An electric pump operation mode optimization method and system based on data analysis

Abstract

This invention discloses a data analysis-based method and system for optimizing the operation mode of electric pumps, specifically relating to the field of electric pump monitoring and control and operation data processing. The method includes reading monitoring and control acquisition records, extracting the main pump load-bearing segment of the previous control cycle and the dual-pump access segment of the next control cycle, using the standby pump start-up position as the boundary, calculating the single-pump power consumption and residual power consumption value from the main pump power and the total flow rate of the pump group, and subtracting the residual power consumption value from the single-pump power consumption value to obtain the unloading amount. This invention calculates the main pump unloading amount and standby pump replacement deviation by extracting the operation segments before and after the standby pump start-up, and generates the standby pump access sequence and main pump exit control actions through a three-state replacement diagram, confidence transfer, and posterior inference. This solves the problem that it is impossible to determine whether the standby pump truly shares the main pump load after startup, leading to invalid access still participating in subsequent rotation and parallel operation.

Description

Technical Field

[0001] This invention relates to the field of electric pump monitoring and control and operation data processing technology, and more specifically, to a method and system for optimizing electric pump operation modes based on data analysis. Background Technology

[0002] In the operation and management of electric pumping stations, the existing solutions are mainly aimed at the automatic maintenance of liquid level, pressure or flow targets. Typically, the monitoring and control and data acquisition system reads the start-stop status, current, power, outlet pressure, total flow and liquid level changes of the main pump and standby pump, and starts the standby pump according to the upper limit of liquid level, puts the parallel pump into operation when the pressure is insufficient, and performs the main and standby rotation according to the running time. Taking the peak discharge period of sewage lifting pump stations or process circulation pump stations as an example, the site requires the liquid level to drop or the pressure to recover in a short time. The control system can only make the replacement judgment based on the connected electrical quantities, hydraulic quantities, and start-up and shutdown status. It cannot shut down for a long time to check the status of pipelines, valves, or pumps. Under these conditions, the successful start of the standby pump does not mean that the effective replacement has been completed. On-site, there will be repeated phenomena where the standby pump has been put into operation but the main pump current and power have not decreased, the total flow rate has not increased, the liquid level drop rate has not increased, and even the outlet pressure fluctuation has increased. The reason is that the existing control logic only confirms "whether the standby pump has started" and "whether the overall target has been met", without using the data collected before and after the connection to determine whether the standby pump has actually shared the load of the main pump. As a result, the standby pump that has been ineffectively connected is still included in the subsequent rotation and parallel operation mode. Therefore, the technical problem to be solved by this application is: how to determine the effectiveness of load sharing after the standby pump is put into operation based on the operation data before and after the access obtained from supervision and control and data acquisition, and optimize the operation mode of the electric pump accordingly. Summary of the Invention

[0003] To overcome the aforementioned deficiencies in the prior art, embodiments of the present invention provide a data analysis-based method and system for optimizing the operation mode of electric pumps. By extracting operation segments before and after the start-up of the standby pump, calculating the unloading amount of the main pump and the standby pump replacement deviation, and generating the standby pump access sequence and main pump exit control actions through a three-state replacement diagram, confidence transfer, and posterior inference, the problems mentioned in the background art are solved.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for optimizing the operation mode of an electric pump based on data analysis, comprising: S1. Read the monitoring and control acquisition records, and take the standby pump start position as the boundary to extract the main pump bearing plate of the previous control cycle and the dual pump access plate of the next control cycle. Calculate the power consumption value of a single pump and the residual power consumption value from the power of the main pump and the total flow of the pump group. Subtract the residual power consumption value from the power consumption value of a single pump to obtain the unloading amount. S2. Using the standby pump power, total flow rate increment of the pump group and liquid level drop rate increment in the dual-pump access plate as input, calculate the pipe connection power consumption value and the discharge power consumption value, and calculate the replacement deviation from the pipe connection power consumption value, the discharge power consumption value and the unloading amount. S3. Generate a replacement code based on the unloading amount and replacement deviation, and read the replacement position of the standby pump start position from the replacement code. The replacement position includes the virtual replacement position, the real replacement position and the drag replacement position to obtain the replacement status. S4. Construct a three-state replacement diagram based on the replacement state, unloading amount, pipe connection power consumption value, drain power consumption value and replacement deviation. Generate region bits from the Kikuchi pipe connection region and the Kikuchi drain region, and obtain the real replacement confidence value, virtual replacement confidence value and drag replacement confidence value through TRW-S message transmission. S5. Construct a posterior row table using real replacement confidence, virtual replacement confidence, and drag replacement confidence. Determine the posterior selected row using a posterior inference algorithm, generate mode actions based on the posterior selected row, and output the optimized result of the electric pump operation mode.

[0005] In a preferred embodiment, S1 includes: S1-1. Define the number that increments by one bit in each data frame acquisition completed in the monitoring and control acquisition record as the sampling sequence number, define the sampling sequence number interval between two adjacent pump group control command issuance positions as the control cycle, and determine the control cycle before the standby pump start position as the main pump's operating segment. S1-2. After the standby pump starts, the control cycle is determined to be the dual-pump access plate. The power of the main pump and the total flow of the pump group in the main pump support plate and the dual-pump access plate are summed according to the sampling sequence number to obtain the sum of single pump power, single pump flow, residual power and residual flow. S1-3. The power consumption of a single pump is obtained by dividing the power of a single pump by the flow rate of a single pump. The residual power consumption is obtained by dividing the residual power by the flow rate of a single pump. The unloading amount is obtained by subtracting the residual power consumption from the power consumption of a single pump.

[0006] In a preferred embodiment, S2 includes: S2-1. Subtract the sampling points of the dual pump connection plate and the main pump bearing plate according to the sampling sequence number to obtain the flow difference of each sampling point. Set the negative values ​​of the flow difference to zero and calculate the average value to obtain the total flow increment of the pump group. S2-2. Subtract the liquid level at the tail of the plate from the liquid level at the beginning of the plate and the liquid level at the end of the plate from the liquid level at the beginning of the plate and the liquid level at the end of the plate respectively, and divide by the plate length to obtain the liquid level reduction rate of the single pump and the liquid level reduction rate of the dual pump. Subtract the liquid level reduction rate of the single pump from the liquid level reduction rate of the dual pump and set the negative value to zero to obtain the liquid level reduction rate increment. S2-3. The average power of the standby pump connected to the dual pumps is obtained by averaging the power of the standby pump. The power consumption of the connecting pipe is obtained by dividing the average power of the standby pump by the total flow rate increment of the pump group. The power consumption of the draining pump is obtained by dividing the average power of the standby pump by the liquid level drop rate increment. The unloading amount is subtracted from the higher of the power consumption of the connecting pipe and the draining pump to obtain the replacement deviation. When the total flow rate increment of the pump group is zero, the power consumption of the connecting pipe is taken as the average power of the standby pump. When the liquid level drop rate increment is zero, the power consumption of the draining pump is taken as the average power of the standby pump.

[0007] In a preferred embodiment, S3 includes: S3-1. Using unloading amount and replacement deviation as input, convert unloading amount greater than zero to unloading position value 1 and unloading amount not greater than zero to unloading position value 0. Convert replacement deviation greater than zero to deviation position value 1 and replacement deviation not greater than zero to deviation position value 0. Output unloading position value and deviation position value. S3-2. Multiply the unloading position value by 2 and add the deviation position value to obtain the replacement code. Read the code table row with the same replacement code from the replacement code table and use the replacement position in the code table row as the replacement position of the standby pump start position. The replacement code table includes the row record of replacement code 1 and virtual replacement position, the row record of replacement code 2 and real replacement position, and the row record of replacement code 3 and drag replacement position.

[0008] In a preferred embodiment, S3 further includes: S3-3. Subtract the absolute value of the replacement deviation from the absolute value of the unloading amount to obtain the replacement amplitude difference, and combine the replacement amplitude difference with the replacement position of the standby pump start position to obtain the amplitude replacement position. S3-4. Generate the replacement state based on the amplitude replacement bit, where the amplitude replacement bit of replacement code 1 indicates that the standby pump consumption has not formed the main pump unloading, the amplitude replacement bit of replacement code 2 indicates that the standby pump is connected to form the main pump unloading, and the amplitude replacement bit of replacement code 3 indicates that the standby pump is connected to form the main pump unloading and still retains the standby pump consumption.

[0009] In a preferred embodiment, S4 includes: S4-1. Taking the replacement state, unloading amount, pipe connection power consumption value, drain power consumption value and replacement deviation as input, take unloading target position 1 and deviation target position 0 as the real replacement row, take unloading target position 0 and deviation target position 1 as the virtual replacement row, take unloading target position 1 and deviation target position 1 as the drag replacement row, and form a three-state replacement diagram by real replacement row, virtual replacement row, drag replacement row, unloading amount column, pipe connection power consumption value column, drain power consumption value column and replacement deviation column. S4-2. Convert the unloading amount greater than zero to unloading input bit 1, and the unloading amount not greater than zero to unloading input bit 0. Convert the replacement deviation greater than zero to deviation input bit 1, and the replacement deviation not greater than zero to deviation input bit 0. Perform XOR operation between unloading input bit and unloading target bit and XOR operation between deviation input bit and deviation target bit on each row of the three-state replacement diagram to obtain unloading difference bit and deviation difference bit.

[0010] In a preferred embodiment, S4 further includes: S4-3. In the three-state succession diagram, the Kikuchi pipe-connecting region is composed of unloading difference, deviation difference, pipe power consumption value, and unloading amount; the Kikuchi draining region is composed of unloading difference, deviation difference, draining power consumption value, and unloading amount. When both the unloading difference and deviation difference in the Kikuchi pipe-connecting region are 0 and the pipe power consumption value is not greater than the unloading amount, the pipe-connecting region is set to 1; otherwise, the pipe-connecting region is set to 0. When both the unloading difference and deviation difference in the Kikuchi draining region are 0 and the draining power consumption value is not greater than the unloading amount, the draining region is set to 1; otherwise, the draining region is set to 0. S4-4. Perform TRW-S message passing row by row on the three-state succession diagram. Multiply the takeover region bit by half to obtain the takeover message value, multiply the drain region bit by half to obtain the drain message value, add the takeover message value and the drain message value to obtain the row message value, and normalize the sum of the row message values ​​of the real succession row, the virtual succession row, and the drag-over succession row to obtain the real succession confidence value, the virtual succession confidence value, and the drag-over succession confidence value.

[0011] In a preferred embodiment, S5 includes: S5-1. Using the real replacement confidence value, the virtual replacement confidence value, and the dragged replacement confidence value as inputs, establish a posterior table. The posterior table includes the real replacement row, the virtual replacement row, and the dragged replacement row. Fill the three confidence values ​​into the confidence value columns of the real replacement row, the virtual replacement row, and the dragged replacement row, respectively. S5-2. The posterior inference algorithm is used to perform pairwise difference on each row of the posterior row table. The confidence value of the current row is subtracted from the confidence values ​​of the other two rows respectively. If the difference is greater than zero, the winning position is 1; if the difference is not greater than zero, the winning position is 0. The winning number of the current row is obtained by adding the two winning positions.

[0012] In a preferred embodiment, S5 further includes: S5-3. Determine the successive row with a winning number of 2 as the successive selected row. If the winning numbers of all three rows are not 2, read the first row with the same high value as the successive selected row in the order of real replacement row, drag replacement row, and virtual replacement row, and generate the pattern action based on the successive selected row. S5-4. When the selected operation is a real replacement, the backup pump connection order is moved forward by one position. When the selected operation is a virtual replacement, the backup pump connection order is moved backward by one position. When the selected operation is a towed replacement, the backup pump connection is retained and the main pump exit position is set to 0. The optimized result of the electric pump operation mode is output.

[0013] In a preferred embodiment, a data analysis-based electric pump operation mode optimization system includes: The segmentation module is used to read the monitoring and control acquisition records, and to extract the main pump bearing segment of the previous control cycle and the dual pump access segment of the next control cycle with the standby pump start position as the boundary. The power consumption value of a single pump and the residual power consumption value are calculated from the power of the main pump and the total flow of the pump group. The unloading amount is obtained by subtracting the residual power consumption value from the power consumption value of a single pump. The replacement quantization module calculates the connection power consumption value and the discharge power consumption value by taking the backup pump power, the total flow rate increment of the pump group and the liquid level reduction speed increment in the dual-pump access chip as inputs, and calculates the replacement deviation from the connection power consumption value, the discharge power consumption value and the unloading amount. The replacement code module generates a replacement code based on the unloading amount and replacement deviation, and reads the replacement position of the standby pump start position from the replacement code. The replacement position includes a virtual replacement position, a real replacement position, and a drag replacement position to obtain the replacement status. The confidence transmission module constructs a three-state succession diagram based on the succession status, unloading amount, pipe connection power consumption value, drain power consumption value, and succession deviation. It generates region bits from the Kikuchi pipe connection region and the Kikuchi drain region, and obtains the real succession confidence value, virtual succession confidence value, and drag succession confidence value through TRW-S message transmission. The mode write-back module constructs a posterior row table using real replacement confidence, virtual replacement confidence, and drag-and-drop replacement confidence. It then determines the posterior selected row using a posterior inference algorithm, generates mode actions based on the posterior selected row, and outputs the optimized results of the electric pump operation mode.

[0014] The technical effects and advantages of this invention are as follows: 1. By calculating the unloading amount through the main pump bearing plate and dual pump access plate before and after the standby pump start position, the unit delivery power consumption of the standby pump is converted into data results, which can relatively reduce the risk of invalid access caused by judging the success of the switchover based solely on successful start-up or the achievement of the overall target. 2. By calculating the replacement deviation through the total flow rate increment of the pump group, the liquid level speed reduction increment, and the standby pump power, the standby pump consumption is offset against the main pump unloading. This can identify the virtual replacement state of "multiple pumps running but no effective delivery is achieved", making the operation mode adjustment basis closer to the load sharing result. 3. By converting the unloading amount and replacement deviation into virtual replacement position, real replacement position and drag replacement position through the replacement code table, the standby pump access result is refined from a single start-stop state to three types of replacement states, which can relatively improve the problem of indistinguishable replacement states in the main standby rotation. 4. By constraining the power consumption of the control and drainage processes respectively through the three-state succession diagram, the Kikuchi control region, and the Kikuchi drainage region, and then generating three types of confidence quantities through TRW-S message passing, the stability of the succession determination can be relatively improved with the joint participation of the control and drainage sides. 5. Perform pairwise differences on the actual replacement confidence, virtual replacement confidence, and drag replacement confidence using the post-verification table, and generate mode actions accordingly, so that the standby pump access priority adjustment and the main pump exit restriction are directly driven by the replacement judgment result. Attached Figure Description

[0015] Figure 1 This is a flowchart of the method steps of the present invention.

[0016] Figure 2 This is a schematic diagram of the system modules of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Refer to the instruction manual appendix Figure 1-2 The present invention provides a data analysis-based method for optimizing the operation mode of an electric pump, comprising: S1. Read the monitoring and control acquisition records, and take the standby pump start position as the boundary to extract the main pump bearing plate of the previous control cycle and the dual pump access plate of the next control cycle. Calculate the power consumption value of a single pump and the residual power consumption value from the power of the main pump and the total flow of the pump group. Subtract the residual power consumption value from the power consumption value of a single pump to obtain the unloading amount. In this embodiment, the monitoring and control data acquisition records save the pump group operation process in data frames. Each data frame includes at least the sampling sequence number, pump group control command, main pump power, standby pump start / stop status, total pump group flow rate, and liquid level. The standby pump start position is used to cut off the operation segment before and after the standby pump is put into operation. The main pump undertaking segment is used to indicate the operation segment in which the main pump undertakes the transportation task before the standby pump is put into operation. The dual pump access segment is used to indicate the operation segment in which the main pump still participates in the operation after the standby pump is put into operation. By comparing the data of adjacent control cycles before and after the same standby pump start position, it is calculated whether the main pump has a unit transportation power consumption decrease after the standby pump is connected, thereby obtaining the unloading amount. The implementation process includes the following steps: In S1-1, each time the supervisory control data acquisition record completes a data frame acquisition, a numbered data frame is generated. The numbering starts from the initial sampling position and increments by one bit to form a sampling sequence number. The sampling sequence number where the pump group control command field changes is recorded as the pump group control command issuance position. The sampling sequence number interval between two adjacent pump group control command issuance positions is a control cycle. The first sampling sequence number when the standby pump's start / stop status changes from the stop state to the running state is recorded as the standby pump start position. When multiple standby pumps start at the same sampling sequence number, standby pump start positions are formed according to the standby pump number and calculated separately. The sampling sequence numbers in the control cycle before the standby pump start position are arranged in chronological order to form the main pump support plate. If the standby pump start position falls within a control cycle, the control cycle is cut off by the standby pump start position. The sampling sequence number interval before the start position is merged into the main pump support plate, and the sampling sequence number interval after the start position is reserved for subsequent dual-pump access plate values. In S1-2, the sampling sequence numbers within one control cycle after the standby pump start position are arranged in chronological order to form a dual-pump access segment, and the sampling order within the segment is represented by the segment sequence number. The common segment length of the main pump support segment and the dual-pump access segment is the number of segment sequence numbers that both have. Sampling bits exceeding the common segment length are not included in the summation in this step. The main pump power is read and summed for each segment sequence number involved in the calculation in the main pump support segment to obtain the single pump power sum. The total pump flow rate is read and summed for each segment sequence number involved in the calculation in the same segment to obtain the single pump flow rate sum. The main pump power is read and summed for each segment sequence number involved in the calculation in the dual-pump access segment to obtain the residual power sum. The total pump flow rate is read and summed for each segment sequence number involved in the calculation in the same segment to obtain the residual flow rate sum. Each summation object is based on the data frame attached to the sampling sequence number, and the main pump power and the total pump flow rate are the recorded values ​​under the same sampling sequence number. In S1-3, when both the sum of single pump flow rates and the sum of residual flow rates are greater than zero, the single pump power consumption value is obtained by dividing the sum of single pump power by the sum of single pump flow rates, and the residual power consumption value is obtained by dividing the sum of residual power by the sum of residual flow rates. The unloading amount is obtained by subtracting the residual power consumption value from the single pump power consumption value. When the unloading amount is positive, it indicates that the unit delivery power consumption of the main pump decreases after the standby pump is connected. When the unloading amount is not greater than zero, it indicates that the unit delivery power consumption of the main pump does not decrease after the standby pump is connected. When the sum of single pump flow rates is zero or the sum of residual flow rates is zero, the standby pump start position is not included in the unloading amount calculation, and a no-delivery start position is generated. The standby pump connection sequence and the main pump exit position remain unchanged. The no-delivery start position is not included in the subsequent calculation of replacement deviation, replacement status, and three-state replacement diagram. Through the above process, the operating data before and after the backup pump is connected are limited to adjacent segments of the same backup pump start position. The main pump support segment provides the reference before connection, and the dual-pump connection segment provides the observation after connection. The unloading amount directly reflects whether the unit delivery power consumption of the main pump is stripped away by the backup pump connection. Subsequent steps continue to calculate the replacement deviation and replacement status based on this. In practical application: the sewage lifting pump station issues the backup pump connection command at sampling number 8, and sampling number 9 detects that the backup pump start / stop status changes from 0 to 1. Then sampling number 9 is used as the backup pump start position. The control cycle in which sampling numbers 7 and 8 are located forms the main pump support segment, and the next control cycle after number 9 forms the dual-pump connection segment. The single pump power consumption value and the residual power consumption value are calculated and subtracted to obtain the unloading amount of the main pump on the backup pump connection.

[0019] S2. Using the standby pump power, total flow rate increment of the pump group and liquid level drop rate increment in the dual-pump access plate as input, calculate the pipe connection power consumption value and the discharge power consumption value, and calculate the replacement deviation from the pipe connection power consumption value, the discharge power consumption value and the unloading amount. In this embodiment, the replacement deviation is used to measure whether the consumption generated after the standby pump is connected is offset by the unloading of the main pump; the main pump bearing plate is used as the reference before the standby pump is connected, and the dual pump connection plate is used as the observation after the standby pump is connected. The two are calculated in the same position according to the sequence within the plate. Only the increased delivery volume and discharge volume relative to the main pump bearing plate after the standby pump is connected are retained to avoid the decrease value after connection offsetting the new contribution of other sampling positions; before the liquid level reduction rate is included in the calculation of the discharge power consumption value, it is converted into the liquid level equivalent flow increment according to the effective cross-sectional area of ​​the water collection well, so that the connection power consumption value, discharge power consumption value and unloading volume are in the power / flow scale; The implementation process includes the following steps: In S2-1, the sampling positions in the main pump support plate and the dual-pump access plate that participate in the calculation are rearranged into intra-plate positions according to the sampling sequence number, and the common plate length of the two is taken as the subtraction range of the same position; for the j-th intra-plate position, the total flow rate of the pump group in the dual-pump access plate is read and the total flow rate of the pump group in the main pump support plate is subtracted to obtain the j-th flow rate difference; when the flow rate difference is greater than zero, the original value is retained, and when the flow rate difference is not greater than zero, it is taken as 0; the flow rate difference after processing all intra-plate positions is summed and divided by the common plate length to obtain the total flow rate increment of the pump group; this processing means that only the new flow rate formed after the standby pump is connected is counted, and the sampling positions where the flow rate drops after the dual pump is connected are not included in the new flow rate accumulation. In S2-2, the liquid levels at the beginning and end of the main pump bearing plate are read. The liquid level at the beginning is subtracted from the liquid level at the end, and then divided by the length of the main pump bearing plate to obtain the single pump liquid level reduction rate. The liquid levels at the beginning and end of the dual pump connection plate are read. The liquid level at the beginning is subtracted from the liquid level at the end, and then divided by the length of the dual pump connection plate to obtain the dual pump liquid level reduction rate. The liquid level reduction rate difference is obtained by subtracting the single pump liquid level reduction rate from the dual pump liquid level reduction rate. When the liquid level reduction rate difference is greater than zero, the original value is retained; when the liquid level reduction rate difference is not greater than zero, it is taken as 0, to obtain the liquid level reduction rate increment. Before entering the calculation of the drainage power consumption value, the liquid level reduction rate increment is multiplied by the effective cross-sectional area of ​​the sump well to obtain the liquid level equivalent flow rate increment. The subsequent drainage power consumption value uses the liquid level equivalent flow rate increment as the denominator in the calculation. In S2-3, the average power of the backup pumps at each sampling point within the dual-pump connection area is obtained by summing the backup pump power and dividing by the length of the dual-pump connection area. When the total flow rate increment of the pump group is greater than zero, the average backup pump power is divided by the total flow rate increment to obtain the inlet power consumption value. When the total flow rate increment of the pump group is zero, the average backup pump power is used as the inlet power consumption value to represent the no-incremental consumption when the backup pump connection does not generate new flow. When the liquid level equivalent flow rate increment is greater than zero, the average backup pump power is divided by the liquid level equivalent flow rate increment to obtain the drainage power consumption value. When the liquid level equivalent flow rate increment is zero, the average backup pump power is used as the drainage power consumption value to represent the no-incremental consumption when the backup pump connection does not generate drainage improvement. Among the inlet power consumption value and the drainage power consumption value, the value that is not lower than the other is taken as the backup pump consumption value. The replacement deviation is obtained by subtracting the unloading amount from the backup pump consumption value. When the two values ​​are the same, the inlet power consumption value is taken as the backup pump consumption value. Through the above process, the replacement deviation is not directly calculated using the standby pump power. Instead, the total flow increment of the pump group and the liquid level equivalent flow increment are used to check whether the standby pump power constitutes a connection result. Then, it is deducted from the unloading amount to obtain the remaining consumption after the standby pump is connected. This result is entered into the subsequent replacement code calculation to distinguish between actual replacement, virtual replacement, and drag-along replacement. In practical applications: After the standby pump of a certain pumping station is started, the flow rate of some sampling positions in the dual-pump connection area is higher than that of the main pump's connection area, and the flow rate of some sampling positions is lower than that of the main pump's connection area. The lower part is treated as 0 and the total flow increment of the pump group is obtained. When the liquid level drop rate in the same segment is higher than before the connection, it is converted into the liquid level equivalent flow increment based on the effective cross-sectional area of ​​the sump. If the connection power consumption value and drainage power consumption value brought by the average standby pump power are not offset by the unloading amount, the replacement deviation is positive, and subsequent steps will proceed to the virtual replacement position or drag-along replacement position determination accordingly.

[0020] S3. Generate a replacement code based on the unloading amount and replacement deviation, and read the replacement position of the standby pump start position from the replacement code. The replacement position includes the virtual replacement position, the real replacement position and the drag replacement position to obtain the replacement status. In this embodiment, the replacement state is used to convert the unloading amount and replacement deviation into the state result of the standby pump start position; the unloading amount represents the decrease in the unit delivery power consumption of the main pump after the standby pump is connected, and the replacement deviation represents the remaining result of the standby pump consumption after deducting the unloading amount; the two are converted into binary bits to form a replacement code, and then the replacement bit is obtained from the replacement code table, and combined with the replacement amplitude difference to form an amplitude-bound replacement bit, so that the subsequent three-state replacement diagram can simultaneously read the state category and amplitude direction. The implementation process includes the following steps: In S3-1, the unloading amount obtained from S1 and the replacement deviation obtained from S2 are used as inputs. When the unloading amount is greater than zero, it indicates that the unit delivery power consumption of the main pump decreases after the standby pump is connected, and the unloading value is 1. When the unloading amount is not greater than zero, it indicates that the unit delivery power consumption of the main pump does not decrease after the standby pump is connected, and the unloading value is 0. When the replacement deviation is greater than zero, it indicates that the consumption value of the standby pump is not offset by the unloading amount, and the deviation value is 1. When the replacement deviation is not greater than zero, it indicates that the consumption value of the standby pump has been offset by the unloading amount, and the deviation value is 0. The unloading value and the deviation value are both linked to the same standby pump start position for use in the replacement code calculation. In S3-2, the unloading position value is used as the high bit and the deviation position value is used as the low bit. The replacement code is obtained by multiplying the unloading position value by 2 and adding the deviation position value. The replacement code table is set up with rows according to the code value. Replacement code 0 corresponds to the virtual replacement bit, replacement code 1 corresponds to the virtual replacement bit, replacement code 2 corresponds to the real replacement bit, and replacement code 3 corresponds to the drag replacement bit. The code table row with the same replacement code is read, and the replacement bit in the code table row is used as the replacement bit of the standby pump start position. Replacement code 0 indicates that the unloading amount is not greater than zero and the replacement deviation is not greater than zero. Although no additional consumption surplus is formed, the main pump unloading has not occurred. Therefore, it is classified as a virtual replacement bit and does not enter the real replacement bit. In S3-3, the absolute value of the unloading amount is taken to obtain the unloading amplitude, the absolute value of the replacement deviation is taken to obtain the deviation amplitude, and the replacement amplitude difference is obtained by subtracting the deviation amplitude from the unloading amplitude. When the replacement amplitude difference is greater than zero, it means that the unloading amplitude of the main pump is greater than the remaining amplitude consumed by the standby pump. When the replacement amplitude difference is not greater than zero, it means that the unloading amplitude of the main pump has not exceeded the remaining amplitude consumed by the standby pump. The replacement amplitude difference is combined with the replacement position of the standby pump start position to obtain the amplitude replacement position. The amplitude replacement position includes at least three fields: standby pump start position, replacement position, and replacement amplitude difference, and is used as the replacement status field for subsequent three-state replacement diagram reading. In S3-4, the replacement state is generated based on the amplitude replacement bit; the amplitude replacement bits of replacement code 0 and replacement code 1 are both assigned to virtual replacement bits, where replacement code 0 indicates that the backup pump connection does not lead to main pump unloading, and replacement code 1 indicates that the backup pump consumption does not lead to main pump unloading; the amplitude replacement bit of replacement code 2 is assigned to real replacement bits, indicating that the backup pump connection leads to main pump unloading and the backup pump consumption is offset by the unloading amount; the amplitude replacement bit of replacement code 3 is assigned to drag replacement bits, indicating that the backup pump connection leads to main pump unloading and the backup pump consumption is still retained; the replacement state consists of replacement bits and the amplitude difference of replacement, and serves as the state input of the three-state replacement diagram in S4; Through the above process, the unloading amount and replacement deviation are compressed into a replacement code. The replacement code table gives the status of the standby pump start position. The replacement amplitude difference retains the amplitude relationship between unloading and consumption, avoiding the loss of status strength based solely on a single positive or negative relationship. This replacement status is used for the status row values ​​of the subsequent Kikuchi connection area, Kikuchi drainage area, and TRW-S message transmission. In practical applications: after the standby pump starts, if the unloading amount is negative and the replacement deviation is positive, the unloading bit value is 0, the deviation bit value is 1, the replacement code is 1, and the replacement bit is a dummy replacement bit, indicating that the standby pump consumption has not brought about the unloading of the main pump; if the unloading amount is positive and the replacement deviation is not greater than zero, the replacement code is 2, and the replacement bit is a real replacement bit, indicating that the standby pump connection has stripped the main pump's unit delivery power consumption; if the unloading amount is positive and the replacement deviation is positive, the replacement code is 3, and the replacement bit is a drag-over replacement bit, indicating that the standby pump connection forms unloading but the main pump should not be directly withdrawn.

[0021] S4. Construct a three-state replacement diagram based on the replacement state, unloading amount, pipe connection power consumption value, drain power consumption value and replacement deviation. Generate region bits from the Kikuchi pipe connection region and the Kikuchi drain region, and obtain the real replacement confidence value, virtual replacement confidence value and drag replacement confidence value through TRW-S message transmission. In this embodiment, the three-state succession diagram is used to put the succession state obtained by S3 into the same state calculation structure as the unloading amount, pipe connection power consumption value, drainage power consumption value and succession deviation obtained by S1 and S2; the real succession row, virtual succession row and drag succession row represent the three types of standby pump access results respectively; the Kikuchi pipe connection area is used to determine whether the pipe connection power consumption value is supported by the unloading amount; the Kikuchi drainage area is used to determine whether the drainage power consumption value is supported by the unloading amount; the TRW-S message transmission is used to convert the position results of the two areas into three types of succession confidence values. The implementation process includes the following steps: In S4-1, the replacement status output by S3 and the unloading amount, pipe connection power consumption value, drainage power consumption value, and replacement deviation output by S1 and S2 are used as inputs. First, a three-row state structure is set: the row with unloading target bit 1 and deviation target bit 0 is taken as the real replacement row, indicating that the standby pump is connected to form the main pump unloading and the replacement deviation has not formed a positive value; the row with unloading target bit 0 and deviation target bit 1 is taken as the virtual replacement row, indicating that the standby pump is connected but the main pump unloading has not been formed and the standby pump consumption has remaining; the row with unloading target bit 1 and deviation target bit 1 is taken as the drag replacement row, indicating that the standby pump is connected to form the main pump unloading but the standby pump consumption still has remaining. The three-row state structure, together with the unloading amount column, pipe connection power consumption value column, drainage power consumption value column, and replacement deviation column, form a three-state replacement diagram; each row stores its own unloading target bit, deviation target bit, unloading amount, pipe connection power consumption value, drainage power consumption value, and replacement deviation for subsequent row-by-row calculation. In S4-2, the unloading amount is converted into an unloading input bit, which is 1 when the unloading amount is greater than zero and 0 when the unloading amount is not greater than zero; the replacement deviation is converted into a deviation input bit, which is 1 when the replacement deviation is greater than zero and 0 when the replacement deviation is not greater than zero; two sets of XOR operations are performed row by row on the three-state replacement diagram: the first set XORs the unloading input bit with the unloading target bit in the row, and the unloading difference bit is 0 when they are the same and 1 when they are different; the second set XORs the deviation input bit with the deviation target bit in the row, and the deviation difference bit is 0 when they are the same and 1 when they are different; the unloading difference bit and the deviation difference bit are used to indicate whether the current input matches the target bit combination of each state row, and enter the Kikuchi pipe-connecting area and Kikuchi drainage area with the state row. In S4-3, two Kikuchi regions are formed in each row of the three-state succession diagram: the Kikuchi pipe-connecting region is formed by unloading difference bit, deviation difference bit, pipe power consumption value, and unloading amount; the Kikuchi draining region is formed by unloading difference bit, deviation difference bit, draining power consumption value, and unloading amount. Within the Kikuchi pipe-connecting region, the unloading difference bit and deviation difference bit of the current row are read first. If both are 0, the pipe power consumption value and unloading amount are compared. If the pipe power consumption value is not greater than the unloading amount, the pipe-connecting region bit is set to 1; otherwise... In the case of taking over the area, the bit 0 is taken; in the Kikuchi drainage area, the unloading difference bit and the deviation difference bit of this line are read first. When both are 0, the drainage power consumption value and the unloading amount are compared. When the drainage power consumption value is not greater than the unloading amount, the drainage area bit 1 is taken. In other cases, the drainage area bit 0 is taken. If the virtual replacement line or the drag replacement line is 0 because the target bit combination of the unloading amount and the replacement deviation is consistent but the area power consumption is not covered by the unloading amount, then the line retains the status line identity in the subsequent TRW-S message transmission, but does not obtain the positive message of this area. In S4-4, TRW-S message passing is performed row by row on the three-state succession diagram. Each state row is jointly covered by the Kikuchi takeover region and the Kikuchi drainage region. Therefore, the takeover tree weight is halved, and the drainage tree weight is halved. The takeover region bit is multiplied by halved to obtain the takeover message value, and the drainage region bit is multiplied by halved to obtain the drainage message value. The takeover message value and the drainage message value are added together to obtain the row message value for that state row. The row message values ​​for the real succession row, the virtual succession row, and the drag succession row are obtained respectively. Then, calculate the sum of the three line message values; when the sum of the three line message values ​​is greater than zero, divide each line message value by the sum of the three line message values ​​to obtain the real replacement confidence value, the virtual replacement confidence value, and the drag replacement confidence value; when the sum of the three line message values ​​is zero, read the replacement status obtained by S3. When the replacement status is the real replacement bit, the real replacement confidence value is 1 and the other two items are 0; when the replacement status is the virtual replacement bit, the virtual replacement confidence value is 1 and the other two items are 0; when the replacement status is the drag replacement bit, the drag replacement confidence value is 1 and the other two items are 0. Through the above process, the three-state succession diagram binds the succession state with the unloading amount, pipe connection power consumption, drainage power consumption, and succession deviation into a three-row state structure. The Kikuchi pipe connection region and the Kikuchi drainage region provide region positions from the pipe connection side and the drainage side, respectively. TRW-S message passing writes back the two region positions as three types of confidence values ​​according to tree weights. This process ensures that the standby pump connection result is not determined by a single difference value, but by the succession state, pipe connection power consumption, and drainage power consumption all entering the three-state calculation. In practical applications: If a standby pump start position is obtained as the actual replacement position via S3, and the unloading amount is greater than zero and the replacement deviation is not greater than zero, then the unloading difference bit and deviation difference bit of the actual replacement row are both 0; if the power consumption value of the connection and the power consumption value of the drainage are not greater than the unloading amount, then the connection area bit and the drainage area bit are both set to 1, and the row message value of the actual replacement row is 1. After normalization, the actual replacement confidence value takes the dominant value among the three items; if both area bits are 0, then the rollback value is performed according to the replacement status of S3 to ensure that the standby pump start position can still output the confidence value result.

[0022] S5. Construct a posterior row table using real replacement confidence, virtual replacement confidence, and drag replacement confidence. Determine the posterior selected row using the posterior inference algorithm, generate mode actions based on the posterior selected row, and output the optimized result of the electric pump operation mode. In this embodiment, the posterior row table is used to receive the real replacement confidence, virtual replacement confidence, and drag replacement confidence from the S4 output, and converts the three confidence values ​​into the electric pump operation mode optimization results; the posterior inference algorithm does not introduce a separate training model, but generates the winning position and winning number through the pairwise difference of the three rows of confidence values, and then determines the backup pump access order and the main pump exit position by the posterior selected row, so that the mode action directly comes from the replacement calculation result of the same backup pump start position; The implementation process includes the following steps: In S5-1, a posterior row table is established using real replacement confidence, dummy replacement confidence, and dragged replacement confidence as inputs. The posterior row table includes columns for row name, confidence, first difference, second difference, first win position, second win position, number of wins, and posterior selected median. The row name column is filled with real replacement rows, dummy replacement rows, and dragged replacement rows in sequence. In the confidence column, real replacement rows are filled with real replacement confidence, dummy replacement rows are filled with dummy replacement confidence, and dragged replacement rows are filled with dragged replacement confidence. The posterior selected median is initially set to 0 and is subsequently changed to 1 only by the posterior inference result. In S5-2, the posterior inference algorithm is executed row by row in the posterior row table. For any posterior row, the confidence value of this row is subtracted from the confidence values ​​of the other two rows to obtain the first difference and the second difference. When the first difference is greater than zero, the first win position is set to 1; when the first difference is not greater than zero, the first win position is set to 0. When the second difference is greater than zero, the second win position is set to 1; when the second difference is not greater than zero, the second win position is set to 0. The first win position and the second win position are added together to obtain the win count of this row. The actual replacement row, the virtual replacement row, and the dragged replacement row are all calculated according to the same rules to obtain the win count of the three rows. In S5-3, the three rows of winning numbers in the posterior row table are read. The posterior row with a winning number of 2 is determined as the posterior selected row, and the posterior selected bit of the posterior selected row is set to 1. The posterior selected bit of the other posterior rows is kept at 0. When the winning numbers of the three rows are not all 2, the confidence values ​​of the three rows that are not lower than the other two rows are read first. Then, the first posterior row with that confidence value is read in the order of real replacement row, drag replacement row, and virtual replacement row as the row with the same value high bit, and the row with the same value high bit is determined as the posterior selected row. The posterior selected row generates a mode action. When the posterior selected row is a real replacement row, a forward movement action is generated. When the posterior selected row is a virtual replacement row, a backward movement action is generated. When the posterior selected row is a drag replacement row, a reserved access and restricted exit action is generated. In S5-4, a standby pump priority table is established, which includes the standby pump number, access priority, standby pump start position, subsequent verification up row, priority action, and main pump exit position. When the subsequent verification up row is a real replacement row, the target standby pump is swapped with the standby pump in the previous access priority position. If the target standby pump is already in the first position, the access priority remains unchanged. When the subsequent verification up row is a virtual replacement row, the target standby pump is swapped with the standby pump in the next access priority position. If the target standby pump is already in the last position, the access priority remains unchanged. When the subsequent verification up row is a towed replacement row, the standby pump access priority remains unchanged, and the main pump exit position is set to 0. A main pump exit position of 1 indicates that the main pump is allowed to exit in the next control cycle, and a main pump exit position of 0 indicates that the main pump is prohibited from exiting in the next control cycle. The standby pump priority table, subsequent verification up row, priority action, and main pump exit position constitute the result of this electric pump operation mode optimization. Through the above process, S5 converts the three confidence values ​​into subsequent verification rows, and then the subsequent verification rows drive the backup pump access sequence and main pump exit sequence to be written back. In the actual replacement row, the backup pump that has completed the replacement moves to the next higher sequence, while in the virtual replacement row, the backup pump that has not yet formed a replacement moves to the next position. The drag-along replacement row retains the backup pump access but prohibits the main pump exit, thus avoiding judging the replacement completion solely based on the backup pump start-up action. In practical applications: after backup pump B2 starts, if the confidence value of the actual replacement row in the subsequent verification row table is higher than that of the virtual replacement row and the drag-along replacement row, then the success number of the actual replacement row is 2, and B2 is exchanged with the backup pump in the previous sequence. If the success number of the virtual replacement row is 2, then B2 is exchanged with the backup pump in the next sequence. If the success number of the drag-along replacement row is 2, then B2 continues to access, the main pump exit sequence is 0, and the main pump exit is not executed in the next control cycle.

[0023] Furthermore, a data analysis-based electric pump operation mode optimization system includes: The segmentation module is used to read the monitoring and control acquisition records, and to extract the main pump bearing segment of the previous control cycle and the dual pump access segment of the next control cycle with the standby pump start position as the boundary. The power consumption value of a single pump and the residual power consumption value are calculated from the power of the main pump and the total flow of the pump group. The unloading amount is obtained by subtracting the residual power consumption value from the power consumption value of a single pump. The replacement quantization module calculates the connection power consumption value and the discharge power consumption value by taking the backup pump power, the total flow rate increment of the pump group and the liquid level reduction speed increment in the dual-pump access chip as inputs, and calculates the replacement deviation from the connection power consumption value, the discharge power consumption value and the unloading amount. The replacement code module generates a replacement code based on the unloading amount and replacement deviation, and reads the replacement position of the standby pump start position from the replacement code. The replacement position includes a virtual replacement position, a real replacement position, and a drag replacement position to obtain the replacement status. The confidence transmission module constructs a three-state succession diagram based on the succession status, unloading amount, pipe connection power consumption value, drain power consumption value, and succession deviation. It generates region bits from the Kikuchi pipe connection region and the Kikuchi drain region, and obtains the real succession confidence value, virtual succession confidence value, and drag succession confidence value through TRW-S message transmission. The mode write-back module constructs a posterior row table using real replacement confidence, virtual replacement confidence, and drag-and-drop replacement confidence. It then determines the posterior selected row using a posterior inference algorithm, generates mode actions based on the posterior selected row, and outputs the optimized results of the electric pump operation mode.

[0024] Working principle: First, the standby pump start position is found from the operation record formed by supervision and control and data acquisition, and the main pump bearing plate before start-up and the dual-pump access plate after start-up are captured; then, the change in the unit delivery power consumption of the main pump before and after the standby pump access is compared to obtain the unloading amount, which is used to determine whether the main pump load has been shared by the standby pump; then, the replacement deviation is calculated by combining the standby pump power, the total flow increment of the pump group and the liquid level reduction speed increment to determine whether the standby pump access forms an effective replacement, a virtual replacement, or a drag replacement; then, three types of confidence quantities are generated through the three-state replacement diagram, Kikuchi connection area, Kikuchi discharge area and TRW-S message transmission; finally, the standby pump access order and the main pump exit state are adjusted by the posterior inference results, and the electric pump operation mode optimization result is output. In sewage lifting pump stations, when the liquid level rises, the standby pump starts. Traditional control typically only considers whether the standby pump starts and whether the liquid level drops. This solution further compares the main pump power, total pump flow, and liquid level drop rate before and after startup. If the main pump's unit delivery power consumption decreases after the standby pump starts, and the increased flow and liquid level drop rate can cover the standby pump's consumption, it is considered an effective replacement, and the standby pump's subsequent connection priority is increased. If the standby pump's power consumption increases but the main pump load does not decrease, it is considered a false replacement, and the standby pump's connection priority is decreased. If the standby pump does share part of the load but the main pump still cannot be shut down, the standby pump's connection is retained, while the main pump's shutdown is restricted. This allows the pump station to no longer judge whether the replacement is completed solely based on start-stop actions, but rather to judge whether the standby pump is truly effective based on on-site operating data.

[0025] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing the operation mode of an electric pump based on data analysis, characterized in that, include: S1. Read the monitoring and control acquisition records, and take the standby pump start position as the boundary to extract the main pump bearing plate of the previous control cycle and the dual pump access plate of the next control cycle. Calculate the power consumption value of a single pump and the residual power consumption value from the power of the main pump and the total flow of the pump group. Subtract the residual power consumption value from the power consumption value of a single pump to obtain the unloading amount. S2. Using the standby pump power, total flow rate increment of the pump group and liquid level drop rate increment in the dual-pump access plate as input, calculate the pipe connection power consumption value and the discharge power consumption value, and calculate the replacement deviation from the pipe connection power consumption value, the discharge power consumption value and the unloading amount. S3. Generate a replacement code based on the unloading amount and replacement deviation, and read the replacement position of the standby pump start position from the replacement code. The replacement position includes the virtual replacement position, the real replacement position and the drag replacement position to obtain the replacement status. S4. Construct a three-state replacement diagram based on the replacement state, unloading amount, pipe connection power consumption value, drain power consumption value and replacement deviation. Generate region bits from the Kikuchi pipe connection region and the Kikuchi drain region, and obtain the real replacement confidence value, virtual replacement confidence value and drag replacement confidence value through TRW-S message transmission. S5. Construct a posterior row table using real replacement confidence, virtual replacement confidence, and drag replacement confidence. Determine the posterior selected row using a posterior inference algorithm, generate mode actions based on the posterior selected row, and output the optimized result of the electric pump operation mode.

2. The method for optimizing the operation mode of an electric pump based on data analysis according to claim 1, characterized in that: S1 includes: S1-1. Define the number that increments by one bit in each data frame acquisition completed in the monitoring and control acquisition record as the sampling sequence number, define the sampling sequence number interval between two adjacent pump group control command issuance positions as the control cycle, and determine the control cycle before the standby pump start position as the main pump's operating segment. S1-2. After the standby pump starts, the control cycle is determined to be the dual-pump access plate. The power of the main pump and the total flow of the pump group in the main pump support plate and the dual-pump access plate are summed according to the sampling sequence number to obtain the sum of single pump power, single pump flow, residual power and residual flow. S1-3. The power consumption of a single pump is obtained by dividing the power of a single pump by the flow rate of a single pump. The residual power consumption is obtained by dividing the residual power by the flow rate of a single pump. The unloading amount is obtained by subtracting the residual power consumption from the power consumption of a single pump.

3. The method for optimizing the operation mode of an electric pump based on data analysis according to claim 2, characterized in that: S2 includes: S2-1. Subtract the sampling points of the dual pump connection plate and the main pump bearing plate according to the sampling sequence number to obtain the flow difference of each sampling point. Set the negative values ​​of the flow difference to zero and calculate the average value to obtain the total flow increment of the pump group. S2-2. Subtract the liquid level at the tail of the plate from the liquid level at the beginning of the plate and the liquid level at the end of the plate from the liquid level at the beginning of the plate and the liquid level at the end of the plate respectively, and divide by the plate length to obtain the liquid level reduction rate of the single pump and the liquid level reduction rate of the dual pump. Subtract the liquid level reduction rate of the single pump from the liquid level reduction rate of the dual pump and set the negative value to zero to obtain the liquid level reduction rate increment. S2-3. The average power of the standby pump connected to the dual pumps is obtained by averaging the power of the standby pump. The power consumption of the connecting pipe is obtained by dividing the average power of the standby pump by the total flow rate increment of the pump group. The power consumption of the draining pump is obtained by dividing the average power of the standby pump by the liquid level drop rate increment. The unloading amount is subtracted from the higher of the power consumption of the connecting pipe and the draining pump to obtain the replacement deviation. When the total flow rate increment of the pump group is zero, the power consumption of the connecting pipe is taken as the average power of the standby pump. When the liquid level drop rate increment is zero, the power consumption of the draining pump is taken as the average power of the standby pump.

4. The method for optimizing the operation mode of an electric pump based on data analysis according to claim 3, characterized in that: S3 includes: S3-1. Using unloading amount and replacement deviation as input, convert unloading amount greater than zero to unloading position value 1 and unloading amount not greater than zero to unloading position value 0. Convert replacement deviation greater than zero to deviation position value 1 and replacement deviation not greater than zero to deviation position value 0. Output unloading position value and deviation position value. S3-2. Multiply the unloading position value by 2 and add the deviation position value to obtain the replacement code. Read the code table row with the same replacement code from the replacement code table and use the replacement position in the code table row as the replacement position of the standby pump start position. The replacement code table includes the row record of replacement code 1 and virtual replacement position, the row record of replacement code 2 and real replacement position, and the row record of replacement code 3 and drag replacement position.

5. The method for optimizing the operation mode of an electric pump based on data analysis according to claim 4, characterized in that: S3 further includes: S3-3. Subtract the absolute value of the replacement deviation from the absolute value of the unloading amount to obtain the replacement amplitude difference, and combine the replacement amplitude difference with the replacement position of the standby pump start position to obtain the amplitude replacement position. S3-4. Generate the replacement state based on the amplitude replacement bit, where the amplitude replacement bit of replacement code 1 indicates that the standby pump consumption has not formed the main pump unloading, the amplitude replacement bit of replacement code 2 indicates that the standby pump is connected to form the main pump unloading, and the amplitude replacement bit of replacement code 3 indicates that the standby pump is connected to form the main pump unloading and still retains the standby pump consumption.

6. The method for optimizing the operation mode of an electric pump based on data analysis according to claim 5, characterized in that: S4 includes: S4-1. Taking the replacement state, unloading amount, pipe connection power consumption value, drain power consumption value and replacement deviation as input, take unloading target position 1 and deviation target position 0 as the real replacement row, take unloading target position 0 and deviation target position 1 as the virtual replacement row, take unloading target position 1 and deviation target position 1 as the drag replacement row, and form a three-state replacement diagram by real replacement row, virtual replacement row, drag replacement row, unloading amount column, pipe connection power consumption value column, drain power consumption value column and replacement deviation column. S4-2. Convert the unloading amount greater than zero to unloading input bit 1, and the unloading amount not greater than zero to unloading input bit 0. Convert the replacement deviation greater than zero to deviation input bit 1, and the replacement deviation not greater than zero to deviation input bit 0. Perform XOR operation between unloading input bit and unloading target bit and XOR operation between deviation input bit and deviation target bit on each row of the three-state replacement diagram to obtain unloading difference bit and deviation difference bit.

7. The method for optimizing the operation mode of an electric pump based on data analysis according to claim 6, characterized in that: S4 further includes: S4-3. In the three-state succession diagram, the Kikuchi pipe-connecting region is composed of unloading difference, deviation difference, pipe power consumption value, and unloading amount; the Kikuchi draining region is composed of unloading difference, deviation difference, draining power consumption value, and unloading amount. When both the unloading difference and deviation difference in the Kikuchi pipe-connecting region are 0 and the pipe power consumption value is not greater than the unloading amount, the pipe-connecting region is set to 1; otherwise, the pipe-connecting region is set to 0. When both the unloading difference and deviation difference in the Kikuchi draining region are 0 and the draining power consumption value is not greater than the unloading amount, the draining region is set to 1; otherwise, the draining region is set to 0. S4-4. Perform TRW-S message passing row by row on the three-state succession diagram. Multiply the takeover region bit by half to obtain the takeover message value, multiply the drain region bit by half to obtain the drain message value, add the takeover message value and the drain message value to obtain the row message value, and normalize the sum of the row message values ​​of the real succession row, the virtual succession row, and the drag-over succession row to obtain the real succession confidence value, the virtual succession confidence value, and the drag-over succession confidence value.

8. The method for optimizing the operation mode of an electric pump based on data analysis according to claim 7, characterized in that: S5 includes: S5-1. Using the real replacement confidence value, the virtual replacement confidence value, and the dragged replacement confidence value as inputs, establish a posterior table. The posterior table includes the real replacement row, the virtual replacement row, and the dragged replacement row. Fill the three confidence values ​​into the confidence value columns of the real replacement row, the virtual replacement row, and the dragged replacement row, respectively. S5-2. The posterior inference algorithm is used to perform pairwise difference on each row of the posterior row table. The confidence value of the current row is subtracted from the confidence values ​​of the other two rows respectively. If the difference is greater than zero, the winning position is 1; if the difference is not greater than zero, the winning position is 0. The winning number of the current row is obtained by adding the two winning positions.

9. The method for optimizing the operation mode of an electric pump based on data analysis according to claim 8, characterized in that: The S5 also includes: S5-3. Determine the successive row with a winning number of 2 as the successive selected row. If the winning numbers of all three rows are not 2, read the first row with the same high value as the successive selected row in the order of real replacement row, drag replacement row, and virtual replacement row, and generate the pattern action based on the successive selected row. S5-4. When the selected operation is a real replacement, the backup pump connection order is moved forward by one position. When the selected operation is a virtual replacement, the backup pump connection order is moved backward by one position. When the selected operation is a towed replacement, the backup pump connection is retained and the main pump exit position is set to 0. The optimized result of the electric pump operation mode is output.

10. A data analysis-based electric pump operation mode optimization system, characterized in that, include: The segmentation module is used to read the monitoring and control acquisition records, and to extract the main pump bearing segment of the previous control cycle and the dual pump access segment of the next control cycle with the standby pump start position as the boundary. The power consumption value of a single pump and the residual power consumption value are calculated from the power of the main pump and the total flow of the pump group. The unloading amount is obtained by subtracting the residual power consumption value from the power consumption value of a single pump. The replacement quantization module calculates the connection power consumption value and the discharge power consumption value by taking the backup pump power, the total flow rate increment of the pump group and the liquid level reduction speed increment in the dual-pump access chip as inputs, and calculates the replacement deviation from the connection power consumption value, the discharge power consumption value and the unloading amount. The replacement code module generates a replacement code based on the unloading amount and replacement deviation, and reads the replacement position of the standby pump start position from the replacement code. The replacement position includes a virtual replacement position, a real replacement position, and a drag replacement position to obtain the replacement status. The confidence transmission module constructs a three-state succession diagram based on the succession status, unloading amount, pipe connection power consumption value, drain power consumption value, and succession deviation. It generates region bits from the Kikuchi pipe connection region and the Kikuchi drain region, and obtains the real succession confidence value, virtual succession confidence value, and drag succession confidence value through TRW-S message transmission. The mode write-back module constructs a posterior row table using real replacement confidence, virtual replacement confidence, and drag-and-drop replacement confidence. It then determines the posterior selected row using a posterior inference algorithm, generates mode actions based on the posterior selected row, and outputs the optimized results of the electric pump operation mode.

Images