A multi-voltage output precision control method and system of a multi-winding isolated transformer

Abstract

The present application relates to the technical field of transformer precision control, and particularly relates to a multi-voltage output precision control method and system for a multi-winding isolation transformer, which comprises the following steps: configuring an independent sampling channel for each winding of the transformer to collect the running state data of the winding; inputting the running state data into a load prediction model to obtain a load prediction result; setting a coarse adjustment trigger condition and coarsely adjusting the output voltage according to the load prediction result; calculating a voltage compensation amount based on temperature fluctuation and coupling interference, superimposing the voltage compensation amount on the target voltage to obtain a corrected target voltage; and finely adjusting the output voltage after coarse adjustment according to the corrected target voltage. Through the present application, the control bottleneck of the power output precision of the multi-winding transformer under complex working conditions such as cross interference and temperature fluctuation is effectively solved, high-precision and high-reliability control of the multi-winding transformer is realized, invalid switching loss is reduced, and the safety and long-term stability of the equipment under complex experimental scenarios are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of transformer precision control technology, and in particular to a method and system for multi-voltage output precision control of a multi-winding isolation transformer. Background Technology

[0002] With the increasing precision of industrial equipment and the growing complexity of power electronic systems, multi-voltage, multi-winding dry-type isolation transformers are widely used in scientific research laboratories, precision instrument power supply, and medical equipment testing, requiring multiple independent and stable voltage outputs for different loads. Traditional voltage control methods mainly rely on fixed-threshold PID feedback regulation or manual tap switching, passively correcting the output voltage through a single closed-loop circuit. While this achieves basic voltage stabilization, its response speed, anti-interference capability, and multi-objective collaborative control performance cannot adequately meet the demands of high-precision, high-dynamic, and complex experiments. Furthermore, the different voltage outputs of multi-winding isolation transformers are typically provided by multiple secondary windings. Since these windings usually share a core or are arranged in close proximity, electromagnetic coupling between windings can occur. Sudden load changes or ambient temperature fluctuations can affect the voltage accuracy between different windings. In particular, when the load on one winding changes, it may affect the output voltage of other windings through coupling effects, leading to errors in the accuracy of the multi-voltage output. Therefore, a multi-voltage output accuracy control method for multi-winding isolation transformers is urgently needed. Summary of the Invention

[0003] This invention provides a method and system for controlling the accuracy of multi-voltage output of a multi-winding isolation transformer, which can effectively solve the problems in the background art.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for controlling the multi-voltage output accuracy of a multi-winding isolation transformer, the method comprising: Each winding of the transformer is equipped with an independent sampling channel to collect the operating status data of the winding. The operating status data is input into the load prediction model to obtain the load prediction result; Set coarse adjustment trigger conditions and coarsely adjust the output voltage based on the load prediction results; Based on temperature fluctuations and coupling interference, the voltage compensation amount is calculated, and the voltage compensation amount is superimposed on the target voltage to obtain the corrected target voltage. The output voltage after coarse adjustment is finely adjusted based on the target voltage.

[0005] Furthermore, the output voltage is controlled in stages, with coarse and fine adjustment circuits set in the transformer, and a dead time set in the control algorithm.

[0006] Furthermore, configuring an independent sampling channel for each winding of the transformer includes: The architecture of the independent sampling channel is designed, and isolation measures are used to electrically isolate the independent sampling channel; Establish a synchronous sampling mechanism to synchronously sample the operating status data of the windings; Configure a data calibration system for the independent sampling channel to verify the collected operating status data.

[0007] Furthermore, constructing the load prediction model includes: Collect historical operating status data and historical load data of the windings to construct a historical dataset; The TCN-Transformer hybrid time series model was selected as the load forecasting model, and the layered architecture of the hybrid time series model was designed. The hybrid time-series model is trained and optimized using the historical dataset to obtain the load prediction model.

[0008] Further, training the hybrid time-series model using the historical dataset includes: Freeze the other layers in the hybrid temporal model except for the TCN layer, and use the historical dataset to perform basic training on the TCN layer; Unfreeze the Transformer layer and feature fusion layer, expand the historical data, and use the expanded data for joint fine-tuning; The hybrid time series model is completely unfrozen, adversarial examples are generated based on the historical dataset, and adversarial training is performed on the hybrid time series model.

[0009] Furthermore, the setting of coarse adjustment trigger conditions includes using fuzzy logic to set a dynamic threshold, with the load prediction change rate and the predicted load rate constituting the trigger conditions, and setting a minimum action time interval.

[0010] Furthermore, based on temperature fluctuations and coupling interference, the voltage compensation amount is calculated as follows: ; in, To correct the target voltage, The original target voltage, For temperature compensation, For coupling compensation.

[0011] Furthermore, establish safety boundaries and set voltage compensation value limits.

[0012] A multi-voltage output accuracy control system for a multi-winding isolation transformer, the system comprising: Data acquisition module: Each winding of the transformer is equipped with an independent sampling channel to collect the operating status data of the winding; Load forecasting module: Inputs the operating status data into the load forecasting model to obtain the load forecasting results; Voltage coarse adjustment module: Sets coarse adjustment trigger conditions and coarsely adjusts the output voltage based on the load prediction results; Voltage correction module: Based on temperature fluctuations and coupling interference, calculates the voltage compensation amount, and adds the voltage compensation amount to the target voltage to obtain the corrected target voltage; Voltage fine-tuning module: Fine-tunes the output voltage after coarse adjustment based on the target voltage.

[0013] Furthermore, the data acquisition module includes: Channel architecture design unit: Design the architecture of independent sampling channels and use isolation measures to electrically isolate the independent sampling channels; Sampling mechanism establishment unit: Establishes a synchronous sampling mechanism to synchronously sample the operating status data of the winding; Calibration system configuration unit: Configures a data calibration system for the independent sampling channel to verify the collected operating status data.

[0014] The technical solution of this invention can achieve the following technical effects: It effectively solves the control bottleneck of power output accuracy of multi-winding transformers under complex operating conditions such as cross interference and temperature fluctuations, and realizes high-precision and high-reliability control of multi-winding transformers. At the same time, it reduces ineffective switching losses through load prediction and hierarchical coordination mechanism, significantly improves equipment safety and long-term stability in complex experimental scenarios, and provides reliable guarantee for precision scientific research testing and high-requirement industrial applications.

[0015] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a method for controlling the accuracy of multi-voltage output in a multi-winding isolation transformer. Figure 2 A flowchart illustrating the configuration of an independent sampling channel; Figure 3 This is a schematic diagram of a multi-voltage output accuracy control system for a multi-winding isolation transformer. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0020] Example 1: like Figure 1 As shown, a method for controlling the multi-voltage output accuracy of a multi-winding isolation transformer includes: S1: Configure an independent sampling channel for each winding of the transformer to collect the operating status data of the winding; Specifically, to avoid interference or sampling lag caused by multiple windings sharing a channel, and to ensure the independence and sampling accuracy of each winding's data, an independent sampling channel can be configured for each winding of the transformer. This enables real-time monitoring of each winding, acquiring key parameters such as temperature, current, voltage, and switching signals, ensuring the accuracy and timeliness of data acquisition. It is important to note that independent sampling of multiple windings can avoid measurement errors caused by cross-interference, which is particularly suitable for isolation transformers with phase-to-phase coupling. However, attention must be paid to sampling synchronization issues; a timestamp alignment mechanism can be added to ensure data timing consistency.

[0021] S2: Input the operating status data into the load prediction model to obtain the load prediction results; For building a load prediction model, learning models such as Long Short-Term Memory (LSTM) networks, Recurrent Neural Networks (RNNs), and Gated Recurrent Units (GRUs) can be selected, with lightweight models preferred to ensure real-time prediction performance. This step, by predicting future load changes, allows for coarse adjustments before the changes occur, ensuring that the output voltage does not deviate significantly due to drastic load changes, thereby improving the overall system stability and accuracy.

[0022] S3: Set the coarse adjustment trigger condition to coarsely adjust the output voltage based on the load prediction result; S4: Based on temperature fluctuations and coupling interference, calculate the voltage compensation amount, and then add the voltage compensation amount to the target voltage to obtain the corrected target voltage; Specifically, temperature changes can affect resistance, insulation material properties, and thus cause voltage deviations; electromagnetic coupling interference may occur between multiple windings due to their physical proximity, and this interference can cause load changes in one winding to affect other windings; temperature fluctuations and coupling interference are important factors affecting the accuracy of multi-voltage output. Directly measuring or predicting them and compensating for them before adding them to the target voltage can significantly reduce the errors caused by the external environment and mutual coupling effects, ensuring that the output is closer to the expected value.

[0023] S5: Fine-tune the output voltage after coarse adjustment based on the target voltage.

[0024] Coarse adjustment can quickly respond to load changes and rapidly adjust the output voltage to near the target value, laying the foundation for subsequent fine adjustment. The purpose of setting trigger conditions in this step is to initiate the coarse adjustment process only when the load prediction results indicate a significant change, avoiding frequent adjustments during small fluctuations. After coarse adjustment, fine adjustment can eliminate residual minor errors. Based on real-time feedback data, a refined control algorithm is used to achieve closed-loop control of the fine adjustment process. In a hierarchical control system, using load prediction for the coarse adjustment loop and voltage compensation for the fine adjustment loop is an effective method to optimize dynamic response and efficiency. Based on real-time signal prediction of short-term load demand and by predicting future load changes, the coarse adjustment loop adjusts the transformer taps or power supply topology in advance, bringing the system operating point close to the target voltage. The fine adjustment loop can fine-tune the voltage using a closed-loop algorithm based on the coarse adjustment, ensuring the accuracy of the output voltage. Coarse adjustment achieves rapid response through load prediction, while fine adjustment provides micro-compensation for temperature drift and electromagnetic coupling, balancing dynamic response and steady-state accuracy, and improving the overall control efficiency of multi-winding systems.

[0025] This invention establishes an independent sampling channel for each winding to accurately collect real-time operating data. It uses a load prediction model to predict future load changes and performs coarse adjustments based on the prediction results when certain triggering conditions are met, proactively addressing significant load fluctuations. Furthermore, based on actual temperature fluctuations and potential coupling interference between windings, it calculates the corresponding voltage compensation amount, finely adjusting the already coarsely adjusted output to achieve higher output accuracy.

[0026] This invention effectively solves the control bottleneck of power output accuracy of multi-winding transformers under complex operating conditions such as cross-interference and temperature fluctuations, and realizes high-precision and high-reliability control of multi-winding transformers. At the same time, it reduces ineffective switching losses through load prediction and hierarchical coordination mechanism, significantly improves equipment safety and long-term stability in complex experimental scenarios, and provides reliable protection for precision scientific research testing and high-requirement industrial applications.

[0027] To achieve precise control over the regulation of multiple output voltages, the output voltages are controlled in stages. Coarse adjustment circuits and fine adjustment circuits are set in the transformer, and a dead time is set in the control algorithm.

[0028] Specifically, in order to solve the problems of wide-range rapid adjustment and fine voltage regulation, coarse adjustment circuits and fine adjustment circuits can be set in the transformer. The coarse adjustment circuit and the fine adjustment circuit are responsible for different control tasks. The two work in coordination to enable the output voltage to quickly approach the target and achieve high-precision steady-state control. The dead time is set to smooth the switching between the two, thereby avoiding instability caused by the system being adjusted too frequently.

[0029] As a preferred embodiment of this example, Figure 2 As shown, each winding of the transformer is configured with an independent sampling channel, including: S11: Design the architecture of the independent sampling channel and use isolation measures to electrically isolate the independent sampling channel; In this embodiment, a separate sampling signal link is designed for each winding, including a sensor, signal conditioning circuit and ADC subsystem. Each channel uses a separate PCB trace to minimize cross-interference between adjacent channels. Isolation amplifiers, optocouplers or digital isolators are used in each sampling path to ensure that each sampling circuit does not interfere with each other electrically. At the same time, good shielding (such as a metal shield) and grounding method (such as single-point grounding) are designed to reduce common-mode interference.

[0030] S12: Establish a synchronous sampling mechanism to synchronously sample the operating status data of the winding; In multi-channel systems, only by ensuring that all data are acquired at the same time can the true operating status of each winding be reflected. This is especially true when load coupling exists in the system, where synchronous sampling is essential to correctly determine the mutual influence between different channels. Therefore, synchronous sampling mechanisms can be established using methods such as synchronous triggering circuits and clock distribution. This facilitates multi-channel joint correction, error comparison, and composite compensation, and is beneficial for subsequently establishing load prediction or compensation algorithms.

[0031] S13: Configure a data calibration system for the independent sampling channel to verify the collected operational status data.

[0032] Specifically, data calibration can eliminate systematic errors caused by the system and environment, making the collected data closer to the true value, thereby improving the accuracy of the entire voltage control system; and a sampling channel redundancy verification mechanism can be designed so that when a certain channel fails, it automatically switches to the sampling value of the adjacent winding for estimation, and triggers a graded alarm to avoid single point failure leading to system collapse.

[0033] Furthermore, constructing the load prediction model includes: S21: Collect historical operating status data and historical load data of the windings to construct a historical dataset; S22: Select the TCN-Transformer hybrid time series model as the load forecasting model, and design the layered architecture of the hybrid time series model; S23: Use historical datasets to train and optimize the hybrid time series model to obtain the load prediction model.

[0034] Specifically, load data typically contains obvious local periodic fluctuations as well as long-term trends or abrupt changes. TCN excels at extracting local features, while Transformer can capture global temporal dependencies. Combining the two can more comprehensively describe data characteristics and improve prediction accuracy. Therefore, the TCN-Transformer hybrid time series model combines the advantages of both, enabling a more comprehensive modeling of various temporal characteristics in load and operational status data, thus obtaining more accurate load prediction results. After selecting the TCN-Transformer hybrid time series model as the load prediction model, the architecture of the hybrid time series model can be designed, including an input layer, a TCN feature extraction layer, a Transformer global modeling layer, a feature fusion layer, and an output layer. This fully leverages the advantages of both models, ultimately achieving accurate load prediction.

[0035] Based on the above embodiments, the hybrid time-series model is trained using a historical dataset, including: S231: Freeze the other layers in the hybrid temporal model except for the TCN layer, and use the historical dataset to perform basic training on the TCN layer; For model training, if the TCN layer and Transformer layer are trained simultaneously in the early stages, their gradients may interfere with each other, leading to slow or unstable convergence. It is advisable to focus on training the TCN layer first, so that the TCN layer learns the local temporal patterns of the load, initially establishes trend perception capabilities, and avoids early interference from complex modules. Adjusting only the TCN layer parameters in the initial stage can reduce the dimensionality of parameter updates, reduce interference during the training process, and help the model converge faster.

[0036] S232: Unfreeze the Transformer layer and feature fusion layer, expand the historical data, and use the expanded data for joint fine-tuning; Based on the already trained TCN layer, after unfreezing the Transformer layer and feature fusion layer, joint fine-tuning enables the model to simultaneously optimize local features (TCN) and global dependencies (Transformer), achieving better synergy between features. Local features of TCN (such as short-term trends) may conflict with global features of Transformer (such as long-term dependencies). This can be addressed by introducing diversity through data expansion, prompting the fusion layer (such as a gating mechanism) to dynamically adjust its weight allocation strategy.

[0037] S233: Fully unfreeze the hybrid time series model, generate adversarial examples based on historical datasets, and perform adversarial training on the hybrid time series model.

[0038] Specifically, the purpose of adversarial training is to allow the model to globally adjust its parameter distribution under adversarial perturbations, thereby enhancing its robustness to input noise and non-stationary features. During the adversarial training phase, all trainable parameters of the model are in an updatable state. If some layers are frozen, the perturbation generation will not be able to reflect the sensitive areas of the complete model, thus reducing the effectiveness of adversarial training.

[0039] Furthermore, setting coarse adjustment trigger conditions includes using fuzzy logic to set dynamic thresholds, using the load prediction change rate and the predicted load rate as trigger conditions, and setting a minimum action time interval.

[0040] In this embodiment, fuzzy logic is used to set dynamic thresholds. The power change rate and predicted load rate are fuzzified using membership functions. Trigger strength coefficients are generated based on a preset rule base. Coarse adjustment is triggered when the coefficient exceeds the dynamically adjusted threshold. This step dynamically sets the threshold parameters through a fuzzy logic controller. The input variables of the fuzzy logic controller include the predicted load change rate and the predicted load rate. When both the predicted load change rate and the predicted load rate are greater than the set percentage, the triggering condition is met. A minimum gear switching interval of no less than 1 second is set to suppress false triggers caused by transient disturbances. The setting of the coarse adjustment triggering condition ensures that the system only initiates coarse adjustment when a significant load change is detected, avoiding frequent adjustments. It also ensures that a large-scale adjustment is performed first, followed by detailed corrections, when necessary.

[0041] Furthermore, based on temperature fluctuations and coupling interference, the voltage compensation amount is calculated as follows: ; in, To correct the target voltage, The original target voltage, For temperature compensation, For coupling compensation.

[0042] Specifically, temperature compensation can be achieved by deploying distributed temperature sensors at key heat points (such as winding ends and core joints) to construct a three-dimensional thermal field model. The compensation algorithm needs to consider the thermal inertia effect and introduce an advance compensation strategy to avoid phase lag caused by pure feedback control. For coupling interference, a mutual inductance matrix between windings should be established, and coupling interference should be quantified by real-time flux linkage observation. Initial parameters can be obtained by finite element simulation, and then dynamic correction can be performed through online parameter identification to improve the accuracy of compensation calculation.

[0043] Based on the above embodiments, a safety boundary is established and a voltage compensation value limit is set.

[0044] Specifically, in voltage compensation control based on temperature fluctuations and coupling interference, the dynamic adjustment of the compensation amount may lead to overcompensation or undercompensation due to model errors, sensor noise, or extreme operating conditions. A safety boundary can be established by setting the voltage compensation value limit to be less than or equal to 5% of the original target voltage to prevent oscillation caused by overcompensation.

[0045] Example 2: like Figure 3 As shown, a multi-voltage output accuracy control system for a multi-winding isolation transformer includes: Data acquisition module: Each winding of the transformer is equipped with an independent sampling channel to collect the operating status data of the winding; Load forecasting module: Inputs operating status data into the load forecasting model to obtain load forecasting results; Voltage coarse adjustment module: Sets coarse adjustment trigger conditions and coarsely adjusts the output voltage based on load prediction results; Voltage correction module: Based on temperature fluctuations and coupling interference, calculates the voltage compensation amount, and adds the voltage compensation amount to the target voltage to obtain the corrected target voltage; Voltage fine-tuning module: Fine-tunes the output voltage after coarse adjustment based on the target voltage.

[0046] The adjustment system described above in this invention can effectively realize the multi-voltage output accuracy control method for multi-winding isolation transformers, and the technical effects it can achieve are as described in the above embodiments, which will not be repeated here.

[0047] Furthermore, the data acquisition module includes: Channel architecture design unit: Design the architecture of independent sampling channels and use isolation measures to electrically isolate the independent sampling channels; Sampling mechanism establishment unit: Establishes a synchronous sampling mechanism to synchronously sample the operating status data of the winding; Calibration system configuration unit: Configures a data calibration system for independent sampling channels to verify the collected operational status data.

[0048] Similarly, the above-mentioned optimization schemes for the system can also achieve the optimization effects corresponding to the methods in Embodiment 1, which will not be repeated here.

[0049] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and accompanying drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for controlling the multi-voltage output accuracy of a multi-winding isolation transformer, characterized in that, The method includes: Each winding of the transformer is equipped with an independent sampling channel to collect the operating status data of the winding. The operating status data is input into the load prediction model to obtain the load prediction result; Set coarse adjustment trigger conditions and coarsely adjust the output voltage based on the load prediction results; Based on temperature fluctuations and coupling interference, the voltage compensation amount is calculated, and the voltage compensation amount is superimposed on the target voltage to obtain the corrected target voltage. The output voltage after coarse adjustment is finely adjusted according to the target voltage. Building a load prediction model includes: Collect historical operating status data and historical load data of the windings to construct a historical dataset; The TCN-Transformer hybrid time series model was selected as the load forecasting model, and the layered architecture of the hybrid time series model was designed. The hybrid time-series model is trained and optimized using the historical dataset to obtain the load prediction model; The setting of coarse adjustment trigger conditions includes using fuzzy logic to set dynamic thresholds, using the load prediction change rate and the predicted load rate as trigger conditions, and setting a minimum action time interval.

2. The multi-voltage output accuracy control method for a multi-winding isolation transformer according to claim 1, characterized in that, The output voltage is controlled in stages, with coarse and fine adjustment circuits set in the transformer, and a dead time set in the control algorithm.

3. The multi-voltage output accuracy control method for a multi-winding isolation transformer according to claim 1, characterized in that, The provision of an independent sampling channel for each winding of the transformer includes: The architecture of the independent sampling channel is designed, and isolation measures are used to electrically isolate the independent sampling channel; Establish a synchronous sampling mechanism to synchronously sample the operating status data of the windings; Configure a data calibration system for the independent sampling channel to verify the collected operating status data.

4. The multi-voltage output accuracy control method for a multi-winding isolation transformer according to claim 1, characterized in that, Training the hybrid time-series model using the historical dataset includes: Freeze the other layers in the hybrid temporal model except for the TCN layer, and use the historical dataset to perform basic training on the TCN layer; Unfreeze the Transformer layer and feature fusion layer, expand the historical data, and use the expanded data for joint fine-tuning; The hybrid time series model is completely unfrozen, adversarial examples are generated based on the historical dataset, and adversarial training is performed on the hybrid time series model.

5. The multi-voltage output accuracy control method for a multi-winding isolation transformer according to claim 1, characterized in that, Based on temperature fluctuations and coupling interference, the voltage compensation amount is calculated as follows: ； in, To correct the target voltage, The original target voltage, For temperature compensation, For coupling compensation.

6. The multi-voltage output accuracy control method for a multi-winding isolation transformer according to claim 5, characterized in that, Establish safety boundaries and set voltage compensation value limits.

7. A multi-voltage output accuracy control system for a multi-winding isolation transformer, applicable to the method described in any one of claims 1 to 6, characterized in that, The system includes: Data acquisition module: Each winding of the transformer is equipped with an independent sampling channel to collect the operating status data of the winding; Load forecasting module: Inputs the operating status data into the load forecasting model to obtain the load forecasting results; Voltage coarse adjustment module: Sets coarse adjustment trigger conditions and coarsely adjusts the output voltage based on the load prediction results; Voltage correction module: Based on temperature fluctuations and coupling interference, calculates the voltage compensation amount, and adds the voltage compensation amount to the target voltage to obtain the corrected target voltage; Voltage fine-tuning module: Fine-tunes the output voltage after coarse adjustment based on the target voltage.

8. A multi-voltage output accuracy control system for a multi-winding isolation transformer according to claim 7, characterized in that, The data acquisition module includes: Channel architecture design unit: Design the architecture of independent sampling channels and use isolation measures to electrically isolate the independent sampling channels; Sampling mechanism establishment unit: establishes a synchronous sampling mechanism to synchronously sample the winding's operating status data; Calibration system configuration unit: configures a data calibration system for the independent sampling channel to verify the collected operating status data.

Images