Power supply system and power supply method based on modular ups power supply

Abstract

The power supply system and method based on modular UPS power supplies include: collecting operating parameters of UPS modules; obtaining the load carrying priority of each UPS module based on the operating parameters; collecting the operating power of the target load; obtaining the target load power of the powered-on UPS module based on the load carrying priority, operating parameters, and operating power of each UPS module; performing real-time monitoring of the powered-on UPS module; marking the powered-on UPS module as normal or faulty based on the real-time monitoring results; performing trend analysis and secondary monitoring on the powered-on UPS module in the normal state; and performing a UPS module switching operation when the powered-on UPS module is marked as faulty or in a warning state, while performing fluctuation detection on the target load and performing adaptive UPS module scheduling operation based on the detection results, significantly improving the overall efficiency and reliability of the UPS modules.

Description

Technical Field

[0001] This invention relates to the field of uninterruptible power supply (UPS) technology, specifically to a power supply system and method based on modular UPS power supplies. Background Technology

[0002] With the rapid development of digital and information technologies, data centers, industrial production, and other fields are increasingly reliant on the continuity and stability of power supply systems. As a critical power supply guarantee device, the performance of UPS power supplies directly affects the safe operation of the load equipment. Traditional UPS power supplies mostly adopt a centralized architecture, which has the following technical drawbacks:

[0003] Poor scalability: The power and capacity of centralized UPS are fixed. If the load demand increases, the entire equipment needs to be replaced, which is costly and complicated to construct. It cannot flexibly adapt to dynamic changes in load.

[0004] Low reliability: Centralized UPS has a single-unit architecture. Failure of any core component will cause the entire system to fail. It has no redundancy backup capability and cannot meet the requirements of high reliability scenarios.

[0005] Coarse load distribution: It relies on a simple current sharing strategy and cannot integrate multi-dimensional indicators such as module health (temperature rise, cumulative running time) and real-time efficiency, resulting in both overload aging of high-load modules and energy waste of low-load modules.

[0006] Delayed fault response: It relies solely on threshold exceeding to trigger fault isolation, lacks trend analysis based on time-series data, and cannot provide early warnings before indicators approach the threshold, resulting in a high risk of fault propagation.

[0007] Insufficient dynamic adaptability: It lacks adaptive load rate adjustment in the face of load fluctuations.

[0008] Existing technologies lack "data-driven dynamic collaborative control"—they cannot optimize load distribution through multi-source data, nor can they predict faults based on historical time series. Ultimately, the flexibility advantage of modular UPS is offset by its shortcomings in reliability, efficiency, and maintenance costs. Summary of the Invention

[0009] To address the aforementioned technical problems, the present invention aims to provide a power supply system based on a modular UPS power supply, comprising a central controller, which is communicatively connected to several UPS modules, a load selection module, a load distribution module, a power-on UPS detection module, a module fault handling module, and a load detection module; the several UPS modules are connected to the same energy storage unit, which is used to store electrical energy.

[0010] The UPS module includes a rectifier, an inverter, a bidirectional DC / DC converter, a local control module, and a status monitoring module. The rectifier converts AC mains power to DC mains power, the inverter converts DC mains power to AC mains power required by the load, the bidirectional DC / DC converter enables bidirectional energy flow between the UPS module and the energy storage unit, the local control module operates the UPS module based on the target load power, and the status monitoring module collects the UPS module's operating parameters, marks the collection time, and sets the collection period. The operating parameters include rated power, input voltage, input current, output voltage, output current, module temperature rise, cumulative operating time, and real-time efficiency. The real-time efficiency is the ratio of the active power output by the UPS module to the load to the total active power obtained by the UPS module from the input side (the input side includes mains power, photovoltaic power, and energy storage unit).

[0011] The load selection module is used to obtain the load carrying priority of each UPS module based on the operating parameters;

[0012] The load distribution module is used to collect the operating power of the target load and obtain the target load power of the powered-on UPS module based on the load carrying priority, operating parameters and operating power of each UPS module.

[0013] The UPS power-on detection module is used to perform real-time detection on the UPS power-on module. Based on the real-time detection results, the UPS power-on module is marked as normal or faulty. For the UPS power-on module in normal state, trend analysis and secondary detection are performed. Based on the secondary detection results, it is determined whether to mark the UPS power-on module as a warning state or update the evaluation index of the selected module.

[0014] The module fault response module is used to perform a UPS module switching operation when the powered-on UPS module is marked as faulty or in a warning state.

[0015] The load detection module is used to detect fluctuations in the target load and execute adaptive power-on UPS module scheduling operations based on the detection results.

[0016] Furthermore, the process of obtaining the load-bearing priority of each UPS module based on operating parameters includes:

[0017] The rated power, real-time efficiency, module temperature rise, and cumulative operating time of each UPS module are extracted from the operating parameters of each UPS module as evaluation indicators. The indicator weights are set (the indicator weights of each evaluation indicator are determined based on the experience of experts to reduce the uncertainty in the weighted calculation process). Based on the evaluation indicators and indicator weights, the load bearing priority of each UPS module is obtained through a weighted average formula (the smaller the module temperature rise, the larger the cumulative operating time, and the larger the rated power, the greater the load bearing priority).

[0018] Furthermore, the process of obtaining the target load power of the powered-on UPS module based on the load carrying priority, operating parameters, and operating power of each UPS module includes:

[0019] Sort the load carrying priorities of each UPS module in ascending order (the higher the carrying priority, the higher the ranking), generate a carrying queue, obtain the preset load rate and the rated power of each UPS module in the carrying queue, obtain the required number of UPS modules K based on the preset load rate, the rated power of each UPS module and the operating power of the target load (the sum of the rated power of the first K UPS modules in the carrying queue multiplied by the preset load rate is greater than or equal to the operating power of the target load, and the sum of the rated power of the first K-1 UPS modules in the carrying queue multiplied by the preset load rate is less than the operating power of the target load), allocate the load to the first K UPS modules in the carrying queue according to the rated power of each UPS module, the preset load rate and the operating power of the target load, obtain the target load power of the first K UPS modules (the sum of the target load power of the first K UPS modules multiplied by the preset load rate is equal to the operating power of the target load), and mark the first K UPS modules as powered-on UPS modules.

[0020] Furthermore, the process of real-time monitoring of the powered-on UPS module includes:

[0021] The operating parameters of the powered-on UPS module are compared with the preset threshold ranges of each type of indicator. If all types of indicators of the powered-on UPS module are within the corresponding preset threshold ranges, the powered-on UPS module is marked as normal. If any type of indicator of the powered-on UPS module is not within the corresponding preset threshold range, the powered-on UPS module is marked as faulty.

[0022] Furthermore, the process of performing trend analysis on a normally powered-on UPS module includes:

[0023] At the end timestamp of the collection period, features of various indicators of the powered-on UPS module are extracted to obtain a set of feature parameters (including the standard deviation of each type of indicator and the Pearson correlation coefficient between each type of indicator). A running prediction model is built based on deep learning. Historical running logs are obtained. The numerical time series sequence and feature parameter set of various indicators of several UPS modules in several historical continuous collection periods are obtained from the historical running logs as training data. The running prediction model is trained using the training data to obtain the running prediction model after training.

[0024] Input the time series of values ​​and feature parameter sets of various types of indicators of the powered-on UPS module within the acquisition period into the running prediction model, and output the predicted time series of values ​​of various types of indicators of the powered-on UPS module in the next acquisition period based on the running prediction model.

[0025] Furthermore, the process of performing secondary testing on a normally powered-on UPS module includes:

[0026] The estimated time series of various indicators of the powered-on UPS module are compared with the preset threshold range of each indicator. If any indicator is not within the corresponding preset threshold range, the powered-on UPS module is marked as a warning state.

[0027] If all indicators of the powered-on UPS module are within the corresponding preset threshold range, the time series of the estimated values ​​of real-time efficiency and module temperature rise in each type of indicator are compared with the real-time efficiency and module temperature rise in the evaluation indicators of the selected load module, respectively, to obtain the average magnitude of the evaluation indicator change. The average magnitude of the evaluation indicator change includes the average magnitude of the real-time efficiency change and the average magnitude of the module temperature rise change, with a preset upper limit for the magnitude change (based on the fluctuation range of the indicators of the UPS module under normal conditions). If the average magnitude of the evaluation indicator change is greater than the upper limit of the magnitude change, the evaluation indicators of the selected load module are updated according to the average magnitude of the evaluation indicator change.

[0028] Furthermore, the process of performing the power-on UPS module switching operation includes:

[0029] The input and output of the powered-on UPS module are disconnected, and the target load power of the powered-on UPS module is obtained. Based on the target load power of the powered-on UPS module, the preset load rate, and the rated power of each UPS module ranked after the powered-on UPS module in the load sequence, the required number of UPS modules R is obtained (the sum of the rated power of the first R UPS modules ranked after the powered-on UPS module in the load queue multiplied by the preset load rate is greater than or equal to the target load power of the UPS module in the fault state or warning state, and the sum of the rated power of the first R-1 UPS modules ranked after the powered-on UPS module in the load queue multiplied by the preset load rate is less than the target load power of the UPS module in the fault state or warning state). Based on the rated power of the first R UPS modules ranked after the powered-on UPS module, the target load power of the powered-on UPS module, and the preset load rate, the load is allocated, and the target load power of the first R UPS modules is obtained (the sum of the target load power of the first R UPS modules multiplied by the preset load rate is equal to the target load power of the powered-on UPS module). The local control module runs the first R UPS modules based on the target load power of the first R UPS modules and marks the first R UPS modules as powered-on UPS modules.

[0030] Furthermore, the process of detecting fluctuations in the target load and executing adaptive power-on UPS module scheduling operations based on the detection results includes:

[0031] The operating power of the target load at each moment is compared with the operating power at the previous moment. A preset error redundancy threshold (based on the accuracy error setting of the data collected by the load detection module) is used to obtain the operating power difference (the operating power at the current moment minus the operating power at the previous moment). When the operating power difference is greater than the error redundancy threshold, if the operating power difference is positive, the redundant power of each powered UPS module is obtained based on the rated power and preset load rate of each powered UPS module (redundant power = rated power - target load power × preset load rate). It is then determined whether the operating power difference is greater than the sum of the redundant power of each powered UPS module. If it is greater, the system is then adjusted according to the preset load rate and the non-powered UPS modules in the load queue. The required number of UPS modules Q is obtained from the rated power and operating power difference of the UPS modules (the sum of the rated power of the first Q non-powered UPS modules in the load queue multiplied by the preset load rate is greater than or equal to the operating power difference, and the sum of the rated power of the first Q-1 non-powered UPS modules in the load queue multiplied by the preset load rate is less than the operating power difference). The load is allocated according to the preset load rate, the operating power difference, and the rated power of the first Q non-powered UPS modules in the load queue. The target load power of the first Q UPS modules is obtained (the sum of the target load power of the first Q UPS modules multiplied by the preset load rate is equal to the operating power difference), and the first Q UPS modules are marked as powered UPS modules.

[0032] If it is not greater than, then the redundant load is allocated to each powered UPS module according to the current operating power difference, the preset load rate and the redundant power of each powered UPS module, and the redundant load of each powered UPS module is obtained (the sum of the redundant loads of each powered UPS module multiplied by the preset load rate equals the operating power difference). The target load power of the powered UPS module at the current moment is obtained according to the redundant load of each powered UPS module (the target load power at the current moment = the target load power at the previous moment + the redundant load).

[0033] Furthermore, if the operating power difference is negative, the preset load rate is reduced based on the operating power difference and the rated power of each powered UPS module (sum of rated power of each powered UPS module × reduced preset load rate = operating power at the previous moment - operating power difference).

[0034] The power supply method based on modular UPS power supplies includes the following steps:

[0035] Step 1: Collect the operating parameters of the UPS module and mark the collection time, and set the collection period;

[0036] Step 2: Obtain the load carrying priority of each UPS module based on the operating parameters;

[0037] Step 3: Collect the operating power of the target load, and obtain the target load power of the powered-on UPS module based on the load carrying priority, operating parameters and operating power of each UPS module;

[0038] Step 4: Perform real-time monitoring of the powered-on UPS module. Based on the real-time monitoring results, mark the powered-on UPS module as normal or faulty. Perform trend analysis and secondary monitoring on the powered-on UPS module in normal condition. Based on the secondary monitoring results, determine whether to mark the powered-on UPS module as a warning or update the evaluation indicators of the selected load-bearing module.

[0039] Step 5: When the powered-on UPS module is marked as faulty or in a warning state, perform a power-on UPS module switching operation, and perform fluctuation detection on the target load. Based on the detection results, perform an adaptive power-on UPS module scheduling operation.

[0040] Compared with the prior art, the beneficial effects of the present invention are:

[0041] 1. Existing technologies often rely on a single indicator (such as rated power) to determine module load capacity, which can easily lead to modules with poor health and low efficiency being forced to bear high loads, or healthy modules remaining idle at low loads for extended periods. This invention integrates multiple evaluation indicators, including rated power, real-time efficiency, module temperature rise, and cumulative operating time, into the load selection process. It also combines these indicators with weighted metrics to quantify the overall load capacity of modules, generating a scientific load queue. This ensures that modules with "good health, high efficiency, and low loss" are prioritized for power supply, preventing module resource mismatch due to biased decision-making from the outset and ensuring a better match between load capacity and actual module capabilities.

[0042] 2. Existing technologies often employ a simple load-sharing strategy, neglecting individual module differences and load requirements. This can easily lead to problems such as some modules operating under overload (accelerated aging) and others operating under low load and inefficiently (high energy consumption). This invention, through a load distribution module based on a load queue, preset load rate, and target load power, precisely distributes the load to the top K optimal modules. This avoids modules exceeding their hardware tolerance limits due to overload, reduces energy waste from prolonged low-load operation, and maintains the load rate of each participating module within an efficient range. Overall, this extends the module lifespan and improves system energy efficiency.

[0043] 3. Existing technologies for UPS module detection are mostly limited to static threshold judgments of "whether indicators exceed limits," which can only respond passively after a fault occurs and cannot identify the gradual risks of indicators approaching the threshold in advance, easily leading to fault propagation or sudden shutdown. This invention uses a power-on UPS detection module to perform trend analysis and secondary detection on modules in normal state: by using an operational prediction model to predict the changes in indicators in the next cycle, it can identify potential risks of "currently normal but possibly exceeding limits in the future" in advance and mark them as warning states; at the same time, by monitoring the average magnitude of changes in evaluation indicators, it dynamically updates the load-bearing evaluation standards, avoiding decision-making lag caused by the slow degradation of module performance, realizing the transformation from "passive fault response" to "proactive risk prevention and control," and significantly reducing the probability of power outages caused by sudden faults.

[0044] 4. Existing technologies often require manual intervention to determine the load transfer plan for the faulty module in the event of a module failure. This switching process is time-consuming, prone to temporary power loss, or can lead to overload of the newly deployed module due to improper load allocation. This invention automatically disconnects the input and output of the faulty module in response to a module failure. Based on the target load power of the faulty module, the preset load rate, and the rated power of subsequent modules, it quickly calculates the required number of new modules and completes precise load allocation. The entire switching process requires no manual intervention and ensures that the load rate of the newly deployed module is controlled within a safe range. This achieves seamless connection between the faulty module and the backup module, minimizing the impact of the failure on the load power supply and ensuring the continuous and stable operation of the load equipment.

[0045] 5. Existing technologies often rely on fixed module configurations or manual adjustment of module numbers when dealing with load power fluctuations. This leads to either insufficient redundancy causing unstable power supply or excessive redundancy resulting in resource waste, failing to dynamically match load changes. This invention uses a load detection module to monitor load power fluctuations in real time and flexibly schedules modules based on the sign and magnitude of the power difference: When power surges, the redundant power of the powered modules is utilized first; if redundancy is insufficient, backup modules are automatically called from the load queue. When power drops, the preset load rate is reduced to minimize ineffective energy consumption by the modules. This adaptive scheduling method requires no manual intervention and can quickly respond to load fluctuations of varying magnitudes, avoiding both power supply pressure due to insufficient redundancy and module resource waste due to excessive redundancy, ensuring that the system's power supply capacity always matches dynamic load demands.

[0046] 6. Existing module evaluation metrics are mostly statically set, failing to consider the impact of performance degradation (such as efficiency decline and adaptive changes in temperature rise threshold) during long-term module operation. As operating time progresses, static metrics gradually become out of sync with the actual module state, leading to load decision failure. This invention, through a secondary detection step, dynamically updates the evaluation metrics for load-bearing modules based on the estimated average magnitude of changes in real-time efficiency and module temperature rise. If the magnitude exceeds the upper limit, the evaluation criteria for load-bearing selection modules are updated synchronously, ensuring that the basis for load queue generation and load allocation always aligns with the actual module state during long-term operation. This avoids decision-making biases caused by static metrics and guarantees the long-term stability and reliability of the system. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of a power supply system based on a modular UPS power supply, according to an embodiment of this application.

[0048] Figure 2 This is a schematic diagram of a power supply method based on a modular UPS power supply according to an embodiment of this application. Detailed Implementation

[0049] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0050] like Figure 1 As shown, the power supply system based on modular UPS power supply includes a central controller, which is communicatively connected to several UPS modules, a load selection module, a load distribution module, a power-on UPS detection module, a module fault handling module, and a load detection module; several UPS modules are connected to the same energy storage unit, which is used to store electricity.

[0051] The UPS module includes a rectifier, an inverter, a bidirectional DC / DC converter, a local control module, and a status monitoring module. The rectifier converts AC mains power to DC mains power, the inverter converts DC mains power to AC mains power required by the load, the bidirectional DC / DC converter enables bidirectional energy flow between the UPS module and the energy storage unit, the local control module operates the UPS module based on the target load power, and the status monitoring module collects the UPS module's operating parameters, marks the collection time, and sets the collection period. The operating parameters include rated power, input voltage, input current, output voltage, output current, module temperature rise, cumulative operating time, and real-time efficiency. The real-time efficiency is the ratio of the active power output by the UPS module to the load to the total active power obtained by the UPS module from the input side (the input side includes mains power, photovoltaic power, and energy storage unit).

[0052] The load selection module is used to obtain the load carrying priority of each UPS module based on the operating parameters;

[0053] The load distribution module is used to collect the operating power of the target load and obtain the target load power of the powered-on UPS module based on the load carrying priority, operating parameters and operating power of each UPS module.

[0054] The UPS power-on detection module is used to perform real-time detection on the UPS power-on module. Based on the real-time detection results, the UPS power-on module is marked as normal or faulty. For the UPS power-on module in normal state, trend analysis and secondary detection are performed. Based on the secondary detection results, it is determined whether to mark the UPS power-on module as a warning state or update the evaluation index of the selected module.

[0055] The module fault response module is used to perform a UPS module switching operation when the powered-on UPS module is marked as faulty or in a warning state.

[0056] The load detection module is used to detect fluctuations in the target load and execute adaptive power-on UPS module scheduling operations based on the detection results.

[0057] It should be further explained that, in the specific implementation process, the process of obtaining the load carrying priority of each UPS module based on operating parameters includes:

[0058] The rated power, real-time efficiency, module temperature rise, and cumulative operating time of each UPS module are extracted from the operating parameters of each UPS module as evaluation indicators. The indicator weights are set (the indicator weights of each evaluation indicator are determined based on the experience of experts to reduce the uncertainty in the weighted calculation process). Based on the evaluation indicators and indicator weights, the load bearing priority of each UPS module is obtained through a weighted average formula (the smaller the module temperature rise, the larger the cumulative operating time, and the larger the rated power, the greater the load bearing priority).

[0059] It should be further explained that, in the specific implementation process, the process of obtaining the target load power of the powered-on UPS module based on the load carrying priority, operating parameters, and operating power of each UPS module includes:

[0060] Sort the load carrying priorities of each UPS module in ascending order (higher priority, higher ranking), generate a load queue, obtain the preset load rate and the rated power of each UPS module in the load queue, and determine the required number of UPS modules K based on the preset load rate, the rated power of each UPS module, and the operating power of the target load (the sum of the rated power of the first K UPS modules in the load queue multiplied by the preset load rate is greater than or equal to the operating power of the target load, and the sum of the rated power of the first K-1 UPS modules in the load queue multiplied by the preset load rate is less than the operating power of the target load). Then, assign loads according to the rated power of each UPS module, the preset load rate, and the operating power of the target load. The first K UPS modules in the queue are assigned loads, and the target load power of the first K UPS modules is obtained (the sum of the target load power of the first K UPS modules multiplied by the preset load rate equals the operating power of the target load). For example, a power supply system based on modular UPS power supply is applied to a medium-sized data center. The data center load includes a server cluster (resistive load, rated power 600kW), the preset load rate is 66.7%, the rated power of each UPS module is 150kW, the server load of 600kW is allocated to 6 UPS modules, the target load power of each UPS module is 100kW, the load rate is 66.7%, and the first K UPS modules are marked as powered-on UPS modules.

[0061] It should be further explained that, in the specific implementation process, the preset load rate of the present invention is set in the high-efficiency operating range of the UPS module, such as 66.7%. Keeping the preset load rate unchanged allows the first K UPS modules participating in power supply to maintain this high-efficiency state, avoiding inefficient operation caused by load rate fluctuations, reducing energy waste, and the stable load rate makes the operating status of each UPS module predictable, providing a unified calculation benchmark for subsequent fault switching and load fluctuation scheduling.

[0062] It should be further explained that, in the specific implementation process, the real-time monitoring of the powered-on UPS module includes:

[0063] The operating parameters of the powered-on UPS module are compared with the preset threshold ranges of each type of indicator (based on the rated configuration of each UPS module, for example, the rated configuration of a 150kVA UPS module is an input voltage of 380V±10% and an output voltage of 380V±2%, and the maximum percentage fluctuation around the rated value constitutes the threshold range). If all types of indicators of the powered-on UPS module are within the corresponding preset threshold ranges, the powered-on UPS module is marked as normal. If any type of indicator of the powered-on UPS module is not within the corresponding preset threshold range, the powered-on UPS module is marked as faulty.

[0064] It should be further explained that, in the specific implementation process, the trend analysis of the UPS module under normal power-on status includes:

[0065] At the end timestamp of the collection period, features of various indicators of the powered-on UPS module are extracted to obtain a set of feature parameters (including the standard deviation of each type of indicator and the Pearson correlation coefficient between each type of indicator). A running prediction model is built based on deep learning. Historical running logs are obtained. The numerical time series sequence and feature parameter set of various indicators of several UPS modules in several historical continuous collection periods are obtained from the historical running logs as training data. The running prediction model is trained using the training data to obtain the running prediction model after training.

[0066] Building a deep learning-based prediction model is a complex process involving multiple steps such as model selection, training, validation, and testing. The following is a detailed explanation of this process:

[0067] This invention selects a fusion model of Long Short-Term Memory (LSTM) network + Fully Connected (FC) layer, which is suitable for processing numerical time series and fusion feature parameter sets, as the deep learning architecture. Based on the LSTM network, it captures temporal dependencies. Based on the FC layer, it organizes the standard deviation of each indicator and the Pearson correlation coefficient between indicators into vectors of fixed dimensions. Fully connected layer mapping: through 2-3 fully connected layers, the statistical feature vector is mapped to a feature vector with the same dimension as the LSTM output, realizing the dimensional alignment of temporal features (LSTM output) and statistical features (FC output). The aligned two types of features are concatenated and the prediction result is output through the last fully connected layer.

[0068] After determining the model architecture, the next step is to define the loss function. This invention uses a double-weighted mean squared error loss function. The prepared training set is then input into the selected deep learning model to begin training. During training, the weights are continuously updated using the backpropagation algorithm, causing the loss function to gradually decrease until it reaches a stable state. During this period, techniques such as early stopping are used to avoid overfitting. In addition to the basic training process, grid search is used to fine-tune various model parameters, including the learning rate, batch size, and regularization coefficient.

[0069] Once the model training is complete and the parameters are tuned, a final evaluation is performed using a test set to obtain the model's evaluation results. These results include classification metrics such as accuracy, recall, and F1 score. Based on the evaluation results on the test set, it is determined whether the model has met the expected standards. If it meets the requirements, the model parameters are saved and deployment is prepared; otherwise, it is necessary to return to a previous stage to re-examine issues such as data quality, model structure, or training strategy.

[0070] Input the time series of values ​​and feature parameter sets of various types of indicators of the powered-on UPS module within the acquisition period into the running prediction model, and output the predicted time series of values ​​of various types of indicators of the powered-on UPS module in the next acquisition period based on the running prediction model.

[0071] It should be further explained that, in the specific implementation process, the secondary testing of the UPS module under normal power-on condition includes:

[0072] The estimated time series of various indicators of the powered-on UPS module are compared with the preset threshold range of each indicator. If any indicator is not within the corresponding preset threshold range, the powered-on UPS module is marked as a warning state.

[0073] If all indicators of the powered-on UPS module are within their respective preset threshold ranges, the estimated time series values ​​of real-time efficiency and module temperature rise in each indicator are compared with the real-time efficiency and module temperature rise in the evaluation indicators of the load selection module, respectively. The average magnitude of the evaluation indicator change is obtained, which includes the average magnitude of real-time efficiency change and the average magnitude of module temperature rise change. A preset upper limit for the magnitude change is set (based on the fluctuation range of the UPS module under normal conditions). If the average magnitude of the evaluation indicator change is greater than the upper limit, the evaluation indicators of the load selection module are updated according to the average magnitude of the evaluation indicator change. For example, if the average magnitude of real-time efficiency change is greater than the upper limit, a new real-time efficiency is obtained based on the average magnitude of real-time efficiency change and the real-time efficiency in the evaluation indicators of the load selection module (new real-time efficiency = real-time efficiency in the evaluation indicators of the load selection module + average magnitude of real-time efficiency change). After the evaluation indicators of the load selection module are updated, the load carrying priority and load queue of each UPS module are regenerated.

[0074] It should be further explained that, in the specific implementation process, the process of performing the power-on UPS module switching operation includes:

[0075] The input and output of the powered-on UPS module are disconnected, and the target load power of the powered-on UPS module is obtained. Based on the target load power of the powered-on UPS module, the preset load rate, and the rated power of each UPS module ranked after the powered-on UPS module in the load sequence, the required number of UPS modules R is obtained (the sum of the rated power of the first R UPS modules ranked after the powered-on UPS module in the load queue multiplied by the preset load rate is greater than or equal to the target load power of the UPS module in the fault state or warning state, and the sum of the rated power of the first R-1 UPS modules ranked after the powered-on UPS module in the load queue multiplied by the preset load rate is less than the target load power of the UPS module in the fault state or warning state). Based on the rated power of the first R UPS modules ranked after the powered-on UPS module, the target load power of the powered-on UPS module, and the preset load rate, the load is allocated, and the target load power of the first R UPS modules is obtained (the sum of the target load power of the first R UPS modules multiplied by the preset load rate is equal to the target load power of the powered-on UPS module). The local control module runs the first R UPS modules based on the target load power of the first R UPS modules and marks the first R UPS modules as powered-on UPS modules.

[0076] It should be further explained that, in the specific implementation process, the process of detecting fluctuations in the target load and executing adaptive power-on UPS module scheduling operations based on the detection results includes:

[0077] The operating power of the target load at each moment is compared with the operating power at the previous moment. A preset error redundancy threshold (based on the accuracy error setting of the data collected by the load detection module) is used to obtain the operating power difference (the operating power at the current moment minus the operating power at the previous moment). When the operating power difference is greater than the error redundancy threshold, if the operating power difference is positive, the redundant power of each powered UPS module is obtained based on the rated power and preset load rate of each powered UPS module (redundant power = rated power - target load power × preset load rate). It is then determined whether the operating power difference is greater than the sum of the redundant power of each powered UPS module. If it is greater, the system is then adjusted according to the preset load rate and the non-powered UPS modules in the load queue. The required number of UPS modules Q is obtained from the rated power and operating power difference of the UPS modules (the sum of the rated power of the first Q non-powered UPS modules in the load queue multiplied by the preset load rate is greater than or equal to the operating power difference, and the sum of the rated power of the first Q-1 non-powered UPS modules in the load queue multiplied by the preset load rate is less than the operating power difference). The load is allocated according to the preset load rate, the operating power difference, and the rated power of the first Q non-powered UPS modules in the load queue. The target load power of the first Q UPS modules is obtained (the sum of the target load power of the first Q UPS modules multiplied by the preset load rate is equal to the operating power difference), and the first Q UPS modules are marked as powered UPS modules.

[0078] If it is not greater than, then the redundant load is allocated to each powered UPS module according to the current operating power difference, the preset load rate and the redundant power of each powered UPS module, and the redundant load of each powered UPS module is obtained (the sum of the redundant loads of each powered UPS module multiplied by the preset load rate equals the operating power difference). The target load power of the powered UPS module at the current moment is obtained according to the redundant load of each powered UPS module (the target load power at the current moment = the target load power at the previous moment + the redundant load).

[0079] It should be further explained that in the specific implementation process, if the operating power difference is negative, the preset load rate will be reduced according to the operating power difference and the rated power of each powered UPS module (the sum of the rated power of each powered UPS module × the reduced preset load rate = the operating power at the previous moment - the operating power difference).

[0080] like Figure 2 As shown, the power supply method based on a modular UPS power supply includes the following steps:

[0081] Step 1: Collect the operating parameters of the UPS module and mark the collection time, and set the collection period;

[0082] Step 2: Obtain the load carrying priority of each UPS module based on the operating parameters;

[0083] Step 3: Collect the operating power of the target load, and obtain the target load power of the powered-on UPS module based on the load carrying priority, operating parameters and operating power of each UPS module;

[0084] Step 4: Perform real-time monitoring of the powered-on UPS module. Based on the real-time monitoring results, mark the powered-on UPS module as normal or faulty. Perform trend analysis and secondary monitoring on the powered-on UPS module in normal condition. Based on the secondary monitoring results, determine whether to mark the powered-on UPS module as a warning or update the evaluation indicators of the selected load-bearing module.

[0085] Step 5: When the powered-on UPS module is marked as faulty or in a warning state, perform a power-on UPS module switching operation, and perform fluctuation detection on the target load. Based on the detection results, perform an adaptive power-on UPS module scheduling operation.

[0086] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A power supply system based on a modular UPS power supply, characterized in that, It includes a central controller, which is connected to several UPS modules, load selection modules, load distribution modules, power-on UPS detection modules, module fault handling modules, and load detection modules. The UPS module includes a local control module and a status monitoring module. The local control module is used to operate the UPS module based on the target load power of the UPS module after power-on. The status monitoring module is used to collect the operating parameters of the UPS module and mark the collection time, and set the collection cycle. The operating parameters include rated power, module temperature rise, cumulative operating time and real-time efficiency. The load selection module is used to extract the rated power, real-time efficiency, module temperature rise and cumulative running time of each UPS module from the operating parameters of each UPS module as evaluation indicators, set the indicator weights of the evaluation indicators, and obtain the load carrying priority of each UPS module according to the evaluation indicators and indicator weights. The load distribution module is used to sort the load carrying priority of each UPS module in ascending order, generate a load queue, obtain the preset load rate and the rated power of each UPS module in the load queue, obtain the required number of UPS modules K according to the preset load rate, the rated power of each UPS module and the operating power of the target load, distribute the load to the first K UPS modules in the load queue according to the rated power of each UPS module, the preset load rate and the operating power of the target load, obtain the target load power of the first K UPS modules, and mark the first K UPS modules as powered-on UPS modules. The UPS power-on detection module is used to perform real-time detection on the UPS power-on module. Based on the real-time detection results, the UPS power-on module is marked as normal or faulty. For the UPS power-on module in normal state, trend analysis and secondary detection are performed. Based on the secondary detection results, it is determined whether to mark the UPS power-on module as a warning state or update the evaluation index of the selected module. The process of trend analysis for a normally powered-on UPS module includes: At the end timestamp of the collection period, features of various indicators of the powered-on UPS module are extracted to obtain a set of feature parameters. An operation prediction model is constructed. The numerical time series sequence and feature parameter set of various indicators of several UPS modules in several historical continuous collection periods are obtained as training data. The operation prediction model is trained using the training data to obtain a completed operation prediction model. Input the time series of values ​​and feature parameter set of various types of indicators of the UPS module that is powered on during the collection period into the running prediction model, and output the predicted time series of values ​​of various types of indicators of the UPS module in the next collection period according to the running prediction model. The process of performing secondary testing on a normally powered-on UPS module includes: The estimated time series of various indicators of the powered-on UPS module are compared with the preset threshold range of each indicator. If any indicator is not within the corresponding preset threshold range, the powered-on UPS module is marked as a warning state. If all the indicators of the powered-on UPS module are within the corresponding preset threshold range, the time series of the real-time efficiency and the estimated value of the module temperature rise in each type of indicator are compared with the evaluation indicators of the load selection module to obtain the average magnitude of the evaluation indicator change. The upper limit of the magnitude change is preset. If the average magnitude of the evaluation indicator change is greater than the upper limit of the magnitude change, the evaluation indicators of the load selection module are updated according to the average magnitude of the evaluation indicator change. After the evaluation indicators of the load selection module are updated, the load carrying priority and load queue of each UPS module are regenerated. The module fault response module is used to disconnect the input and output of the powered-on UPS module when it is marked as faulty or in a warning state. It also obtains the target load power of the powered-on UPS module, and obtains the required number of UPS modules R based on the target load power, preset load rate, and rated power of each UPS module after the powered-on UPS module in the load sequence. Based on the rated power of the first R UPS modules after the powered-on UPS module, the target load power of the powered-on UPS module, and the preset load rate, it performs load allocation, obtains the target load power of the first R UPS modules, and marks the first R UPS modules as powered-on UPS modules. The load detection module compares the operating power of the target load at each moment with the operating power at the previous moment, presets an error redundancy threshold, and obtains the operating power difference. When the operating power difference is greater than the error redundancy threshold, if the operating power difference is positive, the redundant power of each powered UPS module is obtained based on the rated power of each powered UPS module and the preset load rate. It is then determined whether the operating power difference is greater than the sum of the redundant power of each powered UPS module. If it is greater, the required number of UPS modules Q is obtained based on the preset load rate, the rated power of the non-powered UPS modules in the load queue, and the operating power difference. The load is then allocated based on the preset load rate, the operating power difference, and the rated power of the first Q non-powered UPS modules in the load queue. The target load power of the first Q UPS modules is obtained, and the first Q UPS modules are marked as powered UPS modules. If it is not greater than, then based on the current operating power difference, the preset load rate, and the redundant power of each powered-on UPS module, redundant load is allocated to each powered-on UPS module to obtain the redundant load of each powered-on UPS module, and the target load power of the powered-on UPS module at the current moment is obtained based on the redundant load of each powered-on UPS module.

2. The power supply system based on a modular UPS power supply according to claim 1, characterized in that, The process of real-time monitoring of the powered-on UPS module includes: The operating parameters of the powered-on UPS module are compared with the preset threshold ranges of each type of indicator. If all types of indicators of the powered-on UPS module are within the corresponding preset threshold ranges, the powered-on UPS module is marked as normal. If any type of indicator of the powered-on UPS module is not within the corresponding preset threshold range, the powered-on UPS module is marked as faulty.

3. The power supply system based on a modular UPS power supply according to claim 2, characterized in that, If the operating power difference is negative, the preset load rate will be reduced based on the operating power difference and the rated power of each powered-on UPS module.

4. A power supply method based on a modular UPS power supply, specifically applied to the power supply system based on a modular UPS power supply as described in any one of claims 1 to 3, characterized in that, include Step 1: Collect the operating parameters of the UPS module and mark the collection time, and set the collection period; Step 2: Obtain the load carrying priority of each UPS module based on the operating parameters; Step 3: Collect the operating power of the target load, and obtain the target load power of the powered-on UPS module based on the load carrying priority, operating parameters and operating power of each UPS module; Step 4: Perform real-time monitoring of the powered-on UPS module. Based on the real-time monitoring results, mark the powered-on UPS module as normal or faulty. Perform trend analysis and secondary monitoring on the powered-on UPS module in normal condition. Based on the secondary monitoring results, determine whether to mark the powered-on UPS module as a warning or update the evaluation indicators of the selected load-bearing module. Step 5: When the powered-on UPS module is marked as faulty or in a warning state, perform a power-on UPS module switching operation, and perform fluctuation detection on the target load. Based on the detection results, perform an adaptive power-on UPS module scheduling operation.

Images