card and a redundant dual-controller architecture, pre-stores a digital twin model and a multi-objective optimization algorithm, and is used for executing data processing, optimization calculation and control instruction output.

10. The magnetic drive based in-furnace fan efficiency logic control optimization device as claimed in claim 9, wherein, The air gap active adjustment mechanism adopts piezoelectric ceramic driving, and multiple measurement points are arranged at equal intervals in the circumferential direction of the array type Hall sensor, so as to simultaneously monitor the size and direction of the magnetic induction intensity.

Images