Mobile energy storage emergency power supply non-inductive switching method and device

Abstract

The invention discloses a mobile energy storage emergency power supply non-inductive switching method and device, and belongs to the technical field of power supply equipment, and the method comprises the steps: S1, obtaining a load prediction model through a deep learning algorithm based on first original historical load data and first original historical environment data; s2, inputting the second original historical load data and the second original historical environment data into the load prediction model to obtain a first prediction load curve, and obtaining a corresponding rule between the first prediction load curve and an actual load curve by using fuzzy control based on the first prediction load curve and the actual load curve; s3, inputting the original real-time load data of the user into the load prediction model to obtain a second prediction load curve, and obtaining a final load curve by using the second prediction load curve based on a corresponding rule; and S4, on the basis of the final load curve, switching information is obtained by using fuzzy control, and switching is carried out on the basis of the switching information. The technical problem that the original characteristics of the data are ignored during switching prediction is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply equipment, and in particular to a method and device for non-sensing switching of a mobile energy storage emergency power supply. Background Art

[0002] With the construction of new power systems, energy storage technology plays an increasingly important role in power grids. Mobile energy storage vehicles are often used to ensure power supply and power reliability due to their high flexibility and convenience. However, due to the characteristics of mobile energy storage vehicles, the energy storage capacity is limited. The existing technology mainly switches between multiple mobile energy storage vehicles to meet the capacity requirements, such as a dual power switching device and method with patent number CN115912599A, including: an industrial frequency circuit outputs industrial frequency power; a UPS circuit outputs UPS power; a control unit is used to switch the industrial frequency circuit or UPS circuit for power output to output industrial frequency power or UPS power to the transmitter; the control unit is used to control the industrial frequency circuit to continuously output industrial frequency power to the transmitter, and monitor in real time whether the industrial frequency circuit fails, and when the industrial frequency circuit fails, switch to the UPS circuit to output UPS power to the transmitter; the control unit is also used to switch to the industrial frequency circuit to output industrial frequency power to the transmitter after detecting that the UPS circuit has failed. The above scheme uses the power frequency circuit and UPS circuit for dual power supply, and switches the dual power supply in real time by judging the faulty circuit, which can effectively prevent the switching of the power supply when the UPS power supply fails or does not meet the power supply requirements. However, the above scheme is switched based on the monitoring results through real-time monitoring, and does not predict the state of the circuit, etc., which poses the risk of untimely switching. Current technology mainly predicts the state by building a model, but in the process of building the model, in order to simplify the complexity of the model, historical data is usually processed, but in the process of processing the data, the original characteristics of the data are ignored, which leads to the loss of the original characteristics of the data. Summary of the invention

[0003] In view of the technical problem that the original characteristics of data are ignored when making switching predictions in the prior art, the present invention provides a method for senseless switching of a mobile energy storage emergency power supply, which uses a load prediction model obtained from the original data to retain the original characteristics of the data, and then obtains corresponding rules through the load prediction model and the fuzzy reasoning ability of fuzzy control, and obtains the final load curve according to the corresponding rules, and obtains switching information according to the final load curve and the fuzzy reasoning ability of fuzzy control, thereby realizing the prediction of the final load curve, and then realizing the prediction of the switching information, solving the technical problem that the original characteristics of data are ignored when making switching predictions, and significantly improving the timeliness and accuracy of switching.

[0004] In order to solve the above technical problems, the present invention provides a method for non-sensing switching of a mobile energy storage emergency power supply, comprising the following steps: S1: Based on the user's first original historical load data and the first original historical environmental data of the mobile energy storage emergency power supply, a load forecasting model is obtained using a deep learning algorithm; S2: Input the second original historical load data of the user and the second original historical environmental data of the mobile energy storage emergency power supply into the load forecasting model to obtain a first forecast load curve, and based on the first forecast load curve and the actual load curve, use fuzzy control to obtain a correspondence rule between the first forecast load curve and the actual load curve; S3: inputting the user's original real-time load data into the load forecasting model to obtain a second forecast load curve, and based on the corresponding rule, using the second forecast load curve to obtain a final load curve; S4: Based on the final load curve, the fuzzy control is used to obtain switching information, and switching is performed based on the switching information.

[0005] After adopting the above technical solution, the present invention has the following advantages: Considering that the model constructed by raw data and deep learning algorithm has a lot of uncertainty, and fuzzy control has fuzzy reasoning ability, the corresponding rules between the large amount of uncertainty generated and the actual situation are obtained through fuzzy control, so as to realize the quantification of a large amount of uncertainty, and then retain the original characteristics of the data through quantification. At the same time, by obtaining the first predicted load curve, the actual load curve has a corresponding standard, which makes it easier for fuzzy control to obtain the corresponding rules. The combination of raw data, deep learning algorithm and fuzzy control significantly improves the timeliness and accuracy of switching; The final load curve obtained by the corresponding rules retains the circuit status, which is convenient for timely detection of potential fault risks and taking corresponding preventive measures, thereby improving switching safety and circuit safety; The switching information is obtained according to the final load curve and the fuzzy reasoning ability of fuzzy control, and the switching information can be adjusted in time according to the final load curve, thereby improving the switching flexibility. While improving the continuity and reliability of power supply, it also improves the stability and operation efficiency of the system; The technical problem of ignoring the original characteristics of the data when making switching predictions has been solved.

[0006] Preferably, the S1 comprises: S11: constructing a division standard based on the original historical load data characteristics of the user and the original historical environment data characteristics of the mobile energy storage emergency power supply, dividing the original historical load data into the first original historical load data and the second original historical load data based on the division standard, and dividing the original historical environment data into the first original historical environment data and the second original historical environment data based on the division standard; S12: Build an initial load forecasting model based on deep learning algorithm; S13: Using a cross-validation method, the performance index of the initial load prediction model is obtained based on the first original historical load data and the first original historical environment data. If the performance index meets the preset conditions, the initial load prediction model is the load prediction model, otherwise execute S13.

[0007] In this scheme, the characteristics of the original historical load data and the characteristics of the original historical environment data are both long-term data characteristics, i.e., four-season characteristics, and short-term data characteristics, i.e., monthly characteristics or weekly characteristics. The division standard is constructed through the data characteristics, so that the first original historical load data and the second original historical load data obtained through the division have the same characteristics, thereby ensuring the feature consistency of the first original historical load data and the second original historical load data, so that the load forecasting model can be applied to the second original historical load data, and at the same time covering the long-term and short-term data characteristics, so that the load forecasting model can be applied to the original real-time load data. Similarly, the division standard is constructed through the data characteristics, so that the first original historical environment data and the second original historical environment data obtained through the division have the same characteristics, thereby ensuring the feature consistency of the first original historical environment data and the second original historical load environment, so that the load forecasting model can be applied to the second original historical environment data, and at the same time covering the long-term and short-term data characteristics, so that the load forecasting model can be applied to the original real-time load data, thereby significantly improving the applicability of the load forecasting model.

[0008] Preferably, the S13 includes: S131: dividing the first original historical load data and the first original historical environment data into a plurality of sets based on a first criterion in the division criteria; S132: Input several sets into the initial load prediction model in sequence to train the initial load prediction model, and obtain the performance indicators of the initial load prediction model in sequence, until the average value of the performance indicator is greater than or equal to the preset value, then stop inputting the sets. At this time, the performance indicator meets the preset conditions. When several sets are input and the average value of the performance indicator is still less than the preset value, the second standard in the division standard is used as the first standard, and S131 is executed.

[0009] In this scheme, the first standard is week, and the second standard is month or quarter. The initial load forecasting model is first trained from a small period, i.e., week, to find the weekly characteristics of the first original historical load data and the first original historical environmental data. At the same time, when the first original historical load data and the first original historical environmental data do not have weekly characteristics, the monthly characteristics and quarterly characteristics of the first original historical load data and the first original historical environmental data are found. By obtaining regular data characteristics from small to large, the accuracy and flexibility of the load forecasting model are improved.

[0010] Preferably, in S2, the use of fuzzy control to obtain a correspondence rule between the first predicted load curve and the actual load curve based on the first predicted load curve and the actual load curve includes: Compare the first predicted load curves of the corresponding time periods in different cycles, and compare the actual load curves of the corresponding time periods in different cycles. If the comparisons are successful, bind the first predicted load curves and the actual load curves of the corresponding time periods in different cycles to obtain a binding set; If the comparison is unsuccessful, the maximum actual load curve among the actual load curves of the corresponding time periods in different cycles is bound to the corresponding time periods in different cycles, and then the binding set is updated, and the binding set is the corresponding rule.

[0011] In this scheme, the relationship between the first predicted load curve and the actual load curve is quantified through the fuzzy reasoning ability of fuzzy control, and then the original characteristics of the data are retained by quantification. When the comparison is successful, the first predicted load curve and the actual load curve are bound to the corresponding time period. When the second predicted curve in the corresponding time period matches the first predicted load curve, the corresponding actual load curve is the final load curve. In this way, the original characteristics of the data are retained while the accuracy of the final load curve is guaranteed. At the same time, when the comparison is unsuccessful, the maximum actual load curve in the actual load curve is bound to the corresponding time period in different cycles, so that the switching information obtained subsequently can meet the energy requirements of the system, thereby ensuring the safety of the system.

[0012] Preferably, in S3, the obtaining of the final load curve using the second predicted load curve based on the corresponding rule includes: obtaining the final load curve according to a matching result between an acquisition time period of the second predicted load curve and a corresponding time period in the corresponding rule.

[0013] Preferably, S4 includes: S41: using the original historical load data and the historical capacity of the mobile energy storage emergency power supply as input variables, and using the switching information as an output variable; S42: Acquire a mapping relationship between input variables and output variables based on expert experience in the fuzzy control; S43: Obtain switching information based on the mapping relationship, the final load curve and the actual capacity of the mobile energy storage emergency power supply.

[0014] In this scheme, the mapping relationship between input variables and output variables is obtained through expert experience in fuzzy control fuzzy reasoning ability, which realizes the quantification of the relationship between input variables and output variables, and obtains switching information through quantified information, thereby improving the efficiency and accuracy of switching decisions.

[0015] The beneficial effects of this program: Considering that the model constructed by raw data and deep learning algorithm has a lot of uncertainty, and fuzzy control has fuzzy reasoning ability, the corresponding rules between the large amount of uncertainty generated and the actual situation are obtained through fuzzy control, so as to realize the quantification of a large amount of uncertainty, and then retain the original characteristics of the data through quantification. At the same time, by obtaining the first predicted load curve, the actual load curve has a corresponding standard, which makes it easier for fuzzy control to obtain the corresponding rules. The combination of raw data, deep learning algorithm and fuzzy control significantly improves the timeliness and accuracy of switching; The final load curve obtained by the corresponding rules retains the circuit status, which is convenient for timely detection of potential fault risks and taking corresponding preventive measures, thereby improving switching safety and circuit safety; By training the initial load forecasting model from a small period, i.e., a week, the weekly characteristics of the first original historical load data and the first original historical environmental data are found. At the same time, when the first original historical load data and the first original historical environmental data do not have weekly characteristics, the monthly characteristics and quarterly characteristics of the first original historical load data and the first original historical environmental data are found. By obtaining regular data characteristics from small to large, the accuracy and flexibility of the load forecasting model are improved. The technical problem of ignoring the original characteristics of the data when making switching predictions has been solved.

[0016] The present invention also provides a mobile energy storage emergency power supply non-inductive switching device, which is applicable to the mobile energy storage emergency power supply non-inductive switching method, comprising a plurality of mobile energy storage emergency power supplies, a model acquisition module, a rule curve acquisition module, a final load curve acquisition module and a switching information acquisition module; The model acquisition module is used to obtain a load prediction model using a deep learning algorithm based on the user's first original historical load data and the first original historical environmental data of the mobile energy storage emergency power supply; The rule curve acquisition module is used to input the second original historical load data of the user and the second original historical environmental data of the mobile energy storage emergency power supply into the load forecasting model to obtain a first predicted load curve, and based on the first predicted load curve and the actual load curve, use fuzzy control to obtain the corresponding rule between the first predicted load curve and the actual load curve; The final load curve acquisition module is used to input the user's original real-time load data into the load prediction model to obtain a second predicted load curve, and based on the corresponding rules, use the second predicted load curve to obtain the final load curve; The switching information acquisition module is used to obtain the switching information based on the final load curve using the fuzzy control, and perform switching based on the switching information.

[0017] Preferably, the rule curve acquisition module includes a first predicted load curve acquisition module and a corresponding rule acquisition module; The first predicted load curve acquisition module is used to input the user's second original historical load data and the second original historical environmental data of the mobile energy storage emergency power supply into the load prediction model to obtain a first predicted load curve; The corresponding rule acquisition module is used to obtain the corresponding rule between the first predicted load curve and the actual load curve by using fuzzy control based on the first predicted load curve and the actual load curve.

[0018] Preferably, it also includes a status monitoring module, which is used to obtain the first original historical load data, the first original historical environment data, the second original historical load data and the second original historical environment data.

[0019] Preferably, it also includes a wireless transmission module, which is used to upload the first original historical load data, the first original historical environment data, the second original historical load data and the second original historical environment data to the cloud platform.

[0020] The beneficial effects of this program: Considering that the model constructed by raw data and deep learning algorithm has a lot of uncertainty, and fuzzy control has fuzzy reasoning ability, the corresponding rules between the large amount of uncertainty generated and the actual situation are obtained through fuzzy control, so as to realize the quantification of a large amount of uncertainty, and then retain the original characteristics of the data through quantification. At the same time, by obtaining the first predicted load curve, the actual load curve has a corresponding standard, which makes it easier for fuzzy control to obtain the corresponding rules. The combination of raw data, deep learning algorithm and fuzzy control significantly improves the timeliness and accuracy of switching; The final load curve obtained by the corresponding rules retains the circuit status, which is convenient for timely detection of potential fault risks and taking corresponding preventive measures, thereby improving switching safety and circuit safety; The switching information is obtained according to the final load curve and the fuzzy reasoning ability of fuzzy control, and the switching information can be adjusted in time according to the final load curve, thereby improving the switching flexibility and at the same time improving the stability and operation efficiency of the system; By uploading various data to the cloud platform in real time, data sharing and remote monitoring become possible, facilitating the centralized management and optimized dispatch of large-scale energy storage vehicles; The technical problem of ignoring the original characteristics of the data when making switching predictions has been solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Other features, objects and advantages of the present invention will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings. The drawings are only for the purpose of illustrating preferred embodiments and are not to be considered as limiting the present invention. Also, the same reference symbols are used throughout the drawings to represent the same parts.

[0022] Figure 1 This is a flow chart of a method for non-sensing switching of a mobile energy storage emergency power supply according to the present invention; Figure 2 The present invention is a schematic structural diagram of a mobile energy storage emergency power supply non-sensing switching device. DETAILED DESCRIPTION

[0023] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation method described herein is only an optimal embodiment of the present invention, which is only used to explain the present invention and does not limit the scope of protection of the present invention. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0024] Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow charts describe the operations (or steps) as sequential processes, many of the operations (or steps) therein can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but can also have additional steps not included in the drawings; the process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0025] Embodiment 1: like Figure 1As shown, a method for non-sensing switching of a mobile energy storage emergency power supply comprises the following steps: S1: Based on the user's first original historical load data and the first original historical environmental data of the mobile energy storage emergency power supply, a load forecasting model is obtained using a deep learning algorithm.

[0026] The S1 includes: S11: constructing a division standard based on the original historical load data characteristics of the user and the original historical environment data characteristics of the mobile energy storage emergency power supply, dividing the original historical load data into the first original historical load data and the second original historical load data based on the division standard, and dividing the original historical environment data into the first original historical environment data and the second original historical environment data based on the division standard; S12: Build an initial load forecasting model based on deep learning algorithm; S13: Using a cross-validation method, the performance index of the initial load prediction model is obtained based on the first original historical load data and the first original historical environment data. If the performance index meets the preset conditions, the initial load prediction model is the load prediction model, otherwise execute S13.

[0027] The S13 includes: S131: dividing the first original historical load data and the first original historical environment data into a plurality of sets based on a first criterion in the division criteria; S132: Input several sets into the initial load prediction model in sequence to train the initial load prediction model, and obtain the performance indicators of the initial load prediction model in sequence, until the average value of the performance indicator is greater than or equal to the preset value, then stop inputting the sets. At this time, the performance indicator meets the preset conditions. When several sets are input and the average value of the performance indicator is still less than the preset value, the second standard in the division standard is used as the first standard, and S131 is executed.

[0028] In this embodiment, the original historical environmental data is the ambient temperature and the ambient humidity. The load prediction model obtained by the original historical load data, the ambient temperature and the humidity not only considers the data characteristics of the historical load data, but also considers the environment generated by the historical load data, thereby improving the adaptability and accuracy of the load prediction model. At the same time, the load prediction model can timely discover the problems existing in the circuit, thereby improving the switching safety. The original historical load data characteristics and the original historical environmental data characteristics are both long-term data characteristics, i.e., four-season characteristics, and short-term data characteristics, i.e., monthly characteristics or weekly characteristics. The division standard is constructed by the data characteristics. The division standard is that each group of data obtained by the division has long-term data characteristics and short-term data characteristics, which ensures the consistency of the characteristics of the first original historical load data and the second original historical load data, and ensures the consistency of the characteristics of the first original historical environmental data and the second original historical load environment, thereby making the load prediction model applicable to the second original historical load data and the second original historical environmental data, while covering the long-term and short-term data characteristics, so that the load prediction model can be applicable to the original real-time load data and the original real-time environmental data, thereby significantly improving the applicability of the load prediction model.

[0029] In this embodiment, the initial load forecasting model framework constructed by the deep learning algorithm can effectively capture the temporal dependency of the time series and improve the prediction accuracy. The first standard is a week, and the second standard is a month or a quarter. The initial load forecasting model is first trained from a small period, that is, a week, to find the weekly characteristics of the first original historical load data and the first original historical environmental data. At the same time, when the first original historical load data and the first original historical environmental data do not have weekly characteristics, the monthly characteristics of the first original historical load data and the first original historical environmental data are found. When the first original historical load data and the first original historical environmental data do not have monthly characteristics, the quarterly characteristics of the first original historical load data and the first original historical environmental data are found. By obtaining regular data characteristics from small to large, the accuracy and flexibility of the load forecasting model are improved. The performance index can be the root mean square error and the mean absolute percentage error. When the average value of the performance index meets the condition, the input of the set is stopped immediately. At this time, the initial load forecasting model obtained is the load forecasting model, which further improves the accuracy of the load forecasting model and also improves the efficiency of obtaining the load forecasting model.

[0030] S2: Input the user's second original historical load data and the second original historical environmental data of the mobile energy storage emergency power supply into the load forecasting model to obtain a first predicted load curve. Based on the first predicted load curve and the actual load curve, fuzzy control is used to obtain a corresponding rule between the first predicted load curve and the actual load curve.

[0031] In S2, the corresponding rule between the first predicted load curve and the actual load curve obtained by using fuzzy control based on the first predicted load curve and the actual load curve includes: Compare the first predicted load curves of the corresponding time periods in different cycles, and compare the actual load curves of the corresponding time periods in different cycles. If the comparisons are successful, bind the first predicted load curves and the actual load curves of the corresponding time periods in different cycles to obtain a binding set; If the comparison is unsuccessful, the maximum actual load curve among the actual load curves of the corresponding time periods in different cycles is bound to the corresponding time periods in different cycles, and then the binding set is updated, and the binding set is the corresponding rule.

[0032] In this embodiment, the cycles are four seasons, months and weeks, four seasons are a large cycle, one month is a medium cycle, and one week is a small cycle. The comparison is performed in the order of small cycle-medium cycle-large cycle. At the same time, it must be satisfied that when the comparison of the first predicted load curves of the corresponding time periods in different small cycles is unsuccessful or the comparison of the actual load curves of the corresponding time periods in different small cycles is unsuccessful, the first predicted load curves of the corresponding time periods in different medium cycles are compared, and the actual load curves of the corresponding time periods in different medium cycles are compared. If the comparison of the first predicted load curves of the corresponding time periods in different medium cycles is unsuccessful or the comparison of the actual load curves of the corresponding time periods in different medium cycles is unsuccessful, the first predicted load curves of the corresponding time periods in different large cycles are compared, and the actual load curves of the corresponding time periods in different large cycles are compared. When the comparison of the small cycle-medium cycle-large cycle is unsuccessful, the maximum actual load curve in the actual load curve of the corresponding time period in the small cycle is bound to the corresponding time period in the small cycle. Through multi-level comparison, the relationship between the first predicted load curve and the actual load curve can be fully discovered, thereby improving the accuracy of the corresponding rules. At the same time, from the setting of small cycle-medium cycle-large cycle, the partial data is compared first. When no relationship is found in the partial data, the overall data is compared. When the relationship is found between the partial data, the relationship is bound, thereby saving computing resources. At the same time, when the comparison is unsuccessful, the maximum actual load curve is bound to the corresponding time period to improve the switching safety. Through the fuzzy reasoning ability of fuzzy control, the relationship between the first predicted load curve and the actual load curve is quantified, that is, a binding set is obtained, so that the original characteristics of the data are retained, thereby improving the switching adaptability and accuracy. In this embodiment, the conditions for whether the comparison is successful are flexibly set according to user needs.

[0033] S3: Inputting the user's original real-time load data into the load forecasting model to obtain a second forecast load curve, and based on the corresponding rules, using the second forecast load curve to obtain a final load curve.

[0034] In S3, the step of obtaining the final load curve using the second predicted load curve based on the corresponding rule includes: The final load curve is obtained according to the matching result of the acquisition time period of the second predicted load curve and the corresponding time period in the corresponding rule.

[0035] In this embodiment, the final load curve is obtained according to the matching result of the acquisition time period of the second predicted load curve and the corresponding time period in the corresponding rule: the acquisition time period of the second predicted load curve is matched with the corresponding time period in the corresponding rule, and the curve corresponding to the matching result is the final load curve. By directly matching with the corresponding rule to obtain the final load curve, the switching efficiency and convenience are improved.

[0036] S4: Based on the final load curve, the fuzzy control is used to obtain switching information, and switching is performed based on the switching information.

[0037] The S4 includes: S41: using the original historical load data and the historical capacity of the mobile energy storage emergency power supply as input variables, and using the switching information as an output variable; S42: Acquire a mapping relationship between input variables and output variables based on expert experience in the fuzzy control; S43: Obtain switching information based on the mapping relationship, the final load curve and the actual capacity of the mobile energy storage emergency power supply.

[0038] In this embodiment, the switching information is the switching capacity and the switching time. The mapping relationship between the input variables and the output variables obtained based on the expert experience in the fuzzy control can be: obtain the specific time when the load peak in the original historical load data exceeds 80% of the capacity of the mobile energy storage emergency power supply, and when it is the specific time, switch the energy storage operation mode. At this time, the mapping relationship is: when the load peak in the final load curve exceeds 80% of the capacity of the mobile energy storage emergency power supply, switch to another mobile energy storage emergency power supply. Obtaining the mapping relationship through fuzzy control can effectively deal with the uncertainty of load data and energy storage power supply capacity, realize the quantification of the relationship between input variables and output variables, obtain switching information through quantified information, and improve the efficiency and accuracy of switching decisions.

[0039] Embodiment 2: This embodiment also provides a mobile energy storage emergency power supply non-inductive switching device, which is applicable to the mobile energy storage emergency power supply non-inductive switching method, including a plurality of mobile energy storage emergency power supplies, and also including a model acquisition module, a rule curve acquisition module, a final load curve acquisition module and a switching information acquisition module; The model acquisition module is used to obtain a load prediction model using a deep learning algorithm based on the user's first original historical load data and the first original historical environmental data of the mobile energy storage emergency power supply; The rule curve acquisition module is used to input the second original historical load data of the user and the second original historical environmental data of the mobile energy storage emergency power supply into the load forecasting model to obtain a first predicted load curve, and based on the first predicted load curve and the actual load curve, use fuzzy control to obtain the corresponding rule between the first predicted load curve and the actual load curve; The final load curve acquisition module is used to input the user's original real-time load data into the load prediction model to obtain a second predicted load curve, and based on the corresponding rules, use the second predicted load curve to obtain the final load curve; The switching information acquisition module is used to obtain the switching information based on the final load curve using the fuzzy control, and perform switching based on the switching information.

[0040] The rule curve acquisition module includes a first predicted load curve acquisition module and a corresponding rule acquisition module; The first predicted load curve acquisition module is used to input the user's second original historical load data and the second original historical environmental data of the mobile energy storage emergency power supply into the load prediction model to obtain a first predicted load curve; The corresponding rule acquisition module is used to obtain the corresponding rule between the first predicted load curve and the actual load curve by using fuzzy control based on the first predicted load curve and the actual load curve.

[0041] It also includes a status monitoring module, which is used to obtain the first original historical load data, the first original historical environment data, the second original historical load data and the second original historical environment data.

[0042] It also includes a wireless transmission module, which is used to upload the first original historical load data, the first original historical environment data, the second original historical load data and the second original historical environment data to the cloud platform.

[0043] In this embodiment, if Figure 2As shown, the mobile energy storage emergency power supply non-sensing switching device has two sets of power inputs, namely BT1 and BT2, which are connected to the two incoming line switches of the auxiliary device through the inverter device, namely PCS. W1, W2, W3 and W4 on both sides of the inverter device represent control loops, and the other side is connected to the load through the output switch. When a group of batteries is about to be exhausted, the program issues a switching command to close the second incoming line switch and disconnect the first line. The transmitted power can be calculated by the electrical quantity monitoring module in the control unit, and the information can be transmitted to the intelligent switching decision module in the control unit; the intelligent switching decision module issues a switching command to close the incoming line switch connected to the second mobile energy storage vehicle and disconnect the first line. At this time, the first mobile energy storage vehicle can be exited to achieve non-sensing switching of the input power supply. The mobile energy storage emergency power supply non-sensing switching device is provided with a dual-channel input interface, and has the function of non-sensing switching between two energy storage vehicles. A series of state monitoring modules are installed. In addition to obtaining the first original historical load data, the first original historical environmental data, the second original historical load data and the second original historical environmental data, it can also monitor parameters such as temperature and humidity, current, voltage, calculate the power consumption on the user side and upload it to the system platform. At the same time, an intelligent switching decision module is installed. The intelligent switching decision module includes a model acquisition module, a rule curve acquisition module, a final load curve acquisition module and a switching information acquisition module. The system integrates a deep learning algorithm and a fuzzy control logic, and can give switching information, i.e., switching instructions, according to the final load curve and mapping relationship. Compared with the traditional power switching device, the present invention adopts an intelligent prediction algorithm based on deep learning, which can predict the power exhaustion time of the energy storage vehicle according to historical data and real-time monitoring data, so as to make a switching decision in advance. The predictive switching mechanism significantly improves the timeliness and accuracy of switching, and reduces the risk of sudden power outages caused by power exhaustion; at the same time, the switching logic of traditional power switching devices is fixed, and the switching logic of the present invention can be adaptively adjusted according to the actual load and power consumption characteristics on the user side to ensure the optimal switching effect in different power consumption scenarios, thereby improving the flexibility and adaptability of the system; in addition, the wireless transmission module can upload various monitoring data to the cloud platform in real time, making data sharing and remote monitoring possible, and facilitating the centralized management and optimized scheduling of large-scale energy storage vehicles.

[0044] The specific implementation described above is a preferred implementation of a mobile energy storage emergency power supply non-sensing switching method and device of the present invention, and is not intended to limit the specific implementation scope of the present invention. The scope of the present invention includes but is not limited to this specific implementation. All equivalent changes made in accordance with the shape and structure of the present invention are within the protection scope of the present invention.

Claims

1. A method for non-inductive switching of a mobile energy storage emergency power supply, characterized in that: The following steps are involved: S1: Based on the user's first original historical load data and the first original historical environmental data of the mobile energy storage emergency power supply, a load forecasting model is obtained using a deep learning algorithm; S2: Input the second original historical load data of the user and the second original historical environmental data of the mobile energy storage emergency power supply into the load forecasting model to obtain a first forecast load curve, and based on the first forecast load curve and the actual load curve, use fuzzy control to obtain a correspondence rule between the first forecast load curve and the actual load curve; S3: inputting the user's original real-time load data into the load forecasting model to obtain a second forecast load curve, and based on the corresponding rule, using the second forecast load curve to obtain a final load curve; S4: Based on the final load curve, the fuzzy control is used to obtain switching information, and switching is performed based on the switching information.

2. A method for non-sensing switching of a mobile energy storage emergency power supply according to claim 1, characterized in that: The S1 includes: S11: constructing a division standard based on the original historical load data characteristics of the user and the original historical environment data characteristics of the mobile energy storage emergency power supply, dividing the original historical load data into the first original historical load data and the second original historical load data based on the division standard, and dividing the original historical environment data into the first original historical environment data and the second original historical environment data based on the division standard; S12: Build an initial load forecasting model based on deep learning algorithm; S13: Using a cross-validation method, the performance index of the initial load prediction model is obtained based on the first original historical load data and the first original historical environment data. If the performance index meets the preset conditions, the initial load prediction model is the load prediction model, otherwise execute S13.

3. A method for non-inductive switching of a mobile energy storage emergency power supply according to claim 2, characterized in that: The S13 includes: S131: dividing the first original historical load data and the first original historical environment data into a plurality of sets based on a first criterion in the division criteria; S132: Input several sets into the initial load prediction model in sequence to train the initial load prediction model, and obtain the performance indicators of the initial load prediction model in sequence, until the average value of the performance indicator is greater than or equal to the preset value, then stop inputting the sets. At this time, the performance indicator meets the preset conditions. When several sets are input and the average value of the performance indicator is still less than the preset value, the second standard in the division standard is used as the first standard, and S131 is executed.

4. A method for non-sensing switching of a mobile energy storage emergency power supply according to claim 1, characterized in that: In S2, the corresponding rule between the first predicted load curve and the actual load curve obtained by using fuzzy control based on the first predicted load curve and the actual load curve includes: Compare the first predicted load curves of the corresponding time periods in different cycles, and compare the actual load curves of the corresponding time periods in different cycles. If the comparisons are successful, bind the first predicted load curves and the actual load curves of the corresponding time periods in different cycles to obtain a binding set; If the comparison is unsuccessful, the maximum actual load curve among the actual load curves of the corresponding time periods in different cycles is bound to the corresponding time periods in different cycles, and then the binding set is updated, and the binding set is the corresponding rule.

5. A method for non-inductive switching of a mobile energy storage emergency power supply according to claim 4, characterized in that: In S3, the step of obtaining the final load curve using the second predicted load curve based on the corresponding rule includes: The final load curve is obtained according to the matching result of the acquisition time period of the second predicted load curve and the corresponding time period in the corresponding rule.

6. A method for non-sensing switching of a mobile energy storage emergency power supply according to claim 2, characterized in that: The S4 includes: S41: using the original historical load data and the historical capacity of the mobile energy storage emergency power supply as input variables, and using the switching information as an output variable; S42: Acquire a mapping relationship between input variables and output variables based on expert experience in the fuzzy control; S43: Obtain switching information based on the mapping relationship, the final load curve and the actual capacity of the mobile energy storage emergency power supply.

7. A mobile energy storage emergency power supply non-inductive switching device, applicable to a mobile energy storage emergency power supply non-inductive switching method according to any one of claims 1 to 6, comprising a plurality of mobile energy storage emergency power supplies, characterized in that: It also includes a model acquisition module, a rule curve acquisition module, a final load curve acquisition module and a switching information acquisition module; The model acquisition module is used to obtain a load prediction model using a deep learning algorithm based on the user's first original historical load data and the first original historical environmental data of the mobile energy storage emergency power supply; The rule curve acquisition module is used to input the second original historical load data of the user and the second original historical environmental data of the mobile energy storage emergency power supply into the load forecasting model to obtain a first predicted load curve, and based on the first predicted load curve and the actual load curve, use fuzzy control to obtain the corresponding rule between the first predicted load curve and the actual load curve; The final load curve acquisition module is used to input the user's original real-time load data into the load prediction model to obtain a second predicted load curve, and based on the corresponding rules, use the second predicted load curve to obtain the final load curve; The switching information acquisition module is used to obtain the switching information based on the final load curve using the fuzzy control, and perform switching based on the switching information.

8. A mobile energy storage emergency power supply non-sensing switching device according to claim 7, characterized in that: The rule curve acquisition module includes a first predicted load curve acquisition module and a corresponding rule acquisition module; The first predicted load curve acquisition module is used to input the user's second original historical load data and the second original historical environmental data of the mobile energy storage emergency power supply into the load prediction model to obtain a first predicted load curve; The corresponding rule acquisition module is used to obtain the corresponding rule between the first predicted load curve and the actual load curve by using fuzzy control based on the first predicted load curve and the actual load curve.

9. A mobile energy storage emergency power supply non-sensing switching device according to claim 7, characterized in that: It also includes a status monitoring module, which is used to obtain the first original historical load data, the first original historical environment data, the second original historical load data and the second original historical environment data.

10. A mobile energy storage emergency power supply non-sensing switching device according to claim 7, characterized in that: It also includes a wireless transmission module, which is used to upload the first original historical load data, the first original historical environment data, the second original historical load data and the second original historical environment data to the cloud platform.

Images