Injection molding machine abnormal energy consumption early warning method

Abstract

This invention relates to the field of energy consumption monitoring technology for injection molding equipment, and discloses a method for early warning of abnormal energy consumption in injection molding machines. During continuous production, injection cycle and power sampling objects are established, segmented according to process phase, and a piecewise linear power function is formed. Phase boundaries are determined based on process event signals. Through equal-length discretization, amplitude normalization, curvature energy calculation, and phase energy ratio analysis, a fixed benchmark for removing extreme values ​​is established. Anomaly markers are generated by combining the curvature mutation multiple with the synchronous energy consumption increment, and the abnormal phase, warning level, and warning result are output. This solves the problems of difficulty in locating specific process actions in whole-machine energy consumption monitoring, difficulty in comparing data from different cycles and phases, and the susceptibility to misjudgment based on a single threshold, improving the interpretability of abnormal energy consumption analysis and the targeted nature of maintenance.

Description

Technical Field

[0001] This invention relates to the field of energy consumption monitoring technology for injection molding equipment, specifically a method for early warning of abnormal energy consumption in injection molding machines. Background Technology

[0002] Injection molding equipment typically undergoes several stages in continuous production, including mold closing, injection, pressure holding, plasticizing, cooling and holding, and mold opening and ejection. The load states of the motor, hydraulic unit, heating unit, and control unit differ at each stage, resulting in significant phase differences in the active power waveform within the injection cycle. Current energy consumption monitoring technologies for injection molding machines mostly employ electricity meters, power acquisition modules, energy management platforms, and production data acquisition systems to statistically analyze the total electrical energy, average power, peak power, and energy consumption per unit product for a single machine. The energy consumption status is determined through methods such as fixed thresholds, manually set experience ranges, and comparisons with total cycle energy consumption. Some existing solutions also combine production task records, equipment operating time, and output data to generate energy efficiency reports for production management personnel to conduct energy consumption analysis.

[0003] However, existing technologies primarily focus on total cycle power and average energy consumption of the entire machine, making it difficult to reflect the power waveform variation characteristics of each process phase within the injection molding cycle. When abnormal energy consumption occurs only in a certain phase, the change in total cycle power may not be obvious, leading to the inability to identify abnormal energy consumption states in a timely manner. Existing early warning methods based on total thresholds are also easily affected by variations in production volume, process formulation, mold, materials, and cycle duration. It is difficult to unify the energy consumption benchmark for the same equipment under different production conditions, resulting in insufficient stability of early warning results. Existing methods typically lack refined processing of phase boundaries such as mold closing, injection, holding pressure, plasticizing, cooling and holding, and mold opening and ejection, failing to jointly judge changes in phase power waveform morphology with changes in phase energy consumption ratio. Therefore, it is difficult to distinguish between normal process fluctuations and waveform abrupt changes indicating abnormal energy consumption. Furthermore, some existing methods rely on manual experience thresholds and post-event statistical reports, resulting in insufficient real-time performance and automatic positioning capabilities. Some methods use general model training, requiring a large number of labeled samples, and the model output process is not easily interpreted, which is not conducive to forming deterministic early warning criteria on the injection molding site. For situations such as enhanced local refraction in power waveforms, synchronous increase in phase energy consumption ratio, and flat reference period waveforms but abrupt changes in target period, existing technologies lack abnormal energy consumption early warning methods that can be directly calculated based on sampled power, phase event time, and fixed reference period data.

[0004] Therefore, this case aims to propose a method for early warning of abnormal energy consumption in injection molding machines. First, continuous power change data is generated through power sampling. Then, based on the process event signals output by the injection molding machine controller, an injection cycle is divided into multiple phases. Subsequently, the power waveforms of each phase are processed to uniform length and amplitude normalized, so that there is a basis for comparison between different cycles and different phases. Furthermore, features such as waveform refraction degree, phase energy ratio, fixed reference curvature, fixed reference energy consumption, and curvature deviation are extracted. Finally, the abnormal phase, warning level, and warning result are output by combining curvature mutation and synchronous increase in energy consumption. Summary of the Invention

[0005] This invention provides a method for early warning of abnormal energy consumption in injection molding machines, which helps to solve the problems mentioned in the background art.

[0006] This invention provides the following technical solution: a method for early warning of abnormal energy consumption in injection molding machines, comprising:

[0007] In the continuous production process, injection cycle and power sampling objects are established, each injection cycle is segmented according to the process phase, and a piecewise linear power function is formed based on the original power sampling value.

[0008] Read the time of each process event signal, form a phase boundary according to the time of adjacent process event signals, perform equal-length discrete sampling between the phase start time and the phase end time, and form an equal-length phase power sequence composed of equal-length phase power points arranged in the sampling order;

[0009] For equal-length phase power sequences, a lower and upper power bound are selected, a normalized power value is generated according to the amplitude range, and a normalized phase waveform is formed according to the sampling order.

[0010] Based on the normalized phase waveform, the first-order rate of change of the normalized phase position, the second-order refractive index of the normalized phase position, the discrete curvature, and the phase curvature energy are calculated sequentially.

[0011] The phase energy, total periodic energy, and phase energy percentage are calculated based on the piecewise linear power function, and the phase energy consumption structure is generated according to the process phase number order.

[0012] Set the reference acquisition period and the target period, and perform extreme value removal and averaging on the phase curvature energy and phase electrical energy ratio of the reference acquisition period to generate fixed reference curvature energy, fixed reference electrical energy ratio and fixed reference curvature deviation.

[0013] The curvature mutation multiple and energy consumption synchronous increment are calculated for the target period. Phase anomaly markers and period anomaly energy consumption early warning markers are generated according to the curvature mutation multiple and energy consumption synchronous increment.

[0014] The abnormal intensity is generated based on the phase anomaly marker, curvature mutation multiple, and energy consumption synchronous increment. The output is a triplet of abnormal phase output number, warning level, and abnormal energy consumption warning result.

[0015] Optionally, the step of establishing injection molding cycles and power sampling objects during continuous production, segmenting each injection molding cycle according to the process phase, and forming a piecewise linear power function based on the original power sampling values ​​specifically includes:

[0016] For continuous production processes under the same injection molding machine, mold number, material number, and process formula number, establish a set of injection cycle numbers. The injection cycle numbers are incremented from the beginning according to the production sequence, and the total number of injection cycles is the number of numbers in the injection cycle number set.

[0017] Each injection cycle is divided into six process phases, and a set of process phase numbers is established. Process phase number 1 represents the mold closing phase, process phase number 2 represents the injection phase, process phase number 3 represents the holding pressure phase, process phase number 4 represents the plasticizing phase, process phase number 5 represents the cooling and holding phase, and process phase number 6 represents the mold opening and ejection phase.

[0018] For each injection molding cycle, a set of original sampling point numbers is established. The original sampling point numbers are incremented sequentially from the beginning according to the sampling time, and the total number of original sampling points for each injection molding cycle is limited to no less than two.

[0019] For each injection molding cycle, the positive input active power sequence is collected through the power metering unit, the positive input active power sampling value corresponding to each original sampling time is read, and the positive input active power sampling value is recorded as the original power sampling value of the corresponding injection molding cycle and the corresponding original sampling point;

[0020] For each injection molding cycle, the original power sample values ​​corresponding to adjacent original sampling points are connected by a straight line, so that the power value obtained by connecting any adjacent original sampling time with a straight line changes linearly with time, forming a piecewise linear power function for the current injection molding cycle.

[0021] Optionally, the step of reading the time of each process event signal, forming a phase boundary according to the time of adjacent process event signals, performing equal-length discrete sampling between the phase start time and the phase end time, and forming an equal-length phase power sequence composed of equal-length phase power points arranged in the sampling order, specifically includes:

[0022] Read the timing of the mold closing start event signal, injection start event signal, pressure holding start event signal, plasticizing start event signal, cooling and holding start event signal, mold opening and ejection start event signal, and cycle end event signal output by the injection molding machine controller, and form seven process event signal times according to the occurrence sequence of the process event signals: mold closing start, injection start, pressure holding start, plasticizing start, cooling and holding start, mold opening and ejection start, and cycle end.

[0023] The timing of the mold closing start event signal is limited to after or at the same time as the first original sampling time of the current injection cycle, and the timing of the cycle end event signal is limited to before or at the same time as the last original sampling time of the current injection cycle. The timing of the seven process event signals is also limited to increase in the following order: mold closing start, injection start, holding pressure start, plasticizing start, cooling and holding start, mold opening and ejection start, and cycle end.

[0024] For each process phase of each injection molding cycle, the process event signal time corresponding to the current process phase is taken as the phase start time, and the next process event signal time is taken as the phase end time, thus forming the phase boundary of the current process phase.

[0025] For each process phase of each injection molding cycle, the phase duration is calculated by subtracting the phase start time from the phase end time.

[0026] The number of equal-length phase power points retained for each process phase is set to twenty-one.

[0027] For each process phase of each injection molding cycle, twenty-one equal-length sampling points are generated at the same time interval between the start and end of the phase. The first equal-length sampling point is the start of the phase, the twenty-first equal-length sampling point is the end of the phase, and the time interval between adjacent equal-length sampling points is the phase duration divided by twenty.

[0028] For each equal-length sampling point, the function value of the piecewise linear power function of the corresponding injection cycle is read to form the equal-length phase power point corresponding to the current injection cycle, the current process phase, and the current equal-length sampling point.

[0029] Optionally, the step of selecting a lower and upper power bound for the comparison of equal-length phase power sequences, generating normalized power values ​​according to the amplitude range, and forming a normalized phase waveform according to the sampling order specifically includes:

[0030] For each process phase of each injection molding cycle, traverse all twenty-one equal-length phase power points within the current process phase, and select the minimum power value among all equal-length phase power points as the lower power bound of the current process phase.

[0031] For each process phase of each injection molding cycle, traverse all twenty-one equal-length phase power points within the current process phase, and select the maximum power value among all equal-length phase power points as the upper limit of the power of the current process phase.

[0032] When the upper power bound is the same as the lower power bound, the normalized power value corresponding to each equal-length phase power point within the current process phase is set to zero.

[0033] When the upper power limit is greater than the lower power limit, for each equal-length phase power point within the current process phase, the difference between the current equal-length phase power point and the lower power limit is divided by the difference between the upper power limit and the lower power limit to form the normalized power value corresponding to the current equal-length phase power point.

[0034] For each process phase of each injection molding cycle, the normalized power values ​​are arranged in order from front to back according to the time of equal length sampling points to form the normalized phase waveform of the current process phase.

[0035] Optionally, the step of sequentially calculating the first-order rate of change of the normalized phase position, the second-order refractive index of the normalized phase position, the discrete curvature, and the phase curvature energy based on the normalized phase waveform specifically includes:

[0036] For each normalized phase waveform, starting from the first normalized power value, two adjacent normalized power values ​​are selected sequentially. The previous normalized power value is subtracted from the subsequent normalized power value, and then multiplied by the value obtained by subtracting one from the number of equal-length phase power points to form the first-order change rate of the normalized phase position at the corresponding position. These are then arranged in order of position to form a sequence of the first-order change rates of the normalized phase position.

[0037] For the normalized phase position first-order change rate sequence, starting from the second normalized phase position first-order change rate, the current normalized phase position first-order change rate and the previous normalized phase position first-order change rate are selected sequentially. The current normalized phase position first-order change rate is subtracted from the previous normalized phase position first-order change rate, and then multiplied by the value after subtracting one from the number of equal-length phase power points to form the normalized phase position second-order refractive intensity at the corresponding position. These are then arranged in position order to form the normalized phase position second-order refractive intensity sequence.

[0038] For each position corresponding to the normalized phase position second-order refracted intensity in the normalized phase position second-order refracted intensity sequence, read the normalized phase position second-order refracted intensity, the normalized phase position first-order rate of change, and the normalized phase position first-order rate of change of the previous position. Average the normalized phase position first-order rate of change of the current position and the normalized phase position first-order rate of change of the previous position. Square the average result and add it to one. Then raise the sum to the cube of 2. Divide the normalized phase position second-order refracted intensity of the current position by the result of the power operation formed by adding the square of the average result to one and raising it to the cube of 2 to form the discrete curvature of the current position.

[0039] For each process phase, the discrete curvatures are arranged in order of their positions in the normalized phase position second-order refractive intensity sequence to form a discrete curvature sequence. Starting from the position corresponding to the second equal-length phase power point to the position corresponding to the penultimate equal-length phase power point, each discrete curvature is squared. The sum of the squared results is then divided by the number of discrete curvatures involved in the summation to form the phase curvature energy of the current process phase.

[0040] Optionally, the step of calculating the phase energy, total periodic energy, and phase energy percentage based on the piecewise linear power function, and generating the phase energy consumption structure according to the process phase number sequence, specifically includes:

[0041] For each process phase of each injection cycle, the piecewise linear power function of the current injection cycle is integrated over time from the start to the end of the current process phase to form the phase energy of the corresponding process phase of the current injection cycle.

[0042] For each injection cycle, the phase energy of the mold closing phase, injection phase, holding pressure phase, plasticizing phase, cooling and holding phase, and mold opening and ejection phase are added together to form the total cycle energy of the current injection cycle.

[0043] When the total electrical energy of the cycle is zero, the phase electrical energy ratio of each process phase in the current injection molding cycle is set to zero.

[0044] When the total energy of the cycle is greater than zero, for each process phase, the phase energy of the current process phase is divided by the total energy of the current injection cycle to form the phase energy ratio of the current process phase.

[0045] For each injection cycle, the phase energy ratio of each process phase is arranged in the order of mold closing phase, injection phase, holding pressure phase, plasticizing phase, cooling and holding phase, and mold opening and ejection phase, forming the phase energy consumption structure of the current injection cycle.

[0046] Optionally, the setting of the reference acquisition period and the target period, and the extreme value removal and averaging of the phase curvature energy and phase electrical energy ratio of the reference acquisition period to generate fixed reference curvature energy, fixed reference electrical energy ratio, and fixed reference curvature deviation, specifically include:

[0047] Set the number of fixed reference periods to twelve;

[0048] When the total number of injection molding cycles is not greater than the number of fixed reference cycles, all injection molding cycles are limited to the reference acquisition cycle. The phase curvature energy and phase electrical energy ratio of each process phase in each reference acquisition cycle are recorded, and no abnormal energy consumption warning result triplet is formed.

[0049] When the total number of injection cycles is greater than the number of fixed reference cycles, and the injection cycle number of the current injection cycle is not greater than the number of fixed reference cycles, the current injection cycle is limited to the reference acquisition cycle, and only the phase curvature energy and phase electrical energy ratio of each process phase in the current injection cycle are recorded.

[0050] When the total number of injection cycles is greater than the fixed reference number, and the injection cycle number of the current injection cycle is greater than the fixed reference number, the current injection cycle is set as the target cycle.

[0051] When the total number of injection molding cycles is greater than the number of fixed reference cycles, for each process phase, the maximum and minimum phase curvature energy values ​​of the current process phase are compared among the twelve reference acquisition cycles. The phase curvature energy of the current process phase in the twelve reference acquisition cycles is summed and then a maximum and a minimum phase curvature energy value are subtracted. The remaining phase curvature energy is then averaged to form the fixed reference curvature energy of the current process phase.

[0052] When the total number of injection molding cycles is greater than the number of fixed reference cycles, for each process phase, the maximum and minimum phase energy percentages of the current process phase are compared among the twelve reference acquisition cycles. The phase energy percentages of the current process phase in the twelve reference acquisition cycles are summed and then a maximum and a minimum phase energy percentage are subtracted. The remaining phase energy percentages are then averaged to form the fixed reference energy percentage of the current process phase.

[0053] When the total number of injection molding cycles is greater than the number of fixed reference cycles, for each process phase and each reference acquisition cycle, the phase curvature energy of the current process phase in the current reference acquisition cycle is subtracted from the fixed reference curvature energy of the current process phase to form the curvature deviation sample of the current process phase in the current reference acquisition cycle.

[0054] For each process phase, the maximum and minimum values ​​of curvature deviation samples of the current process phase are compared among the twelve reference acquisition cycles.

[0055] For each process phase, the curvature deviation samples of the current process phase in the twelve reference acquisition cycles are summed, and then the maximum value and minimum value of the curvature deviation sample are subtracted. The remaining curvature deviation samples are then averaged to form the fixed reference curvature deviation of the current process phase.

[0056] Optionally, the step of calculating the curvature mutation multiple and energy consumption synchronization increment for the target period, and generating phase anomaly markers and period anomaly energy consumption early warning markers according to the curvature mutation multiple and energy consumption synchronization increment, specifically includes:

[0057] For each process phase in the target cycle, read the phase curvature energy, fixed reference curvature energy, and fixed reference curvature deviation of the current process phase in the target cycle, and add the fixed reference curvature energy and the fixed reference curvature deviation to form the curvature reference sum;

[0058] When the phase curvature energy of the current process phase in the target cycle is zero and the curvature reference sum is zero, the curvature mutation factor of the current process phase is set to one.

[0059] When the phase curvature energy of the current process phase in the target period is greater than zero and the curvature reference sum is zero, divide the phase curvature energy of the current process phase in the target period by the value formed by adding one to the phase curvature energy of the current process phase in the target period, and then add the two to the obtained ratio to form the curvature mutation multiple of the current process phase.

[0060] When the curvature reference sum is greater than zero, the phase curvature energy of the current process phase in the target cycle is divided by the curvature reference sum to form the curvature mutation multiple of the current process phase.

[0061] For each process phase in the target cycle, the phase energy percentage of the current process phase in the target cycle is subtracted from the fixed reference energy percentage of the current process phase to form the synchronous energy consumption increment of the current process phase.

[0062] When the curvature mutation factor of the current process phase in the target cycle is greater than two and the synchronous energy consumption increment of the current process phase is greater than zero, the phase anomaly flag of the current process phase is set to one.

[0063] When the curvature abrupt change factor of the current process phase in the target cycle is not greater than two, the phase anomaly mark of the current process phase is set to zero.

[0064] When the curvature mutation factor of the current process phase in the target cycle is greater than two and the energy consumption synchronization increment of the current process phase is not greater than zero, the phase anomaly mark of the current process phase is set to zero.

[0065] For the target cycle, the phase anomaly markers of the six process phases are merged. When at least one phase anomaly marker is one, the cycle anomaly energy consumption warning marker is set to one. When all six phase anomaly markers are zero, the cycle anomaly energy consumption warning marker is set to zero.

[0066] Optionally, the step of generating anomaly intensity based on phase anomaly marker, curvature mutation multiple, and energy consumption synchronization increment, and outputting a triplet of anomaly phase output number, warning level, and anomaly energy consumption warning result, specifically includes: for each process phase in the target cycle, multiplying the phase anomaly marker of the current process phase, the curvature mutation multiple of the current process phase, and the value formed by adding the energy consumption synchronization increment of the current process phase to form the anomaly intensity of the current process phase;

[0067] When the abnormal energy consumption warning flag of the target cycle is zero, the abnormal phase output number is set to zero.

[0068] When the abnormal energy consumption warning mark of the target cycle is one, the process phase with the largest abnormal intensity is selected from the mold closing phase, injection phase, pressure holding phase, plasticizing phase, cooling holding phase and mold opening ejection phase to form an abnormal phase output number. When there are multiple process phases with the same abnormal intensity, the process phase with the smallest process phase number is selected.

[0069] When the abnormal energy consumption warning mark for the target cycle is zero, the warning level is set to zero;

[0070] When the abnormal energy consumption warning mark of the target cycle is one, and the abnormal intensity of the process phase corresponding to the abnormal phase output number is not greater than three, the warning level is set to one.

[0071] When the abnormal energy consumption warning mark of the target cycle is one, and the abnormal intensity of the process phase corresponding to the abnormal phase output number is greater than three, the warning level is set to two.

[0072] For the target cycle, output and record a triplet of abnormal energy consumption warning results, consisting of the injection cycle number, abnormal phase output number, and warning level.

[0073] The present invention has the following beneficial effects:

[0074] 1. The analysis of abnormal energy consumption in injection molding machines is limited to continuous production processes under the same injection molding machine, mold number, material number, and process formula number. Under this condition, injection cycle objects and original power sampling objects are established. Compared to directly mixing and analyzing data from different equipment, molds, materials, and formulas, this approach first eliminates the natural energy consumption differences caused by variations in process conditions, allowing subsequent anomaly detection to focus more on abnormal changes in equipment status or the production process itself. This scheme also divides each injection cycle into six distinct process phases and collects the positive input active power sequence through an energy metering unit. The power values ​​between adjacent sampling points are then connected linearly to form a piecewise linear power function. The originally discrete sampling data is organized into a continuous power description that can be read and integrated within any phase boundary, preserving the true sampling values ​​while avoiding data breaks caused by incomplete overlap between sampling time and process event time.

[0075] 2. This scheme does not roughly divide the injection molding process according to manually set fixed time slices. Instead, it reads the process event signals output by the injection molding machine controller, such as the start of mold closing, injection, holding pressure, plasticizing, cooling and holding, mold opening and ejection, and cycle end, and forms the true boundary of each process phase according to the time of adjacent process event signals. The phase boundary obtained in this way is consistent with the actual operation flow of the injection molding machine, and can accurately attribute energy consumption changes to the corresponding process actions. Furthermore, this scheme performs equal-length discrete sampling between the start and end times of each phase, and uniformly forms a fixed number of equal-length phase power points. The duration of the same phase in different injection cycles may vary due to fluctuations in operating conditions. If the original sampling sequence is directly compared, problems such as different number of points, different time scales, and inconsistent waveform positions will occur. After equal-length discretization processing, each phase is transformed into a power sequence of fixed length, so that subsequent normalization, curvature calculation, and inter-phase comparison have a unified data scale.

[0076] 3. For each process phase of each injection molding cycle, a lower and upper power bound are selected, and a normalized power value is generated based on the amplitude range of that phase. This transforms the power amplitude differences between different phases and cycles into waveform morphology differences under a unified scale. During injection molding, the power levels of phases such as mold closing, injection, holding pressure, and plasticizing naturally differ significantly. Directly comparing the original power amplitudes can easily misjudge normal process load differences as abnormalities. Normalization shifts the focus of judgment from absolute power magnitude to waveform fluctuations, abrupt changes, and folding characteristics. This scheme also considers the special case where the upper and lower power bounds are the same, and sets all normalized power values ​​for that phase to zero to avoid invalid calculations when the phase power does not change. This preserves the waveform change trend within the phase while reducing the interference of different rated power, load levels, and mold action intensities on curvature feature extraction.

[0077] 4. The normalized phase waveform is further transformed into curvature features that reflect the severity of local waveform changes. This process does not simply involve statistically analyzing maximum power, average power, or power fluctuation range. Instead, it first forms a first-order change in phase position based on adjacent normalized power values, then forms a second-order refractive index based on the difference between adjacent first-order changes, and subsequently generates discrete curvature by combining the average state of adjacent rates of change. The curvature at each position is then summarized into phase curvature energy. Curvature energy can reflect whether the power waveform exhibits sudden spikes, abnormal inflection points, transient load impacts caused by motion lag, or non-smooth changes during hydraulic or servo execution. These problems may not be obvious in the total electrical energy, but they will form obvious refractive in the waveform shape. Compared to existing monitoring methods that mainly rely on total cycle energy consumption or phase average energy consumption, this solution can capture early anomalies where the energy consumption structure has not yet increased significantly but the power action shape has already deteriorated. It is particularly suitable for scenarios with dynamic waveform characteristics, such as mold opening and closing, injection pressure switching, holding pressure transition, and plasticizing load fluctuations in injection molding machines.

[0078] 5. By combining power waveform analysis with phase energy structure analysis, this method not only calculates the actual electrical energy of each process phase within its phase boundary but also aggregates the electrical energy of the six phases into the total cycle energy, forming the proportion of each phase's energy in the entire injection molding cycle. This processing allows energy consumption anomalies to move beyond simply judging high power consumption in the current cycle; it reflects which specific process phase's proportion changes in the overall cycle energy consumption structure. Different phases of the injection molding machine correspond to different mechanical actions and process loads. For example, the injection phase may reflect injection pressure and runner resistance, the plasticizing phase may reflect screw load and heating status, and the mold opening and ejection phase may reflect mechanical resistance or ejection mechanism status. By using the phase energy proportion, a correlation can be established between abnormal energy consumption and specific process actions. Compared to existing technologies that only count the cumulative power consumption of the entire machine or the total cycle energy, this solution can identify situations where the total energy consumption change is not significant but the phase energy consumption distribution has shifted, and it also avoids simply attributing cycle production fluctuations to equipment malfunctions.

[0079] 6. Instead of setting a uniform global threshold for all phases, a fixed benchmark is established for each process phase. This scheme first uses several consecutive injection molding cycles as benchmark acquisition cycles, recording the curvature energy and electrical energy ratio of each phase during the benchmark stage. Then, extreme value removal and averaging are performed on each phase to form a fixed benchmark curvature energy and a fixed benchmark electrical energy ratio, and further, a fixed benchmark curvature deviation is generated. In injection molding production, occasional interference, sampling noise, and instantaneous load changes may cause data to be too high or too low in a certain benchmark cycle. If a normal average value or a single benchmark cycle is used directly, the benchmark is easily distorted. By removing the maximum and minimum values ​​before forming the benchmark, the stability and anti-interference ability of the benchmark can be improved. Compared with the existing technology that relies on manual experience thresholds, fixed absolute thresholds, or single historical average values, this scheme establishes a phase-level fixed benchmark that is jointly defined by equipment, mold, material, formula, and process phase, which is closer to the normal energy consumption pattern under current production conditions.

[0080] 7. Anomaly markers are generated by combining curvature abrupt changes in the target cycle with synchronous increments in phase energy consumption, rather than relying solely on waveform abrupt changes or increases in energy consumption. This scheme first reads the phase curvature energy of each process phase in the target cycle and compares it with the corresponding phase's fixed reference curvature energy and the reference state formed by the fixed reference curvature deviation to obtain the curvature abrupt change multiple. Simultaneously, the phase energy proportion of the target cycle is compared with the fixed reference energy proportion to obtain the synchronous increment in energy consumption. Only when the curvature change reaches the set abrupt change condition and the energy consumption proportion increases synchronously is the phase marked as an anomaly. This effectively distinguishes between process fluctuations with waveform morphology changes but no increase in energy consumption and true abnormal energy consumption states with waveform morphology abrupt changes and increased energy consumption structure, thereby reducing the false alarm rate. Compared to existing technologies that simply look at power peaks, energy consumption thresholds, or waveform fluctuations, this scheme employs dual constraints: on the one hand, it focuses on whether abnormal changes occur during the operation process; on the other hand, it focuses on whether such changes are accompanied by an increase in phase energy consumption burden.

[0081] 8. This solution not only provides an assessment of whether an anomaly is present, but also generates anomaly intensity based on phase anomaly markers, curvature mutation multiples, and synchronous energy consumption increments. It then outputs a tripartite sequence of anomaly phase number, warning level, and abnormal energy consumption warning result. This expands the definition of anomalies from mere presence to identifying the specific process phase, severity level, and corresponding injection cycle, ensuring traceability and executability of the warning information. On-site personnel no longer need to manually review all power curves; they can pinpoint the anomaly to the specific stage within mold closing, injection, holding pressure, plasticizing, cooling, or mold opening / ejection based on the anomaly phase number, and prioritize inspections according to the warning level. For example, when the anomaly is located in the injection phase, the focus can be on injection resistance, pressure switching, material flow status, or mold gate status; when the anomaly is located in the mold opening / ejection phase, the focus can be on mechanical resistance, ejector pin movement, or lubrication status. Compared to existing technologies that only output overall machine alarms or total power anomaly warnings, this solution provides diagnostic results with process implications, resolving issues such as the need for extensive manual investigation after an alarm, unclear anomaly sources, and delayed maintenance response. Attached Figure Description

[0082] Figure 1 This is a schematic diagram of the injection molding cycle and process phases of the present invention.

[0083] Figure 2 This is a schematic diagram of the equal-length phase power sequence and normalized phase waveform of the present invention.

[0084] Figure 3 This is a schematic diagram of the phase energy consumption structure and fixed reference of the present invention.

[0085] Figure 4 This is a schematic diagram of the abnormal energy consumption early warning results of the present invention. Detailed Implementation

[0086] 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.

[0087] An embodiment of an abnormal energy consumption early warning method for injection molding machines includes:

[0088] In the continuous production process, injection cycle and power sampling objects are established, each injection cycle is segmented according to the process phase, and a piecewise linear power function is formed based on the original power sampling value.

[0089] Read the time of each process event signal, form a phase boundary according to the time of adjacent process event signals, perform equal-length discrete sampling between the phase start time and the phase end time, and form an equal-length phase power sequence composed of equal-length phase power points arranged in the sampling order;

[0090] For equal-length phase power sequences, a lower and upper power bound are selected, a normalized power value is generated according to the amplitude range, and a normalized phase waveform is formed according to the sampling order.

[0091] Based on the normalized phase waveform, the first-order rate of change of the normalized phase position, the second-order refractive index of the normalized phase position, the discrete curvature, and the phase curvature energy are calculated sequentially.

[0092] The phase energy, total periodic energy, and phase energy percentage are calculated based on the piecewise linear power function, and the phase energy consumption structure is generated according to the process phase number order.

[0093] Set the reference acquisition period and the target period, and perform extreme value removal and averaging on the phase curvature energy and phase electrical energy ratio of the reference acquisition period to generate fixed reference curvature energy, fixed reference electrical energy ratio and fixed reference curvature deviation.

[0094] The curvature mutation multiple and energy consumption synchronous increment are calculated for the target period. Phase anomaly markers and period anomaly energy consumption early warning markers are generated according to the curvature mutation multiple and energy consumption synchronous increment.

[0095] The abnormal intensity is generated based on the phase anomaly marker, curvature mutation multiple, and energy consumption synchronous increment. The output is a triplet of abnormal phase output number, warning level, and abnormal energy consumption warning result.

[0096] By establishing injection cycle and power sampling objects and segmenting each injection cycle according to the process phase, this solves the problem that existing energy consumption monitoring usually only focuses on the total energy consumption of the whole machine or the power of a fixed time period, making it difficult to correspond to specific injection process actions. By reading the moment of process event signals and forming phase boundaries, it solves the problem that the data is difficult to compare due to the inconsistency of the duration and phase length of different injection cycles. By using equal-length discretization, amplitude normalization, and curvature energy calculation, it solves the problem that the average power or maximum power alone cannot reflect the detailed changes such as waveform abrupt changes, action lag, and load impact. By combining the phase energy ratio and fixed benchmark, it solves the problem that the normal energy consumption levels of different phases are different and the universal threshold adaptability is insufficient. By generating anomaly markers together with the curvature abrupt change multiple and the synchronous energy consumption increment, it avoids misjudgments caused by simple waveform fluctuations or simple energy consumption increases. The early warning results can not only reflect whether there is abnormal energy consumption in the injection molding cycle, but also further output abnormal phases and warning levels, so that on-site investigation can be refined from the whole machine level to specific process links such as mold closing, injection, pressure holding, plasticizing, cooling and holding or mold opening and ejection, thereby improving the interpretability of abnormal energy consumption analysis and the targeting of maintenance.

[0097] Reference Figure 1 The process of establishing injection molding cycles and power sampling objects during continuous production, segmenting each injection molding cycle according to the process phase, and forming a piecewise linear power function based on the original power sampling values ​​specifically includes:

[0098] For continuous production processes under the same injection molding machine, mold number, material number, and process formula number, establish a set of injection cycle numbers. The injection cycle numbers are incremented from the beginning according to the production sequence, and the total number of injection cycles is the number of numbers in the injection cycle number set.

[0099] Each injection cycle is divided into six process phases, and a set of process phase numbers is established. Process phase number 1 represents the mold closing phase, process phase number 2 represents the injection phase, process phase number 3 represents the holding pressure phase, process phase number 4 represents the plasticizing phase, process phase number 5 represents the cooling and holding phase, and process phase number 6 represents the mold opening and ejection phase.

[0100] For each injection molding cycle, a set of original sampling point numbers is established. The original sampling point numbers are incremented sequentially from the beginning according to the sampling time, and the total number of original sampling points for each injection molding cycle is limited to no less than two.

[0101] For each injection molding cycle, the positive input active power sequence is collected through the power metering unit, the positive input active power sampling value corresponding to each original sampling time is read, and the positive input active power sampling value is recorded as the original power sampling value of the corresponding injection molding cycle and the corresponding original sampling point;

[0102] For each injection molding cycle, the original power sample values ​​corresponding to adjacent original sampling points are connected by a straight line, so that the power value obtained by connecting any adjacent original sampling time with a straight line changes linearly with time, forming a piecewise linear power function for the current injection molding cycle.

[0103] By limiting the use of identical equipment, molds, materials, and process formulas, the problem of significant power level differences under different production conditions and the potential for direct comparison to introduce operational deviations was solved. By establishing injection cycle numbers according to the production sequence, the problem of lacking periodic organization in continuous production data and the difficulty in tracking and comparing data over cycles was solved. By clearly dividing each cycle into six phases—mold closing, injection, pressure holding, plasticizing, cooling and holding, and mold opening and ejection—the problem of the overall machine power data not corresponding to specific process actions was solved. By setting original sampling point numbers and requiring a minimum of two sampling points, the basic data conditions for subsequent linear connections and power function construction were ensured. By forming a piecewise linear power function, the problem that the original sampling points are discrete data and the phase boundaries may fall between the sampling points was solved.

[0104] Reference Figure 1 , Figure 2 The process of reading the time of each process event signal, forming a phase boundary according to the time of adjacent process event signals, performing equal-length discrete sampling between the phase start time and the phase end time, and forming an equal-length phase power sequence composed of equal-length phase power points arranged in the sampling order, specifically includes:

[0105] Read the timing of the mold closing start event signal, injection start event signal, pressure holding start event signal, plasticizing start event signal, cooling and holding start event signal, mold opening and ejection start event signal, and cycle end event signal output by the injection molding machine controller, and form seven process event signal times according to the occurrence sequence of the process event signals: mold closing start, injection start, pressure holding start, plasticizing start, cooling and holding start, mold opening and ejection start, and cycle end.

[0106] The timing of the mold closing start event signal is limited to after or at the same time as the first original sampling time of the current injection cycle, and the timing of the cycle end event signal is limited to before or at the same time as the last original sampling time of the current injection cycle. The timing of the seven process event signals is also limited to increase in the following order: mold closing start, injection start, holding pressure start, plasticizing start, cooling and holding start, mold opening and ejection start, and cycle end.

[0107] For each process phase of each injection molding cycle, the process event signal time corresponding to the current process phase is taken as the phase start time, and the next process event signal time is taken as the phase end time, thus forming the phase boundary of the current process phase.

[0108] For each process phase of each injection molding cycle, the phase duration is calculated by subtracting the phase start time from the phase end time.

[0109] The number of equal-length phase power points retained for each process phase is set to twenty-one.

[0110] For each process phase of each injection molding cycle, twenty-one equal-length sampling points are generated at the same time interval between the start and end of the phase. The first equal-length sampling point is the start of the phase, the twenty-first equal-length sampling point is the end of the phase, and the time interval between adjacent equal-length sampling points is the phase duration divided by twenty.

[0111] For each equal-length sampling point, the function value of the piecewise linear power function of the corresponding injection cycle is read to form the equal-length phase power point corresponding to the current injection cycle, the current process phase, and the current equal-length sampling point.

[0112] By directly using the process event signal moments output by the controller, the problem of inconsistency between the injection molding process divided by a fixed time window in the existing technology and the actual actions such as mold closing, injection, pressure holding, and plasticizing is solved. By limiting the sequential increase of events and their location within the sampling range, the problem of phase division errors caused by abnormal event data or boundary crossings is solved. By forming phase boundaries with adjacent process event signal moments, each power segment can be accurately assigned to the corresponding process action. By uniformly setting the number of sampling points of equal length for each phase, the problems of different durations of the same phase in different cycles, different numbers of original sampling points, and inability to directly compare waveforms are solved.

[0113] Reference Figure 2 The step of selecting a lower and upper power bound for the comparison of equal-length phase power sequences, generating normalized power values ​​according to the amplitude range, and forming a normalized phase waveform according to the sampling order specifically includes:

[0114] For each process phase of each injection molding cycle, traverse all twenty-one equal-length phase power points within the current process phase, and select the minimum power value among all equal-length phase power points as the lower power bound of the current process phase.

[0115] For each process phase of each injection molding cycle, traverse all twenty-one equal-length phase power points within the current process phase, and select the maximum power value among all equal-length phase power points as the upper limit of the power of the current process phase.

[0116] When the upper power bound is the same as the lower power bound, the normalized power value corresponding to each equal-length phase power point within the current process phase is set to zero.

[0117] When the upper power limit is greater than the lower power limit, for each equal-length phase power point within the current process phase, the difference between the current equal-length phase power point and the lower power limit is divided by the difference between the upper power limit and the lower power limit to form the normalized power value corresponding to the current equal-length phase power point.

[0118] For each process phase of each injection molding cycle, the normalized power values ​​are arranged in order from front to back according to the time of equal length sampling points to form the normalized phase waveform of the current process phase.

[0119] By independently selecting the upper and lower power bounds within each phase, the problem of distortion easily caused by directly comparing the original power due to the inherently different power levels of different process phases is solved; by normalizing the amplitude range, the problem of waveform amplitudes not being directly comparable due to differences in equipment load levels, material conditions, or action intensity in different cycles is solved; by separately processing the cases where the upper and lower power bounds are the same, invalid calculations are avoided in subsequent processing of phases with no power change.

[0120] Reference Figure 2 The calculation of the first-order rate of change of the normalized phase position, the second-order refractive index of the normalized phase position, the discrete curvature, and the phase curvature energy based on the normalized phase waveform specifically includes:

[0121] For each normalized phase waveform, starting from the first normalized power value, two adjacent normalized power values ​​are selected sequentially. The previous normalized power value is subtracted from the subsequent normalized power value, and then multiplied by the value obtained by subtracting one from the number of equal-length phase power points to form the first-order change rate of the normalized phase position at the corresponding position. These are then arranged in order of position to form a sequence of the first-order change rates of the normalized phase position.

[0122] For the normalized phase position first-order change rate sequence, starting from the second normalized phase position first-order change rate, the current normalized phase position first-order change rate and the previous normalized phase position first-order change rate are selected sequentially. The current normalized phase position first-order change rate is subtracted from the previous normalized phase position first-order change rate, and then multiplied by the value after subtracting one from the number of equal-length phase power points to form the normalized phase position second-order refractive intensity at the corresponding position. These are then arranged in position order to form the normalized phase position second-order refractive intensity sequence.

[0123] For each position corresponding to the normalized phase position second-order refracted intensity in the normalized phase position second-order refracted intensity sequence, read the normalized phase position second-order refracted intensity, the normalized phase position first-order rate of change, and the normalized phase position first-order rate of change of the previous position. Average the normalized phase position first-order rate of change of the current position and the normalized phase position first-order rate of change of the previous position. Square the average result and add it to one. Then raise the sum to the cube of 2. Divide the normalized phase position second-order refracted intensity of the current position by the result of the power operation formed by adding the square of the average result to one and raising it to the cube of 2 to form the discrete curvature of the current position.

[0124] For each process phase, the discrete curvatures are arranged in order of their positions in the normalized phase position second-order refractive intensity sequence to form a discrete curvature sequence. Starting from the position corresponding to the second equal-length phase power point to the position corresponding to the penultimate equal-length phase power point, each discrete curvature is squared. The sum of the squared results is then divided by the number of discrete curvatures involved in the summation to form the phase curvature energy of the current process phase.

[0125] By performing first-order change analysis on the normalized waveform, the problem that simply observing the power point magnitude cannot describe the rise, fall, and rate of change of power is solved. By further forming a second-order refracting intensity, the problem that existing energy consumption statistics are unable to capture local abrupt changes, impacts, and discontinuous load changes in the power waveform is solved. By forming discrete curvature, the degree of bending of the phase power curve can be transformed into a comparable quantitative feature. By summing the squares of the discrete curvature at each position and taking the average, the problem that individual local points may be greatly affected by noise, making it difficult to evaluate the overall phase morphology, is solved.

[0126] Reference Figure 3 The step of calculating phase energy, total periodic energy, and phase energy percentage based on a piecewise linear power function, and generating a phase energy consumption structure according to the process phase number sequence, specifically includes:

[0127] For each process phase of each injection cycle, the piecewise linear power function of the current injection cycle is integrated over time from the start to the end of the current process phase to form the phase energy of the corresponding process phase of the current injection cycle.

[0128] For each injection cycle, the phase energy of the mold closing phase, injection phase, holding pressure phase, plasticizing phase, cooling and holding phase, and mold opening and ejection phase are added together to form the total cycle energy of the current injection cycle.

[0129] When the total electrical energy of the cycle is zero, the phase electrical energy ratio of each process phase in the current injection molding cycle is set to zero.

[0130] When the total energy of the cycle is greater than zero, for each process phase, the phase energy of the current process phase is divided by the total energy of the current injection cycle to form the phase energy ratio of the current process phase.

[0131] For each injection cycle, the phase energy ratio of each process phase is arranged in the order of mold closing phase, injection phase, holding pressure phase, plasticizing phase, cooling and holding phase, and mold opening and ejection phase, forming the phase energy consumption structure of the current injection cycle.

[0132] By performing power integration within the true phase boundary, the problem of not being able to distinguish the energy consumption contribution of each process action based solely on the total power consumption of the entire machine is solved; by forming the total cycle power, the overall description of the energy consumption level of the entire injection molding cycle is preserved; by forming the phase power ratio, the problem of difficulty in comparing the specific phase energy consumption structure when the total power consumption of different cycles may change is solved; by setting the ratio of each phase to zero when the total cycle power is zero, invalid ratio calculations in abnormal operating conditions or shutdown states are avoided.

[0133] Reference Figure 3 The process of setting a reference acquisition period and a target period, and performing extreme value removal and averaging on the phase curvature energy and phase electrical energy ratio of the reference acquisition period to generate fixed reference curvature energy, fixed reference electrical energy ratio, and fixed reference curvature deviation specifically includes:

[0134] Set the number of fixed reference periods to twelve;

[0135] When the total number of injection molding cycles is not greater than the number of fixed reference cycles, all injection molding cycles are limited to the reference acquisition cycle. The phase curvature energy and phase electrical energy ratio of each process phase in each reference acquisition cycle are recorded, and no abnormal energy consumption warning result triplet is formed.

[0136] When the total number of injection cycles is greater than the number of fixed reference cycles, and the injection cycle number of the current injection cycle is not greater than the number of fixed reference cycles, the current injection cycle is limited to the reference acquisition cycle, and only the phase curvature energy and phase electrical energy ratio of each process phase in the current injection cycle are recorded.

[0137] When the total number of injection cycles is greater than the fixed reference number, and the injection cycle number of the current injection cycle is greater than the fixed reference number, the current injection cycle is set as the target cycle.

[0138] When the total number of injection molding cycles is greater than the number of fixed reference cycles, for each process phase, the maximum and minimum phase curvature energy values ​​of the current process phase are compared among the twelve reference acquisition cycles. The phase curvature energy of the current process phase in the twelve reference acquisition cycles is summed and then a maximum and a minimum phase curvature energy value are subtracted. The remaining phase curvature energy is then averaged to form the fixed reference curvature energy of the current process phase.

[0139] When the total number of injection molding cycles is greater than the number of fixed reference cycles, for each process phase, the maximum and minimum phase energy percentages of the current process phase are compared among the twelve reference acquisition cycles. The phase energy percentages of the current process phase in the twelve reference acquisition cycles are summed and then a maximum and a minimum phase energy percentage are subtracted. The remaining phase energy percentages are then averaged to form the fixed reference energy percentage of the current process phase.

[0140] When the total number of injection molding cycles is greater than the number of fixed reference cycles, for each process phase and each reference acquisition cycle, the phase curvature energy of the current process phase in the current reference acquisition cycle is subtracted from the fixed reference curvature energy of the current process phase to form the curvature deviation sample of the current process phase in the current reference acquisition cycle.

[0141] For each process phase, the maximum and minimum values ​​of curvature deviation samples of the current process phase are compared among the twelve reference acquisition cycles.

[0142] For each process phase, the curvature deviation samples of the current process phase in the twelve reference acquisition cycles are summed, and then the maximum value and minimum value of the curvature deviation sample are subtracted. The remaining curvature deviation samples are then averaged to form the fixed reference curvature deviation of the current process phase.

[0143] By setting a baseline acquisition period, the problem of lacking normal references for the machine, mold, material, and process formula for anomaly judgment is solved; by not outputting abnormal energy consumption warning results during the baseline period, unreliable judgments are avoided before the baseline is established; by establishing a baseline for each phase separately, the problem of different degrees of normal energy consumption and normal waveform changes for the six process phases is solved; by removing extreme values ​​and averaging, the impact of occasional sampling fluctuations, short-term process disturbances, or individual abnormal baseline periods on the baseline value is reduced; by establishing a fixed baseline curvature deviation, information on the normal waveform fluctuation range is further preserved.

[0144] Reference Figure 4 The calculation of the curvature mutation multiple and energy consumption synchronization increment for the target period, and the generation of phase anomaly markers and period anomaly energy consumption early warning markers based on the curvature mutation multiple and energy consumption synchronization increment, specifically includes:

[0145] For each process phase in the target cycle, read the phase curvature energy, fixed reference curvature energy, and fixed reference curvature deviation of the current process phase in the target cycle, and add the fixed reference curvature energy and the fixed reference curvature deviation to form the curvature reference sum;

[0146] When the phase curvature energy of the current process phase in the target cycle is zero and the curvature reference sum is zero, the curvature mutation factor of the current process phase is set to one.

[0147] When the phase curvature energy of the current process phase in the target period is greater than zero and the curvature reference sum is zero, divide the phase curvature energy of the current process phase in the target period by the value formed by adding one to the phase curvature energy of the current process phase in the target period, and then add the two to the obtained ratio to form the curvature mutation multiple of the current process phase.

[0148] When the curvature reference sum is greater than zero, the phase curvature energy of the current process phase in the target cycle is divided by the curvature reference sum to form the curvature mutation multiple of the current process phase.

[0149] For each process phase in the target cycle, the phase energy percentage of the current process phase in the target cycle is subtracted from the fixed reference energy percentage of the current process phase to form the synchronous energy consumption increment of the current process phase.

[0150] When the curvature mutation factor of the current process phase in the target cycle is greater than two and the synchronous energy consumption increment of the current process phase is greater than zero, the phase anomaly flag of the current process phase is set to one.

[0151] When the curvature abrupt change factor of the current process phase in the target cycle is not greater than two, the phase anomaly mark of the current process phase is set to zero.

[0152] When the curvature mutation factor of the current process phase in the target cycle is greater than two and the energy consumption synchronization increment of the current process phase is not greater than zero, the phase anomaly mark of the current process phase is set to zero.

[0153] For the target cycle, the phase anomaly markers of the six process phases are merged. When at least one phase anomaly marker is one, the cycle anomaly energy consumption warning marker is set to one. When all six phase anomaly markers are zero, the cycle anomaly energy consumption warning marker is set to zero.

[0154] By reflecting the degree of change in the target periodic waveform shape relative to the normal reference through the curvature mutation multiple, the problem of existing energy consumption detection being unable to detect waveform mutation anomalies in a timely manner is solved; by reflecting the increase of this phase in the periodic energy consumption structure through the energy consumption synchronous increment, the problem of false alarms caused by simple waveform mutations due to normal process switching or sampling disturbances is solved; by marking anomalies only when both curvature mutation and energy consumption increase are satisfied, the anomaly judgment has dual constraints.

[0155] Reference Figure 4 The method of generating anomaly intensity based on phase anomaly marker, curvature mutation multiple, and energy consumption synchronization increment, and outputting a triplet of anomaly phase output number, warning level, and anomaly energy consumption warning result, specifically includes: for each process phase in the target cycle, multiplying the phase anomaly marker of the current process phase, the curvature mutation multiple of the current process phase, and the value formed by adding the energy consumption synchronization increment of the current process phase to form the anomaly intensity of the current process phase;

[0156] When the abnormal energy consumption warning flag of the target cycle is zero, the abnormal phase output number is set to zero.

[0157] When the abnormal energy consumption warning mark of the target cycle is one, the process phase with the largest abnormal intensity is selected from the mold closing phase, injection phase, pressure holding phase, plasticizing phase, cooling holding phase and mold opening ejection phase to form an abnormal phase output number. When there are multiple process phases with the same abnormal intensity, the process phase with the smallest process phase number is selected.

[0158] When the abnormal energy consumption warning mark for the target cycle is zero, the warning level is set to zero;

[0159] When the abnormal energy consumption warning mark of the target cycle is one, and the abnormal intensity of the process phase corresponding to the abnormal phase output number is not greater than three, the warning level is set to one.

[0160] When the abnormal energy consumption warning mark of the target cycle is one, and the abnormal intensity of the process phase corresponding to the abnormal phase output number is greater than three, the warning level is set to two.

[0161] For the target cycle, output and record a triplet of abnormal energy consumption warning results, consisting of the injection cycle number, abnormal phase output number, and warning level.

[0162] By generating abnormal intensity, the problem of not being able to compare the degree of abnormality of different phases when there is only an abnormality mark is solved; by outputting abnormal phase numbers, the problem of existing whole-machine alarms not being able to point out the source of abnormality is solved; by selecting the phase with the smaller process phase number when the abnormal intensity is the problem of uncertain output results in parallel cases is solved; by classifying the warning level, the severity reference can be provided for on-site handling; by recording the warning results, it is convenient for subsequent query, statistics, maintenance traceability and process review.

[0163] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0164] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for early warning of abnormal energy consumption in an injection molding machine, characterized in that, include: In the continuous production process, injection cycle and power sampling objects are established, each injection cycle is segmented according to the process phase, and a piecewise linear power function is formed based on the original power sampling value. Read the time of each process event signal, form a phase boundary according to the time of adjacent process event signals, perform equal-length discrete sampling between the phase start time and the phase end time, and form an equal-length phase power sequence composed of equal-length phase power points arranged in the sampling order; For equal-length phase power sequences, a lower and upper power bound are selected, a normalized power value is generated according to the amplitude range, and a normalized phase waveform is formed according to the sampling order. Based on the normalized phase waveform, the first-order rate of change of the normalized phase position, the second-order refractive index of the normalized phase position, the discrete curvature, and the phase curvature energy are calculated sequentially. The phase energy, total periodic energy, and phase energy percentage are calculated based on the piecewise linear power function, and the phase energy consumption structure is generated according to the process phase number order. Set the reference acquisition period and the target period, and perform extreme value removal and averaging on the phase curvature energy and phase electrical energy ratio of the reference acquisition period to generate fixed reference curvature energy, fixed reference electrical energy ratio and fixed reference curvature deviation. The curvature mutation multiple and energy consumption synchronous increment are calculated for the target period. Phase anomaly markers and period anomaly energy consumption early warning markers are generated according to the curvature mutation multiple and energy consumption synchronous increment. The abnormal intensity is generated based on the phase anomaly marker, curvature mutation multiple, and energy consumption synchronous increment. The output is a triplet of abnormal phase output number, warning level, and abnormal energy consumption warning result.

2. The method for early warning of abnormal energy consumption in an injection molding machine according to claim 1, characterized in that, The process of establishing injection cycle and power sampling objects during continuous production, segmenting each injection cycle according to process phase, and forming a piecewise linear power function based on the original power sampling values ​​specifically includes: For continuous production processes under the same injection molding machine, mold number, material number, and process formula number, establish a set of injection cycle numbers. The injection cycle numbers are incremented from the beginning according to the production sequence, and the total number of injection cycles is the number of numbers in the injection cycle number set. Each injection cycle is divided into six process phases, and a set of process phase numbers is established. Process phase number 1 represents the mold closing phase, process phase number 2 represents the injection phase, process phase number 3 represents the holding pressure phase, process phase number 4 represents the plasticizing phase, process phase number 5 represents the cooling and holding phase, and process phase number 6 represents the mold opening and ejection phase. For each injection molding cycle, a set of original sampling point numbers is established. The original sampling point numbers are incremented sequentially from the beginning according to the sampling time, and the total number of original sampling points for each injection molding cycle is limited to no less than two. For each injection molding cycle, the positive input active power sequence is collected through the power metering unit, the positive input active power sampling value corresponding to each original sampling time is read, and the positive input active power sampling value is recorded as the original power sampling value of the corresponding injection molding cycle and the corresponding original sampling point; For each injection molding cycle, the original power sample values ​​corresponding to adjacent original sampling points are connected by a straight line, so that the power value obtained by connecting any adjacent original sampling time with a straight line changes linearly with time, forming a piecewise linear power function for the current injection molding cycle.

3. The method for early warning of abnormal energy consumption in an injection molding machine according to claim 2, characterized in that, The process of reading the signal times of each process event, forming phase boundaries according to adjacent process event signal times, performing equal-length discrete point sampling between the phase start time and the phase end time, and forming an equal-length phase power sequence composed of equal-length phase power points arranged in the sampling order, specifically includes: Read the timing of the mold closing start event signal, injection start event signal, pressure holding start event signal, plasticizing start event signal, cooling and holding start event signal, mold opening and ejection start event signal, and cycle end event signal output by the injection molding machine controller, and form seven process event signal times according to the occurrence sequence of the process event signals: mold closing start, injection start, pressure holding start, plasticizing start, cooling and holding start, mold opening and ejection start, and cycle end. The timing of the mold closing start event signal is limited to after or at the same time as the first original sampling time of the current injection cycle, and the timing of the cycle end event signal is limited to before or at the same time as the last original sampling time of the current injection cycle. The timing of the seven process event signals is also limited to increase in the following order: mold closing start, injection start, holding pressure start, plasticizing start, cooling and holding start, mold opening and ejection start, and cycle end. For each process phase of each injection molding cycle, the process event signal time corresponding to the current process phase is taken as the phase start time, and the next process event signal time is taken as the phase end time, thus forming the phase boundary of the current process phase. For each process phase of each injection molding cycle, the phase duration is calculated by subtracting the phase start time from the phase end time. The number of equal-length phase power points retained for each process phase is set to twenty-one. For each process phase of each injection molding cycle, twenty-one equal-length sampling points are generated at the same time interval between the start and end of the phase. The first equal-length sampling point is the start of the phase, the twenty-first equal-length sampling point is the end of the phase, and the time interval between adjacent equal-length sampling points is the phase duration divided by twenty. For each equal-length sampling point, the function value of the piecewise linear power function of the corresponding injection cycle is read to form the equal-length phase power point corresponding to the current injection cycle, the current process phase, and the current equal-length sampling point.

4. The method for early warning of abnormal energy consumption in an injection molding machine according to claim 3, characterized in that, The process of selecting a lower and upper power bound for the comparison of equal-length phase power sequences, generating normalized power values ​​according to the amplitude range, and forming a normalized phase waveform according to the sampling order specifically includes: For each process phase of each injection molding cycle, traverse all twenty-one equal-length phase power points within the current process phase, and select the minimum power value among all equal-length phase power points as the lower power bound of the current process phase. For each process phase of each injection molding cycle, traverse all twenty-one equal-length phase power points within the current process phase, and select the maximum power value among all equal-length phase power points as the upper limit of the power of the current process phase. When the upper power bound is the same as the lower power bound, the normalized power value corresponding to each equal-length phase power point within the current process phase is set to zero. When the upper power limit is greater than the lower power limit, for each equal-length phase power point within the current process phase, the difference between the current equal-length phase power point and the lower power limit is divided by the difference between the upper power limit and the lower power limit to form the normalized power value corresponding to the current equal-length phase power point. For each process phase of each injection molding cycle, the normalized power values ​​are arranged in order from front to back according to the time of equal length sampling points to form the normalized phase waveform of the current process phase.

5. The method for early warning of abnormal energy consumption in an injection molding machine according to claim 4, characterized in that, The calculation of the normalized phase position first-order rate of change, normalized phase position second-order refractive index, discrete curvature, and phase curvature energy based on the normalized phase waveform specifically includes: For each normalized phase waveform, starting from the first normalized power value, two adjacent normalized power values ​​are selected sequentially. The previous normalized power value is subtracted from the subsequent normalized power value, and then multiplied by the value obtained by subtracting one from the number of equal-length phase power points to form the first-order change rate of the normalized phase position at the corresponding position. These are then arranged in order of position to form a sequence of the first-order change rates of the normalized phase position. For the normalized phase position first-order change rate sequence, starting from the second normalized phase position first-order change rate, the current normalized phase position first-order change rate and the previous normalized phase position first-order change rate are selected sequentially. The current normalized phase position first-order change rate is subtracted from the previous normalized phase position first-order change rate, and then multiplied by the value after subtracting one from the number of equal-length phase power points to form the normalized phase position second-order refractive intensity at the corresponding position. These are then arranged in position order to form the normalized phase position second-order refractive intensity sequence. For each position corresponding to the normalized phase position second-order refracted intensity in the normalized phase position second-order refracted intensity sequence, read the normalized phase position second-order refracted intensity, the normalized phase position first-order rate of change, and the normalized phase position first-order rate of change of the previous position. Average the normalized phase position first-order rate of change of the current position and the normalized phase position first-order rate of change of the previous position. Square the average result and add it to one. Then raise the sum to the cube of 2. Divide the normalized phase position second-order refracted intensity of the current position by the result of the power operation formed by adding the square of the average result to one and raising it to the cube of 2 to form the discrete curvature of the current position. For each process phase, the discrete curvatures are arranged in order of their positions in the normalized phase position second-order refractive intensity sequence to form a discrete curvature sequence. Starting from the position corresponding to the second equal-length phase power point to the position corresponding to the penultimate equal-length phase power point, each discrete curvature is squared. The sum of the squared results is then divided by the number of discrete curvatures involved in the summation to form the phase curvature energy of the current process phase.

6. The method for early warning of abnormal energy consumption in an injection molding machine according to claim 5, characterized in that, The phase energy, total periodic energy, and phase energy percentage are calculated based on a piecewise linear power function, and a phase energy consumption structure is generated according to the process phase numbering order, specifically including: For each process phase of each injection cycle, the piecewise linear power function of the current injection cycle is integrated over time from the start to the end of the current process phase to form the phase energy of the corresponding process phase of the current injection cycle. For each injection cycle, the phase energy of the mold closing phase, injection phase, holding pressure phase, plasticizing phase, cooling and holding phase, and mold opening and ejection phase are added together to form the total cycle energy of the current injection cycle. When the total electrical energy of the cycle is zero, the phase electrical energy ratio of each process phase in the current injection molding cycle is set to zero. When the total energy of the cycle is greater than zero, for each process phase, the phase energy of the current process phase is divided by the total energy of the current injection cycle to form the phase energy ratio of the current process phase. For each injection cycle, the phase energy ratio of each process phase is arranged in the order of mold closing phase, injection phase, holding pressure phase, plasticizing phase, cooling and holding phase, and mold opening and ejection phase, forming the phase energy consumption structure of the current injection cycle.

7. The method for early warning of abnormal energy consumption in an injection molding machine according to claim 6, characterized in that, The process of setting a reference acquisition period and a target period, and performing extreme value removal and averaging on the phase curvature energy and phase electrical energy ratio of the reference acquisition period to generate fixed reference curvature energy, fixed reference electrical energy ratio, and fixed reference curvature deviation specifically includes: Set the number of fixed reference periods to twelve; When the total number of injection molding cycles is not greater than the number of fixed reference cycles, all injection molding cycles are limited to the reference acquisition cycle. The phase curvature energy and phase electrical energy ratio of each process phase in each reference acquisition cycle are recorded, and no abnormal energy consumption warning result triplet is formed. When the total number of injection cycles is greater than the number of fixed reference cycles, and the injection cycle number of the current injection cycle is not greater than the number of fixed reference cycles, the current injection cycle is limited to the reference acquisition cycle, and only the phase curvature energy and phase electrical energy ratio of each process phase in the current injection cycle are recorded. When the total number of injection cycles is greater than the fixed reference number, and the injection cycle number of the current injection cycle is greater than the fixed reference number, the current injection cycle is set as the target cycle. When the total number of injection molding cycles is greater than the number of fixed reference cycles, for each process phase, the maximum and minimum phase curvature energy values ​​of the current process phase are compared among the twelve reference acquisition cycles. The phase curvature energy of the current process phase in the twelve reference acquisition cycles is summed and then a maximum and a minimum phase curvature energy value are subtracted. The remaining phase curvature energy is then averaged to form the fixed reference curvature energy of the current process phase. When the total number of injection molding cycles is greater than the number of fixed reference cycles, for each process phase, the maximum and minimum phase energy percentages of the current process phase are compared among the twelve reference acquisition cycles. The phase energy percentages of the current process phase in the twelve reference acquisition cycles are summed and then a maximum and a minimum phase energy percentage are subtracted. The remaining phase energy percentages are then averaged to form the fixed reference energy percentage of the current process phase. When the total number of injection molding cycles is greater than the number of fixed reference cycles, for each process phase and each reference acquisition cycle, the phase curvature energy of the current process phase in the current reference acquisition cycle is subtracted from the fixed reference curvature energy of the current process phase to form the curvature deviation sample of the current process phase in the current reference acquisition cycle. For each process phase, the maximum and minimum values ​​of curvature deviation samples of the current process phase are compared among the twelve reference acquisition cycles. For each process phase, the curvature deviation samples of the current process phase in the twelve reference acquisition cycles are summed, and then the maximum value and minimum value of the curvature deviation sample are subtracted. The remaining curvature deviation samples are then averaged to form the fixed reference curvature deviation of the current process phase.

8. The method for early warning of abnormal energy consumption in an injection molding machine according to claim 7, characterized in that, The calculation of the curvature mutation multiple and energy consumption synchronization increment for the target period, and the generation of phase anomaly markers and period anomaly energy consumption early warning markers based on the curvature mutation multiple and energy consumption synchronization increment, specifically includes: For each process phase in the target cycle, read the phase curvature energy, fixed reference curvature energy, and fixed reference curvature deviation of the current process phase in the target cycle, and add the fixed reference curvature energy and the fixed reference curvature deviation to form the curvature reference sum; When the phase curvature energy of the current process phase in the target cycle is zero and the curvature reference sum is zero, the curvature mutation factor of the current process phase is set to one. When the phase curvature energy of the current process phase in the target period is greater than zero and the curvature reference sum is zero, divide the phase curvature energy of the current process phase in the target period by the value formed by adding one to the phase curvature energy of the current process phase in the target period, and then add the two to the obtained ratio to form the curvature mutation multiple of the current process phase. When the curvature reference sum is greater than zero, the phase curvature energy of the current process phase in the target cycle is divided by the curvature reference sum to form the curvature mutation multiple of the current process phase. For each process phase in the target cycle, the phase energy percentage of the current process phase in the target cycle is subtracted from the fixed reference energy percentage of the current process phase to form the synchronous energy consumption increment of the current process phase. When the curvature mutation factor of the current process phase in the target cycle is greater than two and the synchronous energy consumption increment of the current process phase is greater than zero, the phase anomaly flag of the current process phase is set to one. When the curvature abrupt change factor of the current process phase in the target cycle is not greater than two, the phase anomaly mark of the current process phase is set to zero. When the curvature mutation factor of the current process phase in the target cycle is greater than two and the energy consumption synchronization increment of the current process phase is not greater than zero, the phase anomaly mark of the current process phase is set to zero. For the target cycle, the phase anomaly markers of the six process phases are merged. When at least one phase anomaly marker is one, the cycle anomaly energy consumption warning marker is set to one. When all six phase anomaly markers are zero, the cycle anomaly energy consumption warning marker is set to zero.

9. The method for early warning of abnormal energy consumption in an injection molding machine according to claim 8, characterized in that, The method of generating anomaly intensity based on phase anomaly marker, curvature mutation multiple, and energy consumption synchronization increment, and outputting a triplet of anomaly phase output number, warning level, and anomaly energy consumption warning result, specifically includes: for each process phase in the target cycle, multiplying the phase anomaly marker of the current process phase, the curvature mutation multiple of the current process phase, and the value formed by adding one to the energy consumption synchronization increment of the current process phase to form the anomaly intensity of the current process phase. When the abnormal energy consumption warning flag of the target cycle is zero, the abnormal phase output number is set to zero. When the abnormal energy consumption warning mark of the target cycle is one, the process phase with the largest abnormal intensity is selected from the mold closing phase, injection phase, pressure holding phase, plasticizing phase, cooling holding phase and mold opening ejection phase to form an abnormal phase output number. When there are multiple process phases with the same abnormal intensity, the process phase with the smallest process phase number is selected. When the abnormal energy consumption warning mark for the target cycle is zero, the warning level is set to zero; When the abnormal energy consumption warning mark of the target cycle is one, and the abnormal intensity of the process phase corresponding to the abnormal phase output number is not greater than three, the warning level is set to one. When the abnormal energy consumption warning mark of the target cycle is one, and the abnormal intensity of the process phase corresponding to the abnormal phase output number is greater than three, the warning level is set to two. For the target cycle, output and record a triplet of abnormal energy consumption warning results, consisting of the injection cycle number, abnormal phase output number, and warning level.

Images