Output conversion system and method for UPS (Uninterrupted Power Supply)

Abstract

The invention discloses an output conversion system and method for a UPS (Uninterrupted Power Supply), and the system comprises a main power input end which is connected with a bridge rectifier through a first electromagnetic relay; the energy storage module comprises a super capacitor group and a parallel lithium battery group, and is coupled with the bus capacitor through a bidirectional DC / DC converter; the input end of the multi-path regulation and control module is connected with the bus voltage sensor, and the output end is connected with the switching execution unit through a second electromagnetic relay group; the dynamic switching unit comprises an I-GBT matrix and a reverse blocking diode array which are respectively connected with the inverter and the bypass circuit in parallel; and the real-time monitoring module is provided with a current harmonic analysis unit and a load break variable detection unit, and the output end of the real-time monitoring module is connected with the ADC port of the central processor. According to the invention, systematic improvement is realized in three dimensions of electric energy quality guarantee, equipment reliability and dynamic response capability through an original multi-path regulation and control architecture and a deep cooperative intelligent control strategy.

Description

Technical Field

[0001] The present invention belongs to the field of UPS uninterruptible power supply, and more specifically, relates to an output conversion system for UPS uninterruptible power supply. At the same time, the present invention also relates to an output conversion method for UPS uninterruptible power supply. Background Art

[0002] Existing uninterruptible power supply (UPS) output conversion systems commonly suffer from switching delays and inadequate transient interference control when dealing with complex grid fluctuations and sudden load changes. Traditional technologies often rely on a single battery backup power source. When the main grid experiences an anomaly, the system must undergo a full reconstruction process involving rectification, DC bus voltage regulation, and inverter output. This process results in power outages of at least 5ms, making it unable to meet the demanding continuous power requirements of precision medical equipment and industrial control systems.

[0003] The more prominent problems are: 1) The switching mechanism is simplistic. In the event of grid flicker or short-term interruption, it only relies on mechanical relays to perform channel switching. A multi-modal redundant protection system has not been established, and the switching process is easily affected by contact arcing and sudden changes in line impedance, resulting in voltage sag; 2) The coordination of energy storage components is inefficient. Traditional lead-acid batteries are subject to charging and discharging rate limitations and cannot quickly respond to high-frequency pulse load demands. At the same time, there is a lack of compensation mechanism for the relaxation effect of supercapacitors, resulting in the efficiency of hybrid energy storage systems decaying too quickly; 3) The fault prediction capability is weak. Existing monitoring systems mostly use simple threshold methods to judge abnormal conditions. They lack dynamic modeling capabilities for harmonic distortion and phase shift gradual faults, making it difficult to generate preventive control strategies before the switching critical point; 4) The switching process lacks coupling interference suppression. The transient overvoltage generated by the IGBT switching action can easily affect surrounding sensitive circuits through electromagnetic coupling, and the traditional RC absorption circuit cannot adapt to the filtering requirements of wide-band high-frequency interference.

[0004] For example, the dual-bus UPS system disclosed in CN114726023A employs a multi-stage power conversion topology but fails to address the dynamic prioritization of multiple power channels. The hybrid energy storage solution proposed in CN110829435A incorporates supercapacitors, but still suffers from a capacity drop caused by the exacerbated lithium polarization effect under high-frequency conditions. Therefore, we propose an output conversion system and method for a UPS uninterruptible power supply. Summary of the Invention

[0005] The purpose of the present invention is to address the shortcomings of the prior art and to propose an output conversion system and method for a UPS uninterruptible power supply. Through an original multi-path control architecture and a deeply coordinated intelligent control strategy, the present invention achieves systematic improvements in three dimensions: power quality assurance, equipment reliability, and dynamic response capability.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] Output conversion system for UPS uninterruptible power supply, including: central processing unit, switching execution unit and bus voltage sensor;

[0008] The main power input is connected to the bridge rectifier through the first electromagnetic relay; the energy storage module includes a supercapacitor group and a parallel lithium battery group, which are coupled to the bus capacitor via a bidirectional DC / DC converter; the multi-path control module has an input connected to the bus voltage sensor and an output connected to the switching execution unit through the second electromagnetic relay group; the dynamic switching unit includes an IGBT matrix and a reverse blocking diode array, which are connected in parallel with the inverter and bypass circuit respectively; the real-time monitoring module is equipped with a current harmonic analysis unit and a load sudden change detection unit, and the output is connected to the ADC port of the central processing unit;

[0009] The multipath control module includes a voltage phase pre-correction subunit and a frequency synchronization compensator, which is used to dynamically generate a multipath redundant signal at the switching moment to shorten the response time of the switching execution unit; the real-time monitoring module generates a load dynamic characteristic curve and maps it to the charge and discharge strategy optimizer of the energy storage module.

[0010] Preferably, the multipath control module includes:

[0011] Power allocation strategy generator, which builds a three-dimensional parameter space model based on historical load fluctuation data;

[0012] The path health assessment unit uses a convolutional neural network to analyze the contact resistance change trajectory of the electromagnetic relay;

[0013] A dynamic weight calculator that takes the current ripple factor and the temperature gradient change rate as input variables;

[0014] The output strategy fuser sorts the power path scores by priority and generates an optimal switching path set.

[0015] Preferably, the switching execution unit further includes:

[0016] The submodule decomposition processor decomposes the switching process into a pre-switching preparation period (t1), a mechanical action period (t2), and an electrical stabilization period (t3);

[0017] The pre-switching preparation period t1 includes the phase angle closed-loop adjustment process of the inverter output frequency;

[0018] The mechanical action period t2 adopts the electromagnetic relay contact pressure feedback compensation mechanism to dynamically optimize the contact closing speed;

[0019] During the electrical stabilization period t3, a multi-band impedance matching circuit is configured to suppress transient circulating currents by injecting damping current.

[0020] Preferably, the dynamically optimized contact closing speed satisfies the following formula:

[0021] Among them, v t is the contact closing end speed, E arc is the arc energy of the previous breaking operation, K weibull is the shape coefficient of the relay life distribution, σ abn is the standard deviation of contact resistance anomaly, ρ is the dynamic resistivity of the contact material, T contact Contact real-time temperature, N cycle It is the statistical value of the number of mechanical operations.

[0022] Preferably, the charge and discharge strategy optimizer of the energy storage module includes:

[0023] Dynamic capacity estimation model based on joint calibration of lithium battery differential voltage curve and supercapacitor relaxation time;

[0024] The segmented SOC compensation algorithm establishes compensation coefficients for high-frequency pulse load and DC offset components respectively;

[0025] The hybrid energy storage efficiency controller uses the following formula to calculate the optimal power distribution ratio:

[0026] Among them, C sc is the rated capacity of the supercapacitor, τ disch arge is the current discharge time constant, C li is the available capacity of the lithium battery, Δl pulse is the pulse current change rate, T sc 、T li are the temperatures of the capacitor and lithium battery respectively.

[0027] Preferably, the real-time monitoring module further includes:

[0028] Harmonic distortion trend predictor, based on wavelet packet decomposition to reconstruct high-frequency distortion components;

[0029] The fault feature fuzzy matching unit uses an improved Hausdorff distance algorithm to compare waveform segments;

[0030] Adaptive sampling rate controller, which starts high-speed sampling mode when the load step changes and operates at a reduced frequency in steady state;

[0031] The improved Hausdorff distance algorithm includes:

[0032] Multi-dimensional feature weighted calculation layer, which assigns different weight coefficients to voltage mutation and current derivative change rate;

[0033] Dynamic time warping path optimizer to compensate for phase shifts of non-stationary signals;

[0034] The output value of the abnormality judgment threshold generator satisfies the relationship:

[0035] in, is the average voltage RMS value in the last 100ms, l' k is the first-order difference sequence of the current waveform, Δθ max is the maximum phase angle deviation of current relative to voltage, and N is the number of data points in the sampling window.

[0036] Preferably, the multipath control module generates a redundant instruction set including three orthogonal control signals during the critical switching phase, and implements the following functions through FPGA:

[0037] The first path is the main power supply fast reconnection channel;

[0038] The second path is the energy storage module instantaneous power enhancement channel;

[0039] The third path is the bypass current active suppression channel;

[0040] The activation weight of each channel is dynamically determined by the path transmission cost function.

[0041] An output conversion method for a UPS uninterruptible power supply is provided. The method is implemented based on the above-mentioned output conversion system for a UPS uninterruptible power supply and comprises the following steps:

[0042] S1, receiving an AC input signal through the main power input terminal, triggering the bridge rectifier to perform three-phase full-wave rectification via the first electromagnetic relay;

[0043] S2, a parallel lithium battery pack and supercapacitor pack configured with energy storage modules, uses a bidirectional DC / DC converter to match the voltage fluctuation of the bus capacitor in real time;

[0044] S3, the voltage phase pre-correction subunit performs phase advance compensation on the inverter output, and the frequency synchronization compensator generates an orthogonal reference signal;

[0045] S4. Analyze the feedback data of the bus voltage sensor through the multi-path control module and dynamically generate a three-way redundant switching instruction set; Step S4 includes activating the multi-channel collaborative operation mode of the main power path, energy storage enhancement path and bypass suppression path at the switching critical point at the same time

[0046] S5. Implement mechanical action optimization in the switching execution unit in stages, and use an IGBT matrix and a reverse blocking diode array to build an active switching protection mechanism.

[0047] Preferably, the step S1 further includes:

[0048] S11: The current harmonic analysis unit of the real-time monitoring module continuously detects the amplitude of the third and fifth harmonic components of the input current, and activates the active filtering function when the total distortion rate exceeds the preset threshold;

[0049] S12: Using a load sudden change detection unit to capture the step change characteristics of the load power, a two-dimensional feature vector consisting of the current rise rate ΔI / Δt and the voltage drop depth ΔV is input into a dynamic response decision tree; the generation of the dynamic response decision tree relies on the correlation mapping between the degree of relay contact wear and the IGBT junction temperature change data recorded in the historical fault data;

[0050] S13: Based on the trend prediction result of the load dynamic characteristic curve, the energy distribution ratio of the supercapacitor group and the lithium battery group is reconfigured through the charge and discharge strategy optimizer.

[0051] Technical effects and advantages of the present invention: Compared with the prior art, the output conversion system for a UPS uninterruptible power supply provided by the present invention has the following effects:

[0052] Multi-path collaborative switching mechanism: By constructing a redundant instruction set of orthogonal control signals, the main power path, energy storage enhancement path, and bypass suppression path form switching channels with complementary electromagnetic characteristics, optimizing the dynamic selection process of energy transmission paths in both the time and frequency domains.

[0053] Dynamic adaptation mechanism of hybrid energy storage: Based on the collaborative modeling of the relaxation voltage characteristics of lithium batteries and the dynamic internal resistance parameters of supercapacitors, it achieves optimal energy distribution under time-varying temperature conditions for load power demands in different frequency bands, overcoming the polarity reversal risk of traditional energy storage devices during high-rate charge and discharge.

[0054] Contact degradation transfer learning mechanism: A convolutional neural network is used to extract the multi-dimensional degradation characteristics of relay contact resistance changes, establish a mapping relationship between contact wear status and optimal drive parameters, and convert the physical degradation process of mechanical components into quantifiable control variables;

[0055] Directed transient energy dissipation mechanism: Utilizing the broadband tuning characteristics of the multi-band impedance matching network, the transient circulating current energy generated by switching is directed to the preset heat sink structure, while the resonance effect caused by the line distributed parameters is offset by the feedforward compensation algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1This is a diagram of the output conversion system architecture of the UPS uninterruptible power supply used in the present invention. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below with reference to specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0058] The present invention provides Figure 1 The output conversion system and method for UPS uninterruptible power supply shown in the figure achieves systematic improvement in three dimensions: power quality assurance, equipment reliability, and dynamic response capability through an original multi-path control architecture and a deeply coordinated intelligent control strategy.

[0059] First, the established dynamic redundant switching mechanism enables the system to complete seamless switching of multiple energy channels in sub-millisecond time, effectively eliminating the power supply gaps existing in traditional technologies; second, the hybrid energy storage model based on the collaborative analysis of material properties and operating status significantly delays the performance degradation rate of energy storage components under high-frequency conditions; third, the health management system with contact state prediction as the core improves the life utilization rate of key components to a new dimension; finally, by constructing an adaptive electromagnetic interference suppression network, the industry-wide problem of transient overvoltage coupling propagation during multi-path switching is overcome.

[0060] An output conversion system for a UPS uninterruptible power supply, characterized by comprising: a central processing unit, a switching execution unit and a bus voltage sensor;

[0061] The main power input terminal is connected to the bridge rectifier through the first electromagnetic relay; the energy storage module includes a supercapacitor group and a parallel lithium battery group, which is coupled to the bus capacitor through a bidirectional DC / DC converter;

[0062] The multi-path control module has an input end connected to the bus voltage sensor and an output end connected to the switching execution unit through the second electromagnetic relay group; the multi-path control module includes:

[0063] Power allocation strategy generator, which builds a three-dimensional parameter space model based on historical load fluctuation data;

[0064] The path health assessment unit uses a convolutional neural network to analyze the contact resistance change trajectory of the electromagnetic relay;

[0065] A dynamic weight calculator that takes the current ripple factor and the temperature gradient change rate as input variables;

[0066] The output strategy fuser sorts the power path scores by priority and generates an optimal switching path set.

[0067] The dynamic switching unit includes an IGBT matrix and a reverse blocking diode array, which are connected in parallel with the inverter and the bypass circuit respectively; the real-time monitoring module is equipped with a current harmonic analysis unit and a load sudden change detection unit, and its output end is connected to the ADC port of the central processing unit;

[0068] The multipath control module includes a voltage phase pre-correction subunit and a frequency synchronization compensator, which are used to dynamically generate multipath redundancy signals at the switching moment, shortening the response time of the switching execution unit. The real-time monitoring module generates a load dynamic characteristic curve and maps it to the energy storage module's charge and discharge strategy optimizer. The energy storage module's charge and discharge strategy optimizer includes:

[0069] Dynamic capacity estimation model based on joint calibration of lithium battery differential voltage curve and supercapacitor relaxation time;

[0070] The segmented SOC compensation algorithm establishes compensation coefficients for high-frequency pulse load and DC offset components respectively;

[0071] The hybrid energy storage efficiency controller uses the following formula to calculate the optimal power distribution ratio:

[0072] Among them, C sc is the rated capacity of the supercapacitor, τ disch arge is the current discharge time constant, C li is the available capacity of the lithium battery, Δl pulse is the pulse current change rate, T sc 、T li are the temperatures of the capacitor and lithium battery respectively.

[0073] The switch execution unit also includes:

[0074] The submodule decomposition processor decomposes the switching process into a pre-switching preparation period (t1), a mechanical action period (t2), and an electrical stabilization period (t3);

[0075] The pre-switching preparation period t1 includes the phase angle closed-loop adjustment process of the inverter output frequency;

[0076] The mechanical action period t2 adopts the electromagnetic relay contact pressure feedback compensation mechanism to dynamically optimize the contact closing speed; the dynamically optimized contact closing speed satisfies the following formula:

[0077] Among them, v t is the contact closing end speed, E arc is the arc energy of the previous breaking operation, K weibull is the shape coefficient of the relay life distribution, σabn is the standard deviation of contact resistance anomaly, ρ is the dynamic resistivity of the contact material, T contact Contact real-time temperature, N cycle It is the statistical value of the number of mechanical operations.

[0078] During the electrical stabilization period t3, a multi-band impedance matching circuit is configured to suppress transient circulating currents by injecting damping current.

[0079] Furthermore, the real-time monitoring module also includes:

[0080] Harmonic distortion trend predictor, based on wavelet packet decomposition to reconstruct high-frequency distortion components;

[0081] The fault feature fuzzy matching unit uses an improved Hausdorff distance algorithm to compare waveform segments;

[0082] Adaptive sampling rate controller, which starts high-speed sampling mode when the load step changes and operates at a reduced frequency in steady state;

[0083] The improved Hausdorff distance algorithm includes:

[0084] Multi-dimensional feature weighted calculation layer, which assigns different weight coefficients to voltage mutation and current derivative change rate;

[0085] Dynamic time warping path optimizer to compensate for phase shifts of non-stationary signals;

[0086] The output value of the abnormality judgment threshold generator satisfies the relationship:

[0087] in, is the average voltage RMS value in the last 100ms, l' k is the first-order difference sequence of the current waveform, Δθ max is the maximum phase angle deviation of current relative to voltage, and N is the number of data points in the sampling window.

[0088] The multipath control module generates a redundant instruction set containing three orthogonal control signals during the critical switching phase and implements the following functions through the FPGA:

[0089] The first path is the main power supply fast reconnection channel;

[0090] The second path is the energy storage module instantaneous power enhancement channel;

[0091] The third path is the bypass current active suppression channel;

[0092] The activation weight of each channel is dynamically determined by the path transmission cost function;

[0093] Above, the transmission cost function expression is:

[0094] Among them, α, β, γ are path selection priority weights (satisfying α+β+γ=1), ΔV is the path voltage drop, V nom is the rated voltage, di / dt is the transient current change rate, l rate is the rated current, t latency is the path response delay, δ is the delay sensitivity coefficient, which is 0.8-1.2;

[0095] In addition, the inverter includes: an interleaved parallel full-bridge topology with an asymmetric gate drive circuit; a third harmonic injection unit that improves the output voltage THD characteristics by adjusting the SPWM modulation ratio; a fault-tolerant reconstruction controller that automatically reorganizes to a three-phase four-wire output mode when any bridge arm fails; and the asymmetric gate drive circuit's rise time to fall time ratio is precisely controlled at 1:1.6.

[0096] This embodiment also proposes an output conversion method for a UPS uninterruptible power supply. The method is based on the above-mentioned output conversion system for a UPS uninterruptible power supply and includes the following steps:

[0097] S1, receiving an AC input signal through a main power input terminal, triggering a bridge rectifier to perform three-phase full-wave rectification via a first electromagnetic relay; step S1 also includes:

[0098] S11: The current harmonic analysis unit of the real-time monitoring module continuously detects the amplitude of the third and fifth harmonic components of the input current, and activates the active filtering function when the total distortion rate exceeds the preset threshold;

[0099] S12: A load sudden change detection unit is used to capture the step change characteristics of the load power. The two-dimensional feature vector composed of the current rise rate ΔI / Δt and the voltage drop depth ΔV is input into a dynamic response decision tree. The generation of the dynamic response decision tree relies on the correlation mapping between the degree of relay contact wear and the IGBT junction temperature change data recorded in the historical fault data.

[0100] S13: Based on the trend prediction result of the load dynamic characteristic curve, the energy distribution ratio of the supercapacitor group and the lithium battery group is reconfigured through the charge and discharge strategy optimizer.

[0101] S2, a parallel lithium battery pack and supercapacitor pack configured with energy storage modules, uses a bidirectional DC / DC converter to match the voltage fluctuation of the bus capacitor in real time;

[0102] S3, the voltage phase pre-correction subunit performs phase lead compensation on the inverter output, and the frequency synchronization compensator generates an orthogonal reference signal; step S3 further includes:

[0103] S31. In the phase lead compensation process, the sliding window average value of the voltage RMS is introduced as a correction reference, and the adjustment step of the compensation angle is inversely proportional to the load power factor;

[0104] S32, using a frequency synchronization compensator to simultaneously generate two orthogonal signals with the same frequency as the fundamental wave of the input voltage and a phase difference of 90 degrees, and using an improved Costas loop to achieve fast phase locking;

[0105] S33. Implement real-time impedance matching calculation during the switching preparation period, and dynamically adjust the injection ratio of the multi-band damping current according to the equivalent impedance on the inverter output side.

[0106] S4. Analyzing feedback data from the bus voltage sensor through the multi-path control module to dynamically generate a three-way redundant switching instruction set. Step S4 includes simultaneously activating a multi-channel collaborative operation mode of the main power path, the energy storage enhancement path, and the bypass suppression path at a critical switching point. The specific process of dynamically generating the three-way redundant switching instruction set in step S4 includes:

[0107] S41. Calculate the transient load capacity score of the main power path, the response speed score of the energy storage path, and the failure rate prediction score of the bypass path based on the three-dimensional parameter space model output by the power allocation strategy generator;

[0108] S42. Perform convolutional neural network feature extraction on the reliability weight of each path based on the relay contact resistance change trajectory data provided by the path health assessment unit;

[0109] S43, generating a priority ranking list for each path by fusing the current ripple coefficient and the ambient temperature parameter through a dynamic weight calculator;

[0110] S44. Simultaneously generate three orthogonal control pulse sequences in the FPGA chip, and transmit the selection signal to the switching execution unit through the optocoupler isolator.

[0111] S5. Implementing mechanical motion optimization in stages in the switching execution unit, and constructing an active switching protection mechanism using an IGBT matrix and a reverse blocking diode array. Implementing mechanical motion optimization in stages in step S5 includes:

[0112] S51. During the pre-switching preparation period, the submodule decomposition processor performs closed-loop frequency tracking on the inverter output voltage and controls the phase difference within the range of ±0.5° through phase shift adjustment.

[0113] S52, triggering the electromagnetic relay contact pressure feedback compensation mechanism during the mechanical action period, and dynamically adjusting the excitation current amplitude of the drive coil according to the real-time data of the contact resistance;

[0114] S53. During the electrical stabilization period, the multi-band impedance matching circuit is activated, high-frequency harmonic energy is absorbed through the adjustable reactor group, and the feedforward compensation algorithm is used to suppress common-mode circulating current.

[0115] In summary, the present invention has the following effects:

[0116] Multi-path collaborative switching mechanism: By constructing a redundant instruction set of orthogonal control signals, the main power path, energy storage enhancement path, and bypass suppression path form switching channels with complementary electromagnetic characteristics, optimizing the dynamic selection process of energy transmission paths in both the time and frequency domains.

[0117] Dynamic adaptation mechanism of hybrid energy storage: Based on the collaborative modeling of the relaxation voltage characteristics of lithium batteries and the dynamic internal resistance parameters of supercapacitors, it achieves optimal energy distribution under time-varying temperature conditions for load power demands in different frequency bands, overcoming the polarity reversal risk of traditional energy storage devices during high-rate charge and discharge.

[0118] Contact degradation transfer learning mechanism: A convolutional neural network is used to extract the multi-dimensional degradation characteristics of relay contact resistance changes, establish a mapping relationship between contact wear status and optimal drive parameters, and convert the physical degradation process of mechanical components into quantifiable control variables;

[0119] Directed transient energy dissipation mechanism: Utilizing the broadband tuning characteristics of the multi-band impedance matching network, the transient circulating current energy generated by switching is directed to the preset heat sink structure, while the resonance effect caused by the line distributed parameters is offset by the feedforward compensation algorithm.

[0120] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. Output conversion system for UPS uninterruptible power supply, characterized in that: include: central processing unit, switching execution unit and bus voltage sensor; A main power input terminal is connected to the bridge rectifier via a first electromagnetic relay; The energy storage module includes a supercapacitor bank and a parallel lithium battery bank, coupled to the bus capacitor via a bidirectional DC / DC converter. The multipath control module has its input connected to the bus voltage sensor and its output connected to the switching execution unit via a second electromagnetic relay bank. The dynamic switching unit includes an IGBT matrix and a reverse blocking diode array, connected in parallel with the inverter and bypass circuit, respectively. The real-time monitoring module is equipped with a current harmonic analysis unit and a load mutation detection unit, and its output end is connected to the ADC port of the central processing unit; The multipath control module includes a voltage phase pre-correction subunit and a frequency synchronization compensator, which is used to dynamically generate a multipath redundancy signal at the switching moment to shorten the response time of the switching execution unit; The real-time monitoring module generates a load dynamic characteristic curve and maps it to the charge and discharge strategy optimizer of the energy storage module.

2. The output conversion system for a UPS uninterruptible power supply according to claim 1, characterized in that: The multipath control module includes: Power allocation strategy generator, which builds a three-dimensional parameter space model based on historical load fluctuation data; The path health assessment unit uses a convolutional neural network to analyze the contact resistance change trajectory of the electromagnetic relay; A dynamic weight calculator that takes the current ripple factor and the temperature gradient change rate as input variables; The output strategy fuser sorts the power path scores by priority and generates an optimal switching path set.

3. The output conversion system for a UPS uninterruptible power supply according to claim 1, characterized in that: The switching execution unit further includes: The submodule decomposition processor decomposes the switching process into a pre-switching preparation period (t1), a mechanical action period (t2), and an electrical stabilization period (t3); The pre-switching preparation period t1 includes the phase angle closed-loop adjustment process of the inverter output frequency; The mechanical action period t2 adopts the electromagnetic relay contact pressure feedback compensation mechanism to dynamically optimize the contact closing speed; During the electrical stabilization period t3, a multi-band impedance matching circuit is configured to suppress transient circulating currents by injecting damping current.

4. The output conversion system for a UPS uninterruptible power supply according to claim 3, characterized in that: The dynamically optimized contact closing speed satisfies the following formula: Among them, v t is the contact closing terminal speed, E arc is the arc energy of the previous breaking operation, K weibull is the shape coefficient of the relay life distribution, σ abn is the standard deviation of contact resistance anomaly, ρ is the dynamic resistivity of the contact material, T contact Contact real-time temperature, N cycle It is the statistical value of the number of mechanical operations.

5. The output conversion system for a UPS uninterruptible power supply according to claim 1, characterized in that: The charge and discharge strategy optimizer of the energy storage module includes: Dynamic capacity estimation model based on joint calibration of lithium battery differential voltage curve and supercapacitor relaxation time; The segmented SOC compensation algorithm establishes compensation coefficients for high-frequency pulse load and DC offset components respectively; The hybrid energy storage efficiency controller uses the following formula to calculate the optimal power distribution ratio: Among them, C sc is the rated capacity of the supercapacitor, τ discharge is the current discharge time constant, C li is the available capacity of the lithium battery, Δl pulse is the pulse current change rate, T sc 、T li are the temperatures of the capacitor and lithium battery respectively.

6. The output conversion system for a UPS uninterruptible power supply according to claim 1, characterized in that: The real-time monitoring module also includes: Harmonic distortion trend predictor, based on wavelet packet decomposition to reconstruct high-frequency distortion components; The fault feature fuzzy matching unit uses an improved Hausdorff distance algorithm to compare waveform segments; Adaptive sampling rate controller, which starts high-speed sampling mode when the load step changes and operates at a reduced frequency in steady state; The improved Hausdorff distance algorithm includes: Multi-dimensional feature weighted calculation layer, which assigns different weight coefficients to voltage mutation and current derivative change rate; Dynamic time warping path optimizer to compensate for phase shifts of non-stationary signals; The output value of the abnormality judgment threshold generator satisfies the relationship: in, is the average voltage RMS value in the last 100ms, l ’ k is the first-order difference sequence of the current waveform, Δθ max is the maximum phase angle deviation of current relative to voltage, and N is the number of data points in the sampling window.

7. The output conversion system for a UPS uninterruptible power supply according to claim 1, characterized in that: The multipath control module generates a redundant instruction set containing three orthogonal control signals during the critical switching phase and implements the following functions through the FPGA: The first path is the main power supply fast reconnection channel; The second path is the energy storage module instantaneous power enhancement channel; The third path is the bypass current active suppression channel; The activation weight of each channel is dynamically determined by the path transmission cost function.

8. An output conversion method for a UPS uninterruptible power supply, characterized in that: The method is implemented based on the output conversion system for an uninterruptible power supply (UPS) according to any one of claims 1 to 7, and comprises the following steps: S1, receiving an AC input signal through the main power input terminal, triggering the bridge rectifier to perform three-phase full-wave rectification via the first electromagnetic relay; S2, a parallel lithium battery pack and supercapacitor pack configured with energy storage modules, uses a bidirectional DC / DC converter to match the voltage fluctuation of the bus capacitor in real time; S3, the voltage phase pre-correction subunit performs phase advance compensation on the inverter output, and the frequency synchronization compensator generates an orthogonal reference signal; S4. Analyzing feedback data from the bus voltage sensor through the multi-path control module to dynamically generate a three-way redundant switching instruction set; Step S4 includes simultaneously activating a multi-channel coordinated operation mode of the main power path, the energy storage enhancement path, and the bypass suppression path at the switching critical point; S5. Implement mechanical action optimization in the switching execution unit in stages, and use an IGBT matrix and a reverse blocking diode array to build an active switching protection mechanism.

9. The output conversion method for a UPS uninterruptible power supply according to claim 8, characterized in that: The step S1 further includes: S11: The current harmonic analysis unit of the real-time monitoring module continuously detects the amplitude of the third and fifth harmonic components of the input current, and activates the active filtering function when the total distortion rate exceeds the preset threshold; S12: Using a load sudden change detection unit to capture the step change characteristics of the load power, a two-dimensional feature vector consisting of the current rise rate ΔI / Δt and the voltage drop depth ΔV is input into a dynamic response decision tree; the generation of the dynamic response decision tree relies on the correlation mapping between the degree of relay contact wear and the IGBT junction temperature change data recorded in the historical fault data; S13: Based on the trend prediction result of the load dynamic characteristic curve, the energy distribution ratio of the supercapacitor group and the lithium battery group is reconfigured through the charge and discharge strategy optimizer.

Images