Power grid fault intelligent diagnosis method, system and device based on big data and medium

Abstract

The invention discloses a power grid fault intelligent diagnosis method, system and device based on big data and a medium, and relates to the technical field of data processing, and the method comprises the steps: obtaining power grid electrical parameters, setting a first monitoring time period and a second monitoring time period, and obtaining a first monitoring parameter value sequence and a second monitoring parameter value sequence; obtaining a first increment value and a first attenuation value, obtaining a second increment value and a second attenuation value, obtaining a predicted increment value corresponding to the electrical parameters of the power grid based on the prediction model, the first increment value and the second increment value, and obtaining a predicted attenuation value corresponding to the electrical parameters of the power grid based on the prediction model, the first attenuation value and the second attenuation value; obtaining an upper fluctuation threshold value and a lower fluctuation threshold value corresponding to the electrical parameters of the power grid, and obtaining a fault index according to the upper fluctuation threshold value, the lower fluctuation threshold value, the prediction attenuation value and the prediction increment value; and obtaining a power grid fault diagnosis result according to the fault index. The method has the advantages of accurate fault diagnosis, self-adaptive threshold and full-period management.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method, system, equipment and medium for intelligent diagnosis of power grid faults based on big data. Background Art

[0002] In the field of power system operation and maintenance, timely and accurate diagnosis of power grid faults is directly related to power supply reliability and equipment safety. Existing fault diagnosis methods commonly suffer from the following technical bottlenecks: First, static threshold setting relies on historical empirical data and fails to consider the real-time impact of the equipment operating environment (such as temperature, humidity, or load fluctuations). This leads to frequent false alarms caused by parameter drift in high-temperature and high-humidity environments, and normal operating shocks during sudden load changes are mistakenly interpreted as faults. Furthermore, single-parameter threshold methods lack quantitative analysis of parameter change trends, making it difficult to distinguish between the development of overvoltage and transient disturbances, and are insensitive to the gradual characteristics of early-stage faults. Second, existing incremental attenuation calculations use a linear superposition method, which fails to reflect the nonlinear characteristics of parameter mutations. This leads to inaccurate prediction models due to insulation degradation during the aging stage of equipment, and fault indicator calculations ignore the coupling effects of environmental and electrical parameters. Finally, threshold setting fails to incorporate equipment lifecycle compensation. The threshold range in the later stages of operation is mismatched with the actual tolerance of the equipment, resulting in premature or late warnings of faults in aging equipment and an increased false alarm rate in complex power grid environments. Furthermore, existing methods are insufficiently sensitive to the direction of parameter changes and cannot effectively distinguish between recovery and deterioration trends during the development stage of a fault, leading to further misdiagnosis. Therefore, there is an urgent need for an intelligent diagnosis method that can integrate multi-source dynamic factors, adapt to environmental changes, and accurately quantify fault risks. Summary of the Invention

[0003] In response to the deficiencies in the prior art, the present invention provides a method, system, device and medium for intelligent diagnosis of power grid faults based on big data.

[0004] A method for intelligent diagnosis of power grid faults based on big data comprises: obtaining a plurality of power grid electrical parameters based on big data, setting a continuous first monitoring time period and a second monitoring time period, and obtaining a first monitoring parameter value sequence and a second monitoring parameter value sequence of the power grid electrical parameters in the first monitoring time period and the second monitoring time period, respectively, wherein the time intervals corresponding to two adjacent parameter values in the first monitoring parameter value sequence or the second monitoring parameter value sequence are consistent; obtaining a first incremental value and a first attenuation value according to the first monitoring parameter value sequence, and obtaining a second incremental value and a second attenuation value according to the second monitoring parameter value sequence, and obtaining a predicted incremental value corresponding to the power grid electrical parameter based on a prediction model, the first incremental value and the second incremental value, and obtaining a predicted attenuation value corresponding to the power grid electrical parameter based on the prediction model, the first attenuation value and the second attenuation value; obtaining an upper fluctuation threshold and a lower fluctuation threshold corresponding to the power grid electrical parameter, and obtaining a fault indicator according to the upper fluctuation threshold, the lower fluctuation threshold, the predicted attenuation value and the predicted incremental value; and obtaining a power grid fault diagnosis result according to the fault indicator.

[0005] Optionally, obtaining the upper fluctuation threshold and the lower fluctuation threshold corresponding to the electrical parameters of the power grid includes: determining the initial fluctuation interval based on the historical operating data of the electrical parameters of the power grid, and dynamically adjusting the initial fluctuation interval according to the real-time monitored power grid operating environment parameters, and forming a fluctuation interval to be processed, wherein the power grid operating environment parameters include equipment temperature, ambient humidity and load fluctuation level; performing aging compensation on the fluctuation interval to be processed according to the current life cycle status of the power grid equipment; performing threshold separation on the compensated fluctuation interval to be processed, and generating an upper fluctuation threshold and a lower fluctuation threshold that match the current power grid operating status.

[0006] Optionally, dynamic adjustment of the initial fluctuation range based on the real-time monitored grid operating environment parameters includes: adjusting the upper limit of the initial fluctuation range according to the equipment temperature change trend, lowering the upper limit when the temperature increases, and increasing the upper limit when the temperature decreases; synchronously adjusting the lower limit of the initial fluctuation range according to the ambient humidity change trend, increasing the lower limit when the humidity increases, and lowering the lower limit when the humidity decreases; increasing the upper and lower limits of the initial fluctuation range when the load fluctuates violently, and lowering the upper and lower limits of the initial fluctuation range when the load is stable.

[0007] Optionally, obtaining a power grid fault diagnosis result based on a fault indicator includes: dividing into multiple risk intervals, each of which corresponds to a different fault diagnosis conclusion; obtaining the risk interval into which the fault indicator falls, and obtaining the fault diagnosis conclusion corresponding to the risk interval.

[0008] Optionally, obtaining the first increment value and the first attenuation value according to the first monitoring parameter value sequence is expressed as: , ;in, is the first increment value, is the first attenuation value, is the number of parameter values in the first monitoring parameter value sequence, is the j+1th parameter value in the first monitoring parameter value sequence, is the jth parameter value in the first monitoring parameter value sequence.

[0009] Optionally, the prediction model in obtaining the predicted incremental value corresponding to the grid electrical parameter based on the prediction model, the first incremental value, and the second incremental value is expressed as: ;in, To predict the incremental value, is the first increment value, is the second increment value.

[0010] Optionally, the fault indicator is obtained based on the upper fluctuation threshold, the lower fluctuation threshold, the predicted attenuation value, and the predicted increment value as follows: ;in, is the fault indicator, is the upper fluctuation threshold, is the lower fluctuation threshold, To predict the incremental value, is the predicted attenuation value.

[0011] Also provided is an intelligent diagnosis system for power grid faults based on big data, which includes: an acquisition module for acquiring multiple power grid electrical parameters based on big data, and setting a continuous first monitoring time period and a second monitoring time period, and acquiring a first monitoring parameter value sequence and a second monitoring parameter value sequence of the power grid electrical parameters in the first monitoring time period and the second monitoring time period, respectively, wherein the time intervals corresponding to two adjacent parameter values in the first monitoring parameter value sequence or the second monitoring parameter value sequence are consistent; a first data processing module for acquiring a first incremental value and a first attenuation value according to the first monitoring parameter value sequence, and acquiring a second incremental value and a second attenuation value according to the second monitoring parameter value sequence, and acquiring a predicted incremental value corresponding to the power grid electrical parameter based on a prediction model, the first incremental value and the second incremental value, and acquiring a predicted attenuation value corresponding to the power grid electrical parameter based on the prediction model, the first attenuation value and the second attenuation value; a second data processing module for acquiring an upper fluctuation threshold and a lower fluctuation threshold corresponding to the power grid electrical parameter, and acquiring a fault indicator based on the upper fluctuation threshold, the lower fluctuation threshold, the predicted attenuation value and the predicted incremental value; and a diagnosis module for acquiring a power grid fault diagnosis result based on the fault indicator.

[0012] An electronic device is also provided, characterized in that it includes: a memory on which a computer program is stored; and a processor for executing the computer program in the memory to implement the above-mentioned big data-based intelligent diagnosis method for power grid faults.

[0013] A non-temporary computer-readable storage medium is also provided, on which a computer program is stored. When the program is executed by a processor, the above-mentioned big data-based intelligent diagnosis method for power grid faults is implemented.

[0014] The beneficial effects of the present invention are embodied in: In the entire big data-based intelligent diagnosis method for power grid faults, first of all, based on the standardized data acquisition system of dual time windows, the real-time operating status of the equipment and the historical trend characteristics are integrated. By comparing and analyzing the incremental attenuation trend of the parameter sequence, the limitations of the traditional static threshold method are broken through. It can not only capture the gradual degradation caused by equipment aging, but also identify the transient overvoltage caused by sudden disturbances such as lightning strikes, and achieve accurate distinction between early weak fault characteristics and transient interference; further, a dynamic threshold adjustment algorithm coupled with environmental parameters is introduced to perceive the impact of temperature, humidity and load fluctuations on electrical parameters in real time, and then dynamically correct the tolerance threshold area of aging equipment in combination with service life compensation through equipment life cycle. Furthermore, the prediction model integrates the correlation between historical and current trends and adopts a weight distribution strategy to effectively quantify the nonlinear characteristics of parameter mutations. For example, in the diagnosis of capacitor bank fuse aging, it can accurately identify the essential difference between a sudden drop in capacitance and periodic fluctuations, and combine with a graded early warning mechanism to achieve a precise response from preventive maintenance to emergency treatment. Furthermore, through multi-dimensional quantification and trend trajectory analysis of fault indicators, the problem of insufficient sensitivity to the direction of parameter changes is solved. For example, in the partial discharge monitoring of existing equipment, it can not only identify the deterioration trend of insulation defects with continuously increasing discharge volume, but also distinguish between instantaneous discharge pulses caused by operational overvoltage, significantly reducing the misjudgment rate and improving fault location efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly describes the drawings required for the specific embodiments or the description of the prior art. Similar elements or parts are generally identified by similar reference numerals throughout the drawings. Elements or parts in the drawings are not necessarily drawn to scale.

[0016] Figure 1 This is a schematic diagram of the steps of the big data-based intelligent diagnosis method for power grid faults of the present invention; Figure 2 This is a schematic diagram of a portion of step S3 in the big data-based intelligent diagnosis method for power grid faults of the present invention; Figure 3 This is a schematic diagram of a portion of the steps in S31 of the big data-based intelligent diagnosis method for power grid faults of the present invention; Figure 4 This is a schematic diagram of a portion of step S4 in the big data-based intelligent diagnosis method for power grid faults of the present invention; Figure 5The present invention is a block diagram of an electronic device according to an embodiment of the present invention.

[0017] Reference numerals: 700 - electronic device, 701 - processor, 702 - memory, 703 - multimedia component, 704 - I / O interface, 705 - communication component. DETAILED DESCRIPTION

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0019] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are intended to fall within the scope of protection of the present invention.

[0020] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. In addition, the terms "first," "second," etc. are used only to distinguish the descriptions and should not be understood as indicating or implying relative importance.

[0021] like Figure 1 As shown, a method for intelligent diagnosis of power grid faults based on big data is provided, including: S1. Acquire multiple power grid electrical parameters based on big data, set a first monitoring time period and a second monitoring time period, and acquire a first monitoring parameter value sequence and a second monitoring parameter value sequence of the power grid electrical parameters within the first monitoring time period and the second monitoring time period, respectively, wherein the time intervals corresponding to two adjacent parameter values within the first monitoring parameter value sequence or the second monitoring parameter value sequence are consistent; S2. Obtaining a first incremental value and a first attenuation value according to the first monitoring parameter value sequence, and obtaining a second incremental value and a second attenuation value according to the second monitoring parameter value sequence, and obtaining a predicted incremental value corresponding to the power grid electrical parameter based on the prediction model, the first incremental value, and the second incremental value, and obtaining a predicted attenuation value corresponding to the power grid electrical parameter based on the prediction model, the first attenuation value, and the second attenuation value; S3. Obtain an upper fluctuation threshold and a lower fluctuation threshold corresponding to the electrical parameters of the power grid, and obtain a fault indicator based on the upper fluctuation threshold, the lower fluctuation threshold, the predicted attenuation value, and the predicted increment value; S4. Obtain power grid fault diagnosis results based on the fault indicators.

[0022] In this embodiment, it should be noted that in S1, multi-dimensional grid electrical parameters (such as voltage, current, frequency, and harmonic components) are collected in real time via a big data platform during grid operation, and a dynamic monitoring data sequence is constructed based on a time window partitioning mechanism. Specifically, this method divides the monitoring cycle into a continuous first monitoring period and a second monitoring period. The first period is typically used to establish a historical change baseline, while the second period is used to capture the current operating status. Within each period, a parameter value sequence is generated at a fixed sampling frequency (e.g., every 10 seconds or every minute), ensuring that the time intervals between adjacent data points are strictly consistent. This design not only preserves the temporal characteristics of parameter changes, but also enhances the continuity of trend analysis through dual time window comparison, providing standardized input for subsequent increment and decay calculations. For example, in a transformer overload monitoring scenario, the winding temperature sequence for the last 30 minutes is simultaneously recorded as the second monitoring data and compared with the historical sequence for the previous 30 minutes, thereby eliminating interference from diurnal fluctuations in ambient temperature on equipment status assessment.

[0023] Furthermore, S1's data collection mechanism deeply integrates the multi-source sensing capabilities of power grid equipment. Taking the diagnosis of transmission line insulation faults as an example, the first monitoring time period can be set as a data set within a certain historical operating cycle of the equipment, while the second monitoring time period corresponds to the previous operating cycle. Both use high-precision sensors to collect local discharge signal strength at millisecond intervals. By comparing the amplitude series of discharge pulses in the two time periods, it is possible to effectively distinguish between progressive discharges caused by equipment aging and transient overvoltages caused by lightning strikes. At the same time, this step ensures that the data lengths of different monitoring periods are completely consistent with the sampling conditions through standardized time slicing, providing aligned input feature vectors for subsequent nonlinear prediction models to avoid calculation deviations caused by time series misalignment. For example, when analyzing changes in cable joint contact resistance, the 96 sampling points per hour in the two time periods will be automatically aligned to accurately capture the subtle gradual changes in resistance values.

[0024] In S2, dynamic trend analysis of dual time windows is used to achieve refined modeling of grid parameter changes. Based on the parameter sequences in the first and second monitoring time periods, the increasing and decreasing trend strengths of the parameters in each time period are calculated respectively: the incremental value reflects the persistence and amplitude of the parameter increase, and the attenuation value represents the severity of the parameter decrease. For example, when analyzing the aging of the cable insulation layer, if the leakage current attenuation value in the second time period is significantly higher than that in the first time period, it implies that the insulation performance has accelerated deterioration; and abnormal fluctuations in the incremental value may point to the intensification of partial discharge. This step innovatively adopts cross-period comparison of trend strength, rather than the linear accumulation of a single time period in the existing method, to effectively capture the nonlinear characteristics of the evolution of equipment status.

[0025] Furthermore, S2's prediction model integrates the correlation between historical and current trends. By cross-comparing the incremental attenuation data of two time periods, the model dynamically adjusts the weight distribution of trend predictions: when the incremental value of the second time period increases significantly compared to the previous time period, the prediction model will amplify the contribution of positive mutations; if the attenuation values of the two time periods fluctuate in opposite directions, the mutation interference is suppressed through an exponential function. For example, when evaluating the changes in transformer oil temperature, if the oil temperature rises slowly in the first time period and accelerates in the second time period, the model will identify it as an early sign of overheating. However, for the complex trend of the temperature rising again after a short drop, a nonlinear compensation mechanism is used to avoid misjudging it as a recovery state, accurately distinguishing between abnormal equipment heat dissipation and fluctuations in the external environment.

[0026] In S3, a dynamic threshold generation mechanism is used to implement intelligent fault-tolerant assessment of the grid's operating status. This stage first establishes an initial fluctuation range based on historical device data and environmental parameters. When the ambient temperature rises, the upper limit of the parameter increase is automatically reduced, enhancing the monitoring sensitivity of thermal expansion of the insulation material caused by high temperatures. When the ambient humidity suddenly increases, the lower limit of the parameter decrease is expanded to accurately capture the risk of abnormally increased contact resistance in humid environments. For example, in substation circuit breaker contact monitoring, if the real-time monitoring indicates that the ambient humidity exceeds the 70% threshold, the lower alarm limit for contact resistance decrease will be dynamically adjusted to prevent poor contact caused by moisture on the surface oxide film from being misidentified as normal fluctuations.

[0027] Furthermore, S3 introduces adaptive compensation for the equipment aging coefficient, and elastically corrects the initial threshold by analyzing the service life and performance degradation of the equipment. Taking a high-voltage cable that has been in operation for more than 15 years as an example, the fluctuation threshold of its insulation dielectric strength needs to be gradually relaxed according to the historical degradation trend of the dielectric loss angle to prevent normal parameter fluctuations caused by material fatigue from triggering false alarms. At the same time, in the case of sudden load changes, a buffer zone is formed by expanding the upper and lower limits of the fluctuation range. For example, when it is detected that wind power generation is connected to the grid and causes a sharp power fluctuation, the allowable fluctuation range of the bus voltage is temporarily expanded by 20%, effectively filtering out the instantaneous disturbance caused by the intermittent output of renewable energy, and triggering a fault warning only when the limit is continuously exceeded, significantly improving the diagnostic reliability under complex working conditions.

[0028] In S4, a hierarchical early warning mechanism enables grid fault classification and response decisions. Multiple risk levels are pre-set based on equipment type and operating environment. For example, fault indicators are categorized into three levels: normal fluctuation, potential risk, and emergency failure. Each level corresponds to a differentiated response strategy. For example, when a fault indicator falls within the potential risk range, high-frequency data collection automatically initiates and maintenance recommendations are issued. If the indicator exceeds the emergency failure threshold, the protective device immediately trips and locates the faulty section. This grading mechanism deeply integrates equipment lifecycle characteristics. For example, a narrower potential risk range is set for transformers in mid-service life, while the threshold is appropriately relaxed for equipment nearing retirement, thereby avoiding excessive maintenance and ensuring safety. Analysis of the timeline trajectory of fault indicators can also enhance judgment accuracy. For example, if a cable joint temperature indicator continuously shifts from the normal range to the emergency range within one hour, it is identified as a contact resistance degradation fault rather than a short-term overload. If the indicator exhibits periodic oscillation within the potential risk range, it is identified as harmonic resonance caused by the integration of renewable energy, triggering dynamic reactive power compensation rather than an emergency shutdown.

[0029] To sum up, in the entire intelligent diagnosis method of power grid faults based on big data, first of all, based on the standardized data acquisition system of dual time windows, the real-time operating status and historical trend characteristics of the equipment are integrated, and the incremental attenuation trend of the parameter sequence is compared and analyzed to break through the limitations of the traditional static threshold method. It can not only capture the gradual degradation caused by equipment aging, but also identify the transient overvoltage caused by sudden disturbances such as lightning strikes, and realize the accurate distinction between early weak fault characteristics and transient interference; further, a dynamic threshold adjustment algorithm coupled with environmental parameters is introduced to perceive the impact of temperature, humidity and load fluctuations on electrical parameters in real time, and then dynamically correct the tolerance threshold of aging equipment in combination with service life compensation through equipment life cycle compensation. value range; further, the prediction model integrates the correlation between historical and current trends and adopts a weight distribution strategy to effectively quantify the nonlinear characteristics of parameter mutations. For example, in the diagnosis of capacitor bank fuse aging, it can accurately identify the essential difference between a sudden drop in capacitance and a periodic fluctuation, and combine with the graded early warning mechanism to achieve accurate response from preventive maintenance to emergency disposal; further, through the multi-dimensional quantification and trend trajectory analysis of fault indicators, the problem of insufficient sensitivity to the direction of parameter change is solved. For example, in the partial discharge monitoring of existing equipment, it can not only identify the deterioration trend of insulation defects with continuously increasing discharge volume, but also distinguish the instantaneous discharge pulses caused by operational overvoltage, greatly reducing the misjudgment rate and improving the efficiency of fault location.

[0030] like Figure 2 As shown, in one embodiment, obtaining the upper fluctuation threshold and the lower fluctuation threshold corresponding to the electrical parameter of the power grid in S3 includes: S31. Determine an initial fluctuation range based on historical operating data of the grid electrical parameters, and dynamically adjust the initial fluctuation range according to real-time monitored grid operating environment parameters to form a to-be-processed fluctuation range, wherein the grid operating environment parameters include device temperature, ambient humidity, and load fluctuation level; S32. Perform aging compensation for the fluctuation range to be processed according to the current life cycle status of the power grid equipment; S33 , performing threshold separation on the compensated fluctuation interval to be processed, and generating an upper fluctuation threshold and a lower fluctuation threshold that match the current power grid operation state.

[0031] In this embodiment, it should be noted that, in S31, a dynamic threshold adjustment mechanism linked to environmental parameters is used to achieve precise adaptation of the operating state to external conditions. Real-time analysis of the impact of equipment temperature, ambient humidity, and load fluctuation levels on electrical parameters is performed. For example, when a continuous rise in transformer winding temperature is detected, the upper limit of the allowable oil temperature fluctuation is automatically reduced to enhance the monitoring sensitivity of the risk of thermal cracking of insulating oil. When humidity rises sharply during the rainy season, the lower limit of the alarm threshold for the decrease in cable joint contact resistance is simultaneously increased to avoid surface oxidation caused by moisture from falsely triggering a ground fault alarm. For scenarios with drastic power fluctuations caused by the access of new energy stations, a buffer zone is formed by temporarily expanding the bus voltage fluctuation range. For example, when the output power of a photovoltaic power station oscillates at the minute level due to cloud cover, the upper and lower limits of the voltage threshold are simultaneously relaxed by 15%, effectively suppressing misjudgments caused by intermittent power fluctuations.

[0032] In S32, threshold elastic compensation is implemented based on the performance degradation of the equipment throughout its life cycle to solve the problem of tolerance characteristic deviation of aging equipment. By analyzing the length of time the equipment is put into operation, the cumulative operating load and historical fault records, a material performance attenuation curve is constructed. Taking GIS equipment that has been in operation for ten years as an example, its SF6 gas leakage rate threshold is gradually adjusted according to the seal aging coefficient. In the early stage, it is tightened by 3% each year to capture small leaks. After ten years, it is relaxed by 2% each year to adapt to the natural aging characteristics of the material. For circuit breakers operating under high load, the opening and closing time threshold is dynamically corrected in combination with the contact electrical wear data. For example, after 10,000 cumulative operations, the allowed time deviation is extended from ±5ms to ±8ms, which not only avoids excessive maintenance but also ensures the reliability of operation during the remaining life cycle.

[0033] In S33, during the threshold separation process, determining the baseline value is a key step in dynamically adapting the device's operating status. The baseline value is not a fixed value, but rather the result of intelligent fusion of the device's real-time operating conditions and historical data. For steady-state equipment, the baseline value is typically the weighted average of the parameter within the current monitoring period. For example, for transformer winding temperature, the baseline value is the sliding average of the past hour. For scenarios with periodic fluctuations (such as the grid-connected voltage of a photovoltaic power plant), the median of historical data from the same period under the same operating conditions is used as the dynamic baseline. During the threshold separation process, using cable joint contact resistance monitoring as an example, the compensated fluctuation range is (80, 120). Based on the intelligent fusion of the device's real-time operating conditions and historical data, a baseline resistance of 105 is used as the baseline value. This results in an upper fluctuation threshold of 15 (120-105) and a lower fluctuation threshold of 25 (105-80), which accurately match the abnormal excursion characteristics under high-load conditions. Values of 15 and 25 are used in the subsequent calculation of fault indicators.

[0034] Furthermore, for parameters with significant time correlation (such as capacitor capacitance decay), threshold separation uses the initial baseline value of the equipment after commissioning and adds an aging correction factor. For example, a capacitor with five years of operation uses an initial capacitance of 5000μF as the standard value, and adds a dielectric loss compensation of 1.2% per year, resulting in a dynamic standard value of 5278μF. In this case, if the compensated fluctuation range is (4850μF, 5450μF), the upper fluctuation threshold is 5450-5278=172μF, and the lower fluctuation threshold is 5278-4850=428μF, ensuring that the threshold separation results reflect actual operating conditions.

[0035] like Figure 3 As shown, in one embodiment, dynamically adjusting the initial fluctuation range according to the real-time monitored grid operating environment parameters in S31 includes: S311. Adjust the upper limit of the initial fluctuation range according to the device temperature change trend, lowering the upper limit when the temperature increases and increasing the upper limit when the temperature decreases; S312. Synchronously adjust the lower limit of the initial fluctuation range according to the change trend of the ambient humidity. When the humidity increases, the lower limit is increased, and when the humidity decreases, the lower limit is decreased. S313. When the load fluctuates violently, increase the upper limit and the lower limit of the initial fluctuation range; when the load is stable, reduce the upper limit and the lower limit of the initial fluctuation range.

[0036] In this embodiment, it should be noted that in S311, precise threshold control is achieved through dynamic optimization of temperature sensitivity. When the equipment operating temperature is monitored to be on an upward trend, the upper limit of the parameter fluctuation is automatically reduced. For example, in transformer oil temperature monitoring, the upper limit of the allowable temperature rise threshold is reduced by 1.2% for every 5°C increase in ambient temperature. This nonlinear adjustment mechanism effectively prevents the risk of thermal decomposition of insulating oil. In low-temperature conditions, the upper limit is reversely expanded. For example, when the ambient temperature of the substation drops to 0°C in winter, the upper limit of the circuit breaker tripping time threshold is relaxed to 115% of the standard value to avoid mechanical delays caused by increased grease viscosity being misjudged as operating mechanism failures. This two-way adjustment ensures both monitoring sensitivity in high-temperature environments and diagnostic reliability in low-temperature scenarios.

[0037] In S312, a humidity-related negative mutation monitoring system was established. To address the increased surface conductivity of equipment in high humidity environments, the lower alarm threshold for parameter decreases was raised when the relative humidity exceeded 75%. For example, the contact resistance decrease threshold for cable connectors was adjusted from the standard value of 20% to 35% to prevent false contact alarms caused by condensation-induced transient resistance decreases. In the dry season (humidity <30%), reverse optimization was implemented, such as tightening the lower threshold for SF6 gas pressure decrease in GIS equipment to 90% of the standard value to enhance the ability to detect minor leaks. This humidity-driven asymmetric adjustment addresses the problem of increased false alarm rates in coastal substations during the rainy season, which is often caused by traditional methods.

[0038] In S313, adaptive threshold scaling based on load conditions is implemented. To address the dramatic power fluctuations caused by the grid connection of new energy stations, a wide threshold mode is activated. For example, during periods of PV cluster output power fluctuations, the allowable deviation range of the bus voltage is expanded to 1.8 times the standard value, effectively filtering out second-level voltage flickers caused by cloud cover. When the load stabilizes, the system automatically switches to precise measurement mode. For example, during steady-state operation of a data center power supply circuit, the current fluctuation threshold is compressed to 60% of the standard value, allowing even minor harmonic distortion to trigger an early warning. This mechanism performs exceptionally well in monitoring wind farm collector lines, ignoring random power fluctuations caused by gusts while capturing gradual μA-level leakage current changes before cable insulation breakdown.

[0039] like Figure 4 As shown, in one embodiment, obtaining the power grid fault diagnosis result according to the fault indicator in S4 includes: S41. Divide into multiple risk intervals, each of which corresponds to a different fault diagnosis conclusion; S42: Obtain the risk interval that the fault indicator falls into, and obtain the fault diagnosis conclusion corresponding to the risk interval.

[0040] In this embodiment, it should be noted that in S41, fault level diagnosis is achieved through multiple risk intervals. Differentiated risk intervals are established based on equipment type, operating age, and environmental characteristics. For example, for circuit breaker mechanism boxes in coastal substations, three risk intervals (normal interval, potential risk interval, and emergency fault interval) are set, including a humidity compensation coefficient. Specifically, the fault index is defined as 0-0.8 for the normal interval, 0.8-1.5 for the potential risk interval, and 1.5 or above for the emergency fault interval, providing an intuitive understanding of the equipment status level.

[0041] In S42, a diagnostic conclusion is output through an interval matching mechanism. When the real-time calculated fault indicator falls within a specific numerical range, it is automatically associated with a predefined fault type library. For example, if a cable insulation monitoring indicator enters the potential risk range (e.g., 0.8-1.5), the conclusion "Insulation performance has moderately degraded, and partial discharge testing is recommended within 72 hours" is automatically generated. If it exceeds the emergency threshold (>1.5), the action instruction "High risk of insulation breakdown, emergency power outage and maintenance required" is immediately issued. This mechanism achieves a seamless transition from data monitoring to operation and maintenance decision-making by establishing a direct mapping between indicator ranges and action strategies.

[0042] In one embodiment, obtaining the first increment value and the first attenuation value according to the first monitoring parameter value sequence in S2 is expressed as: , ;in, is the first increment value, is the first attenuation value, is the number of parameter values in the first monitoring parameter value sequence, is the j+1th parameter value in the first monitoring parameter value sequence, is the jth parameter value in the first monitoring parameter value sequence.

[0043] In this embodiment, it should be noted that, in the first increment value and the first attenuation value In the expression of It means only keeping positive changes (upward trend) and filtering negative fluctuations. It means that only negative changes (downward trends) are retained and positive fluctuations are filtered out; the two functions are to decompose parameter changes into independent growth and decline trends, avoiding the trend blurring caused by the positive and negative offset in traditional linear superposition. Furthermore, the differences between adjacent time points are averaged ( ), eliminating the interference of single-point mutations, enhancing sensitivity to sustained trends and reducing the probability of misjudging transient disturbances (such as lightning overvoltage). Furthermore, independent calculations of incremental and attenuation values preserve the asymmetric nature of trends, accurately capturing unidirectional degradation caused by equipment aging (such as the slow increase in insulation material conductivity) while distinguishing bidirectional fluctuations caused by sudden load changes.

[0044] In summary, we can distinguish between gradual and transient faults. Gradual faults manifest as persistently high incremental or attenuation values (e.g., cable insulation aging causes leakage current attenuation values to be above the threshold for a long period of time). Transient disturbances manifest as single-point mutations with normal average values (e.g., lightning overvoltage causes a brief incremental spike, but the subsequent data returns to a stable state). Assuming the cable leakage current sequence is [10mA, 12mA, 15mA, 18mA, 20mA], we can calculate: 2.5 , the continuous increasing trend indicates a gradual degradation of insulation performance rather than a transient disturbance. Furthermore, the nonlinear mutation characteristics are quantified to reflect the severity of the parameter mutation. Assuming that a capacitor capacitance sequence undergoes a mutation in the second time period: [5000μF, 4950μF, 4800μF, 4700μF, 4650μF], the calculation is: 87.5, a high attenuation value indicates that the capacitance decays rapidly, which can be determined as a serious fuse failure.

[0045] For example, assuming the monitoring parameter sequence is [20, 17, 11, 19, 21], substitute it into the expression to calculate the first increment value , the first attenuation value In summary, trend direction separation is achieved. The incremental value (2.5) only captures the upward trend of the parameter (such as the jump from 11→19 and 19→21), ignoring the downward fluctuation; the attenuation value (2.25) only captures the downward trend of the parameter (such as the continuous decline from 20→17 and 17→11), ignoring the subsequent rise. This overcomes the problem of traditional methods' lack of sensitivity to the direction of parameter change and accurately distinguishes between fault deterioration (continuous decline) and recovery (subsequent rise). Therefore, in the sequence, the sudden change of the parameter from 11→19 causes a significant increase in the incremental value (contribution 8), while the early continuous decline (20→17→11) is reflected in the attenuation value (contribution 3+6). It effectively quantifies nonlinear mutations (such as the sudden drop in capacitance before the insulation breakdown of the equipment), avoiding the smoothing loss of mutation characteristics caused by linear superposition methods. In summary, by separating positive and negative trends, sliding average, and asymmetric quantization, accurate trend characteristics are provided for dynamic threshold adjustment and fault prediction, directly solving technical bottlenecks such as parameter drift false alarms and gradual fault omissions.

[0046] Similarly, the second incremental value and the second attenuation value obtained according to the second monitoring parameter value sequence can be calculated using an expression similar to the above expression, such as the second incremental value , where m is the number of parameter values in the second monitoring parameter value sequence.

[0047] In one embodiment, the prediction model in S2 in which the predicted incremental value corresponding to the grid electrical parameter is obtained based on the prediction model, the first incremental value, and the second incremental value is expressed as: ;in, To predict the incremental value, is the first increment value, is the second increment value.

[0048] In this embodiment, it should be noted that, in the entire expression, The benchmark trend quantification is implemented to calculate the average value of the incremental value of two time periods as a benchmark reference for trend changes, balance the weight of historical and current data, and avoid interference from sudden changes in a single period. The trend direction is strengthened and the trend direction is judged by the symbol function. > A positive direction indicates that the parameter is accelerating (such as overvoltage deterioration); < A negative direction indicates that the parameter growth rate is slowing down or reversing (such as fault recovery). It is a nonlinear weight adjustment based on the difference of incremental values: when the difference is small ( When the difference is large ( Larger): The exponential term approaches 1 and the adjustment amplitude decreases (suppressing mutation noise).

[0049] In summary, the capture of nonlinear mutations is achieved. The linear superposition in the existing technology cannot distinguish between gradual changes and sudden changes (such as a sudden drop in capacitor capacitance vs. periodic fluctuations). In this embodiment, when the increment in the second period is significantly higher than that in the first period (V_{in}^2ggV_{in}^1), the exponential term weight amplifies the positive trend and promptly warns of nonlinear degradation. Furthermore, the trend direction is distinguished. The static threshold in the existing technology cannot determine the direction of parameter change (such as the temperature continues to rise after a brief drop). In this embodiment, the sign function clearly distinguishes between accelerated increases ( ) and deceleration / reversal ( ), avoiding false positives. Furthermore, dynamic environmental adaptation is achieved. In existing technologies, high temperatures can cause parameter drift and lead to false alarms. In this implementation, nonlinear weight adjustment terms dynamically suppress large sudden disturbances (such as lightning overvoltage) while sensitively capturing subtle gradual changes (such as insulation aging).

[0050] For example, the scenario is transformer oil temperature monitoring, the incremental value of the first period =2, the second period increment value =5. Substitute into the expression to calculate the predicted incremental value In summary, the predicted increment of 6.226 is significantly higher than the single-period value, indicating an accelerated increase in oil temperature. Combined with subsequent temperature compensation (such as shrinking the upper threshold as the ambient temperature rises), this triggers an overpressure warning rather than misinterpreting it as normal fluctuation.

[0051] For further example, the scenario is cable leakage current monitoring, the incremental value of the first period =4, the second period increment value =1. Substitute into the expression to calculate the predicted incremental value In summary, the predicted incremental value of 0.553 has decreased significantly, indicating that the leakage current growth trend has reversed, identifying it as a short-term disturbance (such as load switching) rather than continuous insulation degradation, thus avoiding unnecessary maintenance shutdowns.

[0052] In one embodiment, the fault indicator is obtained in S2 according to the upper fluctuation threshold, the lower fluctuation threshold, the predicted attenuation value, and the predicted increment value as follows: ;in, is the fault indicator, is the upper fluctuation threshold, is the lower fluctuation threshold, To predict the incremental value, is the predicted attenuation value.

[0053] In this embodiment, it should be noted that, in the entire expression, Only the predicted incremental values that exceed the upper threshold are retained, and those that do not exceed the threshold are ignored; Only the predicted attenuation value that exceeds the lower threshold is retained, and the under-limit situation is ignored; the two functions are to focus on abnormal deviations and avoid interference from normal fluctuations. Furthermore, the excess part is divided by its corresponding threshold and This converts absolute deviations into relative proportions, eliminating dimensional differences between parameters and achieving unified risk quantification across devices and parameters. Furthermore, bidirectional risk superposition is achieved, independently calculating and adding together the risks of overvoltage (incremental overrun) and undervoltage (attenuation overrun) to comprehensively assess the overall failure probability. This helps capture complex failure scenarios (such as alternating overvoltage and undervoltage caused by voltage fluctuations).

[0054] For example, the scenario is wind turbine grid-connected voltage monitoring, dynamic threshold (compensated for ambient temperature and equipment aging): upper fluctuation threshold =400V, lower fluctuation threshold =380V; predicted incremental value =405V, predicted attenuation value =375V. Substitute into the expression to calculate the fault index Then the diagnostic decision is made and the risk interval is defined: Normal: 0.02, potential risk: 0.02 0.03, Emergency Failure: 0.03 ; Diagnosis results: =0.0125 is in the normal range and does not trigger an alarm.

[0055] For further example, the scenario is a transformer overload causing a voltage surge. The dynamic threshold (compensated for ambient temperature and equipment aging) is: upper fluctuation threshold =410V, lower fluctuation threshold =385V; predicted incremental value =425V, predicted attenuation value =382V. Substitute into the expression to calculate the fault index Then complete the diagnosis decision and the diagnosis result: =0.0366 is in the emergency fault range, triggering an alarm, prompting "overvoltage trend accelerates, load reduction immediately."

[0056] A big data-based intelligent power grid fault diagnosis system is also provided, which includes: an acquisition module, configured to acquire a plurality of power grid electrical parameters based on big data, and to set a first monitoring time period and a second monitoring time period continuously, and to acquire a first monitoring parameter value sequence and a second monitoring parameter value sequence of the power grid electrical parameters within the first monitoring time period and the second monitoring time period, respectively, wherein the time intervals corresponding to two adjacent parameter values within the first monitoring parameter value sequence or the second monitoring parameter value sequence are consistent; a first data processing module, configured to obtain a first incremental value and a first attenuation value according to a first monitoring parameter value sequence, obtain a second incremental value and a second attenuation value according to a second monitoring parameter value sequence, obtain a predicted incremental value corresponding to a power grid electrical parameter based on a prediction model and the first incremental value and the second incremental value, and obtain a predicted attenuation value corresponding to the power grid electrical parameter based on the prediction model and the first attenuation value and the second attenuation value; The second data processing module is used to obtain the upper fluctuation threshold and the lower fluctuation threshold corresponding to the electrical parameters of the power grid, and obtain the fault indicator according to the upper fluctuation threshold, the lower fluctuation threshold, the predicted attenuation value and the predicted increment value; The diagnosis module is used to obtain the power grid fault diagnosis result according to the fault indicators.

[0057] In this embodiment, it should be noted that, regarding the above-mentioned big data-based intelligent diagnosis system for power grid faults, the specific method of performing operations has been described in detail in the implementation of the big data-based intelligent diagnosis method for power grid faults, and will not be elaborated here.

[0058] Figure 5 1 is a block diagram of an electronic device showing a method for intelligent diagnosis of power grid faults based on big data according to an exemplary embodiment. Figure 5 As shown, the electronic device 700 may include: a processor 701 , a memory 702 , and may further include one or more of a multimedia component 703 , an I / O interface 704 (input / output interface), and a communication component 705 .

[0059] The processor 701 is used to control the overall operation of the electronic device 700 to complete all or part of the steps in the above-mentioned big data-based intelligent power grid fault diagnosis method. The memory 702 is used to store various types of data to support the operation of the electronic device 700. This data may include, for example, instructions for any application or method operating on the electronic device 700, as well as application-related data, such as contact information, sent and received messages, images, audio, video, etc. The memory 702 can be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The multimedia component 703 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signals may be further stored in the memory 702 or transmitted via the communication component 705. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 704 provides an interface between the processor 701 and other interface modules, such as a keyboard, a mouse, and buttons. These buttons may be virtual or physical. The communication component 705 is used for wired or wireless communication between the electronic device 700 and other devices. Wireless communication may include Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, 4G, NB-IOT, eMTC, or other 5G networks, or a combination thereof, without limitation. Accordingly, the communication component 705 may include a Wi-Fi module, a Bluetooth module, an NFC module, and the like.

[0060] In an exemplary embodiment, the electronic device 700 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to execute the above-mentioned big data-based intelligent power grid fault diagnosis method.

[0061] In another exemplary embodiment, a computer-readable storage medium including program instructions is further provided. When executed by a processor, the program instructions implement the steps of the above-mentioned method for intelligent diagnosis of power grid faults based on big data. For example, the computer-readable storage medium may be the above-mentioned memory 702 including the program instructions. The program instructions may be executed by the processor 701 of the electronic device 700 to implement the above-mentioned method for intelligent diagnosis of power grid faults based on big data.

[0062] In another exemplary embodiment, a computer program product is also provided, which includes a computer program that can be executed by a programmable device, and the computer program has a code portion for executing the above-mentioned big data-based intelligent diagnosis method for power grid faults when executed by the programmable device.

[0063] The preferred embodiments of the present disclosure are described in detail above in conjunction with the accompanying drawings. However, the present disclosure is not limited to the specific details of the above embodiments. Within the technical concept of the present disclosure, various simple modifications can be made to the technical solutions of the present disclosure, and these simple modifications all fall within the scope of protection of the present disclosure.

[0064] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the present disclosure will not further describe various possible combinations.

[0065] In addition, the various embodiments of the present disclosure may be arbitrarily combined, and as long as they do not violate the concept of the present disclosure, they should also be regarded as the contents disclosed by the present disclosure.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and description of the present invention.

Claims

1. A method for intelligent diagnosis of power grid faults based on big data, characterized in that: include: Acquiring multiple power grid electrical parameters based on big data, setting a first monitoring time period and a second monitoring time period that are continuous, and acquiring a first monitoring parameter value sequence and a second monitoring parameter value sequence of the power grid electrical parameters within the first monitoring time period and the second monitoring time period, respectively, wherein the time intervals corresponding to two adjacent parameter values within the first monitoring parameter value sequence or the second monitoring parameter value sequence are consistent; Obtaining a first incremental value and a first attenuation value according to a first monitoring parameter value sequence, and obtaining a second incremental value and a second attenuation value according to a second monitoring parameter value sequence, and obtaining a predicted incremental value corresponding to a power grid electrical parameter based on a prediction model, the first incremental value, and the second incremental value, and obtaining a predicted attenuation value corresponding to the power grid electrical parameter based on the prediction model, the first attenuation value, and the second attenuation value; Obtaining upper and lower fluctuation thresholds corresponding to electrical parameters of the power grid, and obtaining fault indicators based on the upper and lower fluctuation thresholds, the predicted attenuation value, and the predicted increment value; Obtain power grid fault diagnosis results based on fault indicators.

2. The method for intelligent diagnosis of power grid faults based on big data according to claim 1, characterized in that: The obtaining of the upper fluctuation threshold and the lower fluctuation threshold corresponding to the electrical parameters of the power grid includes: Determining an initial fluctuation range based on historical operating data of the electrical parameters of the power grid, and dynamically adjusting the initial fluctuation range according to real-time monitored power grid operating environment parameters to form a fluctuation range to be processed, wherein the power grid operating environment parameters include equipment temperature, ambient humidity, and load fluctuation level; Perform aging compensation for the fluctuation range to be processed based on the current life cycle status of the power grid equipment; The compensated fluctuation interval to be processed is threshold-separated, and an upper fluctuation threshold and a lower fluctuation threshold that match the current grid operation state are generated.

3. The method for intelligent diagnosis of power grid faults based on big data according to claim 2, characterized in that: The dynamic adjustment of the initial fluctuation range according to the real-time monitored grid operating environment parameters includes: Adjust the upper limit of the initial fluctuation range according to the device temperature change trend. When the temperature is increasing, the upper limit is lowered, and when the temperature is decreasing, the upper limit is increased. The lower limit of the initial fluctuation range is adjusted synchronously according to the change trend of the ambient humidity. When the humidity increases, the lower limit is increased, and when the humidity decreases, the lower limit is lowered. When the load fluctuates violently, the upper and lower limits of the initial fluctuation range are increased, and when the load is stable, the upper and lower limits of the initial fluctuation range are reduced.

4. The method for intelligent diagnosis of power grid faults based on big data according to claim 3 is characterized in that: The obtaining of the power grid fault diagnosis result according to the fault indicator includes: Divide into multiple risk intervals, each of which corresponds to a different fault diagnosis conclusion; Obtain the risk interval that the fault indicator falls into, and obtain the fault diagnosis conclusion corresponding to the risk interval.

5. The method for intelligent diagnosis of power grid faults based on big data according to claim 1, characterized in that: The first incremental value and the first attenuation value are obtained according to the first monitoring parameter value sequence as follows: , ;in, is the first increment value, is the first attenuation value, is the number of parameter values in the first monitoring parameter value sequence, is the j+1th parameter value in the first monitoring parameter value sequence, is the jth parameter value in the first monitoring parameter value sequence.

6. The method for intelligent diagnosis of power grid faults based on big data according to claim 1, characterized in that: The prediction model in obtaining the predicted incremental value corresponding to the electrical parameter of the power grid based on the prediction model, the first incremental value and the second incremental value is expressed as: ;in, To predict the incremental value, is the first increment value, is the second increment value.

7. The method for intelligent diagnosis of power grid faults based on big data according to claim 1, characterized in that: The fault indicator obtained according to the upper fluctuation threshold, the lower fluctuation threshold, the predicted attenuation value and the predicted increment value is expressed as follows: ;in, is the fault indicator, is the upper fluctuation threshold, is the lower fluctuation threshold, To predict the incremental value, is the predicted attenuation value.

8. A big data-based intelligent power grid fault diagnosis system, characterized in that: The system comprises: an acquisition module, configured to acquire a plurality of power grid electrical parameters based on big data, and to set a first monitoring time period and a second monitoring time period continuously, and to acquire a first monitoring parameter value sequence and a second monitoring parameter value sequence of the power grid electrical parameters within the first monitoring time period and the second monitoring time period, respectively, wherein the time intervals corresponding to two adjacent parameter values within the first monitoring parameter value sequence or the second monitoring parameter value sequence are consistent; a first data processing module, configured to obtain a first incremental value and a first attenuation value according to a first monitoring parameter value sequence, obtain a second incremental value and a second attenuation value according to a second monitoring parameter value sequence, obtain a predicted incremental value corresponding to a power grid electrical parameter based on a prediction model and the first incremental value and the second incremental value, and obtain a predicted attenuation value corresponding to the power grid electrical parameter based on the prediction model and the first attenuation value and the second attenuation value; The second data processing module is used to obtain the upper fluctuation threshold and the lower fluctuation threshold corresponding to the electrical parameters of the power grid, and obtain the fault indicator according to the upper fluctuation threshold, the lower fluctuation threshold, the predicted attenuation value and the predicted increment value; The diagnosis module is used to obtain the power grid fault diagnosis result according to the fault indicators.

9. An electronic device, characterized in that: include: a memory having a computer program stored thereon; A processor, configured to execute the computer program in the memory to implement the big data-based intelligent diagnosis method for power grid faults as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for intelligent diagnosis of power grid faults based on big data as described in any one of claims 1 to 7 is implemented.

Images