A data center intelligent harmonic dynamic compensation method and system

Abstract

The present application provides a method and system for intelligent harmonic dynamic compensation in a data center. Among them, a dynamic coupling model is established through multi-dimensional data fusion. First, the equipment vibration, noise and power quality data are collected, and a coupling model containing the mapping relationship between vibration modal characteristics, noise characteristics and harmonic distortion rate is constructed. Based on the model, the operating conditions of the HVAC equipment are analyzed, and the energy distribution caused by the working fluid flow and the motor is quantified as an operating condition coefficient through spectrum energy calculation and fed back to the model. Combined with the nonlinear characteristics of electric energy harmonics changing with load, the inverter data is used to dynamically match the speed range of the equipment, so that the equipment speed difference is mapped to the harmonic offset component generated by the frequency conversion equipment. Finally, the equipment parameter adjustment value that matches the current working condition is generated to achieve dynamic adaptation of harmonic compensation. The technical solution provided by the present application realizes the coordinated optimization of harmonic compensation and equipment operating conditions in the data center.

Description

Technical Field

[0001] The present application relates to the technical field of data center energy efficiency optimization and power quality management, and in particular to a method and system for intelligent harmonic dynamic compensation in a data center. Background Art

[0002] As data center equipment power density and load dynamics increase, harmonic distortion and mechanical vibration noise generated by power electronics, nonlinear loads, and high-frequency switching devices in the power supply system are strongly coupled, exacerbating the time-varying nature of the harmonic spectrum in the distribution network. An intelligent harmonic suppression method that can sense the harmonic-vibration coupling characteristics and dynamically adjust compensation parameters is urgently needed to ensure stable equipment operation in highly volatile scenarios.

[0003] Currently, a common solution is an active filter system based on fast Fourier transforms and preset harmonic templates. This system uses low-sampling-rate current sensors and single-point vibration sensors. A central controller periodically performs spectrum analysis, driving the active filter to inject compensation current according to a pre-stored reverse current template. To address mechanical vibration interference, the system incorporates a built-in fixed-frequency notch filter to suppress the amplitude of preset mechanical resonance frequencies.

[0004] However, this solution has some drawbacks: The fixed time window results in insufficient spectral resolution, making it impossible to accurately separate harmonics and vibration components at adjacent frequencies. If the mechanical resonance frequency shifts beyond a certain threshold due to temperature drift or aging, the notch filter will fail and even amplify the interference. The preset harmonic template cannot dynamically adapt to load changes. When the harmonic phase offset or amplitude fluctuation exceeds the preset threshold, the compensation current will mismatch the actual harmonics, potentially increasing the total harmonic distortion. Summary of the Invention

[0005] The present application provides a data center intelligent harmonic dynamic compensation method and system to solve the problem of low efficiency in suppressing harmonic vibration coupling interference in the prior art.

[0006] In a first aspect, the present application provides a method for intelligent harmonic dynamic compensation in a data center, comprising:

[0007] A dynamic coupling model is established by collecting vibration signals and noise data from data center infrastructure equipment and combining it with power quality data from input devices. The dynamic coupling model includes mapping relationships between vibration modal characteristics, noise amplitude-frequency characteristics, and harmonic distortion rates of various power sub-wavelengths.

[0008] Based on the correlation results between the vibration modal characteristics and the noise amplitude-frequency characteristics in the dynamic coupling model, the operating conditions of the HVAC equipment in the data center are analyzed. The spectrum energy distribution caused by the working fluid flow and motor in the equipment is calculated based on the vibration and noise characteristics of the equipment. The spectrum energy distribution is converted into equipment operating condition coefficients and fed back to the dynamic coupling model.

[0009] Based on the nonlinear characteristics of the load-dependent distortion rate of each harmonic of electric energy in the dynamic coupling model and the operating condition coefficient of the equipment, the operating speed range of the fan, water pump, and chiller is matched by frequency converter frequency and torque feedback data, so that the speed difference of the fan, water pump, and chiller is dynamically mapped to the offset components of each harmonic of electric energy generated by the superposition of pulsation of the frequency converter and the variable frequency motor in the matching equipment, and the offset components of each harmonic of electric energy are input into the dynamic coupling model as parameters;

[0010] Dynamically summing and optimizing are performed based on the harmonic offset components of each order of electric energy in the dynamic coupling model and the harmonic components collected during equipment operation, and combining the data center load fluctuation prediction data to generate equipment parameter adjustment values ​​corresponding to the equipment operating condition coefficient.

[0011] Optionally, the device is adjusted using the device parameter adjustment value, and the various electric energy harmonic offset components generated after the device operating parameters are adjusted are superimposed on the harmonic components generated by the currently running device, so that the time-varying amplitude and phase of the electric energy harmonic offset components dynamically offset the harmonic distortion components generated by the coordinated adjustment of load fluctuations by the device motor and inverter.

[0012] Optionally, based on the nonlinear characteristics of the harmonic distortion rate of electric energy and load changes in the dynamic coupling model and the operating condition coefficient of the equipment, the operating speed ranges of the fans, water pumps, and chillers are matched by frequency converter frequency data and torque feedback data. When it is detected that the operating speed range of any device exceeds a preset speed threshold interval, the speed difference of the devices in adjacent time windows is calculated;

[0013] Decomposing the speed difference into a fundamental component and multiple frequency components in a time series, and aligning the phases of the decomposed components with the inherent pulsation waveform of the power device to generate a superimposed waveform containing an amplitude variation;

[0014] Determining a frequency weight coefficient based on the correlation between the harmonic distortion rate and the equipment load in the dynamic coupling model, multiplying the amplitude changes of the fundamental component and each harmonic component in the superimposed waveform by the frequency weight coefficient, and dynamically mapping to generate each harmonic offset component of the electric energy corresponding to the pulsation superposition of the power device and the speed adjustment device;

[0015] The offset components of each harmonic of the electric energy are input as parameters into the dynamic coupling model.

[0016] Optionally, based on the equipment parameter adjustment value, the speed adjustment devices of the fan, water pump, and chiller are synchronously adjusted with the operating parameters of the power device;

[0017] extracting the offset components of each harmonic of electric energy generated by the coordinated operation of the speed adjustment device and the power device after the equipment parameters are adjusted;

[0018] Decomposing the waveform of each harmonic offset component of the electric energy into a fundamental component and a frequency multiplication component, and performing a phase inversion operation on the decomposed components and the harmonic components output by the operating device to generate a superposition result of the harmonic offset components of each harmonic offset component of the electric energy and the harmonic components;

[0019] According to the superposition result, dynamically matching the time-varying amplitude of the offset component of each harmonic of the electric energy with the distortion amplitude of the harmonic component, and generating an inverse relationship between the changing direction of the time-varying amplitude and the changing direction of the distortion amplitude;

[0020] Through the inverse relationship, the time-varying amplitude of each harmonic offset component of the electric energy is continuously updated within a preset time window until the time-varying amplitude offsets the harmonic distortion component generated by the equipment motor and inverter cooperating to adjust the load fluctuation.

[0021] Optionally, obtaining the harmonic offset components of each order of electric energy recorded in the dynamic coupling model and the harmonic components generated during the operation of the collected equipment;

[0022] Superimposing the waveform of each harmonic offset component of the electric energy with the waveform of the harmonic component according to a time window to generate a candidate superposition result including the time-varying amplitude and phase of each harmonic;

[0023] Based on the data center load fluctuation prediction data, the amplitude change trend of each time window in the candidate superposition results is correlated with the change trend of the equipment operating condition coefficient, and the target superposition results whose amplitude change direction is consistent with the load growth direction in the load fluctuation prediction data are screened out;

[0024] According to the corresponding relationship between the time-varying amplitude and the phase in the superposition result, an equipment parameter adjustment value corresponding to the equipment operating condition coefficient is generated.

[0025] Optionally, extracting device vibration mode data and device noise frequency data recorded in the dynamic coupling model and associating the vibration mode data and noise frequency data according to a time window;

[0026] The frequency corresponding to the peak amplitude in the correlated vibration mode data is overlapped and compared with the main frequency band in the noise frequency data to screen out the characteristic interval where the amplitude and frequency change synchronously;

[0027] In the characteristic interval, the square value of the amplitude in the vibration mode data at each time point is multiplied by the weight coefficient of the corresponding frequency and the calculation results of all time points are accumulated to generate a frequency spectrum energy distribution corresponding to the fluid flow state and the power unit operation state;

[0028] generating an equipment operating condition coefficient according to a ratio of an energy proportion of a fluid flow state to an energy proportion of an operating state of a power device in the spectrum energy distribution;

[0029] The equipment operating condition coefficient is fed back to the dynamic coupling model and the correlation parameters between the vibration mode data and the noise frequency data in the dynamic coupling model are updated.

[0030] Optionally, extracting a load change direction mark corresponding to each time window in the data center load fluctuation prediction data, the load change direction mark including a load increase direction and a load decrease direction;

[0031] Obtaining time-varying amplitude change data of each harmonic in each time window in the candidate superposition result and time-varying trend data of the equipment operating condition coefficient;

[0032] Matching the slope direction of the time-varying amplitude change data with the load change direction mark, and determining that the match is successful if the slope direction of the time-varying amplitude change data has the same sign as the slope direction of the load change direction mark;

[0033] According to the determination result of direction consistency, a target superposition result whose slope direction of the time-varying amplitude change data is consistent with the load growth direction is screened from the superposition results.

[0034] In a second aspect, the present application provides a data center intelligent harmonic dynamic compensation system, comprising:

[0035] Establish a module that uses the collected vibration signals and noise data of data center infrastructure equipment and combines it with the power quality data of the input equipment to establish a dynamic coupling model, wherein the dynamic coupling model includes the mapping relationship between vibration modal characteristics, noise amplitude-frequency characteristics, and harmonic distortion rates of various power levels;

[0036] a correlation module that analyzes the operating conditions of the HVAC equipment in the data center based on the correlation results between the vibration modal characteristics and the noise amplitude-frequency characteristics in the dynamic coupling model, calculates the spectrum energy distribution caused by the working fluid flow and motor in the equipment based on the vibration and noise characteristics of the equipment, and converts the spectrum energy distribution into equipment operating condition coefficients to feed back to the dynamic coupling model;

[0037] An input module, based on the nonlinear characteristics of the load-dependent distortion rate of each harmonic of electric energy in the dynamic coupling model and the operating condition coefficient of the equipment, matches the operating speed range of the fan, water pump, and chiller through the frequency converter and torque feedback data, dynamically maps the speed difference of the fan, water pump, and chiller to each harmonic offset component of electric energy generated by the superposition of pulsation of the frequency converter and the variable frequency motor in the matching equipment, and inputs the each harmonic offset component of electric energy as a parameter into the dynamic coupling model;

[0038] The generation module dynamically adds and optimizes the harmonic offset components of each power in the dynamic coupling model and the harmonic components collected during equipment operation, and generates equipment parameter adjustment values ​​corresponding to the equipment operating condition coefficient in combination with the data center load fluctuation prediction data.

[0039] In a third aspect, an embodiment of the present application provides a computing device comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a data center intelligent harmonic dynamic compensation method as described in the first aspect above.

[0040] In a fourth aspect, an embodiment of the present application provides a computer storage medium storing a computer program. When the computer program is executed by a computer, it implements a data center intelligent harmonic dynamic compensation method as described in the first aspect.

[0041] In an embodiment of the present application, a dynamic coupling model is established by collecting vibration signals and noise data of data center infrastructure equipment and combining them with the power quality data of the input equipment, wherein the dynamic coupling model includes a mapping relationship between vibration modal characteristics, noise amplitude-frequency characteristics and each harmonic distortion rate of electric energy; according to the correlation results between the vibration modal characteristics and the noise amplitude-frequency characteristics in the dynamic coupling model, the operating conditions of the HVAC equipment in the data center are analyzed, and the spectrum energy distribution caused by the working fluid flow and the motor in the equipment is calculated in combination with the vibration and noise characteristics of the equipment, and the spectrum energy distribution is converted into the equipment operating condition coefficient and fed back to the dynamic coupling model; based on the power quality data in the dynamic coupling model, the power quality data of the input equipment is fed back to the dynamic coupling model; based on the power quality data in the dynamic coupling model, the power quality data of the input equipment is fed back to the dynamic coupling model; and based on the power quality data in the dynamic coupling model, the power quality data of the input equipment is fed back to the dynamic coupling model. The nonlinear characteristics of each harmonic distortion rate changing with load and the operating condition coefficient of the equipment are considered. The operating speed range of the fan, water pump and chiller is matched by the frequency converter and torque feedback data. The speed difference of the fan, water pump and chiller is dynamically mapped to the various electric energy harmonic offset components generated by the superposition of the pulsation of the frequency converter and the variable frequency motor in the matching equipment, and the various electric energy harmonic offset components are input into the dynamic coupling model as parameters. Dynamic addition and optimization are performed according to the various electric energy harmonic offset components in the dynamic coupling model and the harmonic components collected during equipment operation, and the equipment parameter adjustment value corresponding to the equipment operating condition coefficient is generated in combination with the load fluctuation prediction data of the data center.

[0042] This application has the following beneficial effects:

[0043] This application establishes a dynamic coupling model through multi-source data fusion to accurately characterize the time-varying correlation between vibration noise and electrical harmonics, and realize cross-domain modeling of mechanical and electrical coupling interference; based on the spectrum energy distribution and operating condition coefficient conversion, the contribution of equipment mechanical excitation to harmonic distortion is quantified to improve the accuracy of harmonic tracing; through the dynamic mapping of inverter data and speed difference, the transfer function of mechanical pulsation and harmonic offset is established to realize closed-loop analysis of the harmonic generation mechanism; combined with the dynamic optimization of harmonic components and load prediction, the working condition adaptive equipment parameter adjustment value is generated, so that the harmonic suppression strategy synchronously matches the load fluctuation and the equipment operating status, and finally forms a full-link closed-loop control of "mechanism modeling-source suppression-dynamic compensation", which improves the harmonic suppression efficiency and system stability of the data center.

[0044] Furthermore, by decomposing the speed difference and aligning it with the inherent pulsation phase, the transmission effect of mechanical dynamic characteristics on harmonic distortion can be accurately quantified; combined with the dynamic mapping of frequency weight coefficients, a quantitative transmission link of mechanical and electrical coupling interference is established, thereby improving the analytical accuracy of the harmonic generation mechanism; through closed-loop feedback of the harmonic offset component, the decoupling and matching of the operating status of the frequency conversion equipment and the harmonic distortion of the electric energy is achieved, so that the harmonic compensation strategy can synchronously track the influence of mechanical vibration and load fluctuation, ultimately reducing the harmonic suppression delay and improving the compensation accuracy.

[0045] These and other aspects of the present application will become more readily apparent from the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0047] Figure 1 A flow chart of a method for intelligent harmonic dynamic compensation of a data center provided by the present application is shown;

[0048] Figure 2 The present invention provides a structural diagram of a data center intelligent harmonic dynamic compensation system;

[0049] Figure 3 A schematic structural diagram of a computing device provided by the present application is shown. DETAILED DESCRIPTION

[0050] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0051] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this document or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to being different types.

[0052] Researchers have found that existing data center harmonic suppression methods struggle to effectively address the coupling effects of mechanical vibration noise and power harmonic distortion. Traditional models lack cross-domain dynamic correlation, resulting in lagging compensation strategies and insufficient accuracy. Based on this, a method for intelligent harmonic dynamic compensation in data centers is proposed. This method constructs a dynamic coupling model by integrating vibration, noise, and power quality data. This model analyzes the mapping relationship between mechanical excitation and harmonic generation, dynamically generates harmonic offset components based on the operating status of variable-frequency equipment, and achieves closed-loop optimization of compensation parameters through load prediction and harmonic component optimization, significantly improving the efficiency and accuracy of harmonic suppression.

[0053] The technical solution of this application can be applied to data center scenarios with high load fluctuations and dense frequency conversion equipment, and is especially suitable for the dynamic compensation needs of harmonics under strong coupling interference between HVAC system equipment fans, water pumps, chillers and power supply networks.

[0054] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0055] Figure 1 The present invention provides a flow chart of a method for intelligent harmonic dynamic compensation of a data center, such as Figure 1 As shown, the method includes:

[0056] 101. Establish a dynamic coupling model by combining the collected vibration signals and noise data of the data center infrastructure equipment with the power quality data of the input equipment, wherein the dynamic coupling model includes a mapping relationship between vibration modal characteristics, noise amplitude-frequency characteristics, and each harmonic distortion rate of the power;

[0057] In step 101, the infrastructure equipment vibration signal refers to the mechanical vibration waveform data collected by the acceleration sensor; the noise data refers to the sound pressure level data in the 30Hz-20kHz frequency band obtained by the acoustic sensor; the power quality data refers to the set of electrical parameters including the total harmonic distortion rate of voltage, the harmonic components of current and the power factor; the dynamic coupling model refers to the multivariate correlation mathematical model that integrates the vibration spectrum, acoustic characteristics and harmonic parameters; the vibration modal characteristics refer to the amplitude-phase distribution pattern corresponding to the natural frequency of the equipment; the noise amplitude-frequency characteristics refer to the intensity distribution of noise energy in different frequency bands; the harmonic distortion rate of each subharmonic of electric energy refers to the proportion of voltage or current of the subharmonics that are integer multiples of the fundamental frequency.

[0058] In an embodiment of the present application, a triaxial acceleration sensor is first deployed at a key node of the HVAC equipment casing, for example, at the location of the chiller compressor to obtain a vibration signal at a sampling frequency of 10kHz, and an A-weighted sound level meter (an instrument for measuring sound intensity) is simultaneously used to collect noise data at a sampling rate of 48kHz. The input side voltage and current waveforms are obtained through a power quality analyzer, for example, the 3rd, 5th, and 7th harmonic distortion rates are extracted. The wavelet packet decomposition algorithm is used to extract the modal energy ratio of the vibration signal in the 3-5kHz frequency band as the vibration modal feature, and the noise 1 / 3 octave spectrum is calculated as the amplitude-frequency characteristic through fast Fourier transform. Finally, a dynamic coupling model is established. For example, when the vibration modal energy ratio of the chiller exceeds 15%, the corresponding 5th harmonic distortion rate increases by 0.8%.

[0059] The specific process of establishing a dynamic coupling model involves the following logical steps: First, a multi-source data synchronization acquisition system is deployed. A triaxial vibration accelerometer array and a high-precision acoustic microphone array are installed at key nodes of the data center's HVAC equipment (fans, pumps, and chillers). Simultaneously, power quality monitoring devices are connected to obtain the harmonic spectrum of the three-phase voltage and current. A timestamp alignment mechanism is used to synchronize the sampling of vibration signals, noise waveforms, and power quality data at the millisecond level. The sampling frequency is set to at least 10 times the fundamental frequency based on the mechanical characteristics of the equipment.

[0060] Subsequently, signal feature decoupling processing is performed: empirical mode decomposition is used on the vibration signal to extract the vibration modal characteristic parameters of the equipment's mechanical structure, including the first six natural frequencies, modal damping ratio, and vibration mode energy ratio; short-time Fourier transform is performed on the noise signal to obtain the time-frequency matrix, and the amplitude-frequency characteristic parameters of the 200Hz-4kHz characteristic frequency band are extracted, with a focus on analyzing the sound pressure level distribution of the blade passing frequency and its multiples; the power quality data is decomposed into 50 harmonic components through discrete Fourier transform, and the harmonic distortion rate and interharmonic content of each harmonic are calculated.

[0061] A dynamic weight allocation mechanism is used to establish the coupling relationship: a three-dimensional feature space is constructed based on the equipment operating condition database. The modal confidence matrix of the vibration modes and the noise spectrum coherence function are subjected to singular value decomposition, and the first three principal components are extracted as mechanical state characteristics. A nonlinear mapping relationship between these characteristics and the harmonic distortion rates is established through partial least squares regression, and the coupling coefficient matrix is ​​updated using a sliding time window (window length 30 seconds, step size 5 seconds). The modal energy fraction of the vibration fundamental frequency shows a strong correlation with the 5th / 7th harmonic distortion rate (Pearson coefficient > 0.85), while the high-frequency noise amplitude slope shows an exponential correlation with the harmonic clustering effect near the switching frequency (e.g., 2.5kHz±500Hz).

[0062] Ultimately, a dynamic coupling model was constructed using a long-short-term memory network. The input layer contains time-series vibration modal parameters, noise spectrum envelope, and fundamental load factor, while the output layer predicts the changing trends of each harmonic distortion rate. The model was trained using an adaptive moment estimation optimizer and a gradient penalty mechanism for equipment startup and shutdown transients, ensuring that the model maintains a prediction error of less than 3% over a load fluctuation range of 80%-120%. This dynamic coupling model performs online parameter identification every five minutes, fusing monitoring data with predicted values ​​through a Kalman filter to form a closed-loop update mechanism.

[0063] For example, within a large data center, the operations and maintenance team deployed a multi-channel vibration sensor array and acoustic acquisition devices to conduct 24 / 7 vibration and noise monitoring of key equipment, including chillers, water pumps, and variable-frequency fans. Simultaneously, a power quality analyzer was used to collect voltage and current harmonic data at the equipment input, including third-, fifth-, and seventh-order harmonic distortion rates. Using a joint time-frequency domain analysis method, the modal characteristics of the vibration signal, such as the amplitude and phase of the chiller compressor's fundamental frequency of 45 Hz and its 90 Hz harmonic harmonic, were correlated with characteristic frequency bands in the noise spectrum, such as the broadband noise energy from 200 to 800 Hz. This dynamic coupling model with harmonic distortion was established. When the chiller's vibration energy suddenly increased at 90 Hz, the model automatically correlated to a nonlinear response relationship, indicating a 0.8% increase in the fifth-order harmonic distortion rate, providing a basis for subsequent harmonic compensation.

[0064] 102. Analyze the operating conditions of the HVAC equipment in the data center based on the correlation results between the vibration modal characteristics and the noise amplitude-frequency characteristics in the dynamic coupling model. Calculate the spectrum energy distribution caused by the working fluid flow and motor in the equipment based on the vibration and noise characteristics of the equipment. Convert the spectrum energy distribution into equipment operating condition coefficients and feed them back to the dynamic coupling model.

[0065] In step 102, the HVAC equipment operating condition refers to the load status of the air-conditioning terminal, cooling water pump and compressor; the spectral energy distribution caused by the working fluid flow refers to the broadband vibration component caused by the flow of refrigerant; the spectral energy distribution caused by the motor refers to the characteristic frequency vibration excited by the electromagnetic force of the motor stator and rotor; the equipment operating condition coefficient refers to a 0-1 standardized parameter that quantifies the equipment load status.

[0066] In an embodiment of the present application, first, based on the correlation results between the vibration modal characteristics and the noise amplitude-frequency characteristics in the dynamic coupling model, the vibration signal of the chiller is subjected to order analysis. For example, the 24th-order vibration component corresponding to the compressor speed is separated as the working fluid flow characteristic, and the 2nd frequency of the motor fundamental frequency 50Hz is extracted as the electromagnetic vibration characteristic. The noise source is identified by acoustic array positioning technology, for example, the water pump outlet pipe section is determined to be the main source of 500-800Hz high-frequency noise. The working condition coefficient is calculated by weighting the proportion of working fluid flow and motor vibration energy, for example, the working fluid flow vibration accounts for 60% and the motor vibration accounts for 40%. When the total vibration energy reaches the threshold, the working condition coefficient is set to 0.82, and the spectrum energy distribution is converted into the equipment operation condition coefficient and fed back to the dynamic coupling model.

[0067] Continuing with the above example, based on the dynamic coupling model, the operations and maintenance team discovered that at a 75% load factor, a particular chiller experienced strong coupling between the 200Hz high-frequency vibration caused by the internal fluid flow and the 37Hz low-frequency noise from the motor bearings, resulting in an abnormal spectral energy distribution. Using a wavelet packet decomposition algorithm, they calculated that the fluid flow energy accounted for 62% and the motor vibration energy accounted for 38%. This was quantified as an operating condition coefficient of 0.82, with a normal range of 0.75 to 0.85. This coefficient was fed back into the dynamic coupling model, triggering an early warning mechanism: when the operating condition coefficient fell below 0.78, the fifth harmonic distortion rate would exceed the 4% threshold. Based on this information, the system automatically adjusted the chiller's operating parameters, limiting the load factor to a safe range of 70% to 80% to prevent harmonics from exceeding the specified limit.

[0068] 103. Based on the nonlinear characteristics of the load-dependent distortion rate of each harmonic of electric energy in the dynamic coupling model and the operating condition coefficient of the equipment, the operating speed range of the fan, water pump, and chiller is matched by frequency converter frequency and torque feedback data, so that the speed difference of the fan, water pump, and chiller is dynamically mapped to the offset components of each harmonic of electric energy generated by the superposition of pulsation of the frequency converter and the variable frequency motor in the matching equipment, and the offset components of each harmonic of electric energy are input into the dynamic coupling model as parameters;

[0069] In step 103, the nonlinear characteristics of the harmonic distortion rate of electric energy in the dynamic coupling model changing with load refer to that the harmonic distortion rate and the equipment load are exponentially or polynomially related rather than simply linearly correlated; the equipment operating condition coefficient refers to a normalized parameter that quantifies the operating status of the equipment; the inverter frequency refers to the fundamental frequency of the PWM waveform output by the inverter; the torque feedback data refers to the torque value of the motor output shaft; the operating speed range refers to the minimum to maximum speed range allowed by the equipment; the speed difference refers to the deviation between the actual speed and the set value of multiple devices in the same system; the pulsation superposition of the inverter and the variable frequency motor refers to the combined effect of the high-frequency switching of the switching device and the pulsation of the motor magnetic field; the harmonic offset component of each electric energy refers to the change value of the specific harmonic distortion caused by the dynamic operation of the equipment.

[0070] In an embodiment of the present application, a correlation curve between the harmonic distortion rate of electric energy and the load is first extracted from the dynamic coupling model. For example, when the load rate of the chiller is 70%, the 5th harmonic distortion rate rises nonlinearly to 4.2%. Combined with the equipment operating condition coefficient, such as the water pump operating condition coefficient of 0.75, the speed range of the fan, water pump, and chiller is matched by a fuzzy PID control algorithm: the cooling tower fan speed upper limit is set to 1800rpm and the lower limit is set to 1200rpm. The water pump speed dynamically follows the load rate of the chiller according to a quadratic function. The inverter output frequency, such as 45Hz, and torque data, such as 85% of the rated torque, are collected, and the root mean square value of the speed difference is calculated, such as ±2.1%. The speed difference is mapped to the harmonic offset component through the transfer function. For example, every 1% increase in the speed difference causes the 5th harmonic offset component to increase by 0.12% and the 7th harmonic offset component to increase by 0.08%. Finally, the harmonic offset component matrix is ​​generated and input into the dynamic coupling model. For example, the 5th harmonic offset component in the updated model is +0.25%.

[0071] Continuing with the above example, to address the sudden increase in the seventh harmonic during variable frequency speed regulation of a pump group, the system invoked the nonlinear characteristic curve in the dynamic coupling model. This curve showed that for every 10% increase in load factor, the seventh harmonic offset component ΔH7 increased by 0.3%. Based on the collected data indicating an increase in the inverter output frequency from 45Hz to 48Hz and a fluctuation range of 320Nm±5% in the torque feedback data, the system dynamically adjusted the speed control strategy, limiting the speed difference between adjacent pumps from the originally designed 100rpm to within 50rpm. This adjustment offset the pulsating harmonic components of the variable frequency motors in multiple pumps, ultimately reducing the seventh harmonic offset component from 1.2% to 0.6%. The model also updated parameters simultaneously, enhancing the accuracy of harmonic prediction.

[0072] 104. Dynamically sum and optimize the harmonic offset components of each power in the dynamic coupling model and the harmonic components collected during equipment operation, and generate equipment parameter adjustment values ​​corresponding to the equipment operating condition coefficient in combination with the data center load fluctuation prediction data.

[0073] In step 104, dynamic summation optimization refers to the optimization process of adaptively weighted fusion of harmonic components and predicted data; harmonic components refer to the distortion rates of each harmonic currently measured by the power quality analyzer; load fluctuation prediction data refers to the output value of the ARIMA time series model established based on historical power consumption; and the equipment parameter adjustment value refers to the optimized inverter carrier frequency, dead time and speed compensation parameter set.

[0074] In the embodiment of the present application, harmonic data is first collected in a 1-minute cycle, for example, the current 3rd harmonic distortion rate is 3.2%, the 5th harmonic is 4.5%, and the 7th harmonic is 2.8%. Load forecast data is obtained synchronously, for example, the power consumption of IT equipment is expected to increase by 12% in the next 5 minutes. The Kalman filter algorithm is used to dynamically fuse the value and the predicted value: the predicted weight is 0.6, the measured weight is 0.4, and the target harmonic component is generated, such as the 5th harmonic target value of 4.3%. The equipment parameter adjustment plan is calculated by the multi-objective particle swarm optimization algorithm, for example: the carrier frequency of the cooling water pump inverter is increased from 8kHz to 10kHz to suppress the 5th harmonic; the dead time of the refrigeration unit compressor is adjusted from 2μs to 2.5μs to reduce the switching pulsation; the fan speed compensation is increased by +1.2% to balance the system impedance. Finally, the parameter adjustment instruction set is output and sent to each device controller via the Modbus protocol.

[0075] Continuing with the above example, during the data center's evening peak load fluctuations, the system activated a dynamic summation optimization algorithm based on the upward trend in the predicted load factor data from 55% to 82%. The algorithm weighted the model's output harmonic offset components (H3 = 1.2%, H5 = 2.1%, H7 = 0.8%) with the monitored harmonic components (H3 = 1.5%, H5 = 2.4%, H7 = 1.1%), using a weighting factor of 0.7. Combined with the load forecast for the next 15 minutes, the system generated equipment parameter adjustments: the chiller speed was increased to 485 rpm, the cooling water pump capacitor compensation was increased by 3%, and the variable frequency fan PWM modulation frequency was optimized to 4.8 kHz. After implementation, the data center's total harmonic distortion (THD) dropped from 4.7% to 3.1%, validating the effectiveness of the dynamic compensation strategy under complex operating conditions.

[0076] In summary, steps 101 to 104 implement the multi-physics field collaborative compensation technology for data center power quality. Through cross-domain correlation modeling of equipment vibration signals, noise spectra, and power harmonic data, a multi-dimensional dynamic coupling relationship network of vibration mode-acoustic characteristics-harmonic distortion rate is constructed. Based on the spectrum energy decomposition and operating condition coefficient dynamic feedback mechanism of the HVAC equipment operating conditions, the harmonic offset components caused by the electromechanical coupling effect of the variable frequency equipment are analyzed. Combining the load fluctuation prediction model with the dynamic superposition compensation algorithm of harmonic components, harmonic cancellation parameters are generated for equipment speed difference and pulsation superposition characteristics, forming a closed-loop optimization chain of "mechanical vibration perception-harmonic offset prediction-parameter adaptive matching", breaking through the response lag bottleneck of traditional static harmonic control, and realizing dynamic tracking and intelligent suppression of the harmonic distortion rate of the data center power supply system.

[0077] In order to solve the technical problems of equipment collaborative adjustment lag and insufficient harmonic distortion cancellation accuracy in dynamic compensation of power harmonics in data centers, in some embodiments, the method includes: using the equipment parameter adjustment value to adjust the equipment, and superimposing the various power harmonic offset components generated after the equipment operating parameters are adjusted to the harmonic components generated by the currently operating equipment, so that the time-varying amplitude and phase of the power harmonic offset components dynamically cancel out the harmonic distortion components generated by the collaborative adjustment of load fluctuations by the equipment motor and inverter.

[0078] In the steps, the equipment parameter adjustment value refers to the set of control parameters such as the inverter carrier frequency, dead time and speed compensation generated by the optimization algorithm; each harmonic offset component of electric energy refers to the change value of the specific harmonic distortion caused by the dynamic operation of the equipment; the harmonic component refers to the actual harmonic distortion rate measured by the power quality analyzer at the current moment; the harmonic distortion component refers to the voltage or current harmonic increment caused by load fluctuations and nonlinear characteristics of the equipment; the dynamic cancellation of time-varying amplitude and phase refers to adjusting the amplitude and phase angle of the harmonic component to make it superimposed inversely with the target harmonic to achieve cancellation.

[0079] In an embodiment of the present application, the device parameter adjustment value generated in step 104 is first sent to the device controller, for example, the carrier frequency of the chiller inverter is set to 10kHz, the dead time is 2.5μs, and the fan speed is compensated by +1.2%. After the adjustment is performed, the offset components of each harmonic are calculated by the dynamic coupling model. For example, the offset component of the 5th harmonic is -0.3%, indicating that 0.3% of the distortion can be offset. The harmonic component data is collected synchronously. For example, the current 5th harmonic distortion rate is 4.7%. The harmonic offset component is superimposed on the harmonic component: 4.7% is calculated plus -0.3%, and the compensated distortion rate is 4.4%. The phase angle of the harmonic offset component is adjusted by the phase synchronization controller so that it differs by 180 degrees from the harmonic phase. For example, if the 5th harmonic phase is detected to be 30 degrees, the phase of the offset component is adjusted to 210 degrees to achieve amplitude superposition and anti-phase cancellation.

[0080] Here's a specific example:

[0081] In intelligent dynamic harmonic compensation scenarios in data centers, a harmonic sensor array and IGBT active compensation devices are deployed to establish a closed-loop control system. When a sudden load change in a cabinet group is detected, generating 12.7% fifth-order harmonic distortion, the system executes the following process: First, a baseline harmonic spectrum is generated based on the equipment's historical operating parameters, identifying the combined load fluctuations of the variable-frequency refrigeration unit and the server power module as the primary cause of the distortion. A dynamic parameter optimization algorithm then calculates the required compensation harmonic component with an amplitude of 3.2A and a phase angle of 178° that must be injected into the IGBT. Time series superposition technology is used to time-domain match the compensation component with the harmonics, reducing the fifth-order harmonic distortion from 12.7% to 2.3% within 300ms. The synchronous inverter dynamically adjusts the output frequency from 47.5Hz to 49.8Hz, stabilizing the total harmonic distortion (THD) within the range of 3.1±0.5%. After 24 hours of continuous operation verification, this method improves the harmonic cancellation efficiency by 11 times compared with the traditional static filtering solution, improves the comprehensive power quality index by 37%, and shortens the compensation response delay to 80μs through the adaptive phase tracking algorithm.

[0082] In summary, the above steps have achieved intelligent compensation technology based on dynamic harmonic phasor synthesis. This system constructs a dynamic compensation system for harmonic distortion in variable-frequency equipment by combining harmonic offset components triggered by equipment parameter adjustments with harmonic components in a closed-loop superposition mechanism. Dynamic phasor synthesis is performed using the time-varying amplitude and phase parameters of the harmonic offset components, combined with the spectral characteristics of the harmonic components. A vector cancellation algorithm is then used to generate a harmonic compensation waveform that is in phase with and inversely proportional to the equipment load fluctuations. This technology overcomes the phase lag inherent in traditional fixed-parameter filters, enabling tracking and adaptive cancellation of pulsating harmonic components during the operation of variable-frequency motors and chiller groups. This creates a dynamic compensation closed loop of "parameter adjustment-harmonic generation-vector synthesis," effectively suppressing power quality degradation caused by electromechanical coupling.

[0083] In order to solve the technical problem of insufficient accuracy in equipment coordinated control and harmonic component matching in dynamic compensation of harmonics in data centers, in some embodiments, step 103 is based on the nonlinear characteristics of the harmonic distortion rate of each order of electric energy changing with load in the dynamic coupling model and the equipment operating condition coefficient, and the operating speed range of the fan, water pump, and chiller is matched by the frequency converter frequency and torque feedback data, so that the speed difference of the fan, water pump, and chiller is dynamically mapped to the harmonic offset components of each order of electric energy generated by the pulsation superposition of the frequency converter and the variable frequency motor in the matching equipment, and the harmonic offset components of each order of electric energy are input into the dynamic coupling model as parameters, including:

[0084] 301. Based on the nonlinear characteristics of the harmonic distortion rate of electric energy and load changes in the dynamic coupling model and the equipment operating condition coefficient, the operating speed ranges of the fans, water pumps, and chillers are matched using the inverter frequency data and torque feedback data. When it is detected that the operating speed range of any device exceeds a preset speed threshold range, the speed difference of the device in adjacent time windows is calculated;

[0085] In step 301, the nonlinear characteristics of the harmonic distortion rate of electric energy and the load change in the dynamic coupling model refer to the correlation relationship between the harmonic distortion rate and the equipment load showing an exponential or polynomial change; the equipment operating condition coefficient refers to the equipment operating status parameter quantified by vibration and acoustic characteristics; the inverter frequency data refers to the fundamental frequency of the inverter output PWM waveform; the torque feedback data refers to the dynamic torque measurement value of the motor output shaft; the operating speed range refers to the lowest to highest speed range allowed by the equipment; the preset speed threshold range refers to the upper and lower limits of the speed set based on the safe operation of the equipment; the speed difference refers to the absolute value of the difference between the actual speeds of the equipment in adjacent time windows.

[0086] In the embodiment of the present application, the process of "matching the operating speed ranges of the fans, pumps, and chillers by using inverter frequency data and torque feedback data based on the nonlinear characteristics of the power harmonic distortion rate and load changes in the dynamic coupling model and the equipment operating condition coefficient" includes:

[0087] 3011. Based on the nonlinear characteristics of the harmonic distortion rate of electric energy and load changes in the dynamic coupling model and the equipment operating condition coefficient, obtain the inverter frequency data corresponding to the fan, water pump, and chiller and the torque feedback data of the power output end;

[0088] It should be noted that, since the dynamic coupling model reveals the correlation between the electrical harmonics and the mechanical operating status of the equipment under different loads through the preset nonlinear relationship between the harmonic distortion rate and the load, and combines the mechanical vibration, noise and working fluid flow characteristics represented by the equipment operating condition coefficient, it can directly map the dynamic coupling effect of load changes on the equipment speed and power output by collecting the electrical signal frequency of the inverter at the output end of the power unit and the torque fluctuation data of the mechanical transmission shaft, thereby providing the underlying data that matches the actual operating status of different equipment for the generation of harmonic offset components, namely the inverter frequency data corresponding to the fan, water pump and chiller and the torque feedback data at the power output end.

[0089] 3012. Calculate the load factors of the fans, water pumps, and chillers based on the proportional relationship between the inverter frequency data and the torque feedback data, and determine the target speed adjustment direction and adjustment range of the corresponding devices based on the difference between the load factors and the preset load range;

[0090] 3013. Perform a sliding window comparison on the target speed adjustment amplitude and the speed history data of the device in the current operation cycle. When the target speed adjustment amplitude exceeds the preset speed threshold interval, extract the actual speed change sequence of the device in the adjacent time window.

[0091] 3014. Extract periodic components from the actual speed variation sequence to obtain a speed difference decomposition result consisting of a fundamental frequency component and integer multiple frequency components, wherein the fundamental frequency component is the main frequency of speed fluctuation of the device during steady-state operation, and the integer multiple frequency components are additional frequency components generated by sudden load changes or resonance of the power unit;

[0092] 3015. Matching the amplitude and phase information of the fundamental frequency component and the integer multiple frequency component with the waveform characteristics of the power device in a preset pulsating waveform library, and generating a synthetic pulsating waveform including an amplitude accumulation effect by waveform superposition;

[0093] 3016. According to the associated weights of each harmonic distortion rate and the load in the dynamic coupling model, the amplitudes of the fundamental frequency component and the integer multiple frequency component in the synthetic pulsating waveform are proportionally scaled to generate each harmonic offset component of the electric energy corresponding to the equipment speed adjustment process.

[0094] Furthermore, the inverter output frequency data, such as 45Hz, and torque feedback data, such as 85% of the rated torque, are collected. When the actual speed of the cooling water pump is detected to be 1850rpm and exceeds the upper limit of the preset speed threshold range of 1800rpm, the speed difference within adjacent time windows, such as 5 minutes, is calculated to be ±3.2%.

[0095] 302. Decompose the rotational speed difference into a fundamental component and multiple frequency components in a time series, and align the phases of the decomposed components with the inherent pulsating waveform of the power device to generate a superimposed waveform including an amplitude variation;

[0096] In step 302, the fundamental component refers to the low-frequency component in the speed difference corresponding to the fundamental frequency of the equipment; the harmonic component refers to the high-frequency component that is an integer multiple of the fundamental frequency; the inherent pulsating waveform refers to the characteristic waveform caused by the periodic vibration of the power device; phase alignment refers to adjusting the time offset to make the decomposed component consistent in phase with the inherent waveform; the superimposed waveform refers to the time domain signal after the fundamental wave and the harmonic component are synthesized.

[0097] In an embodiment of the present application, the speed difference of ±3.2% is first decomposed by fast Fourier transform to extract the fundamental component of 0.8Hz, the 2nd harmonic component of 1.6Hz, and the 3rd harmonic component of 2.4Hz. The phase offset of the inherent pulsating waveform of the power unit, such as the 48Hz characteristic waveform corresponding to the 24th order vibration of the motor rotor, is calculated by the cross-correlation algorithm. For example, the fundamental component needs to be delayed by 1.5ms to align with the phase of the inherent waveform. The amplitude of the decomposed fundamental component of 0.5%, the amplitude of the 2nd harmonic component of 0.3%, and the amplitude of the 3rd harmonic component of 0.2% are superimposed in the time domain to generate a synthetic waveform containing the amplitude change, for example, the peak value of the superimposed waveform is 0.8%.

[0098] 303. Determine a frequency weight coefficient based on the correlation between the harmonic distortion rate and the equipment load in the dynamic coupling model, multiply the amplitude change of the fundamental component and each harmonic component in the superimposed waveform by the frequency weight coefficient, and dynamically map to generate each harmonic offset component of the electric energy corresponding to the superposition of the pulsation of the power device and the speed adjustment device;

[0099] In step 303, the frequency weight coefficient refers to the weight values ​​of different frequency bands allocated according to the contribution of the harmonic distortion rate; the correlation between the harmonic distortion rate and the equipment load refers to the mathematical model of the change of the specific harmonic distortion amount with the load; the amplitude change refers to the amplitude proportion of each component in the superimposed waveform; the pulsation superposition of the power unit and the speed adjustment device refers to the interaction between the electromagnetic pulsation of the motor and the switching harmonics of the inverter; the harmonic offset component of each order of electric energy refers to the change value of the harmonic distortion amount caused by the dynamic adjustment of the equipment.

[0100] In the embodiment of the present application, first, based on the quadratic function relationship between the 5th harmonic distortion rate and the load in the dynamic coupling model, for example, a 10% increase in the load rate leads to a 0.3% increase in the 5th harmonic distortion rate, the frequency weight coefficients are set as follows: the fundamental component weight is 0.6, the 2nd frequency component weight is 0.3, and the 3rd frequency component weight is 0.1. The amplitude of the fundamental component in the superimposed waveform is 0.5%. 0.6, 2 times frequency component 0.3% 0.3, 3 times frequency component 0.2% 0.1, calculated comprehensive harmonic offset component = 0.5% 0.6+0.3% 0.3+0.2% 0.1 = 0.41%. Finally, the integrated offset components are mapped to power harmonic parameters through the transfer function. For example, the 5th harmonic offset component is negative 0.33%, and the 7th harmonic offset component is negative 0.15%.

[0101] 304. Input the offset components of each harmonic of the electric energy as parameters into the dynamic coupling model.

[0102] In step 304 , each harmonic offset component of electric energy refers to the correction value of each harmonic distortion generated in step 203 ; and the dynamic coupling model refers to a multivariate correlation mathematical model integrating vibration, acoustic and electric energy parameters.

[0103] In an embodiment of the present application, the generated 5th harmonic offset component of -0.33% and the 7th harmonic offset component of -0.15% are first input into the dynamic coupling model to update the harmonic parameter library in the model. For example, the original model predicts that the 5th harmonic distortion rate is 4.7% under the current load, and after superimposing the offset component, it is corrected to 4.7% minus 0.33%, which equals 4.37%. Through the feedback mechanism, the model outputs new equipment parameter adjustment instructions, such as adjusting the dead time of the chiller inverter from 2.5 microseconds to 3.0 microseconds, and reducing the water pump speed to 1750rpm. Monitoring after execution shows that the 5th harmonic distortion rate has dropped from 4.7% to 4.3%, and the total harmonic distortion rate has dropped from 6.8% to 5.5%.

[0104] In summary, steps 301 to 304 demonstrate a multi-parameter fusion prediction technique for harmonic offsets in variable-frequency equipment. A dynamic analytical model of the harmonic generation mechanism is constructed through speed threshold monitoring and a spectrum decomposition algorithm. A fundamental-multiple frequency joint analysis mechanism based on equipment speed differences, combined with the inherent pulsating waveform characteristics of the power unit, performs time-frequency phase synchronization to generate a harmonic superposition waveform that incorporates amplitude modulation. A frequency-domain optimization method dynamically weights harmonic distortion and load correlation to establish a nonlinear mapping between the amplitude of the multiple frequency components and the harmonic offset. Adaptive adjustment of the frequency weight coefficients enables cross-domain conversion of electromechanical pulsations to electrical harmonics. This technique overcomes the mechanism-separation limitations of traditional harmonic source modeling, forming a closed-loop prediction system of "speed monitoring-spectrum reconstruction-weight mapping." This system tracks and phase-locks harmonic components induced by load fluctuations in variable-frequency generator sets, providing a high-precision offset prediction data foundation for dynamic harmonic compensation.

[0105] In order to solve the technical problems of lag in dynamic compensation of harmonic distortion and insufficient phase matching accuracy in the coordinated control of multiple devices in a data center, in some embodiments, the device parameter adjustment value is used to adjust the device in step 201, and the various power harmonic offset components generated after the device operating parameters are adjusted are superimposed on the harmonic components generated by the currently running device, so that the time-varying amplitude and phase of the power harmonic offset components dynamically offset the harmonic distortion components generated by the coordinated adjustment of load fluctuations by the device motor and inverter, including:

[0106] 401. Based on the equipment parameter adjustment value, synchronously adjust the speed adjustment devices of the fan, water pump, and chiller and the operating parameters of the power device;

[0107] In step 401, the device parameter adjustment values ​​generated by the dynamic coupling model refer to a set of inverter carrier frequency, deadtime, and speed compensation parameters generated by an optimization algorithm. These device parameter adjustment values ​​correspond to the fan, pump, and chiller. These adjustment values ​​are generated by matching the operating speed ranges of the fan, pump, and chiller and analyzing the dynamic mapping relationship between their speed differences and harmonic offset components. These adjustment values ​​specifically target the operating parameters (such as speed and torque) of the fan, pump, and chiller, achieving dynamic harmonic compensation by adjusting the device's operating state. The speed adjustment device refers to the variable frequency controller that drives the fan, pump, and chiller speed changes. The power unit operating parameters refer to the motor torque output, PWM modulation index, and deadtime configuration.

[0108] In the embodiment of the present application, the device parameter adjustment values ​​are first adjusted based on the output of the dynamic coupling model. For example, the carrier frequency of the chiller inverter is set to increase from 8 kHz to 10 kHz, the dead time is increased from 2 microseconds to 3 microseconds, and the cooling water pump speed compensation is adjusted to +1.5%. Parameter instructions are then synchronously sent to the controllers of the fan, water pump, and chiller via the Modbus-TCP protocol. For example, the instruction "SetFrequency 10000Hz" is sent to the chiller inverter, and the speed closed-loop control deviation threshold is set to + / - 1%.

[0109] 402. Extracting the offset components of each harmonic of electric energy generated by the coordinated operation of the speed adjustment device and the power device after the equipment parameters are adjusted;

[0110] In step 402, the harmonic offset components of each order of electric energy refer to the change in harmonic distortion caused by dynamic adjustment of the equipment; the coordinated operation of the speed adjustment device and the power device refers to the linkage control of the inverter parameters and the motor torque output.

[0111] In this embodiment of the present application, a power quality analyzer is used to collect the adjusted voltage and current waveforms, for example, extracting the 3rd, 5th, and 7th harmonic distortion rates at a sampling rate of 1000 times per second. The motor torque output data at a 10 kHz inverter carrier frequency is simultaneously recorded, for example, with the torque fluctuation range being within ±5% of the rated value. The harmonic spectrum is analyzed using a fast Fourier transform, detecting, for example, a 5th harmonic offset component of negative 0.35% and a 7th harmonic offset component of negative 0.18%.

[0112] 403. Decompose the waveform of each harmonic offset component of the electric energy into a fundamental component and a frequency multiplication component, and perform a phase inversion operation on the decomposed components and the harmonic components output by the operating device to generate a superposition result of the harmonic offset components of each harmonic offset component of the electric energy and the harmonic components;

[0113] In step 403, waveform decomposition refers to converting the time domain signal into the fundamental wave and integer multiple frequency components in the frequency domain; the phase inversion operation refers to increasing the phase angle of the decomposed harmonic component by 180 degrees; and the superposition result refers to the vector sum of the antiphase harmonic and the original harmonic.

[0114] In the embodiment of the present application, the waveform of each harmonic offset component of the electric energy is first decomposed into a fundamental component and a frequency multiplication component. The 5th harmonic offset component of -0.35% is decomposed by fast Fourier transform, and the amplitude of the fundamental component corresponding to 50 Hz is 0.05% and the amplitude of the 5th harmonic corresponding to 250 Hz is 0.3%. The 5th harmonic component is phase-inverted by a digital signal processor, for example, the original phase angle of 30 degrees is adjusted to 210 degrees. The inverted harmonic component is superimposed on the harmonic component. For example, the original 5th harmonic distortion rate of 4.7% is superimposed on the inverted component of -0.35%, and the actual distortion rate is reduced to 4.35%.

[0115] 404. Dynamically match the time-varying amplitude of each harmonic offset component of the electric energy with the distortion amplitude of the harmonic component according to the superposition result, and generate an inverse relationship between the changing direction of the time-varying amplitude and the changing direction of the distortion amplitude;

[0116] In step 404, the time-varying amplitude refers to the amplitude of the harmonic component changing dynamically over time; the distortion amplitude refers to the absolute value of the distortion rate of the harmonic; and the opposite relationship in the direction of amplitude change refers to the amplitude change trend of the compensated harmonic being opposite to that of the original distorted harmonic.

[0117] In this embodiment, the fifth harmonic distortion amplitude is first monitored, fluctuating from 4.7% to 4.9%. Simultaneously, the time-varying amplitude of the corresponding harmonic offset component is dynamically adjusted from -0.35% to -0.42%. A proportional-integral-derivative control algorithm is used to dynamically match the changing trends of the two. For example, when the distortion amplitude increases by 0.2%, the time-varying amplitude is controlled to increase by 0.25% inverse compensation, ensuring that the amplitude changes in the opposite direction.

[0118] 405. Through the inverse relationship, the time-varying amplitude of each harmonic offset component of the electric energy is continuously updated within a preset time window until the time-varying amplitude offsets the harmonic distortion component generated by the equipment motor and the inverter cooperating to adjust the load fluctuation.

[0119] In step 405, the preset time window refers to the closed-loop control period set by the system; and the harmonic distortion component cancellation refers to the amplitude of the compensation harmonic completely neutralizing the original distortion.

[0120] In the embodiment of the present application, the time-varying amplitudes of the offset components of each power harmonic are first continuously updated within a preset time window through the opposite relationship of the amplitude change direction. The time window is set to 5 minutes, and the time-varying amplitudes of the harmonic offset components are updated every 10 seconds. For example, if the initial 5th harmonic offset component is negative 0.35%, and the distortion rate is still higher than the target value of 4.3%, the offset component is gradually increased to negative 0.45% until the 5th harmonic distortion rate is monitored to be stable below 4.3%. When the distortion rate meets the standard for three consecutive cycles, the offset is determined to be complete and the current compensation parameters are locked.

[0121] In summary, steps 401 to 405 implement a closed-loop optimization technique for dynamic harmonic cancellation. The generation and compensation mechanism for harmonic offset components is triggered by the coordinated adjustment of device parameters, establishing a dynamic cancellation chain of "parameter adjustment-harmonic extraction-reverse superposition." A fundamental-multiple frequency joint decomposition and phase reversal algorithm are used to reconstruct the waveform of the harmonic components. A dynamic distortion amplitude matching mechanism is combined to establish an inverse relationship between the compensation amplitudes. Continuous iterative updates of the time-varying amplitudes form a self-convergent process for the harmonic cancellation amount. This technology overcomes the spectrum fixation bottleneck of traditional static harmonic suppression, achieving phase synchronization tracking and dynamic vector cancellation of harmonic components caused by load fluctuations in variable-frequency equipment. This creates a negative feedback regulation loop for the harmonic distortion rate and compensation amplitude, providing data center power supply systems with adaptive harmonic management capabilities with dynamic tracking capabilities.

[0122] In order to solve the problem of insufficient coordination between data and load prediction and delayed parameter tuning in data center harmonic dynamic compensation, in some embodiments, the step 104 dynamically adds and optimizes the harmonic offset components of each power harmonic in the dynamic coupling model and the harmonic components collected during equipment operation, and generates equipment parameter adjustment values ​​corresponding to the equipment operating condition coefficient in combination with the data center load fluctuation prediction data, including:

[0123] 501. Obtain the harmonic offset components of each order of electric energy recorded in the dynamic coupling model and the harmonic components generated during the operation of the collected equipment;

[0124] In step 501, the power harmonic offset components recorded by the dynamic coupling model refer to the harmonic distortion changes caused by device parameter adjustment stored in the model; the harmonic component refers to the actual harmonic distortion rate measured by the power quality analyzer at the current moment.

[0125] In this embodiment, the recorded 5th harmonic offset component of -0.35% and 7th harmonic offset component of -0.18% are first extracted from the historical database of the dynamic coupling model. Simultaneously, a high-precision power quality analyzer is used to collect the harmonic components of the equipment during operation at a sampling rate of 1000 times per second. For example, the current 5th harmonic distortion rate is 4.7% and the 7th harmonic distortion rate is 2.3%. To ensure data timing consistency, the network time protocol is used to synchronize the model data with the collected data, with the time error controlled to within plus or minus 10 milliseconds.

[0126] 502. Superimpose the waveforms of the harmonic offset components of each order of electric energy and the waveforms of the harmonic components according to a time window to generate a candidate superposition result including the time-varying amplitude and phase of each harmonic;

[0127] In step 502, the time window refers to a preset fixed time length or a dynamically adjusted data processing period; the superposition result refers to the synthetic signal after the waveform of the harmonic offset component and the harmonic component are superimposed in the time domain or frequency domain; the time-varying amplitude and phase refer to the characteristics of the amplitude and phase of the harmonic component in the superimposed waveform changing with time.

[0128] In this embodiment, a 5-minute time window is first selected. The waveform of the 5th harmonic offset component (-0.35%, amplitude 0.35%, phase 180 degrees) is superimposed in the time domain with the waveform of the 5th harmonic distortion component (4.7%, amplitude 4.7%, phase 30 degrees). A waveform synthesis algorithm is used to calculate the composite amplitude to be 4.35%, with a phase shift of 35 degrees. Within each time window, a candidate superposition result matrix is ​​generated containing the time-varying amplitudes and phases of each harmonic. For example, the 5th harmonic amplitude fluctuates from 4.35% to 4.5%, with a phase shift from 35 degrees to 40 degrees.

[0129] 503. Based on the data center load fluctuation prediction data, correlate the amplitude change trend of each time window in the candidate superposition results with the change trend of the equipment operating condition coefficient, and select target superposition results whose amplitude change direction is consistent with the load growth direction in the load fluctuation prediction data;

[0130] In step 503, the load fluctuation prediction data refers to the load change trend output by the ARIMA model established based on historical power consumption; the amplitude change trend refers to the law of the increase or decrease of the harmonic amplitude in the superimposed waveform over time; the equipment operating condition coefficient refers to the normalized parameter that quantifies the load state of the equipment; and the consistent load growth direction means that the harmonic amplitude change trend is in the same direction as the increase or decrease in the load forecast.

[0131] In this embodiment, an ARIMA model is used to predict that IT equipment power consumption will increase by 12% over the next 10 minutes. The 5th harmonic amplitude trend in the superposition results is analyzed. Within the time window, the amplitude increases from 4.35% to 4.5%, a change consistent with the load growth forecast. The operating condition coefficient of the associated equipment, for example, increases from 0.75 to 0.82. Superposition results that meet directional consistency are selected, such as marking time windows T1 to T3 as abnormal operating condition windows.

[0132] 504. Generate an equipment parameter adjustment value corresponding to the equipment operating condition coefficient according to the corresponding relationship between the time-varying amplitude and the phase in the superposition result.

[0133] In step 504, the correspondence between the time-varying amplitude and the phase refers to the correlation characteristic between the harmonic amplitude change and the phase offset; the device parameter adjustment value refers to the optimized inverter carrier frequency, dead time and speed compensation parameter set.

[0134] In an embodiment of the present application, first, a mapping model of the time-varying amplitude and phase is established based on the corresponding relationship between the time-varying amplitude and the phase in the superposition result. For example, for every 0.1% increase in the 5th harmonic amplitude, a phase shift of 1.2 degrees is corresponding. When an amplitude increase of 0.15% and a phase shift of 1.8 degrees are detected, the device parameter adjustment value is generated by a multi-objective particle swarm optimization algorithm: the carrier frequency of the chiller inverter is increased from 10 kHz to 12 kHz to suppress high-frequency harmonics, the dead time is adjusted from 3 microseconds to 3.5 microseconds to reduce switching losses, and the water pump speed compensation is reduced from positive 1.5% to positive 1.0% to balance the system impedance. The parameter adjustment instruction is sent to the device controller via industrial Ethernet. After execution, monitoring shows that the 5th harmonic distortion rate has dropped from 4.5% to 4.1%.

[0135] Here's a specific example:

[0136] In a data center intelligent harmonic compensation scenario, the system detected a surge in load on a distribution unit, causing the 5th harmonic distortion rate to rise from 4.7% to 9.2%. It immediately extracted the 7th harmonic compensation component (2.3A / 168°) and the 5th harmonic component (8.1A / 52°) preset in the dynamic coupling model. Using a time-domain superposition algorithm, it generated a 5th harmonic dynamic spectrum, showing amplitudes of 6.5-8.9A and phase shifts of 45-58° within a 50ms window. Combined with the load forecasting model, the system identified 32 time windows that matched the load growth trend. Of these, windows 15-23 showed an amplitude increase of 0.18A / ms. Based on the phase-amplitude correlation model, a mapping rule was established: every 1° increase in phase corresponds to a 0.15rpm reduction in fan speed, and an amplitude increase of 0.1A / ms triggers a 0.3Hz frequency modulation of the water pump. Within 300ms, the system phased in the fan speed from 3150rpm to 2870rpm and increased the water pump frequency by 2.4Hz, reducing the fifth harmonic distortion rate to 3.1% and stabilizing the total distortion rate within 4.2%. The response speed is 8 times faster than that of traditional solutions, and the parameter fluctuation is compressed to ±0.15%.

[0137] In summary, steps 501 to 504 implement adaptive optimization technology for harmonic control parameters. A time-varying amplitude-phase joint characterization model is constructed through dynamic coupling analysis of harmonic offset components and harmonic amplitude-phase components. Combined with load fluctuation prediction trends, a correlation screening mechanism is established between harmonic evolution direction and equipment operating conditions. Based on the spatiotemporal matching principle between load growth direction and harmonic amplitude variation trends, a dynamic mapping relationship between harmonic compensation parameters and equipment operating condition coefficients is established. Adaptive adjustment values ​​for the equipment control strategy are generated through closed-loop iteration of time-varying amplitude-phase parameters. This technology breaks through the traditional static parameter setting model of harmonic control and forms an intelligent decision-making chain of "harmonic superposition analysis-load trend correlation-amplitude-phase dynamic solution." This enables the coordinated optimization of compensation parameters with load fluctuations and equipment operating conditions, building a dynamic harmonic control system with trend prediction capabilities for data centers.

[0138] In order to solve the technical problems of insufficient correlation between vibration and noise characteristics and low accuracy of energy distribution analysis in the evaluation of the operating conditions of HVAC equipment in data centers, an intelligent evaluation system based on vibration-noise coupling analysis and dynamic analysis of spectral energy has been developed. This system achieves accurate quantification and dynamic feedback optimization of equipment operating condition coefficients, significantly improving the accuracy of equipment status assessment and system energy efficiency management capabilities. In some embodiments, step 103 analyzes the operating conditions of HVAC equipment in the data center based on the correlation results of the vibration modal characteristics and noise amplitude-frequency characteristics in the dynamic coupling model, calculates the spectral energy distribution caused by the working fluid flow and motor in the equipment in combination with the vibration and noise characteristics of the equipment, and converts the spectral energy distribution into the equipment operating condition coefficient and feeds it back to the dynamic coupling model, including:

[0139] 601. Extracting device vibration mode data and device noise frequency data recorded in the dynamic coupling model and associating the vibration mode data and noise frequency data according to a time window;

[0140] In step 601, the device vibration mode data recorded in the dynamic coupling model refers to the spectral feature data collected by the acceleration sensor and obtained through fast Fourier transform processing, which contains the amplitude and phase information of each frequency component; the device noise frequency data refers to the sound pressure level distribution data in the frequency band of 30 Hz to 10 kHz obtained by the acoustic sensor; and time window association refers to aligning the vibration data and noise data according to the same timestamp.

[0141] In an embodiment of the present application, the vibration mode data of the chiller compressor for 24 consecutive hours is first extracted from the dynamic coupling model. For example, in the time window from 10:00 to 10:05 a.m., the 125 Hz frequency amplitude is detected to be 0.5 mm per square second, the 250 Hz frequency amplitude is 1.2 mm per square second, and the 375 Hz frequency amplitude is 0.8 mm per square second. The noise frequency data of the same time window is synchronously acquired. For example, the sound pressure level in the 250 Hz frequency band is 68 decibels, and that in the 375 Hz frequency band is 63 decibels. The network time protocol is used to align the timestamps of the vibration and noise data, and the time synchronization error is controlled within plus or minus 5 milliseconds to ensure that one associated data unit is generated every minute.

[0142] 602. Overlap and compare the frequency corresponding to the peak amplitude in the correlated vibration mode data with the main frequency band in the noise frequency data to screen out a characteristic interval in which the amplitude and frequency change synchronously;

[0143] In step 602, the frequency corresponding to the peak amplitude refers to the characteristic frequency point with the largest amplitude in the vibration spectrum; the main frequency band refers to the continuous frequency range in the noise spectrum where the sound pressure level exceeds the preset threshold; the synchronous change of amplitude and frequency means that the increase or decrease trend of the vibration amplitude is consistent with the change trend of the sound pressure level in the corresponding frequency band.

[0144] In an embodiment of the present application, the vibration pattern data is first analyzed to identify that the peak frequency in the time window from 10:00 a.m. to 10:05 a.m. is 250 Hz, corresponding to an amplitude of 1.2 mm per square second. The noise data of the same time window is checked, and it is found that the sound pressure level in the frequency band of 250 Hz to 300 Hz is continuously higher than 65 decibels. When the vibration amplitude increases from 1.2 mm per square second to 1.5 mm per square second, the sound pressure level of the corresponding noise main frequency band rises from 68 decibels to 72 decibels, and the 250 Hz to 300 Hz interval is determined to be a characteristic interval. The screening rule sets the amplitude change rate to be greater than or equal to 10% and the sound pressure level change to be greater than or equal to 3 decibels as the synchronous change threshold.

[0145] 603. Multiplying the squared amplitude value in the vibration mode data at each time point by the weight coefficient of the corresponding frequency within the characteristic interval and accumulating the calculation results at all time points to generate a frequency spectrum energy distribution corresponding to the fluid flow state and the power plant operation state;

[0146] In step 603, the square of the amplitude multiplied by the frequency weight coefficient refers to quantifying the contribution of vibration energy in different frequency bands through weighted calculation; the spectrum energy distribution refers to a frequency domain energy proportion model that reflects the fluid flow state and the mechanical operation state of the power device.

[0147] In the embodiment of the present application, the frequency weight coefficient is first set: the low frequency band is less than 200 Hz with a weight of 0.3, the medium frequency band is 200 Hz to 500 Hz with a weight of 0.5, and the high frequency band is greater than 500 Hz with a weight of 0.2. In the characteristic interval of 250 Hz to 300 Hz, energy calculation is performed on the vibration data per minute: for example, the amplitude of 1.5 mm per square second at the time point 10:01 is 2.25, which is multiplied by the medium frequency band weight of 0.5 to obtain an energy value of 1.125; the total energy of all time points in the cumulative 10-minute window is 15.6. The fluid flow energy is 9.2 through the 250 Hz component energy value, and the power unit energy is 6.4 through the 375 Hz component energy value. The ratio of the two is calculated to be 1.437.

[0148] 604. Generate an equipment operating condition coefficient according to a ratio of an energy proportion of a fluid flow state to an energy proportion of an operating state of a power device in the spectrum energy distribution;

[0149] In step 604, the energy proportion of the fluid flow state refers to the proportion of broadband vibration energy caused by refrigerant turbulence; the energy proportion of the power unit operation state refers to the proportion of characteristic frequency energy caused by motor bearing rotation; and the equipment operating condition coefficient refers to the normalized ratio of the two types of energy proportions.

[0150] In this embodiment, the total fluid flow energy within the characteristic interval is calculated to be 45.3, while the total power unit energy is 32.7. The energy ratio is 45.3 divided by 32.7, which is approximately 1.386. This ratio is normalized to a working condition coefficient ranging from 0 to 1. A maximum ratio of 2.0 corresponds to a coefficient of 1.0, so 1.386 corresponds to a working condition coefficient of 0.82. When the ratio exceeds 1.5, the working condition coefficient is marked as abnormal and is greater than or equal to 0.9.

[0151] 605. Feedback the equipment operating condition coefficient to the dynamic coupling model and update the correlation parameters between the vibration mode data and the noise frequency data in the dynamic coupling model.

[0152] In step 605 , the correlation parameters between the vibration mode data and the noise frequency data in the dynamic coupling model include frequency band correlation weight, time window length, and abnormality determination threshold.

[0153] In this embodiment of the present application, the operating condition coefficient of 0.82 is first fed back to the dynamic coupling model, and the frequency band correlation weight of the updated characteristic interval of 250 Hz to 300 Hz is increased from 0.5 to 0.6. At the same time, the time window length is shortened from 5 minutes to 3 minutes to improve the response speed, and the abnormality judgment threshold is adjusted from 0.9 to 0.85. The updated model detects an operating condition coefficient of 0.88 in the next monitoring cycle, triggering a warning signal and indicating the risk of refrigerant flow abnormality 15 minutes in advance.

[0154] Here's a specific example:

[0155] In a data center equipment status monitoring scenario, the system collects chiller vibration pattern data and noise spectrum data, and uses a timestamp alignment algorithm to correlate 1,200 vibration peaks per minute with the main frequency band of the soundprint. When the system detects that the peak vibration amplitude of a water pump increases from 0.12mm to 0.35mm within 15 seconds and the main noise frequency shifts from 850Hz to 920Hz, it selects a characteristic interval with synchronous amplitude-frequency changes and selects 18 key time points within a 22-second period to calculate the spectrum energy. The system multiplies the square of the vibration amplitude at each time point by the corresponding frequency weight coefficients: 1.2 for 850Hz and 1.8 for 920Hz. This generates a spectrum distribution in which fluid flow energy accounts for 58% and mechanical transmission energy accounts for 42%. Based on a flow-mechanical energy ratio of 3:2, the equipment operating condition coefficient of 0.67 is generated. The dynamic coupling model adjusts the vibration compensation parameters accordingly, adjusting the inverter output frequency from 45Hz to 42.5Hz, reducing the vibration peak to 0.18mm and stabilizing the noise main frequency at 880±5Hz. The equipment's operating energy efficiency ratio is improved by 23%, and the abnormal state warning response time is shortened to 8 seconds.

[0156] In summary, steps 601 to 605 implement multi-source feature fusion perception technology for equipment operating status. A joint mechanical-acoustic characterization model is constructed through cross-domain correlation analysis of vibration modal spectra and noise band energy. A feature interval screening mechanism based on synchronous amplitude-frequency variation is employed, and a frequency-domain energy-weighted integration algorithm is used to quantify the energy distribution ratio between fluid dynamics and electromechanical operation. Dynamic weight allocation and energy proportion decoupling techniques are used to generate equipment operating condition coefficients, and a closed-loop feedback channel for characteristic parameters is constructed to achieve online optimization of coupling model parameters. This method overcomes the limitations of traditional single-source state assessment, forming an intelligent perception chain of "joint vibration-noise analysis - dynamic energy solution - iterative model evolution," providing highly accurate equipment operating condition characterization capabilities for harmonic dynamic compensation.

[0157] In order to solve the technical problems of insufficient correlation between load fluctuation and harmonic amplitude trend and poor synchronization of equipment control in data center harmonic dynamic compensation, in some embodiments, the step 503 is based on the data center load fluctuation prediction data, and the amplitude change trend of each time window in the candidate superposition result is associated with the change trend of the equipment operating condition coefficient, and the target superposition result whose amplitude change direction is consistent with the load growth direction in the load fluctuation prediction data is screened out, including:

[0158] 701. Extract a load change direction mark corresponding to each time window in the data center load fluctuation prediction data, where the load change direction mark includes a load increase direction and a load decrease direction;

[0159] In step 701, the data center load fluctuation prediction data refers to the load change trend output by the ARIMA model established based on historical power consumption; the time window refers to a preset fixed duration or a dynamically adjusted data processing cycle; the load change direction mark refers to quantifying the load trend as a symbolic growth +1 or decrease -1 state.

[0160] In this embodiment, the ARIMA model is first used to predict load fluctuations over the next 24 hours. The load change direction marker corresponding to each time window in the data center load fluctuation forecast data is extracted. For example, from 10:00 AM to 11:00 AM, the IT equipment power consumption is expected to increase by 8%, which is marked as a load increase direction of +1; from 3:00 PM to 4:00 PM, the power consumption is expected to decrease by 5%, which is marked as a load decrease direction of -1. Each time window is one hour long, and the direction marker is stored as a table corresponding to timestamps and symbol values.

[0161] 702. Obtain time-varying amplitude change data of each harmonic in each time window in the candidate superposition result and time-varying trend data of the equipment operating condition coefficient;

[0162] In step 702, the time-varying amplitude change data of each harmonic in each time window in the superposition result refers to the time fluctuation curve of the harmonic amplitude extracted by fast Fourier transform; the time-varying trend data of the equipment operating condition coefficient refers to the linear regression slope of the operating condition coefficient in the time window.

[0163] In this embodiment, the time-varying amplitude change data of each harmonic within each time window and the time-varying trend data of the equipment operating condition coefficient are first extracted from the superposition results. For example, the amplitude change data of the fifth harmonic in the time window from 10:00 to 11:00 am is linearly increased from 4.3% to 4.7%, and the time-varying amplitude slope is calculated to be +0.4% per hour. The equipment operating condition coefficient data is obtained simultaneously. For example, the operating condition coefficient increases from 0.75 to 0.82, and the time-varying trend slope is +0.07 per hour.

[0164] 703. Match the slope direction of the time-varying amplitude change data with the load change direction mark. If the slope direction of the time-varying amplitude change data has the same sign as the slope direction of the load change direction mark, then determine that the match is successful.

[0165] In step 703, the slope direction of the time-varying amplitude change data refers to the sign of the amplitude change trend calculated by linear regression; the slope direction sign of the load change direction mark refers to the +1 or -1 value of the predicted load increase or decrease; the same sign judgment means that both are positive or both are negative.

[0166] In this embodiment, the time-varying amplitude slope of the fifth harmonic is first calculated to be +0.4% per hour, sign +1. The load change direction in the corresponding time window is marked as +1 increasing. The slope direction of the time-varying amplitude change data is matched with the load change direction mark. Since the two have the same sign, the match is determined to be successful. If the amplitude slope is -0.3% per hour, sign -1, and the load mark is +1, the determination fails.

[0167] 704. According to the determination result of direction consistency, select a target superposition result whose slope direction of the time-varying amplitude change data is consistent with the load growth direction from the superposition results.

[0168] In step 704, the determination result of consistent direction refers to the record whose time-varying amplitude slope matches the load mark sign; the target superposition result of consistent load growth direction refers to the time window data set whose sign is all +1.

[0169] In this example, the results of the overlay analysis during the 10:00 AM to 11:00 AM time window were first screened based on the direction-consistent determination results. The 5th harmonic amplitude slope of +0.4% per hour successfully matched the load marker +1, and the 7th harmonic slope of +0.2% per hour also successfully matched the load marker +1. This data was marked as a valid association set and stored in the training database of the dynamic coupling model for subsequent harmonic suppression strategy optimization.

[0170] Here's a specific example:

[0171] In a data center intelligent harmonic dynamic compensation scenario, the system analyzed load fluctuation forecast data for a power distribution unit over a 300-millisecond window, identifying 68% of time periods with increasing load. Simultaneously, the system extracted the time-varying amplitude data of the fifth harmonic from the compensation superposition results and found that its amplitude slope was +0.15A / ms, consistent with the load growth direction. Using a direction matching algorithm, the superposition results of 18 time windows were screened. Among them, the window with a fifth harmonic amplitude growth rate of 0.21A / ms matched the predicted load growth direction 92% of the time. Based on this information, the system triggered a dynamic adjustment strategy: when the fifth harmonic amplitude slope exceeded 0.2A / ms, the variable-frequency water pump speed was reduced from 2950 rpm to 2730 rpm, and the cooling tower fan frequency was increased by 1.8Hz. After the adjustment, the fifth harmonic distortion rate dropped from 8.7% to 3.4% within 40ms, the total harmonic distortion rate stabilized below 4.1%, and the equipment energy efficiency ratio increased by 19%. Through the slope direction matching mechanism, the system improves the harmonic suppression response speed to 15ms, which is 6 times more efficient than the traditional threshold trigger solution, and compresses the fluctuation range of key parameters to ±0.12%.

[0172] In summary, steps 701 to 704 implement a dynamic correlation screening technique for harmonic control strategies and load trends, and construct an intelligent matching engine through the analysis of the trend evolution equidirectionality of the load direction mark and the time-varying amplitude slope. Based on the sign consistency criterion of the harmonic amplitude change slope and the load growth direction, a spatiotemporal correlation rule for harmonic dynamic characteristics and load fluctuation trends is established, and a trend direction coupling screening mechanism is used to extract superposition results with equidirectional evolution characteristics. This method breaks through the traditional control mode of harmonic compensation and load decoupling, forming an intelligent decision chain of "load trend mark-amplitude slope solution-direction sign matching", achieving precise adaptation of the dynamic weight distribution of compensation parameters to the load fluctuation trend, and providing a harmonic dynamic control enhancement mechanism with the ability to predict trend evolution for data center power supply systems.

[0173] Figure 2 The present invention provides a structural diagram of a data center intelligent harmonic dynamic compensation system. Figure 2 As shown, the system includes:

[0174] Establishing module 21, establishing a dynamic coupling model by combining the collected vibration signals and noise data of the data center infrastructure equipment with the power quality data of the input equipment, wherein the dynamic coupling model includes a mapping relationship between vibration modal characteristics, noise amplitude-frequency characteristics, and various harmonic distortion rates of the power;

[0175] Correlation module 22 analyzes the operating conditions of the HVAC equipment in the data center based on the correlation results between the vibration modal characteristics and the noise amplitude-frequency characteristics in the dynamic coupling model, calculates the spectrum energy distribution caused by the working fluid flow and motor in the equipment based on the vibration and noise characteristics of the equipment, and converts the spectrum energy distribution into equipment operating condition coefficients and feeds them back to the dynamic coupling model;

[0176] An input module 23, based on the nonlinear characteristics of the load-dependent distortion rate of each harmonic of electric energy in the dynamic coupling model and the equipment operating condition coefficient, matches the operating speed range of the fan, water pump, and chiller by using the frequency converter and torque feedback data, dynamically maps the speed difference of the fan, water pump, and chiller to each harmonic offset component of electric energy generated by the superposition of pulsation of the frequency converter and the variable frequency motor in the matching equipment, and inputs the each harmonic offset component of electric energy as a parameter into the dynamic coupling model;

[0177] The generation module 24 dynamically adds and optimizes the harmonic offset components of each power in the dynamic coupling model and the harmonic components collected during equipment operation, and generates an equipment parameter adjustment value corresponding to the equipment operating condition coefficient in combination with the data center load fluctuation prediction data.

[0178] Figure 2 The intelligent harmonic dynamic compensation system for a data center can be implemented Figure 1 The implementation principle and technical effects of the intelligent harmonic dynamic compensation method for a data center described in the illustrated embodiment are not further elaborated. The specific manner in which each module and unit performs operations in the intelligent harmonic dynamic compensation system for a data center in the above embodiment has been described in detail in the embodiments of the method and will not be elaborated on here.

[0179] In one possible design, Figure 2 The data center intelligent harmonic dynamic compensation system of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0180] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0181] The processing component 32 is used for the above Figure 1The embodiment provides a method for intelligent harmonic dynamic compensation in a data center.

[0182] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above method.

[0183] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0184] Of course, a computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0185] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0186] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0187] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0188] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 The embodiment shown is a method for intelligent harmonic dynamic compensation in a data center.

[0189] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0190] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0191] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0192] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A data center intelligent harmonic dynamic compensation method, characterized in that: include: A dynamic coupling model is established by collecting vibration signals and noise data from data center infrastructure equipment and combining it with power quality data from input devices. The dynamic coupling model includes mapping relationships between vibration modal characteristics, noise amplitude-frequency characteristics, and harmonic distortion rates of various power sub-wavelengths. Based on the correlation results between the vibration modal characteristics and the noise amplitude-frequency characteristics in the dynamic coupling model, the operating conditions of the HVAC equipment in the data center are analyzed. The spectrum energy distribution caused by the working fluid flow and motor in the equipment is calculated based on the vibration and noise characteristics of the equipment. The spectrum energy distribution is converted into equipment operating condition coefficients and fed back to the dynamic coupling model. Based on the nonlinear characteristics of the load-dependent distortion rate of each harmonic of electric energy in the dynamic coupling model and the operating condition coefficient of the equipment, the operating speed range of the fan, water pump, and chiller is matched by frequency converter frequency and torque feedback data, so that the speed difference of the fan, water pump, and chiller is dynamically mapped to the offset components of each harmonic of electric energy generated by the superposition of pulsation of the frequency converter and the variable frequency motor in the matching equipment, and the offset components of each harmonic of electric energy are input into the dynamic coupling model as parameters; Dynamically summing and optimizing are performed based on the harmonic offset components of each order of electric energy in the dynamic coupling model and the harmonic components collected during equipment operation, and combining the data center load fluctuation prediction data to generate equipment parameter adjustment values ​​corresponding to the equipment operating condition coefficient.

2. The method according to claim 1, characterized in that Also includes: The device is adjusted using the device parameter adjustment value, and the various electric energy harmonic offset components generated after the device operating parameters are adjusted are superimposed on the harmonic components generated by the currently running device, so that the time-varying amplitude and phase of the electric energy harmonic offset components dynamically offset the harmonic distortion components generated by the coordinated adjustment of load fluctuations by the device motor and inverter.

3. The method according to claim 1, characterized in that Based on the nonlinear characteristics of the power harmonic distortion rate changing with load in the dynamic coupling model and the equipment operating condition coefficient, the operating speed range of the fan, water pump, and chiller is matched by the frequency converter and torque feedback data, so that the speed difference of the fan, water pump, and chiller is dynamically mapped to the power harmonic offset components generated by the pulsation superposition of the frequency converter and the variable frequency motor in the matching equipment, including: Based on the nonlinear characteristics of the harmonic distortion rate of electric energy and load changes in the dynamic coupling model and the operating condition coefficient of the equipment, the operating speed ranges of the fans, pumps, and chillers are matched by using the frequency converter frequency data and torque feedback data. When it is detected that the operating speed range of any device exceeds a preset speed threshold range, the speed difference of the device in adjacent time windows is calculated; Decomposing the speed difference into a fundamental component and multiple frequency components in a time series, and aligning the phases of the decomposed components with the inherent pulsation waveform of the power device to generate a superimposed waveform containing an amplitude variation; The frequency weight coefficient is determined based on the correlation between the harmonic distortion rate and the equipment load in the dynamic coupling model, and the amplitude change of the fundamental component and each harmonic component in the superimposed waveform is multiplied by the frequency weight coefficient, and dynamic mapping is used to generate the harmonic offset components of each order of electric energy corresponding to the pulsation superposition of the power device and the speed adjustment device.

4. The method according to claim 2, characterized in that The device is adjusted using the device parameter adjustment value, and each of the electric energy harmonic offset components generated after the device operating parameter adjustment is added to the harmonic components generated by the currently running device, so that the time-varying amplitude and phase of the electric energy harmonic offset component dynamically offset the harmonic distortion component generated by the device motor and inverter collaboratively adjusting the load fluctuation, including: Based on the equipment parameter adjustment value, synchronously adjust the speed adjustment devices of the fan, water pump, and chiller and the operating parameters of the power unit; extracting the offset components of each harmonic of electric energy generated by the coordinated operation of the speed adjustment device and the power device after the equipment parameters are adjusted; Decomposing the waveform of each harmonic offset component of the electric energy into a fundamental component and a frequency multiplication component, and performing a phase inversion operation on the decomposed components and the harmonic components output by the operating device to generate a superposition result of the harmonic offset components of each harmonic offset component of the electric energy and the harmonic components; According to the superposition result, dynamically matching the time-varying amplitude of the offset component of each harmonic of the electric energy with the distortion amplitude of the harmonic component, and generating an inverse relationship between the changing direction of the time-varying amplitude and the changing direction of the distortion amplitude; Through the inverse relationship, the time-varying amplitude of each harmonic offset component of the electric energy is continuously updated within a preset time window until the time-varying amplitude offsets the harmonic distortion component generated by the equipment motor and inverter cooperating to adjust the load fluctuation.

5. The method according to claim 1, characterized in that Dynamically summing and optimizing the harmonic offset components of each power harmonic in the dynamic coupling model and the collected harmonic components during equipment operation, and combining the data center load fluctuation prediction data to generate equipment parameter adjustment values ​​corresponding to the equipment operating condition coefficient, including: Obtaining the harmonic offset components of each order of electric energy recorded in the dynamic coupling model and the harmonic components generated during the operation of the collected equipment; Superimposing the waveform of each harmonic offset component of the electric energy with the waveform of the harmonic component according to a time window to generate a candidate superposition result including the time-varying amplitude and phase of each harmonic; Based on the data center load fluctuation prediction data, the amplitude change trend of each time window in the candidate superposition results is correlated with the change trend of the equipment operating condition coefficient, and the target superposition results whose amplitude change direction is consistent with the load growth direction in the load fluctuation prediction data are screened out; According to the corresponding relationship between the time-varying amplitude and the phase in the target superposition result, an equipment parameter adjustment value corresponding to the equipment operating condition coefficient is generated.

6. The method according to claim 1, wherein Based on the correlation results between the vibration modal characteristics and the noise amplitude-frequency characteristics in the dynamic coupling model, the operating conditions of the HVAC equipment in the data center are analyzed. The spectrum energy distribution caused by the working fluid flow and motor in the equipment is calculated based on the vibration and noise characteristics of the equipment. The spectrum energy distribution is converted into equipment operating condition coefficients and fed back to the dynamic coupling model, including: Extracting device vibration mode data and device noise frequency data recorded in the dynamic coupling model and associating the vibration mode data and noise frequency data according to a time window; The frequency corresponding to the peak amplitude in the correlated vibration mode data is overlapped and compared with the main frequency band in the noise frequency data to screen out the characteristic interval where the amplitude and frequency change synchronously; In the characteristic interval, the square value of the amplitude in the vibration mode data at each time point is multiplied by the weight coefficient of the corresponding frequency and the calculation results of all time points are accumulated to generate a frequency spectrum energy distribution corresponding to the fluid flow state and the power unit operation state; generating an equipment operating condition coefficient according to a ratio of an energy proportion of a fluid flow state to an energy proportion of an operating state of a power device in the spectrum energy distribution; The equipment operating condition coefficient is fed back to the dynamic coupling model and the correlation parameters between the vibration mode data and the noise frequency data in the dynamic coupling model are updated.

7. The method according to claim 5, characterized in that Based on the data center load fluctuation prediction data, the amplitude change trend of each time window in the candidate superposition results is correlated with the change trend of the equipment operating condition coefficient, and a target superposition result whose amplitude change direction is consistent with the load growth direction in the load fluctuation prediction data is screened out, including: Extracting a load change direction mark corresponding to each time window in the data center load fluctuation prediction data, wherein the load change direction mark includes a load increase direction and a load decrease direction; Obtaining time-varying amplitude change data of each harmonic in each time window in the candidate superposition result and time-varying trend data of the equipment operating condition coefficient; Matching the slope direction of the time-varying amplitude change data with the load change direction mark, and determining that the match is successful if the slope direction of the time-varying amplitude change data has the same sign as the slope direction of the load change direction mark; According to the determination result of direction consistency, a target superposition result whose slope direction of the time-varying amplitude change data is consistent with the load growth direction is screened from the superposition results.

8. A data center intelligent harmonic dynamic compensation system, characterized in that: include: Establish a module that uses the collected vibration signals and noise data of data center infrastructure equipment and combines it with the power quality data of the input equipment to establish a dynamic coupling model, wherein the dynamic coupling model includes the mapping relationship between vibration modal characteristics, noise amplitude-frequency characteristics, and harmonic distortion rates of various power levels; a correlation module that analyzes the operating conditions of the HVAC equipment in the data center based on the correlation results between the vibration modal characteristics and the noise amplitude-frequency characteristics in the dynamic coupling model, calculates the spectrum energy distribution caused by the working fluid flow and motor in the equipment based on the vibration and noise characteristics of the equipment, and converts the spectrum energy distribution into equipment operating condition coefficients to feed back to the dynamic coupling model; An input module, based on the nonlinear characteristics of the load-dependent distortion rate of each harmonic of electric energy in the dynamic coupling model and the operating condition coefficient of the equipment, matches the operating speed range of the fan, water pump, and chiller through the frequency converter and torque feedback data, dynamically maps the speed difference of the fan, water pump, and chiller to each harmonic offset component of electric energy generated by the superposition of pulsation of the frequency converter and the variable frequency motor in the matching equipment, and inputs the each harmonic offset component of electric energy as a parameter into the dynamic coupling model; The generation module dynamically adds and optimizes the harmonic offset components of each power in the dynamic coupling model and the harmonic components collected during equipment operation, and generates equipment parameter adjustment values ​​corresponding to the equipment operating condition coefficient in combination with the data center load fluctuation prediction data.

9. A computing device, characterized in that It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a data center intelligent harmonic dynamic compensation method as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, a data center intelligent harmonic dynamic compensation method as described in any one of claims 1 to 7 is implemented.

Images