Port high-precision load and new energy output prediction fusion method and system

Abstract

The invention discloses a port high-precision load and new energy output prediction fusion method and system, and relates to the technical field of port energy management, and the method comprises the steps: obtaining the operation data of port heavy equipment, and extracting the behavior feature information of the operation of the equipment based on the operation data; constructing an equipment operation mode feature set based on the behavior feature information; continuously comparing the equipment operation mode feature set with a standard equipment operation mode feature set to identify whether the statistical distribution of the equipment operation mode has continuous deviation or not; and according to the continuous deviation identification result, generating mode deviation early warning information, and according to the mode deviation early warning information, adjusting a fusion strategy of the port load prediction result and the new energy output prediction result. The output prediction precision and reliability of the port energy management system are improved, and the problem of low energy scheduling efficiency caused by low adaptability of the prediction fusion strategy in the state of slow and continuous change of the heavy equipment operation mode in port energy management is solved.

Description

Technical Field

[0001] This application relates to the field of port energy management technology, specifically a method and system for integrating high-precision port load and new energy output prediction. Background Technology

[0002] In modern port energy management, accurate forecasting of electricity demand and renewable energy generation is crucial for the efficient and stable operation of the system. Port environments are complex, and electricity loads are easily affected by factors such as the operating status of equipment, operator habits, and external weather conditions. Especially when these factors undergo slow and continuous changes, traditional forecasting methods struggle to capture them, easily leading to forecasting biases and impacting the economy and reliability of energy dispatch. In particular, the operating patterns of heavy port equipment (such as quay cranes) are prone to subtle and continuous changes as operators accumulate experience, resulting in slowly accumulating behavioral drift. This leads to statistically significant differences between actual load fluctuations and the "typical" patterns relied upon by the forecasting system.

[0003] Conventional data verification and cleaning mechanisms primarily target drastic abnormal fluctuations such as sudden failures and missing data, but they cannot identify slow, subtle, yet consistent overall load data shifts. Load data with systematic biases is misjudged as valid data and continuously input into the load forecasting system, leading to a clear and persistent directional bias in the forecast results.

[0004] However, the fusion strategy of the port energy forecasting system relies on initially set static weighting coefficients, which cannot adapt to the aforementioned asymmetric input conditions. For example, the system fuses the results of skewed load forecasts and accurate renewable energy forecasts at a fixed ratio, leading to a significant decrease in overall forecast accuracy and causing the fusion strategy of the port energy forecasting system to fail. Furthermore, during troubleshooting, maintenance personnel often mistakenly attribute problems to the renewable energy forecasting system due to the inherent uncertainty of renewable energy output. Adjustments made by modifying parameters or reducing its weights not only fail to solve the fundamental problem of load data deviation but also introduce new errors, further reducing forecast accuracy. This results in multi-energy coordinated scheduling relying on erroneous data for long-term decisions, ultimately leading to low port energy scheduling efficiency. Summary of the Invention

[0005] The purpose of this application is to address the problem of low energy dispatch efficiency in port energy management due to the low adaptability of prediction fusion strategies under the slow and continuous changes in the operating modes of heavy equipment. A high-precision load and renewable energy output prediction fusion method and system for ports is proposed. By acquiring the behavioral characteristics of port heavy equipment, it can effectively identify the continuous deviation of equipment operating modes, improving the output prediction accuracy and reliability of the port energy management system. Based on this, the fusion strategy of load and renewable energy output prediction is dynamically adjusted to avoid the decline in energy dispatch efficiency caused by prediction deviations.

[0006] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows: In a first aspect, embodiments of this application provide a method for fusing high-precision port load and new energy output prediction, the method comprising: Acquire operational data of heavy equipment in the port, and extract behavioral characteristic information of equipment operation based on the operational data; Construct a feature set of equipment operation modes based on the aforementioned behavioral feature information; The equipment operation mode feature set is continuously compared with the standard equipment operation mode feature set to identify whether there is a continuous deviation in the statistical distribution of the equipment operation mode. Based on the persistent deviation identification results, a pattern offset early warning information is generated, and based on the pattern offset early warning information, the fusion strategy of port load forecast results and new energy output forecast results is adjusted.

[0007] This solution acquires behavioral characteristic information of equipment operation and constructs a feature set of equipment operation modes. By continuously comparing this feature set with the feature set of standard equipment operation modes, it can accurately identify slow and continuous deviations in operation modes caused by factors such as the operating habits of heavy equipment in ports. This slow change is quantified, so that pattern deviation early warning information can be generated based on the quantification results. This effectively corrects the systematic deviation of load forecasting and avoids its negative impact on the final fusion forecast results. It significantly improves the overall accuracy of high-precision load and new energy output forecasting in ports. At the same time, it avoids unnecessary adjustments to the originally accurate new energy forecasting system due to misjudgment by operation and maintenance personnel. This ensures that the port's multi-energy collaborative scheduling system can make decisions based on more accurate forecast data in the long term, thereby improving the economy and reliability of energy scheduling.

[0008] Preferably, before acquiring the operating data of the port heavy equipment and extracting behavioral feature information of the equipment operation based on the operating data, the process includes: Acquire sensor operating status information of the device to determine the measurement starting point offset; A real-time dynamic correction factor is generated by measuring the starting point offset to compensate for the original current or voltage data of the device, and a sensor health score is generated based on the change of the dynamic correction factor. Simultaneously, network transmission status information of the device is acquired, and network transmission quality anomaly indicators are determined based on the frequency and duration of abnormal transmission of sensor data packets.

[0009] Preferably, the step of generating a real-time dynamic correction factor by measuring the starting point offset to compensate for the original current or voltage data of the device, and generating a sensor health score based on the change state of the dynamic correction factor, includes: The deviation of the measurement starting point is corrected based on sensor measurement information during low-load operation of the equipment; The corrected sensor operating status information is obtained, and a real-time dynamic correction factor is generated based on the current measurement starting point offset and the measurement offset at the previous moment. Then, when the equipment is operating, the original current or voltage data is compensated based on the real-time dynamic correction factor. Sensor health is scored based on the comparison between the dynamic correction factor and the preset correction threshold.

[0010] Preferably, the step of acquiring the operating data of heavy port equipment and extracting behavioral feature information of equipment operation based on the operating data includes: Calculate the power ramp-up speed, steady-state operating time, and braking energy feedback intensity based on the equipment's operating data; Behavioral characteristic information is obtained based on the power ramp speed, steady-state operating time, and braking energy feedback intensity.

[0011] Preferably, the step of continuously comparing the equipment operation mode feature set with the standard equipment operation mode feature set to identify whether there is a persistent deviation in the statistical distribution of the equipment operation modes includes: The credibility weight of behavioral characteristics is calculated based on sensor health scores and network transmission quality anomaly indicators. The equipment operation mode feature set is compared with the standard equipment operation mode feature set, and the results are weighted and statistically analyzed based on the confidence weight comparison. Based on the weighted statistical analysis results, it is determined whether there is a persistent deviation in the statistical distribution of the equipment operation mode, and the cause of the persistent deviation is attributed based on sensor operating status information, network transmission quality information, and stability information of operator behavior patterns.

[0012] Preferably, the calculation of the credibility weight of behavioral characteristics based on sensor health scores and network transmission quality anomaly indicators includes: Analyze the temporal co-occurrence of the sensor health score and the network transmission quality anomaly indicators; If temporal co-occurrence is detected, then a source tracing analysis is performed on the network transmission quality anomaly indicators; Based on the source tracing analysis results, determine whether the abnormal network transmission quality is the direct cause of the sensor measurement starting point offset; Based on the determination of the measurement starting point offset, adjust the relative contributions of sensor operating status information, network transmission quality information, and operator behavior pattern stability information in the confidence weight calculation. The credibility weight of the behavioral feature information is calculated based on the adjusted relative contribution.

[0013] Preferably, the step of identifying whether there is a persistent deviation in the statistical distribution of the equipment operation mode based on the weighted statistical analysis results, and attributing the persistent deviation based on sensor operating status information, network transmission quality information, and stability information of operator behavior patterns, includes: A multi-dimensional feature space is constructed, which maps the device's sensor operating status information, the device's network transmission quality information, and the stability information of the operator's behavior patterns to the multi-dimensional feature space. Identify the degree of mutual influence between information in the multidimensional feature space, and decouple each piece of information according to the degree of mutual influence; Calculate the contribution of each decoupled information dimension to the persistent deviation of the equipment operation mode; An attribution report is generated based on the contribution level to obtain the identification results of whether there is a persistent deviation in the statistical distribution of the equipment operation mode, and the persistent deviation is attributed according to the sensor operating status information, network transmission quality information, and stability information of operator behavior mode.

[0014] Preferably, identifying the degree of mutual influence between information in the multidimensional feature space includes: Continuously acquire sensor operating status information, network transmission quality information, and stability information of operator behavior patterns; Within a preset time window, calculate the statistical distribution characteristics of the sensor operating status information, network transmission quality information, and operator behavior pattern stability information in different time periods. By comparing the changes in the statistical distribution characteristics of various information within different time windows, it is possible to identify whether there are non-stationary or time-varying characteristics. When non-stationary or time-varying characteristics are identified, the length and overlap rate of the time window are adjusted, and nonlinear correlation analysis is performed on the sensor operating status information, network transmission quality information, and operator behavior pattern stability information within the adjusted time window to quantify the degree of their nonlinear mutual influence.

[0015] Preferably, generating the attribution report based on the contribution includes: Obtain the historical fluctuation range and statistical dispersion related to each contribution level; Based on the currently acquired contribution level, combined with the historical fluctuation range and statistical dispersion, the instantaneous uncertainty of the current contribution level is calculated. The confidence level of the attribution report is adjusted based on the degree of transient uncertainty: when the degree of transient uncertainty increases, the confidence level is decreased; when the degree of transient uncertainty decreases, the confidence level is increased. The recommended strength of the suggested intervention is adjusted based on the adjusted confidence level, thereby generating an attribution report that includes confidence level information and suggested intervention.

[0016] Secondly, embodiments of this application provide a port high-precision load and new energy output prediction fusion system, including: The data feature module is used to acquire the operating data of heavy equipment in the port and extract behavioral feature information of equipment operation based on the operating data; The pattern construction module is used to construct a set of equipment operation pattern features based on the behavioral feature information; The deviation identification module is used to continuously compare the equipment operation mode feature set with the standard equipment operation mode feature set to identify whether there is a continuous deviation in the statistical distribution of the equipment operation mode. The early warning adjustment module is used to generate pattern offset early warning information based on the persistent deviation identification results, and adjust the fusion strategy of port load forecast results and new energy output forecast results based on the pattern offset early warning information.

[0017] The beneficial effects of this application are: 1. By identifying slow and persistent deviations in the operating patterns of heavy equipment in ports and dynamically adjusting the fusion strategy of load and new energy output forecasting accordingly, the limitations of traditional forecasting methods in dealing with such deviations are overcome, significantly improving the accuracy and reliability of output and load forecasting in port energy management systems and avoiding a decline in energy dispatch efficiency due to forecasting deviations. 2. By dynamically correcting the offset of the starting point of the equipment sensor measurement, the accuracy of the original current or voltage data is ensured, and the health status of the sensor is evaluated in real time, thereby effectively improving the reliability and accuracy of data acquisition and further improving the accuracy of continuous deviation identification of the operation mode of heavy equipment in the port. 3. By dynamically adjusting the fusion strategy of port load forecast results and new energy output forecast results, the limitations of the traditional static fusion strategy in the face of load forecast deviation are avoided. The weight of different forecast sources can be flexibly adjusted according to the actual situation, thereby effectively improving the overall forecast accuracy and reliability. Attached Figure Description

[0018] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings. The drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings.

[0019] Figure 1 A flowchart of a method for fusing high-precision port load and new energy output prediction provided in this application embodiment.

[0020] Figure 2 Provided for the embodiments of this application Figure 1 A flowchart illustrating the specific steps of step S3.

[0021] Figure 3 A schematic diagram of the calculation method for the credibility weight of a consistent behavioral feature provided in the embodiments of this application.

[0022] Figure 4 This is a schematic diagram of a port high-precision load and new energy output prediction fusion system module provided in an embodiment of this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely one preferred embodiment of this application and are only used to explain this application. They do not limit the scope of protection of this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] Example 1: As Figure 1 As shown, a method for fusing high-precision port load and new energy output prediction includes steps S1-S4, wherein: S1. Obtain the operating data of heavy equipment in the port, and extract behavioral feature information of equipment operation based on the operating data; S2. Construct a device operation mode feature set based on the behavioral feature information; S3. Continuously compare the equipment operation mode feature set with the standard equipment operation mode feature set to identify whether there is a continuous deviation in the statistical distribution of the equipment operation mode; S4. Generate pattern offset early warning information based on the persistent deviation identification results, and adjust the fusion strategy of port load forecast results and new energy output forecast results based on the pattern offset early warning information.

[0025] In this embodiment, heavy port equipment refers to large mechanical equipment that plays a major role in port operations, such as quay cranes, yard cranes, forklifts, and reach stackers. These devices generate significant electrical loads during operation, and their operating modes directly affect the overall power demand of the port. The equipment operation mode feature set, constructed based on behavioral characteristic information, describes the set of equipment operation modes within a specific time period; it can be understood as the "fingerprint" of equipment operating behavior. Persistent deviation refers to a slow, cumulative, and non-sudden difference between the statistical distribution of equipment operation modes and the standard mode. This difference may be caused by factors such as changes in operator habits and equipment wear. Mode deviation early warning information is used to indicate potential deviations in load forecasting.

[0026] In this embodiment, by acquiring behavioral characteristic information of equipment operation and constructing a feature set of equipment operation modes, and continuously comparing it with a standard feature set of equipment operation modes, it is possible to accurately identify the slow and continuous deviation of operation modes caused by factors such as the operating habits of heavy equipment in the port. This slow change is quantified so that pattern deviation early warning information can be generated based on the quantification results. This effectively corrects the systematic deviation of load forecasting and avoids its negative impact on the final fusion forecast results, significantly improving the overall accuracy of high-precision load and new energy output forecasting in the port. At the same time, it avoids unnecessary adjustments to the originally accurate new energy forecasting system due to misjudgment by operation and maintenance personnel, thereby ensuring that the port multi-energy collaborative scheduling system can make decisions based on more accurate forecast data in the long term, improving the economy and reliability of energy scheduling.

[0027] As an optional implementation, before step S1, the following steps are included: Acquire sensor operating status information of the device to determine the measurement starting point offset; A real-time dynamic correction factor is generated by measuring the starting point offset to compensate for the original current or voltage data of the device, and a sensor health score is generated based on the change of the dynamic correction factor. Simultaneously, network transmission status information of the device is acquired, and network transmission quality anomaly indicators are determined based on the frequency and duration of abnormal transmission of sensor data packets.

[0028] In this embodiment, the measurement starting point offset refers to the systematic deviation between the sensor's measured value and the true zero point or standard reference value under no-load or known stable conditions. Generating a real-time dynamic correction factor by measuring the starting point offset to compensate for the device's original current or voltage data means calculating a correction parameter that can adjust the original measurement data in real time using the determined measurement starting point offset.

[0029] In some embodiments, the real-time dynamic correction factor can be an additive or multiplicative factor used to offset or reduce systematic errors in sensor measurements, ensuring that subsequent processed current or voltage data is closer to the true value. For example, when the sensor has a positive offset, the correction factor can be a negative value, used to subtract the offset from the original measurement value. Generating a sensor health score based on the changing state of the dynamic correction factor refers to evaluating the sensor's operating status and reliability by continuously monitoring the magnitude, trend, and stability of the real-time dynamic correction factor. For example, if the correction factor continuously increases or fluctuates drastically, it may indicate that the sensor performance is declining or about to fail, in which case its health score can be lowered.

[0030] Furthermore, it is possible to obtain network transmission status information of the device, including but not limited to monitoring network performance indicators such as packet loss rate, transmission delay, and jitter. By obtaining abnormal network transmission quality indicators, the health status of network transmission can be accurately reflected, thereby timely discovering and quantifying potential problems in the data transmission process, such as data loss or delay. These problems can also affect the integrity and real-time performance of the operating data.

[0031] In this embodiment, by compensating for the raw current or voltage data of the device, the limitation of decreased prediction accuracy caused by data source quality issues in traditional methods is effectively addressed. Specifically, by acquiring the sensor operating status information of the device and determining the measurement starting point offset, the systematic errors of the sensor itself can be identified and quantified. Through dynamic correction factors, the operating data input to subsequent analysis stages is ensured to have higher accuracy and reliability, thereby avoiding deviations in behavioral feature extraction caused by inaccurate sensor measurements. By comprehensively considering sensor health scores and network transmission quality anomaly indicators, the credibility of the raw operating data is comprehensively evaluated, providing a more reliable data foundation for subsequent behavioral feature extraction and pattern recognition.

[0032] As an optional implementation, the step of generating a real-time dynamic correction factor by measuring the starting point offset to compensate for the original current or voltage data of the device, and generating a sensor health score based on the change state of the dynamic correction factor, includes: The deviation of the measurement starting point is corrected based on sensor measurement information during low-load operation of the equipment; The corrected sensor operating status information is obtained, and a real-time dynamic correction factor is generated based on the current measurement starting point offset and the measurement offset at the previous moment. Then, when the equipment is operating, the original current or voltage data is compensated based on the real-time dynamic correction factor. Sensor health is scored based on the comparison between the dynamic correction factor and the preset correction threshold.

[0033] It should be noted that after the initial correction of the measurement starting point deviation, the system continuously monitors the sensor's operating status. The real-time dynamic correction factor is not a fixed value, but is dynamically generated based on the relationship between the measurement starting point deviation at the current moment and the measurement deviation at the previous moment. This dynamic generation mechanism allows the correction factor to adapt to the slow drift or sudden changes in sensor performance over time, ensuring that the compensation for the original current or voltage data is real-time and highly accurate during equipment operation. Furthermore, by compensating for possible errors in real-time sensor measurements, it ensures that the compensated data can more accurately reflect the true electrical parameters of the equipment, reducing the prediction risk caused by data quality issues. This effectively avoids prediction errors caused by sensor data quality problems and improves the robustness of the overall prediction system.

[0034] As an optional implementation, step S1 includes: Calculate the power ramp-up speed, steady-state operating time, and braking energy feedback intensity based on the equipment's operating data; Behavioral characteristic information is obtained based on the power ramp speed, steady-state operating time, and braking energy feedback intensity.

[0035] Specifically, by deploying sensors on the equipment, electrical parameters such as current, voltage, and power, as well as mechanical parameters such as operating speed, position, and work cycle time, are collected in real time. Signal processing techniques (such as filtering and noise reduction) are used to preprocess the collected raw data to eliminate noise and outliers. Based on the preprocessed data, behavioral characteristics that can quantify the equipment's operating characteristics and reflect the operator's working habits or the equipment's operating status are extracted, including but not limited to: power ramp-up speed, steady-state operating time, and braking energy feedback intensity. Among them, power ramp-up speed represents the time and rate required for power to rise from zero to a stable value, which is obtained by differentiating the equipment's current or voltage data and normalizing it in combination with the equipment's rated power parameters; steady-state operating time represents the time the equipment runs continuously at a certain load level, which is determined by identifying continuous time periods in the equipment's operating data where the power fluctuation amplitude is less than a preset threshold; braking energy feedback intensity represents the intensity of energy fed back to the grid by the equipment during braking, which is quantified by monitoring the amplitude and duration of the feedback current or voltage.

[0036] Furthermore, based on the behavioral feature information obtained above, a feature set of equipment operation mode is constructed, including: aggregating the behavioral feature information based on the equipment operation cycle to form a multi-dimensional vector or time series, which serves as the feature set of equipment operation mode within that time period; the feature set includes statistical indicators such as the average, maximum, and standard deviation of power ramp-up speed, the distribution of steady-state operation time, and the cumulative value of braking energy feedback intensity.

[0037] In other implementations, clustering algorithms (such as K-means and DBSCAN) are used to analyze behavioral feature information, classifying equipment operating states with similar behavioral characteristics into different operating modes, and generating a representative feature vector for each operating mode. These feature vectors together constitute the equipment operating mode feature set. For example, the operating modes of quay cranes can be divided into rapid loading and unloading mode, slow and precise operation mode, and idle standby mode, each with its unique combination of behavioral features.

[0038] In this embodiment, by acquiring behavioral characteristic information such as power ramp-up speed, steady-state running time, and braking energy feedback intensity, the operating mode of the equipment can be comprehensively characterized from both dynamic and static dimensions. For example, power ramp-up speed can reflect the equipment's initiative in starting and accelerating, steady-state running time reveals the efficiency and stability of the equipment's continuous operation, and braking energy feedback intensity reflects the equipment's energy management strategy during deceleration or braking. This allows for more accurate capture of the equipment's unique behavioral patterns under different operating scenarios, thus providing high-quality, high-dimensional data support for the subsequent construction of the equipment's operating mode feature set.

[0039] It should be noted that, compared to traditional methods that rely solely on average power or simple switching states, incorporating power ramp-up speed, steady-state operating time, and braking energy feedback intensity can more accurately reflect the dynamic changes and energy interaction characteristics of equipment during actual operation. Therefore, the extracted behavioral feature information has higher discriminative power and representativeness, helping to more effectively identify whether there are persistent deviations in the statistical distribution of equipment operating modes, thereby improving the accuracy and timeliness of the strategy adjustment for integrating port load forecasting results with new energy output forecasting results.

[0040] As an optional implementation method, combined with Figure 2 As shown, step S3 includes: The credibility weight of behavioral characteristics is calculated based on sensor health scores and network transmission quality anomaly indicators. The equipment operation mode feature set is compared with the standard equipment operation mode feature set, and the results are weighted and statistically analyzed based on the confidence weight comparison. Based on the weighted statistical analysis results, it is determined whether there is a persistent deviation in the statistical distribution of the equipment operation mode, and the cause of the persistent deviation is attributed based on sensor operating status information, network transmission quality information, and stability information of operator behavior patterns.

[0041] In some embodiments, continuously comparing the equipment operation mode feature set with a standard equipment operation mode feature set includes: Statistical methods, such as the chi-square test, Kolmogorov-Smirnov test, or Wasserstein distance, are used to compare the differences in statistical distributions between two feature sets. If the statistical distribution difference exceeds a preset threshold, it is considered that there is a deviation.

[0042] As an optional implementation method, combined with Figure 3 As shown, the credibility weights for calculating behavioral characteristics based on sensor health scores and network transmission quality anomaly indicators include: Analyze the temporal co-occurrence of the sensor health score and the network transmission quality anomaly indicators; If temporal co-occurrence is detected, then a source tracing analysis is performed on the network transmission quality anomaly indicators; Based on the source tracing analysis results, determine whether the abnormal network transmission quality is the direct cause of the sensor measurement starting point offset; Based on the determination of the measurement starting point offset, adjust the relative contributions of sensor operating status information, network transmission quality information, and operator behavior pattern stability information in the confidence weight calculation. The credibility weight of the behavioral feature information is calculated based on the adjusted relative contribution.

[0043] In some embodiments, temporal co-occurrence refers to the overlap or close correlation between abnormal changes in sensor health scores and the occurrence or changes in abnormal network transmission quality indicators in time. If temporal co-occurrence is detected, the root cause of the abnormal network transmission quality can be investigated in depth. For example, network logs, device connection status, packet loss rate, latency jitter, etc. can be checked to determine the specific type and scope of the network anomaly, so as to provide a basis for subsequent judgment on whether the network anomaly directly caused the sensor measurement starting point offset.

[0044] Furthermore, the source analysis results are used to determine whether network transmission quality anomalies are the direct cause of sensor measurement start-point offset. This aims to distinguish between independent events and causally related events, avoiding misjudgments. For example, if the source analysis shows that network congestion caused severe loss or delay of sensor data packets, thus affecting the accuracy of sensor data and ultimately manifesting as measurement start-point offset, then network transmission quality anomalies can be considered the direct cause. Conversely, if there is no direct causal relationship between network anomalies and sensor measurement start-point offset, then it is not considered a direct cause. Once this causal relationship is determined, the relative contributions of sensor operating status information, network transmission quality information, and operator behavior pattern stability information in the reliability weight calculation can be adjusted accordingly.

[0045] In this embodiment, by calculating the credibility weights of behavioral feature information, the problem of inaccurate identification due to the potential neglect of data source reliability in pattern comparison is addressed. Sensor health scores and network transmission quality anomaly indicators, as direct reflections of data quality, are used to quantify the reliability of behavioral features. When these indicators show a decline in data quality, the credibility weight of the corresponding behavioral feature decreases accordingly, thus appropriately weakening the impact of these low-reliability behavioral features on pattern deviation identification in subsequent weighted statistical analysis. Conversely, high-quality behavioral features are assigned higher weights, ensuring that the pattern deviation identification results more accurately reflect the actual operating status of the device. Furthermore, by incorporating sensor health scores and network transmission quality anomaly indicators into the calculation of the credibility weights of behavioral features, the pattern comparison and deviation identification process can effectively avoid misjudgments caused by data quality issues.

[0046] As an optional implementation, the step of identifying whether there is a persistent deviation in the statistical distribution of the equipment operation mode based on the weighted statistical analysis results, and attributing the persistent deviation based on sensor operating status information, network transmission quality information, and stability information of operator behavior patterns, includes: A multi-dimensional feature space is constructed, which maps the device's sensor operating status information, the device's network transmission quality information, and the stability information of the operator's behavior patterns to the multi-dimensional feature space. Identify the degree of mutual influence between information in the multidimensional feature space, and decouple each piece of information according to the degree of mutual influence; Calculate the contribution of each decoupled information dimension to the persistent deviation of the equipment operation mode; An attribution report is generated based on the contribution level to obtain the identification results of whether there is a persistent deviation in the statistical distribution of the equipment operation mode, and the persistent deviation is attributed according to the sensor operating status information, network transmission quality information, and stability information of operator behavior mode.

[0047] In this embodiment, by constructing a multi-dimensional feature space, the sensor operating status information, network transmission quality information, and operator behavior pattern stability information of the equipment are uniformly represented, thus laying the foundation for subsequent comprehensive analysis and enabling the system to systematically identify the complex degree of mutual influence among them. By decoupling these mutual influences, redundancy and interference between factors can be effectively eliminated, thereby more clearly identifying the independent contributions of sensor operating status information, network transmission quality information, and operator behavior pattern stability information to the continuous deviation of equipment operation patterns.

[0048] As an optional implementation, identifying the degree of mutual influence between information in the multidimensional feature space includes: Continuously acquire sensor operating status information, network transmission quality information, and stability information of operator behavior patterns; Within a preset time window, calculate the statistical distribution characteristics of the sensor operating status information, network transmission quality information, and operator behavior pattern stability information in different time periods. By comparing the changes in the statistical distribution characteristics of various information within different time windows, it is possible to identify whether there are non-stationary or time-varying characteristics. When non-stationary or time-varying characteristics are identified, the length and overlap rate of the time window are adjusted, and nonlinear correlation analysis is performed on the sensor operating status information, network transmission quality information, and operator behavior pattern stability information within the adjusted time window to quantify the degree of their nonlinear mutual influence.

[0049] In this embodiment, the limitations of identifying mutual influences of information in the aforementioned multidimensional feature space are overcome by introducing dynamic time window analysis and nonlinear correlation quantification. Specifically, firstly, continuous acquisition of multi-source information ensures comprehensive real-time monitoring of the system status. Secondly, statistical distribution characteristics are calculated within a preset time window, enabling the system to capture the inherent change patterns of information over different time periods. It is precisely by comparing the changes in statistical distribution characteristics within different time windows that the system can identify non-stationary or time-varying characteristics that traditional methods may overlook. Once these dynamic characteristics are identified, the analysis can better adapt to the actual dynamics of the data by adaptively adjusting the length and overlap of the time window. Based on this, nonlinear correlation analysis is employed to overcome the limitations of linear models and accurately quantify the complex nonlinear mutual influence between sensor operating status information, network transmission quality information, and operator behavior pattern stability information. Thus, this embodiment enables a deeper and more accurate understanding of how these factors work together to influence equipment operating modes, providing a solid foundation for subsequent attribution analysis.

[0050] Specifically, compared to methods that only use static or linear analysis, the degree of nonlinear interaction can effectively capture and quantify the non-stationary, time-varying, and nonlinear relationships between these factors, thereby avoiding attribution bias caused by inconsistent model assumptions. It can also provide more accurate and detailed attribution reports for the continuous deviation of equipment operation modes, thereby supporting more precise early warning of mode shifts and adjustments to the strategy of integrating port load and new energy output forecasts, ultimately improving the intelligence level and forecast accuracy of port energy management.

[0051] As an optional implementation, generating the attribution report based on the contribution includes: Obtain the historical fluctuation range and statistical dispersion related to each contribution level; Based on the currently acquired contribution level, combined with the historical fluctuation range and statistical dispersion, the instantaneous uncertainty of the current contribution level is calculated. The confidence level of the attribution report is adjusted based on the degree of transient uncertainty: when the degree of transient uncertainty increases, the confidence level is decreased; when the degree of transient uncertainty decreases, the confidence level is increased. The recommended strength of the suggested intervention is adjusted based on the adjusted confidence level, thereby generating an attribution report that includes confidence level information and suggested intervention.

[0052] In some embodiments, the instantaneous uncertainty of the current contribution is calculated based on the currently acquired contribution, combined with historical fluctuation range and statistical dispersion. This includes determining whether the current contribution is in an abnormal fluctuation state by comparing its current contribution with its historical performance. For example, if the current contribution deviates from its historical average by more than a preset statistical threshold (such as two standard deviations), its instantaneous uncertainty can be considered high. The instantaneous uncertainty reflects the potential risk that the current contribution may change in a short period of time.

[0053] In this embodiment, by introducing an assessment of the instantaneous uncertainty of contribution information, the problem of inaccurate confidence levels in traditional attribution reports due to fluctuations in contribution is effectively solved. When a high degree of instantaneous uncertainty in contribution is identified, the system can adjust the confidence level of the attribution report in a timely manner, thereby avoiding giving overly arbitrary or erroneous attribution conclusions under conditions of high uncertainty. This ensures that the attribution report can more accurately reflect the root cause and reliability of the deviation from the current equipment operating mode.

[0054] In this embodiment, the attribution report not only identifies the potential causes of persistent deviations in equipment operating patterns but also quantifies the confidence levels of these attribution results, enabling decision-makers to more clearly understand the reliability of the attribution conclusions. Furthermore, by adjusting the recommended strength of suggested interventions based on the confidence level, more targeted and practical intervention strategies can be provided, avoiding unnecessary or high-risk actions when uncertainty is high. This significantly improves the effectiveness and economy of adjusting the fusion strategy between port load forecasting and renewable energy output forecasting results.

[0055] In some examples, it is assumed that, over a certain period, the stability information of operator behavior patterns, calculated using the above method, contributes 60% to the persistent deviation of equipment operating patterns. However, analysis of historical data reveals that the contribution of this operator behavior pattern fluctuated significantly over the past week, and its instantaneous uncertainty was calculated as high. Based on this, the system lowers the confidence level of the "operator behavior pattern" attribution item in the attribution report, for example, from the default 90% to 70%. Simultaneously, the recommendation strength for interventions suggested regarding the instability of operator behavior patterns (e.g., retraining operators or optimizing operating procedures) is adjusted from "strongly recommended" to "suggest further observation and evaluation." Thus, the generated attribution report will clearly indicate that operator behavior patterns are one of the potential causes, but also indicate that the confidence level of this attribution is low, and recommend more cautious interventions, thereby avoiding overreaction when information is incomplete or uncertainty is high, and improving the robustness of decision-making.

[0056] As an optional implementation, step S4 includes: When a persistent deviation from the equipment's operating mode is detected, an early warning message is generated that includes the degree of deviation, duration, and potential impact range. Based on the aforementioned early warning information, the contribution ratios of load forecasting and renewable energy output forecasting in the final fusion result are dynamically adjusted according to the real-time identified pattern deviation.

[0057] It should be noted that dynamically adjusting the contribution ratio of load forecasting and renewable energy output forecasting in the final fusion result ensures that when there are deviations in load forecasting, the system can rely more on the relatively accurate renewable energy output forecasting, thereby improving the overall forecasting accuracy. For example, if the load forecasting is systematically underestimated due to deviations in equipment operating modes, the weight of the load forecasting result in the fusion strategy can be appropriately increased, or a correction based on the degree of deviation can be made to the load forecasting result.

[0058] In this embodiment, by dynamically adjusting the fusion strategy of load forecasting and new energy output forecasting, the problem of reduced overall forecast accuracy caused by statically weighting and fusing biased load forecasting results with accurate new energy output forecasting results can be effectively avoided, ensuring that the port multi-energy collaborative scheduling system can make decisions based on more accurate forecasting data.

[0059] Based on the same inventive concept, this application also provides a port high-precision load and new energy output prediction fusion system corresponding to a port high-precision load and new energy output prediction fusion method, such as... Figure 4 As shown, the system includes: The data feature module is used to acquire the operating data of heavy equipment in the port and extract behavioral feature information of equipment operation based on the operating data; The pattern construction module is used to construct a set of equipment operation pattern features based on the behavioral feature information; The deviation identification module is used to continuously compare the equipment operation mode feature set with the standard equipment operation mode feature set to identify whether there is a continuous deviation in the statistical distribution of the equipment operation mode. The early warning adjustment module is used to generate pattern offset early warning information based on the persistent deviation identification results, and adjust the fusion strategy of port load forecast results and new energy output forecast results based on the pattern offset early warning information.

[0060] In this embodiment, by acquiring behavioral characteristic information of equipment operation and constructing a feature set of equipment operation modes, and continuously comparing it with a standard feature set of equipment operation modes, it is possible to accurately identify the slow and continuous deviation of operation modes caused by factors such as the operating habits of heavy equipment in the port. This slow change is quantified so that pattern deviation early warning information can be generated based on the quantification results. This effectively corrects the systematic deviation of load forecasting and avoids its negative impact on the final fusion forecast results, significantly improving the overall accuracy of high-precision load and new energy output forecasting in the port. At the same time, it avoids unnecessary adjustments to the originally accurate new energy forecasting system due to misjudgment by operation and maintenance personnel, thereby ensuring that the port multi-energy collaborative scheduling system can make decisions based on more accurate forecast data in the long term, improving the economy and reliability of energy scheduling.

[0061] The above-described embodiments are preferred embodiments of this application and are not intended to limit the specific scope of this application. The scope of this application includes but is not limited to the specific embodiments described above. All equivalent changes made in accordance with the shape, structure, and method of this application are within the protection scope of this application.

Claims

1. A method for fusing high-precision port load and new energy output prediction, characterized in that, Includes the following steps: Acquire operational data of heavy equipment in the port, and extract behavioral characteristic information of equipment operation based on the operational data; Construct a feature set of equipment operation modes based on the aforementioned behavioral feature information; The equipment operation mode feature set is continuously compared with the standard equipment operation mode feature set to identify whether there is a continuous deviation in the statistical distribution of the equipment operation mode. Based on the persistent deviation identification results, a pattern offset early warning information is generated, and based on the pattern offset early warning information, the fusion strategy of port load forecast results and new energy output forecast results is adjusted.

2. The method for fusing high-precision port load and new energy output prediction according to claim 1, characterized in that, Before acquiring the operating data of heavy port equipment and extracting behavioral feature information of equipment operation based on the operating data, the process includes: Acquire sensor operating status information of the device to determine the measurement starting point offset; A real-time dynamic correction factor is generated by measuring the starting point offset to compensate for the original current or voltage data of the device, and a sensor health score is generated based on the change of the dynamic correction factor. Simultaneously, network transmission status information of the device is acquired, and network transmission quality anomaly indicators are determined based on the frequency and duration of abnormal transmission of sensor data packets.

3. The method for fusing high-precision port load and new energy output prediction according to claim 2, characterized in that, The process of generating a real-time dynamic correction factor by measuring the starting point offset to compensate for the original current or voltage data of the device, and generating a sensor health score based on the changes in the dynamic correction factor, includes: The deviation of the measurement starting point is corrected based on sensor measurement information during low-load operation of the equipment; The corrected sensor operating status information is obtained, and a real-time dynamic correction factor is generated based on the current measurement starting point offset and the measurement offset at the previous moment. Then, when the equipment is operating, the original current or voltage data is compensated based on the real-time dynamic correction factor. Sensor health is scored based on the comparison between the dynamic correction factor and the preset correction threshold.

4. The method for fusing high-precision port load and new energy output prediction according to claim 2, characterized in that, The acquisition of operational data of heavy port equipment, and the extraction of behavioral characteristic information of equipment operation based on the operational data, includes: Calculate the power ramp-up speed, steady-state operating time, and braking energy feedback intensity based on the equipment's operating data; Behavioral characteristic information is obtained based on the power ramp speed, steady-state operating time, and braking energy feedback intensity.

5. The method for fusing high-precision port load and new energy output prediction according to claim 2, characterized in that, The step of continuously comparing the equipment operation mode feature set with the standard equipment operation mode feature set to identify whether there is a persistent deviation in the statistical distribution of the equipment operation modes includes: The credibility weight of behavioral characteristics is calculated based on sensor health scores and network transmission quality anomaly indicators. The equipment operation mode feature set is compared with the standard equipment operation mode feature set, and the results are weighted and statistically analyzed based on the confidence weight comparison. Based on the weighted statistical analysis results, it is determined whether there is a persistent deviation in the statistical distribution of the equipment operation mode, and the cause of the persistent deviation is attributed based on sensor operating status information, network transmission quality information, and stability information of operator behavior patterns.

6. The method for fusing high-precision port load and new energy output prediction according to claim 5, characterized in that, The credibility weights for calculating behavioral characteristics based on sensor health scores and network transmission quality anomaly indicators include: Analyze the temporal co-occurrence of the sensor health score and the network transmission quality anomaly indicators; If temporal co-occurrence is detected, then a source tracing analysis is performed on the network transmission quality anomaly indicators; Based on the source tracing analysis results, determine whether the abnormal network transmission quality is the direct cause of the sensor measurement starting point offset; Based on the determination of the measurement starting point offset, adjust the relative contributions of sensor operating status information, network transmission quality information, and operator behavior pattern stability information in the confidence weight calculation. The credibility weight of the behavioral feature information is calculated based on the adjusted relative contribution.

7. The method for fusing high-precision port load and new energy output prediction according to claim 6, characterized in that, The step involves identifying whether there is a persistent deviation in the statistical distribution of the equipment's operating mode based on the weighted statistical analysis results, and attributing the persistent deviation to factors based on sensor operating status information, network transmission quality information, and operator behavior pattern stability information, including: A multi-dimensional feature space is constructed, which maps the device's sensor operating status information, the device's network transmission quality information, and the stability information of the operator's behavior patterns to the multi-dimensional feature space. Identify the degree of mutual influence between information in the multidimensional feature space, and decouple each piece of information according to the degree of mutual influence; Calculate the contribution of each decoupled information dimension to the persistent deviation of the equipment operation mode; An attribution report is generated based on the contribution level to obtain the identification results of whether there is a persistent deviation in the statistical distribution of the equipment operation mode, and the persistent deviation is attributed according to the sensor operating status information, network transmission quality information, and stability information of operator behavior mode.

8. The method for fusing high-precision port load and new energy output prediction according to claim 7, characterized in that, The identification of the degree of mutual influence between information in the multidimensional feature space includes: Continuously acquire sensor operating status information, network transmission quality information, and stability information of operator behavior patterns; Within a preset time window, calculate the statistical distribution characteristics of the sensor operating status information, network transmission quality information, and operator behavior pattern stability information in different time periods. By comparing the changes in the statistical distribution characteristics of various information within different time windows, it is possible to identify whether there are non-stationary or time-varying characteristics. When non-stationary or time-varying characteristics are identified, the length and overlap rate of the time window are adjusted, and nonlinear correlation analysis is performed on the sensor operating status information, network transmission quality information, and operator behavior pattern stability information within the adjusted time window to quantify the degree of their nonlinear mutual influence.

9. The method for fusing high-precision port load and new energy output prediction according to claim 7, characterized in that, The generation of the attribution report based on the contribution includes: Obtain the historical fluctuation range and statistical dispersion related to each contribution level; Based on the currently acquired contribution level, combined with the historical fluctuation range and statistical dispersion, the instantaneous uncertainty of the current contribution level is calculated. The confidence level of the attribution report is adjusted based on the degree of transient uncertainty: when the degree of transient uncertainty increases, the confidence level is decreased; when the degree of transient uncertainty decreases, the confidence level is increased. The recommended strength of the suggested intervention is adjusted based on the adjusted confidence level, thereby generating an attribution report that includes confidence level information and suggested intervention.

10. A port high-precision load and new energy output prediction fusion system, applicable to the port high-precision load and new energy output prediction fusion method as described in any one of claims 1-9, characterized in that, include: The data feature module is used to acquire the operating data of heavy equipment in the port and extract behavioral feature information of equipment operation based on the operating data; The pattern construction module is used to construct a set of equipment operation pattern features based on the behavioral feature information; The deviation identification module is used to continuously compare the equipment operation mode feature set with the standard equipment operation mode feature set to identify whether there is a continuous deviation in the statistical distribution of the equipment operation mode. The early warning adjustment module is used to generate pattern offset early warning information based on the persistent deviation identification results, and adjust the fusion strategy of port load forecast results and new energy output forecast results based on the pattern offset early warning information.

Images