UPS uninterruptible power supply parallel redundancy system and control method thereof

Abstract

The application discloses a UPS uninterruptible power supply parallel redundancy system and a control method thereof, and relates to the field of uninterruptible power supplies.The UPS uninterruptible power supply parallel redundancy system comprises a plurality of groups of UPS modules, a plurality of groups of UPS modules being connected in parallel, a monitoring unit, a control unit and a switching unit; the voltage sampling points of each group of UPS modules are connected to the AD conversion module of the monitoring unit via double-shielded coaxial lines; the control unit is connected to the monitoring unit through a PCIe bus; and the switching unit receives signals output from the control unit through a fiber channel.The application realizes real-time dynamic optimization of UPS module load distribution by means of a multi-factor coupling mechanism of a dynamic redundancy capacity distribution model, deeply fuses a temperature correction index and a time attenuation factor, effectively suppresses performance attenuation and local overload in a high-temperature environment, and improves the cooperative adaptation capability of the system to gradual faults and sudden load fluctuations through a joint criterion of historical fault integration and instantaneous over-threshold values.

Description

Technical Field

[0001] This invention belongs to the field of uninterruptible power supplies (UPS), and more specifically, relates to a parallel redundancy system for UPS. Furthermore, this invention also relates to a method for controlling the parallel redundancy of UPS. Background Technology

[0002] UPS (Uninterruptible Power Supply) parallel redundancy systems play a crucial role in ensuring the continuity of power supply. Traditional solutions often employ a fixed-ratio load distribution strategy, allocating the same proportion of load to each parallel UPS module. However, this static distribution mechanism is ill-suited to dynamic load fluctuations, especially prone to causing localized overloads during sudden load changes or module performance degradation. Existing redundancy control systems are mostly based on simple threshold comparisons of instantaneous load rates for switching, lacking compensation mechanisms for historical module failure rates and environmental temperature factors, resulting in insufficiently granular scheduling of redundant resources.

[0003] In existing technologies, the dynamic adjustment coefficient is often updated by superimposing historical fault parameters with fixed weights, failing to incorporate the logarithmic characteristics of the fault accumulation function and the nonlinear response of the overload trigger function, resulting in insufficient sensitivity in weight allocation. Furthermore, the temperature compensation mechanism lacks adaptive parameter settings, making it difficult to quickly converge to a stable range under high-temperature conditions, exacerbating the uneven temperature distribution between modules. Data transmission between the control unit and monitoring unit uses a single cable, making it susceptible to interference in complex electromagnetic environments, causing deviations between AD sampling values ​​and actual load conditions. The triggering conditions for the switching unit rely on fixed thresholds preset by human experience, without dynamic adjustment based on the statistical characteristics of load fluctuations, leading to premature or delayed redundancy switching and reduced overall system reliability. Therefore, there is an urgent need for an adaptive parallel redundancy system that deeply integrates environmental parameters, fault history, and load fluctuation characteristics to overcome the shortcomings of existing technologies in dynamic scheduling, temperature compensation, and redundancy triggering accuracy. Therefore, we propose a UPS uninterruptible power supply parallel redundancy system and its control method. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a machine control method for parallel redundant UPS systems. This method utilizes a multi-factor coupling mechanism of a dynamic redundancy capacity allocation model, deeply integrating temperature correction exponents and time decay factors to achieve real-time dynamic optimization of UPS module load allocation. This effectively suppresses performance degradation and local overload under high-temperature environments. The adaptive adjustment coefficient, based on the logarithmic characteristics of the fault accumulation function and the nonlinear response of the overload trigger function, uses a joint criterion of historical fault integrals and instantaneous threshold exceedances to ensure the stability of weight updates under projection operator constraints. This significantly improves the system's ability to adapt collaboratively to gradual faults and sudden load fluctuations.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A UPS uninterruptible power supply parallel redundancy system includes: several UPS modules connected in parallel, a monitoring unit, a control unit, and a switching unit; the voltage sampling point of each UPS module is connected to the AD conversion module of the monitoring unit via a double-shielded coaxial cable, the control unit and the monitoring unit are connected via a PCIe bus, and the switching unit receives signals output from the control unit via a fiber optic channel;

[0007] The monitoring unit collects the current load rate L of each UPS module in real time. i (t), temperature parameter Θ i (t) and historical failure rate F i (t);

[0008] The control unit calculates the real-time allocation weight U of each module based on a dynamic redundancy capacity allocation formula. i (t), the calculation formula is:

[0009] In the formula, The dynamic adjustment coefficient is determined by L. i The time-series derivative of (t) and F i The adaptive parameters are constructed from the product of the moving averages of (t), where α is the temperature correction exponent, β is the time decay factor, Δt is the duration parameter, and Ω(Θ) is the time duration parameter. i (t) is the temperature compensation function, which satisfies the following relationship:

[0010] Ω(Θ i (t))=1+γ·exp(-k(Θ m -Θ i (t))), where γ is the temperature compensation gain coefficient, k is the temperature attenuation coefficient, and Θ m This is the critical value for the rated operating temperature;

[0011] The control unit outputs the weight allocation result to the switching unit, and the switching unit, upon meeting the trigger condition ∑U i (t)·L i (t)≥δL max Redundancy switching is initiated at a specific time, where δ is the redundancy trigger threshold determined based on the load fluctuation variance, δL max Indicates based on maximum load capacity L max The set trigger conditions.

[0012] Preferably, the dynamic adjustment coefficient Update using the following method:

[0013] 1) Set feature parameters and The constraint boundary constitutes the parameter variation;

[0014] 2) Real-time calculation of instantaneous dynamic coefficient φ s (t)=ω1·f1(F i (t))+ω2·f2(ΔL(t)), where:

[0015] f1(F i (t) is the fault accumulation function, defined as follows:

[0016] f2(ΔL(t)) is the overload trigger function, defined as follows:

[0017] In the formula, φ s (t) Instantaneous dynamic coefficient, reflecting the dynamic adjustment needs of the system at the current moment, used to dynamically adjust the load distribution weight of the UPS modules; ω1 and ω2 weight coefficients are used to adjust the historical failure rate f1(F) respectively. i (t)) and the overload trigger function f2(ΔL(t)) affect the instantaneous dynamic coefficient φ s The degree of influence of (t); This represents the historical failure rate (F) within the time window [t-τ, t]. i The integral of (t) represents the cumulative number of faults during that time period; maxL i (t) represents the maximum load rate of all UPS modules at the current moment; L th τ is the load threshold, representing the maximum allowable load rate of the system. Exceeding this value may trigger overload protection; v is the slope parameter, controlling the sensitivity of the overload trigger function; τ is the time window length.

[0018] 3) Finally, φ(t) is constrained by the projection operator. Interval.

[0019] Preferably, the redundancy triggering threshold determined based on the load fluctuation variance includes:

[0020] 1) Establish a load forecasting model:

[0021] 2) Calculate the second derivative index Statistical characteristics;

[0022] 3) When satisfied The redundancy trigger threshold δ will be adjusted as follows:

[0023]

[0024] In the formula, L pre(t+Δt) represents the predicted load value at time t+Δt, and L(t) represents the current load value at time t. Acceleration α represents the rate of change of load. L and β L These are used to adjust the contributions of the first and second derivatives to the load prediction value, respectively. The statistical characteristics representing the degree of fluctuation in load changes, where X represents the threshold for judging abnormal fluctuations, and δ s This represents the adjusted redundancy trigger threshold, an adaptive threshold adjusted based on load fluctuations. ξ is the adjustment step size coefficient used to control the adjustment range of the redundancy trigger threshold. This indicates the direction of load change, and Δt is the prediction time interval, representing the time range for predicting future load.

[0025] Preferably, the constraint boundary is set to the interval [0.5, 2.5], the projection operator adopts the L2 norm constraint strategy, the time window τ adopts the sliding time window mechanism, the initial length of the sliding time window is 10 seconds and has an adaptive expansion function, the integral operation of the fault accumulation function adopts the trapezoidal integral method, and the overload trigger function activates the adaptive adjustment mechanism of the slope parameter when the load threshold is reached.

[0026] Preferably, the adjustment operation of the redundancy trigger threshold δ triggers the preloading mechanism when the load fluctuation is in the rising edge stage, the value of the adjustment step size coefficient ξ is inversely proportional to the statistical characteristics of the second derivative index, and the correction factor of the load prediction model is dynamically and jointly calibrated by the temperature decay coefficient η and the time decay factor β.

[0027] A UPS (Uninterruptible Power Supply) parallel redundancy control method, the method being implemented based on the aforementioned system, includes the following steps:

[0028] S1. Real-time output current waveforms of each UPS module are simultaneously acquired using no fewer than three sets of LVDT sensors;

[0029] S2. Construct the load balancing evaluation matrix Q = [q ij N×N, where:

[0030] element q ij =Corr(L i (t), L j (t) represents the load correlation between modules, Corr(L) i (t), L j (t) is the correlation coefficient function used to calculate the load variable L. i (t) and L jThe correlation between (t) is in the range of [-1,1][-1,1], and N represents the total number of UPS modules in the matrix;

[0031] S3. Solve the optimization problem using the improved alternating direction multiplier method;

[0032] S4. Initiate the tiered switching procedure based on monitoring data, and simultaneously determine the activation priority level. When ∑U is detected... i (t)·L i (t)≥δ·L max When the time comes, initiate the graded switching procedure;

[0033] Prioritize activating the health indicator H i (t)=e -λ The highest-level module;

[0034] S5. After the switch is completed, perform closed-loop verification: inject a test current pulse into the activated module;

[0035] The above, U i (t) represents the weight allocation value for i, U j (t) represents the weight allocation value for j; L i (t) represents the weight load value of i, and L j (t) represents the weighted load value of j; δ represents the trigger threshold coefficient; L max This represents the maximum load capacity.

[0036] Preferably, the correlation coefficient function is calculated based on covariance analysis within a preset time window, and a moving average algorithm is introduced to eliminate high-frequency interference. An adaptive penalty factor is used in the iterative process of the improved alternating direction multiplier method, which is dynamically adjusted according to the load fluctuation amplitude.

[0037] Preferably, the amplitude of the test current pulse is 10%-15% of the rated current of the activated module, the pulse width is 2-5 power cycles, and the health index is determined by weighted calculation of historical running time, number of failures and temperature rise rate, with a weight ratio of 3:2:1.

[0038] Technical effects and advantages of the present invention: Compared with the prior art, the UPS uninterruptible power supply parallel redundancy system and its control method provided by the present invention have the following effects:

[0039] Dynamic weight allocation optimization: By integrating the temperature correction index and the time decay factor, the control unit dynamically compensates for the performance degradation under high temperature environment when allocating weights in real time. Combined with the logarithmic accumulation characteristics of historical failure rates, the load distribution of each UPS module is more balanced, reducing the risk of single-point overload.

[0040] Adaptive adjustment coefficient update: Based on the nonlinear combination of fault accumulation function and overload trigger function, the instantaneous dynamic coefficient can simultaneously reflect the long-term fault trend and the instantaneous overload degree. By constraining the boundary value through the projection operator, the weight coefficient update process is ensured to converge stably.

[0041] Dynamic adjustment of redundancy trigger threshold: By introducing the second derivative index and abnormal fluctuation judgment threshold from the load prediction model, the system captures the acceleration characteristics and fluctuation statistics of load changes, and adaptively adjusts the trigger threshold, thereby improving the timeliness and accuracy of redundancy switching.

[0042] Temperature compensation gain mechanism: The product design of the exponential temperature compensation function and the temperature decay coefficient enables the compensation gain to increase rapidly near the high temperature operating point, which improves the uniformity of temperature distribution between modules and avoids accelerated device aging caused by thermal stress concentration.

[0043] Anti-interference signal transmission guarantee: The voltage sampling point and the AD conversion module are connected by a double-shielded coaxial cable, and the monitoring data and control signals are transmitted separately by PCIe bus and fiber optic channel, which suppresses the impact of electromagnetic interference on load rate measurement and improves the accuracy of system status perception.

[0044] Enhanced fault tolerance: By integrating the historical failure rate in the dynamic redundancy capacity allocation model and using the threshold cross-judgment of the overload trigger function, the system identifies potential faulty modules in real time and reduces their allocation weight, thereby achieving soft redundancy before triggering a switchover and enhancing the system's fault tolerance in progressive failure scenarios. Attached Figure Description

[0045] Figure 1 This is a diagram of the parallel redundant UPS system architecture of the present invention.

[0046] Figure 2 This is a flowchart of the UPS uninterruptible power supply parallel redundancy control method of the present invention. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0048] This invention provides a parallel redundant UPS system and its control method. Through a multi-factor coupling mechanism of a dynamic redundancy capacity allocation model, it deeply integrates the temperature correction exponent and the time decay factor to achieve real-time dynamic optimization of UPS module load allocation, effectively suppressing performance degradation and local overload under high-temperature environments. The adaptive adjustment coefficient is based on the logarithmic characteristics of the fault accumulation function and the nonlinear response of the overload trigger function. Through a joint criterion of historical fault integrals and instantaneous over-threshold, the stability of weight updates is ensured under projection operator constraints, significantly improving the system's ability to adapt to gradual faults and sudden load fluctuations. The redundancy trigger threshold relies on the acceleration feature extraction of the second derivative in the load prediction model and the statistical characteristics of abnormal fluctuations. Combined with a dynamic iterative mechanism of direction-sensitive parameters and adjustment step size coefficients, it solves the lag problem of traditional threshold strategies in nonlinear scenarios. Through an exponential temperature compensation function and a double-shielded coaxial-fiber composite transmission architecture, the high-temperature compensation gain is improved while suppressing the impact of electromagnetic interference on monitoring accuracy, optimizing the temperature uniformity between modules and ensuring signal transmission reliability. Furthermore, based on the soft redundancy prediction mechanism of historical failure rate integral and overload threshold, the system can complete the dynamic degradation allocation of load weight before hardware switching, thus constructing a hierarchical fault-tolerant system from the mechanism level and expanding the fault protection dimension from a single hardware switching to the full-domain collaboration of dynamic software scheduling.

[0049] UPS uninterruptible power supply parallel redundancy system, such as Figure 1 As shown, it includes: several UPS modules connected in parallel, a monitoring unit, a control unit, and a switching unit; the voltage sampling point of each UPS module is connected to the AD conversion module of the monitoring unit via a double-shielded coaxial cable, the control unit and the monitoring unit are connected via a PCIe bus, and the switching unit receives the signal output from the control unit via a fiber optic channel;

[0050] The monitoring unit collects the current load rate L of each UPS module in real time. i (t), temperature parameter Θ i (t) and historical failure rate F i (t);

[0051] The control unit calculates the real-time allocation weight U of each module based on a dynamic redundancy capacity allocation formula. i (t), the calculation formula is:

[0052] In the formula, The dynamic adjustment coefficient is determined by L. i The time-series derivative of (t) and F i The adaptive parameters are constructed from the product of the moving averages of (t), where α is the temperature correction exponent, β is the time decay factor, Δt is the duration parameter, and Ω(Θ) is the time duration parameter. i(t) is the temperature compensation function, which satisfies the following relationship:

[0053] Ω(Θ i (t))=1+γ·exp(-k(Θ m -Θ i (t))), where γ is the temperature compensation gain coefficient, k is the temperature attenuation coefficient, and Θ m This is the critical value for the rated operating temperature;

[0054] Among them, the dynamic adjustment coefficient Update using the following method:

[0055] 1) Set feature parameters and The constraint boundary constitutes the parameter variation;

[0056] 2) Real-time calculation of instantaneous dynamic coefficient φ s (t)=ω1·f1(F i (t))+ω2·f2(ΔL(t)), where:

[0057] f1(F i (t) is the fault accumulation function, defined as follows:

[0058] f2(ΔL(t)) is the overload trigger function, defined as follows:

[0059] In the formula, φ s (t) Instantaneous dynamic coefficient, reflecting the dynamic adjustment needs of the system at the current moment, used to dynamically adjust the load distribution weight of the UPS modules; ω1 and ω2 weight coefficients are used to adjust the historical failure rate f1(F) respectively. i (t)) and the overload trigger function f2(ΔL(t)) affect the instantaneous dynamic coefficient φ s The degree of influence of (t); This represents the historical failure rate (F) within the time window [t-τ, t]. i The integral of (t) represents the cumulative number of faults during that time period; maxL i (t) represents the maximum load rate of all UPS modules at the current moment; L th τ is the load threshold, representing the maximum allowable load rate of the system. Exceeding this value may trigger overload protection; v is the slope parameter, controlling the sensitivity of the overload trigger function; τ is the time window length.

[0060] 3) Finally, φ(t) is constrained by the projection operator. Interval.

[0061] The control unit outputs the weight allocation result to the switching unit, and the switching unit, upon meeting the trigger condition ∑U,...i (t)·L i (t)≥δL max Redundancy switching is initiated at a specific time, where δ is the redundancy trigger threshold determined based on the load fluctuation variance, δL max Indicates based on maximum load capacity L max The set trigger conditions;

[0062] It should be noted that the redundancy trigger threshold determined based on the load fluctuation variance includes:

[0063] 1) Establish a load forecasting model:

[0064] 2) Calculate the second derivative index Statistical characteristics;

[0065] 3) When satisfied The redundancy trigger threshold δ will be adjusted as follows:

[0066]

[0067] In the formula, L pre (t+Δt) represents the predicted load value at time t+Δt, and L(t) represents the current load value at time t. Acceleration α represents the rate of change of load. L and β L These are used to adjust the contributions of the first and second derivatives to the load prediction value, respectively. The statistical characteristics representing the degree of fluctuation in load changes, where X represents the threshold for judging abnormal fluctuations, and δ s This represents the adjusted redundancy trigger threshold, an adaptive threshold adjusted based on load fluctuations. ξ is the adjustment step size coefficient used to control the adjustment range of the redundancy trigger threshold. This indicates the direction of load change, and Δt is the prediction time interval, representing the time range for predicting future load.

[0068] In addition, temperature parameters include junction temperature parameters and hot spot temperature parameters. The historical failure rate is calculated through the fault code classification weight matrix. The switching unit synchronously triggers the phase change cooling module during redundancy switching. The start-up threshold of the phase change cooling module is associated with the temperature compensation gain coefficient γ. The constraint boundary is set in the interval [0.5, 2.5]. The projection operator adopts the L2 norm constraint strategy. The time window τ adopts the sliding time window mechanism. The initial length of the sliding time window is 10 seconds and has an adaptive expansion function. The integral operation of the fault accumulation function adopts the trapezoidal integral method. The overload trigger function activates the adaptive adjustment mechanism of the slope parameter when the load threshold is reached. The adjustment operation of the redundancy trigger threshold δ triggers the preloading mechanism when the load fluctuation is in the rising edge stage. The value of the adjustment step size coefficient ξ is inversely proportional to the statistical characteristics of the second derivative index. The correction factor of the load prediction model is dynamically jointly calibrated through the temperature decay coefficient η and the time decay factor β.

[0069] This embodiment also provides a UPS uninterruptible power supply parallel redundancy control method, such as Figure 2 As shown, the method is based on the aforementioned UPS uninterruptible power supply parallel redundancy system and includes the following steps:

[0070] S1. Real-time output current waveforms of each UPS module are simultaneously acquired using no fewer than three sets of LVDT sensors;

[0071] S2. Construct the load balancing evaluation matrix Q = [q ij N×N, where:

[0072] element q ij =Corr(L i (t), L j (t) represents the load correlation between modules, Corr(L) i (t), L j (t) is the correlation coefficient function used to calculate the load variable L. i (t) and L j The correlation between (t) is in the range of [-1,1][-1,1], where N represents the total number of UPS modules in the matrix. The correlation coefficient function is calculated based on covariance analysis within a preset time window, and a moving average algorithm is introduced to eliminate high-frequency interference. An adaptive penalty factor is used in the iterative process of the improved alternating direction multiplier method, which is dynamically adjusted with the load fluctuation amplitude.

[0073] S3. Solve the optimization problem using the improved alternating direction multiplier method;

[0074] S4, when ∑U is detected i (t)·L i (t)≥δ·L maxAt that time, the tiered switching procedure is initiated, prioritizing the activation of the health indicator H. i (t)=e -λ The highest-level module;

[0075] S5. After the switch is completed, perform closed-loop verification: inject a test current pulse into the activated module; the amplitude of the test current pulse is 10%-15% of the rated current of the activated module, the pulse width is 2-5 power cycles, and the health index is determined by weighted calculation of historical running time, number of failures and temperature rise rate, with a weight ratio of 3:2:1.

[0076] The above, U i (t) represents the weight allocation value for i, U j (t) represents the weight allocation value for j; L i (t) represents the weight load value of i, and L j (t) represents the weighted load value of j; δ represents the trigger threshold coefficient; L max This represents the maximum load capacity.

[0077] In summary, the present invention has the following effects:

[0078] Dynamic weight allocation optimization: By integrating the temperature correction index and the time decay factor, the control unit dynamically compensates for the performance degradation under high temperature environment when allocating weights in real time. Combined with the logarithmic accumulation characteristics of historical failure rates, the load distribution of each UPS module is more balanced, reducing the risk of single-point overload.

[0079] Adaptive adjustment coefficient update: Based on the nonlinear combination of fault accumulation function and overload trigger function, the instantaneous dynamic coefficient can simultaneously reflect the long-term fault trend and the instantaneous overload degree. By constraining the boundary value through the projection operator, the weight coefficient update process is ensured to converge stably.

[0080] Dynamic adjustment of redundancy trigger threshold: By introducing the second derivative index and abnormal fluctuation judgment threshold from the load prediction model, the system captures the acceleration characteristics and fluctuation statistics of load changes, and adaptively adjusts the trigger threshold, thereby improving the timeliness and accuracy of redundancy switching.

[0081] Temperature compensation gain mechanism: The product design of the exponential temperature compensation function and the temperature decay coefficient enables the compensation gain to increase rapidly near the high temperature operating point, which improves the uniformity of temperature distribution between modules and avoids accelerated device aging caused by thermal stress concentration.

[0082] Anti-interference signal transmission guarantee: The voltage sampling point and the AD conversion module are connected by a double-shielded coaxial cable, and the monitoring data and control signals are transmitted separately by PCIe bus and fiber optic channel, which suppresses the impact of electromagnetic interference on load rate measurement and improves the accuracy of system status perception.

[0083] Enhanced fault tolerance: By integrating historical failure rates in the dynamic redundancy capacity allocation model and applying threshold cross-judgment of the overload trigger function, the system identifies potentially faulty modules in real time and reduces their allocation weight. This achieves soft redundancy before a switchover is triggered, enhancing the system's fault tolerance in progressive failure scenarios.

[0084] Finally, it should be noted that the above description 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 foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An uninterruptible power supply parallel redundant system, characterized in that, The UPS module comprises a monitoring unit, a control unit, and a switching unit, voltage sampling points of each group of the UPS module are connected to an AD conversion module of the monitoring unit via double-shielded coaxial lines, the control unit is connected to the monitoring unit through a PCIe bus, and the switching unit receives a signal output from the control unit through a fiber channel; The control unit outputs a weight distribution result to the switching unit, and the switching unit starts a redundancy switching when a trigger condition is met. The monitoring unit collects the current load rate of each UPS module in real time , temperature parameters and historical failure rate ; The control unit calculates real-time allocation weights of each module based on a dynamic redundant capacity allocation calculation formula The real-time allocation weights The dynamic adjustment coefficient is used in the calculation The calculation is performed; the dynamic adjustment factor by updating: 1) Set characteristic parameters and constitute constraint boundaries for parameter variations; 2) Real-time calculation of instantaneous dynamic coefficients + wherein: For the failure accumulation function, define as, = log; The overload trigger function is defined as follows: = ; In the formula, Instantaneous dynamic coefficient, reflecting the dynamic adjustment requirement of the system at the current moment, used for dynamically adjusting the load distribution weight of the UPS module; And Weight coefficient, used for adjusting the influence degree of the instantaneous dynamic coefficient on the failure accumulation function And overload triggering function ; The influence degree of the instantaneous dynamic coefficient Indicates the integral of the historical failure rate within the time window Indicates the cumulative failure amount within the time window ; The maximum value of the load rate of all UPS modules at the current moment; The load threshold value, indicating the maximum load rate allowed by the system, exceeding which triggers the overload protection; The slope parameter, controlling the sensitivity of the overload triggering function; τ is the time window length; 3) final by the projection operator interval; The redundancy trigger threshold is determined according to a fluctuation variance of the load, and the redundancy trigger threshold comprises:

2. The UPS uninterruptible power supply parallel redundant system of claim 1, wherein, The temperature parameters comprise a junction temperature parameter and a hot spot temperature parameter, the historical failure rate is calculated through a failure code classification weight matrix, the switching unit synchronously triggers a phase change cooling module when the redundancy switching is started, and a start threshold of the phase change cooling module is associated with a temperature compensation gain coefficient γ. 1) Establishing a load prediction model: ; 2) Calculate the second derivative index of the statistical properties; 3) the redundancy trigger threshold δ is adjusted to: when the following is true: ； In the formula, is the predicted load value, indicating the predicted load value at time is the current load value, indicating the actual load value at time t, is the acceleration indicating the rate of change of the load, and are respectively used to adjust the contribution of the first derivative and the second derivative to the predicted load value, is the statistical characteristic indicating the fluctuation degree of the load change, and X is the abnormal fluctuation determination threshold, is the adjusted redundancy triggering threshold, the threshold is adaptively adjusted based on the load fluctuation, and ξ is the adjustment step coefficient, used to control the adjustment range of the redundancy triggering threshold, is the direction of the load change, is the prediction time interval, indicating the time range of predicting the future load.

3. The UPS uninterruptible power supply parallel redundant system of claim 1, wherein, The constraint boundary is set as an interval of [0.5, 2.5], the projection operator adopts an L2 norm constraint strategy, the time window τ adopts a sliding time window mechanism, an initial length of the sliding time window is 10 seconds and the sliding time window has an adaptive expansion function, an integral operation of the failure accumulation function adopts a trapezoidal integral method, and an adaptive adjustment mechanism of a slope parameter is activated when the overload trigger function reaches a load threshold.

4. The UPS uninterruptible power supply parallel redundant system of claim 1, wherein, The adjustment operation of the redundancy trigger threshold δ triggers a preloading mechanism when the load fluctuation is in a rising edge stage, a value of the adjustment step coefficient ξ is inversely proportional to statistical characteristics of the second derivative index, and a correction factor of the load prediction model is dynamically jointly calibrated through a temperature attenuation coefficient and a time attenuation factor.

5. The UPS uninterruptible power supply parallel redundant system of claim 2, wherein, The method is implemented based on the system according to any one of claims 1-5, and comprises the following steps:

6. A method for redundant control of parallel UPSs, characterized in that: S1, synchronously collecting real-time output current waveforms of each UPS module through no less than three groups of LVDT sensors; S2, constructing a load balancing degree evaluation matrix; S3, solving an optimization problem by using an improved alternating direction multiplier method; S5, after the switching is completed, performing closed-loop verification: injecting a test current pulse into the activated module; S4. Start the hierarchical switching procedure according to the monitoring data, and determine the activation priority level. When monitoring data is , start the hierarchical switching procedure, and activate the health index with the highest module first; An amplitude of the test current pulse is 10%-15% of a rated current of the activated module, a pulse width is 2-5 power supply periods, the health degree index is determined by weighted calculation of historical running time, failure times, and temperature rise rate, and a weight ratio is 3:2:

1. The above, is a weight assignment value for i, ; is a weight load value for i, is a weight load value for j; δ is a triggering threshold coefficient; is the maximum load capacity.

7. The UPS uninterruptible power supply parallel redundancy control method of claim 6, wherein, ​

Images